CONTENTS LIST OF AUTHORS ............................................... PREFACE ..................... Peter D. Vickery and James R. Herkert INTRODUCTION Conservation of grassland birds in the Western Hemisphere ........... ... Peter D. Vickery, Pablo L. Tubaro, Jos6 Maria Cardosa da Silva, Bruce G. Peterjohn, James R. Herkert, and Roberto B. Cavalcanti 2 ECOLOGY Population status of North American grassland birds from the North American Breeding Bird Survey, 1966-1996 ...................... ........................... Bruce G. Peterjohn and John R. Sauer 27 Linking continental climate, land use, and land patterns with grassland bird distribution across the conterminous United States ............. ...... Raymond J. O'Connor, Malcolm T. Jones, Randall B. Boone, and T Bruce Lauber 45 History of grassland birds in eastern North America .................. ............................................. Robert A. Askins 60 Grassland bird conservation in northeastern North America ........... ...................... Jeffrey V. Wells and Kenneth V. Rosenberg 72 Use of cultivated fields by breeding Mountain Plovers in Colorado .... ............................ Fritz L. Knopf and Jeffery R. Rupert 81 Changes in bird populations on Canadian grasslands ................. ......................... C. Stuart Houston and Josef K. Schmutz 87 Multiscale habitat associations of the Sage Sparrow: implications for con- servation biology ......... John T. Rotenberry and Steven T. Knick 95 Spatial distribution of breeding passerine bird habitats in a shrubsteppe region of southwestern Idaho .................................... ......................... Steven T Knick and John T. Rotenberry 104 BREEDING ECOLOGY Habitat Selection Habitat relations and breeding biology of grassland birds in New York ............... Christopher J. Norment, Charles D. Ardizzone, and Kathleen Hartman 112 Experimental analysis of nest predation in a New York grassland: effects of habitat and nest distribution ................................... ................ Charles D. Ardizzone and Christopher J. Norment 122 Satellite burrow use by Burrowing Owl chicks and its influence on nest fate ..................... Martha J. Desmond and Julie A. Savidge 128 Songbird abundance in grasslands at a suburban interface on the Colorado High Plains .... Carl E. Bock, Jane H. Bock, and Barry C. Bennett 131 Thermal aspects of nest-site location for Vesper Sparrows and Horned Larks in British Columbia ........ Kari J. Nelson and Kathy Martin 137 Fire The effects of summer burns on breeding Florida Grasshopper and Bach- man's sparrows ................................................ ...... W. Gregory Shriver, Peter D. Vickery, and Dustin W. Perkins 144 Effects of fire and herbicide treatment on habitat selection in grassland birds in southern Maine ......................................... ..... Peter D. Vickery, Malcolm L. Hunter, Jr., and Jeffrey V. Wells 149 Henslow's Sparrow response to prescribed fire in an Illinois prairie rem- nant ..................... James R. Herkert and William D. Glass 160 Effects of prescribed burning and grazing on nesting and reproductive success of three grassland passerine species in tallgrass prairie ...... ..... Ronald W. Rohrbaugh, Jr., Dan L. Reinking, Donald H. Wolfe, Steve K. Sherrod, and M. Alan Jenkins 165 Relationship of fire history to territory size, breeding density, and habitat of Baird's Sparrow in North Dakota ................ Maiken Winter 171 Conservation Reserve Program Le Conte's Sparrows breeding in Conservation Reserve Program fields: precipitation and patterns of population change .................... ........................ Lawrence D. Igl and Douglas H. Johnson 178 Density and fledging success of grassland birds in Conservation Reserve Program fields in North Dakota and west-central Minnesota ......... ............................................... Rolf R. Koford 187 Management Nesting birds and grazing cattle: accommodating both on Midwestern pastures ...... Stanley A. Temple, Brick M. Fevold, Laura K. Paine, Daniel J. Undersander, and David W. Sample 196 Bird populations of seeded grasslands in the Aspen Parkland of Alberta ...................... David R. C. Prescott and Andrew J. Murphy 203 Grassland songbird occurrence in native and crested wheatgrass pastures of southern Saskatchewan .. Stephen K. Davis and David C. Duncan 211 Data Collection and Analysis Monitoring grassland birds in nocturnal migration .................... ....................... William R. Evans and David K. Mellinger 219 Design and duration of perturbation experiments: implications for data interpretation .............. Kenneth L. Petersen and Louis B. Best 230 Sampling considerations for estimating density of passerines in grasslands ........ Jay J. Rotella, Elizabeth M. Madden, and Andrew J. Hansen 237 LATIN AMERICA Bird species richness and conservation in the Cerrado region of central Brazil .................................... Roberto B. Cavalcanti 244 The decline of the Pampas Meadowlark: difficulties of applying the IUCN criteria to neotropical grassland birds ............................. ................... Pablo Luis Tubaro and Fabifin Marcelo Gabelli 250 A preliminary assessment of distributions and conservation needs of grassland birds in Mexico ....................................... ..................... A. Townsend Peterson and Mark B. Robbins 258 Grassland birds in prairie-dog towns in northwestern Chihuahua, Mexico ......... Patricia Manzano-Fischer, Rurik List, and Gerardo Ceballos 263 Seasonal movements and conservation of seedeaters of the genus Spo- rophila in South America ............ Jos6 Maria Cardosa da Silva 272 Demographic characteristics of Dickcissels in winter ................. ..................... Gianfranco D. Basili and Stanley A. Temple 281 Winter ecology, behavior, and conservation needs of Dickcissels in Venezuela ............ Gianfranco D. Basili and Stanley A. Temple 289 LIST OF AUTHORS CHARLES D. ARDIZZONE Department of Biological Sciences State University of New York College at Brockport Brockport, NY 14420 (present address: 1011 East Tudor Road Anchorage, AK 99503) ROBERT A. ASKINS Department of Zoology Connecticut College New London, CT 06320 GIANFRANCO D. BASILI Department of Wildlife Ecology University of Wisconsin Madison, WI 53706 (present address: Florida Audubon Society 1331 Palmetto Avenue Winter Park, FL 32789) BARRY C. BENNEWI' Department of Environmental, Population, and Organismic Biology University of Colorado Boulder, CO 80309-0334 LouIs B. BEST Department of Animal Ecology Iowa State University Ames, IA 50011 CARE E. BOCK Department of Environmental, Population, and Organismic Biology University of Colorado Boulder, CO 80309-0334 JANE H. BOCK Department of EnvironmentaL, Population, and Organismic Biology University of Colorado Boulder, CO 80309-0334 RANDALL B. BOONE Department of Wildlife Ecology Nutting Hall University of Maine Orono, ME 04469 ROBERTO B. CAVALCANTI Departamento de Zoologia Universidade de Brasflia 70910-900 Brasflia, D.E, Brazil GERARDO CEBALLOS Instituto de Ecologia Universidad Nacional Aut6noma de Mxico, C.U. Apartado Postal 70-275 Mxico, D.E, C.P. 04510 Mexico STEPHEN K. DAVIS Saskatchewan Wetland Conservation Corporation 202-2050 Cornwall Street Regina, SK S4P 2K5 Canada MARTHA J. DESMOND Department of Forestry, Fisheries and Wildlife University of Nebraska Lincoln, NE 68583-0819 (present address: Caesar Kleberg Wildlife Research Institute Texas A&M University Kingsville, TX 78363) DAVID C. DUNCAN Saskatchewan Wetland Conservation Corporation 202-2050 Cornwall Street Regina, SK S4P 2K5 Canada WILLIAM R. EVANS Cornell Laboratory of Ornithology 159 Sapsucker Woods Road Ithaca, NY 14850 (present address: EO. Box 46 Mecklenberg, NY 14863) BRICK M. FEVOLD Department of Wildlife Ecology University of Wisconsin Madison, WI 53706 FABIAN MARCLEO GABELLI Laboratorio de Biologa del Comportamiento Instituto de Biologfa y Medicina Experimental Obligado 2490 1428 Buenos Aires, Argentina and Facultad de Psicologfa Universidad de Buenos Aires Buenos Aires, Argentina WILLIAM D. GLASS Illinois Department of Natural Resources Division of Natural Heritage P.O. Box 88 Wilmington, 1L 6048 l ANDREW J. HANSEN Fish & Wildlife Management Program Biology Department Montana State University Bozeman, MT 59717 KATHLEEN HARTMAN Department of Biological Sciences State University of New York College at Brockport Brockport, NY 14420 JAMES R. HERKERT Illinois Endangered Species Protection Board 524 South Second Street Springfield, IL 62701 C. STUART HOUSTON 853 University Drive Saskatoon, SK S7N 0J8 Canada MALCOLM L. HUNTER, JR. Department of Wildlife Ecology Nutting Hall University of Maine Orono, ME 04469 LAWRENCE D. IGL Northern Prairie Wildlife Research Center U.S. Geological Survey, Biological Resources Division 8711 37th Street SE Jamestown, ND 58401 M. ALAN JENKINS George M. Sutton Avian Research Center P.O. Box 2007 Bartlesville, OK 74005-2007 DOUGLAS H. JOHNSON Northern Prairie Wildlife Research Center U.S. Geological Survey, Biological Resources Division 8711 37th Street SE Jamestown, ND 58401 MALCOLM T. JONES Department of Wildlife Ecology Nutting Hall University of Maine Orono, ME 04469 STEVEN T. KNICK USGS Forest and Rangeland Ecosystem Science Center Snake River Field Station 970 Lusk Street Boise, ID 83706 FRI1Z k. KNOPF U.S. Geological Survey, Biological Resources Division 4512 McMurry Avenue Fort Collins, CO 80525-3400 ROLF R. KOFORD Northern Prairie Wildlife Research Center U.S. Geological Survey, Biological Resources Division 8711 37th Street SE Jamestown, ND 58401 (present address: Iowa Cooperative Fish and Wildlife Research Unit Science Hall II Iowa State University Ames, IA 50011) T. BRUCE LAUBER Department of Wildlife Ecology Nutting Hall University of Maine Orono, ME 04469 (present address: Department of Natural Resources Fernow Hall Cornell University Ithaca, NY 14853) RURIK LIST Department of Zoology, Oxford University South Parks Road Oxford 0X1 3PS United Kingdom ELIZABETH M. MADDEN Fish & Wildlife Management Program Biology Department Montana State University Bozeman, MT 59717 (present address: J. Clark Salyer National Wildlife Refuge EO. Box 66 Upham, ND 58789) PATRICIA MANZANo-FISCHER Department of Zoology, Oxford University South Parks Road Oxford 0X1 3PS United Kingdom (present address: Apartado Postal 32-F Toluca, Mdxico 50190 Mexico) KATHY MARTIN Centre for Applied Conservation Biology Department of Forest Sciences #270-2357 Main Mall University of British Columbia Vancouver, BC V6T 1Z4 Canada (present address: Canadian Wildlife Service 5421 Robertson Road, R.R. 1 Delta, BC V4K 3N2 Canada) DAVID K. MELL1NGER Cornell Laboratory of Ornithology 159 Sapsucker Woods Road Ithaca, NY 14850 (present address: Monterey Bay Aquarium Research Institute 7700 Sandholdt Road Moss Landing, CA 95039-0628) ANDREW J. MURPHY North American Waterfowl Management Plan c/o Ducks Unlimited Canada #8, 5580-45th Street Red Deer, AB T4N 1L1 Canada KARI J. NELSON Centre for Applied Conservation Biology Department of Forest Sciences #270-2357 Main Mall University of British Columbia Vancouver, BC V6T IZA Canada (present address: 1895 Sea Lion Crs. Nanoose Bay, BC V9P 9J3 Canada) CHRISTOPHER J. NORMENT Department of Biological Sciences State University of New York College at Brockport Brockport, NY 14420 RAYMOND J. O'CONNOR Department of Wildlife Ecology Nutting Hall University of Maine Orono, ME 04469 LAURA K. PAINE Department of Agronomy University of Wisconsin Madison, WI 53706 DUSTIN W. PERKINS Department of Forestry and Wildlife Conservation Holdsworth Natural Resource Center University of Massachusetts Amherst, MA 01003 BRUCE G. PETERJOHN U.S. Geological Survey, Biological Resources Division Patuxent Wildlife Research Center 12100 Beech Forest Road Laurel, MD 20708 KENNETH L. PETERSEN Department of Environmental Studies Dordt College Sioux Center, IA 51250 A. TOWNSEND PETERSON Natural History Museum University of Kansas Lawrence, KS 66045 DAVID R. C. PRESCOTT Land Stewardship Centre of Canada 13 Mission Avenue St. Albert, AB T8N 1H6 Canada (present address: Alberta Conservation Association P.O. Box 40027, Baker Centre Postal Outlet Edmonton, AB T5J 4M9 Canada) DAN L. REINKING George M. Sutton Avian Research Center P.O. Box 2007 Bartlesville, OK 74005-2007 MARK B. ROBBINS Natural History Museum University of Kansas Lawrence, KS 66045 RONALD W. ROHRBAUGH, JR. George M. Sutton Avian Research Center P.O. Box 2007 Bartlesville, OK 74005-2007 (present address: Cornell Laboratory of Ornithology 159 Sapsucker Woods Road Ithaca, NY 14850) KENNETH V. ROSENBERG Cornell Laboratory of Ornithology 159 Sapsucker Woods Road Ithaca, NY 14850 JAY J. ROTELLA Fish & Wildlife Management Program Biology Depm-tment Montana State University Bozeman, MT 59717 JOHN T. ROTENBERRY Natural Reserve System and Department of Biology University of California Riverside, CA 92521 JEFFERY R. RUPERT U.S. Geological Survey, Biological Resources Division 4512 McMurry Avenue Fort Collins, CO 80525-3400 DAVID W. SAMPLE Bureau of Research Wisconsin Department of Natural Resources Monona, WI 53716 JOHN R. SAUER U.S. Geological Survey, Biological Resources Division Patuxent Wildlife Research Center 12100 Beech Forest Road Laurel, MD 20708 JULIE A. SAVIDGE Department of Forestry, Fisheries and Wildlife University of Nebraska Lincoln, NE 68583-0819 JOSEF K. SCHMUTZ Department of Biology University of Saskatchewan 112 Science Place Saskatoon, SK S7N 5E2 Canada STEVE K. SHERROD George M. Sutton Avian Research Center P.O. Box 2007 Bartlesville, OK 74005-2007 W. GREGORY SHRIVER Department of Forestry and Wildlife Conservation Holdsworth Natural Resource Center University of Massachusetts Amherst, MA 01003 (present address: College of Environmental Science and Forestry State University of New York 1 Forestry Drive Syracuse, NY 13210) Josg MARIA CARDOSO D^ SILVA Universidade Federal de Pernambuco Centro de Cincias Biologicas Departamento de Zoologia Av. Prof. Morals Rego, 1235 50670-420 Recife, PE Brazil STANLEY A. TEMPLE Department of Wildlife Ecology University of Wisconsin Madison, WI 53706 PABLO LUIS TUBARO Laboratorio de Biologfa del Comportamiento Instituto de Biologfa y Medicina Experimental Obligado 2490 1428 Buenos Aires Argentina DANIEL J. UNDERSANDER Department of Agronomy University of Wisconsin Madison, WI 53706 PETER D. VICKERY Department of Wildlife Ecology Nutting Hall University of Maine Orono, ME 04469 (present address: Center for Biological Conservation Massachusetts Audubon Society Lincoln, MA 01773 and Department of Forestry and Wildlife Conservation University of Massachusetts Amherst, MA 01003) JEFFREY V. WELLS Cornell Laboratory of Ornithology 159 Sapsucker Woods Road Ithaca, NY 14850 (present address: National Audubon Society Cornell Laboratory of Ornithology 159 Sapsucker Woods Road Ithaca, NY 14850) MAIKEN WINTER Department of Behavioral Physiology University-Ttibingen 72072 Ttibingen Germany (present address: 611 Winston Court, Apt. #4 Ithaca, NY 14850-1953) DONALD H. WOLFE George M. Sutton Avian Research Center EO. Box 2007 Bartlesville, OK 74005-2007 Studies in Avian Biology No. 19:1, 1999. PREFACE This book had its genesis in 1994, when the Council of the Association of Field Ornitholo- gists and the staff of the George M. Sutton Avi- an Research Center recognized the need to con- vene a conference on the ecology, status, and conservation of grassland birds in the Western Hemisphere. This two-day conference, convened in Tulsa, Oklahoma, in October 1995, reflected the deep concern held by many avian biologists that populations of many grassland bird species are declining throughout the Western Hemi- sphere. Generous support from the U.S. Fish and Wildlife Service, the Association of Field Or- nithologists, the Sutton Avian Research Center, and the National Fish and Wildlife Foundation made it possible to invite a broad international contingent, especially from South America. Steve Sherrod and the Sutton Avian Research Center staff facilitated conference arrangements and field trips for this productive meeting. The Council of the Association of Field Or- nithologists, notably presidents Greg Butcher, Elissa Landre, and Charles Duncan, provided leadership and support throughout this process. The commitment of the AFO council to both the conference and the publication of this volume is warmly and gratefully acknowledged. We also thank Steve Lewis and the Office of Migratory Bird Management of the U.S. Fish and Wildlife Service for their financial support of this vol- ume. The Center for Biological Conservation of the Massachusetts Audubon Society, especially Christopher Leahy, and the Illinois Endangered Species Protection Board provided logistical support and encouragement to Vickery and Her- kert, respectively. We thank the more than 40 reviewers whose insights measurably improved the manuscripts in this volume. We also thank Andrea Jones, Dustin Perkins, Jan Pierson, Vanessa Rule, and Greg Shriver for their help and suggestions on a variety of issues. Elizabeth Pierson meticu- lously copyedited the entire manuscript and brought greater clarity to every manuscript here- in; that she was able to do this with wit and grace and without offending anyone seems re- markable. We thank Eugenia Wheelwright, who translated all abstracts into Spanish, and Rosita Moore, who provided assistance with graphics. We are immeasurably grateful to Barbara, Simon, and Gabriel Vickery and to Linda, Na- than, and Nicholas Herkert for their collective patience and support. We especially thank John Rotenberry, editor of the Studies in Avian Bi- ology series, for his cheerful guidance, encour- agement, and good counsel throughout. This volume is dedicated to John A. Wiens, whose research on grassland and shrubsteppe birds has had a profound influence not only on both of us but on countless other ecologists of many different disciplines. John's ecological perspicacity and intellectual brilliance continue to inspire and serve as a model. This volume is also dedicated to our children and their millions of cohorts throughout this hemisphere, that they may all have the opportunity to admire prairie- chickens and buntings, or rheas, canasteros, and seedeaters, in wonder, joy, and we hope, curi- osity. Peter D. Vickery Center for Biological Conservation Massachusetts Audubon Society Lincoln, Massachusetts James R. Herkert Illinois Endangered Species Protection Board Springfield, Illinois Studies in Avian Biology No. 19:2-26, 1999. CONSERVATION OF GRASSLAND BIRDS IN THE WESTERN HEMISPHERE PETER D. VICKERY, PABLO L. TUBARO, Jos MARIA CARDOSO DA SILVA, BRUCE O. PETERJOHN, JAMES R. HERKERT, AND ROBERTO B. CAVALCANTI "The sweeping vista of the world's natural grasslands--be they steppes, savannas, range- lands, punas or prairies--occupy nearly seven billion hectares; over half of the earth's land surface. Add to that figure the vast area converted to... habitats of low intensity agriculture and grasslands become second only to the oceans in terms of direct dominance of the planet's ecosystems. They govern, directly, the livelihoods of hundreds of millions of people." --C. Imboden (1988:vii). Research on and interest in grassland birds have increased considerably in the past 20 yr. There are several reasons for this heightened in- terest. Foremost, it is clear that populations of many grassland birds have declined sharply throughout the Western Hemisphere (e.g., Buch- er and Nores 1988, Cavalcanti 1988, Fjelds5 1988, McNicholl 1988, Knopf 1994, Peterjohn and Sauer 1999). In North America, populations of at least 13 species of grassland birds declined significantly between 1966 and 1995 (Peterjohn and Sauer 1999). And as a group, North Amer- ican grassland birds have experienced "steeper, more consistent, and more geographically wide- spread declines than any other behavioral or ecological guild," largely because of habitat loss and degradation (Knopf 1994:251). Similar de- clines are also occurring in South America, where species such as Pampas Meadowlark (Sturnella defilippii; Tubaro and Gabelli 1999), Saffron-cowled Blackbird (Agelaius fiavus; Fra- ga et al. 1998), and Sporophila seedeaters (Silva 1999) have declined in the past 20 yr. Indeed, Collar et al. (1992:35) describe the "near-total destruction of open grasslands in south-east Bra- zil... and in the vast central planalto... as one of the great ecological catastrophes in South America." Another reason for the increased research in- terest in grassland birds is changing agricultural practices. For example, the U.S. Department of Agriculture's Conservation Reserve Program (CRP), which has taken more than 14 million ha of cropland out of production under 10-yr con- tracts, has made it possible to examine regional, and even continental, effects of changing land- scapes on grassland birds (e.g., Lauber 1991, Reynolds et al. 1994, Herkert 1998). Addition- ally, the CRP has provided excellent opportu- nities to study bird colonization, habitat use, and nesting success in different regions and under different ecological conditions. Finally, grass- land birds are also fascinating from ecological and evolutionary perspectives. Distinctive or un- usual adaptations, such as large body size and cursorial habits, have evolved in grassland birds. And the ability to readily observe many behav- iors makes these species ideal for research (e.g., Wheelwright and Mauck 1998). GRASSLAND HABITATS IN THE WESTERN HEMISPHERE Grassland ecosystems occur in a variety of forms and are affected by geology, geography, moisture, soil type, elevation, climate, and dis- turbance regime (Kantrud 1981, Vickery et al. in press). In this volume, we define a grassland habitat as any extensive area that is dominated by more than 50% grass (Poaceae) or sedge (Cy- peraceae) cover and that generally has few scat- tered shrubs (< 4 m high) and trees. We have generally excluded habitats that are dominated by more than 50% shrub cover, such as chap- In addition to such obvious grassland habitats as tallgrass and shortgrass prairies, pampas, and Patagonian grassland, we include sedge-domi- nated tundra, alpine ridges and barrens, puna, and pararno. We also include the longleaf pine (Pinus palustris) ecosystems of the southeastern United States and the pine (Pinus spp.) forests and savannas of Mexico because it is clear that several species of birds, among them Bachman's Sparrow (Aimophila aestivalis), Striped Sparrow (Oriturus superciliosus), and Sierra Madre Spar- row (Xenospiza baileyi), have adapted to the graminoid ground cover beneath these forests. Although these ecosystems are generally viewed as forests, the above species appear to occupy them as a form of grassland, not forest, habitat. Bachman's Sparrow, for example, continues to occupy clear-cut glades after forest removal (Dunning 1993). In North America, we also in- clude as grassland wet-mesic upland habitats where the soil is often saturated but not inun- dated for long periods; we do not include fresh- water, brackish, and saltwater wetlands where 2 INTRODUCTION--Vickery et al. 3 FIGURE 1. Distribution of major grassland ecosystems in North America and Mexico prior to European settlement. Alpine zones above tree line have not been depicted. This map was adapted and modified from two primary sources, Risser et al. 1981 and Environment Canada 1998. standing water is present for long periods, how- ever. Native grasslands in the Western Hemisphere extend from high-arctic sedge meadows in the tundra of North America to pampas and Pata- gonian grasslands in southern South America (Figs. 1 and 2). In North America, a mosaic of tundra/barrens habitats forms the northernmost grassland component. In the temperate region, the most extensive grasslands historically in- cluded the shortgrass prairie and southern mixed prairie of the western Great Plains and the tall- grass prairie and northern mixed prairie of the midwestern United States and Canada (Knopf 1988; Fig. 1). Although they were less exten- sive, bunchgrass shrubsteppe (including palouse prairie) and California grasslands in the west, desert grasslands in the southern United States and Mexico, and palmetto (Serenoa repens) dry prairie in Florida were historically all major grassland types in North America (Fig. 1). In South America, major native grassland ecosystems include high-altitude paramo and puna grasslands (listed as Andean grasslands; Fig. 2) and mid-elevation monte grasslands (Fig. 2). Low-elevation grasslands include ParagonJan grasslands in southern Argentina and Chile and pampas in eastern Argentina, Uruguay, and southernmost Brazil. Brushier savanna grass- lands include chaco, cerrado (particularly "cam- po limpo" and "campo sujo" in central Brazil), Beni savannas, Amazonian savannas, Guianan savannas, and espinal. Native South American grasslands also include such roesic ecosystems as the llanos of Venezuela and Colombia and the Pantanal of southwestern Brazil, where seasonal flooding for several months each year is fol- lowed by pronounced dry seasons when most surface water disappears (Soriano 1991, Diner- stein et al. 1995, Stotz et al. 1996; Fig. 2). DEFINING GRASSLAND BIRDS "The difficulty ... in defining grassland species ß.. results from the fact that grassland itself is not easy to define precisely. How small may a prairie be before it is a mere opening? Where does grass- land stop and very open woodland begin? ... How much sage is required before grassland be- comes some form of desert scrub?" --R. M. Mengel (1970:283) Few would argue that species such as Lesser Rhea (Rhea pennata), Sprague's Pipit (Anthus 4 STUDIES IN AVIAN BIOLOGY NO. 19 Onian FIGURE 2. Distribution of major grassland ecosystems in South America prior to European settlement. Puna and paramo grasslands have been classified as Andean grasslands. This map was adapted and modified from two primary sources, Cabrera and Willink 1980 and Dinerstein et al. 1995. spragueii), McCown's Longspur (Calcarius mccownii), and Wedge-tailed Grass-Finch (Em- berizoides herbicola) are completely adapted to grassland habitats and should be considered grassland specialists. Classification seems obvi- ous in these cases, as all of these species use grassland habitat for all their life-history needs. But for many other species, determining which ones should be considered grassland birds quick- ly becomes complicated and invariably some- what subjective. Are Western Kingbirds (Tyran- nus verticalis), Red-winged Blackbirds (Age- laius phoeniceus), and Blue-black Grassquits (Volatinia jacarina), for instance, also grassland birds? What about jaegers (Stercorarius spp.)? Although each of the three jaeger species spends 9 mo a year on the open ocean, all require open tundra for nesting. And nest success in Pomarine Jaegers (S. pomarinus), as in Snowy Owls (Nyc- tea scandiaca), depends strongly on collared lemming (Dicrostonyx torquatus) populations (Pitelka et al. 1955). Mengel (1970) recognized the difficulties in- herent in trying to define grassland birds. He re- alized that grasslands extend along a moisture gradient--from arid prairies to wet meadows and marshes--and that defining the limits of this gradient in relation to the birds that occupy these habitats can be, and is, somewhat arbitrary. In addition, he noted that grassland ecosystems fre- quently intergrade with forested and other hab- itat types, making it difficult to define the limits of some grassland types. In the Cerrado of cen- tral Brazil, for example, "campo limpo," or open grasslands, are interspersed with "campo sujo," or grasslands with scattered trees and shrubs; and campo sujo may blend into "cerra- d5o," which is even more densely forested (El- ten 1972). In the United States, tallgrass prairie intergrades into oak (Quercus) savannas in the Midwest, and in the Southeast the dry palmetto prairies of central Florida merge into longleaf pine savannas, called "flatwoods." Consequent- ly, it is often difficult to delineate where grass- land ends and forest begins. Furthermore, dif- ferent species of birds may respond differently to the same ecotone. In Florida, Grasshopper Sparrows (Ammodramus savannarum floridan- INTRODUCTIONVickery et al. 5 us) breed only on treeless palmetto prairies and do not )ccupy savanna flatwoods. Bachman's Sparrows, however, breed commonly in both habitats. From the perspective of these two sym- patric grassland sparrows, the definition of grassland habitat is quite different. This process is further complicated by the fact that some grassland species use different habi- tats in different parts of their ranges. Savannah Sparrows (Passerculus sandwichensis) are known to use an extraordinary array of open habitats throughout their extensive range (Wheelwright and Rising 1993). In eastern Tex- as, Bachman's Sparrows typically breed in open pine forests, but in central Florida they com- monly breed on treeless palmetto prairies (Dun- ning 1993, Shriver et al. 1999). Although there are similarities in these habitats, notably the pre- dominant graminoid ground cover, the differ- ences are also obvious and striking. Finally, the fact that so many grassland hab- itats have been severely altered by modern ag- ricultural practices further complicates efforts to define grassland birds. Many grassland species in the Western Hemisphere are presently occu- pying artificial habitats that did not exist 200- 300 yr ago. For example, Northern Hamers (Circus cyaneus), Short-eared Owls (Asio flarn- rneus), Henslow's Sparrows (Amrnodrarnus hen- slowii), and many other grassland birds now breed on reclaimed surface coal mines in west- ern Pennsylvania, West Virginia, Ohio, and In- diana. These newly created "prairies" did not exist 100 yr ago, but they appear to be providing important refugia for threatened species in these regions (D. Brauning, pers. comm.). Conversely, some steppe or forest birds are invading open habitats because as early settlers cleared the land for agriculture, they provided the perches and refuges these species require (Gochfeld 1979, McNicholl 1988). Thus, it is necessary to have some understanding of habitat preferences prior to European settlement to determine whether present-day habitat use reflects long-term evo- lutionary patterns. Given the complexities in defining grassland habitats, how does one define the birds that use this variety of habitats? Are there common threads that help define grassland birds? And are these similarities consistent spatially and across taxa? In midwestern North America, Mengel (1970) recognized two groups of grassland birds based on distribution and habitat selection. He relied on limited geographic range and endemism to determine "primary" grassland birds, which were restricted to the central Great Plains. He identified as "secondary" grassland birds those species that had "strong affinities with the grass- lands, although [were] not restricted to them" (Mengel 1970:283). This geographic emphasis created ecological inconsistencies. Wilson's Phalarope (Phalaropus tricolor) and Franklin's Gull (Larus pipixcan), for instance, were con- sidered "primary" grassland species, but the ecological connections to grassland habitat for either species are limited. Wilson's Phalarope, for example, generally breeds along the edges of prairie potholes and open marshes but makes lit- tle use of the surrounding grassland habitat. We prefer an ecological basis for defining grassland birds. We thus define a grassland bird as any species that has become adapted to and reliant on some variety of grassland habitat for part or all of its life cycle, be it breeding (either nesting or feeding), migration, or wintering. Grassland birds often, but not necessarily, nest on the ground. Thus, we consider Swainson's Hawk (Buteo swainsoni), Mountain Plover (Charadrius rnontanus), and Long-billed Curlew (Numenius americanus) to be grassland birds, despite the fact that Swainsoh's Hawks nest in trees and that curlews often use a variety of in- tertidal habitats in the nonbreeding seasons. Along the moisture gradient, we include as grassland birds four species of South American geese (Chloephaga spp.), Sedge Wren (Cistotho- rus platensis), Henslow's Sparrow, and Le Conte's Sparrow (Amrnodrarnus leconteii), but we exclude birds that normally breed over or adjacent to standing water, among them Swamp Sparrow (Melospiza georgiana), Nelson's Sharp-tailed Sparrow (Amrnodrarnus nelsoni), Seaside Sparrow (A. rnaritirna), some waterfowl (Anatidae), and most rails (Rallidae) and herons (Ardeidae; but see Sample and Mossman 1997 for a different perspective). Along the shrub gra- dient, we consider Rufous-winged Sparrow (Abnophila carpalis) and Lark Sparrow (Chond- estes gramrnacus) to be grassland birds but not Brewer's Sparrow (Spizella breweri). We ex- clude species that occur commonly in grassland habitats but do not use the graminoid compo- nents of these habitats; examples include Pinyon Jay (Gyrnnorhinus cyanocephalus), which feeds almost exclusively on shrub seeds, and aerial in- sectivores such as swifts (Apodidae) and swal- lows (Hirundinidae), which only feed over grasslands. Finally, we include species that occupy wet- land, shrub, and forest edges adjacent to grass- land habitats only when they make regular use of the grassland habitat away from edge (> 100 m). For example, we consider the American Bit- tern (Botaurus lentiginosus), which nests in prai- rie fragments and fields, and the various puddle ducks that nest in upland fields far from wet- lands to be grassland birds. 6 STUDIES IN AVIAN BIOLOGY NO. 19 OBLIGATE AND FACULTAT[VE GRASSLAND BIRDS Within our ecological definition of grassland birds, two groups can be reasonably separated. Obligate grassland specialists are species that are exclusively adapted to and entirely depen- dent on grassland habitats and make little or no use of other habitat types. Examples include Lesser Rhea, Baird's Sparrow (Ammodramus bairdii), and Pampas Meadowlark (Tables 1 and 3). Obligate grassland birds would almost cer- tainly become extinct without the appropriate grassland habitat. Facultative grassland specialists use grass- lands as part of a wider array of habitats. In general, these species are not entirely dependent on grasslands but use them commonly and reg- ularly. If the appropriate types of grassland hab- itat were destroyed, populations of some facul- tative grassland birds would diminish but prob- ably would not completely disappear. Examples of facultative grassland birds include Barn Owl (Tyto alba), Loggerhead Shrike (Lanius ludovi- cianus), Clay-colored Sparrow (Spizella palli- da), and Blue-black Grassquit (Tables 2 and 4). The number of obligate species found in grasslands is not especially great compared with other habitats. In North America, Mexico, and the Caribbean, for example, there are 59 species of obligate grassland species from 35 genera (Table 1) compared with more than 180 species of obligate forest-dwelling species. With 124 species from 59 genera (Table 3), South Amer- ica supports many more obligate grassland spe- cies than do North America, Mexico, and the Caribbean. Not surprisingly, facultative grass- land species are more numerous than obligates; there are 97 species of facultative grassland birds in North America, Mexico, and the Carib- bean (Table 2) and 164 in South America (Table 4). DISTRIBUTION OF GRASSLAND BIRDS Obligate grassland specialists have a wide geographic distribution. They occur from north of the Arctic Circle to the southern tip of Ar- gentina and Chile and as far offshore as the Islas Malvinas (Falkland Islands) and, 1770 km east of Tierra del Fuego, South Georgia Island (Ta- bles 1 and 3). As a genus, pipits (Antbus spp.) have the widest breeding range of any Western Hemisphere passefines, extending from arctic Canada (American Pipit [A. rubestens]) to South Georgia Island (South Georgia Pipit [A. antarc- ticus]). Only three obligate grassland species are widely distributed across the Americas, howev- er The Short-eared Owl breeds discontinuously from the arctic regions of Canada and Alaska to Tierra del Fuego; the Burrowing Owl (Athene cunicularia) breeds from southern Canada and Florida to the southern pampas of Argentina; and the Sedge Wren, currently classified as a sin- gle, widely distributed species, occurs from east- ern North America to southern South America (AOU 1998). Only seven obligate grassland spe- cies in North America breed in both arctic/alpine and temperate regions (Table 1). Although there are differences between arctic/ alpine breeders in North America (e.g., ptarmi- gan [Lagopus spp.], jaegers, and buntings [Plec- trophenax spp.]) and temperate or steppe breed- ers (e.g., prairie-chickens [Tympanuchus spp.], sparrows [Aimophila spp.], and meadowlarks [Sturnella spp.]), the similarities between grass- land birds of these regions are pronounced. Many genera are shared between the arctic/al- pine and temperate regions, despite the fact that the breeding ranges of most species are restrict- ed to either the arctic/alpine or temperate region (Table 1). For example, McCown's Longspurs and Chestnut-collared Longspurs (Calcarius or- natus), both of which occur in shortgrass and mixed prairies, are replaced by Smith's Long- spurs (C. pictus) and Lapland Longspurs (C. lapponicus) farther north. The same allopatric relationships are found among hawks (Buteo spp.), falcons (Falco spp.), plovers (Charadrius spp.), curlews (Numenius spp.), godwits (Limosa spp.), shrikes (Lanius spp.), and pipits. In South America, taxonomic affinities be- tween high-altitude and lowland temperate birds occur in hawks (Buteo spp.), caracaras (Phal- coboenus spp.), seedsnipes (Attagis and Thino- corus spp.), doves (Metriopelia and Zenaida spp.), tyrant flycatchers (Tyrannidae), and seed- eaters (Emberizinae). It should be noted that the geographic scope of research in this volume is limited to birds that breed in the temperate regions of North, Central, and South America. In North America, the geographic separation between arctic/alpine and temperate breeders largely disappears in the nonbreeding season. Although a few species such as ptarmigan are largely resident, many arctic/alpine species mi- grate medium to long distances and can be found wintering with temperate grassland breeding birds. A few arctic breeders, such as American Golden-Plovers (Pluvialis dominicus) and Eski- mo Curlews (Numenius borealis), join more temperate breeders such as Upland Sandpipers (Bartramia longicauda) and Bobolinks (Doli- chonyx oryzivorus) to winter on the pampas in Argentina and southern Brazil. LOSS OF GRASSLAND HABITAT Since the early 1800s, most grassland ecosys- tems in North America have been profoundly INTRODUCTION--Vickery et al. 7 altered by agricultural activities, and many are now among the continent's most endangered ecosystems (Table 5; Noss et al. 1995). In most areas, habitat loss has exceeded 80% (Samson and Knopf 1994, Noss et al. 1995), and where soil and topography are well suited for crops, less than 0.1% of native prairie remains (Samson and Knopf 1994). Since 1850, for example, the decline of tallgrass prairie (estimated to be 88- 99%) exceeds that reported for any other major ecosystem in North America (Samson and Knopf 1994, Noss et al. 1995). Similarly, in Florida only 19% of the original palmetto dry prairie remains, with most of this habitat having been converted to citrus groves and improved cattle pastures since about 1950 (Shriver and Vickery 1999). Native temperate grasslands in the Western Hemisphere have experienced major, sometimes profound, losses from agriculture, range man- agement, and urban development. Some grass- land species, however, notably Picazuro Pigeon (Columba picazuro), Spot-winged Pigeon (C. maculosa), Eared Dove (Zenaida auriculata), Grasshopper Sparrow, Dickcissel (Spiza ameri- cana), Bobolink, and meadowlarks have adapted successfully to these modified landscapes (Gra- ber and Graber 1963, Bucher and Nores 1988, Rodenhouse et al. 1995, O'Connor et al. 1999). In the midwestern United States, agricultural lands have provided adequate breeding habitat for many species, but in the past 50 yr conver- sion of pastures and hayfields into rowcrops (e.g., corn [Zea mays] and soybeans [Glycine max]) and shortened cutting rotations of hay have made much of this habitat unsuitable and have become major threats to grassland bird populations (Herkert 1991, 1997; Warner 1994; Herkert et al. 1996). In Canada, approximately 25% of native grasses remain, but losses continue; 570,000 ha, or approximately 6% of what remained, were lost between 1991 and 1996 (Statistics Canada 1997). Southeastern Alberta and southwestern Saskatchewan contain much of the remaining native prairie, and several grassland bird species, among them Baird's Sparrow and Sprague's Pip- it, are abundant there (Price et al. 1995). Grazing pressure has generally increased on remaining native grasslands (Gayton 1991). In South America, modernization and me- chanical changes in agricultural practices have had similarly adverse effects on breeding birds (Bucher and Nores 1988, Cavalcanti 1999b, Tu- baro and Gabelli 1999). Horses and cattle were introduced to the Pampas in 1535, and by 1750 feral populations were so common that they sup- ported a growing industry of exporting hides. The effects of grazing and burning to improve pastures and to deter aboriginal Indians trans- formed the Pampas and were commented on by Darwin (1876). The most profound changes, however, occurred after 1890 with the expansion of agriculture in South America. During the first quarter of the twentieth century, the negative ef- fect of agriculture on grassland species such as the Strange-tailed Tyrant (Alectrurus risora) be- came evident (Wilson 1926). Since 1970, in- creased use of agrochemicals and technology has contributed to the intensive use of grass- lands. In the northern Pampas, silviculture is also reducing grassland area. In Brazil, more than 50% of the Cerrado has been converted for human uses since 1950 (Sil- va 1995), and today the region is seen as a promising area for "carbon bank" mitigation (planting trees to absorb and convert carbon di- oxide) against deforestation in Amazonia (Cav- alcanti 1999a). The trend in the Cerrado is an ever-growing rate of destruction of natural hab- itats. Recent estimates indicate that approxi- mately 75% of this biome can be converted to pastures and agriculture fields to produce about 100 million ton of crops and meat annually (Ma- cedo 1994). An analysis of satellite images from 1987 to 1993 covering the entire Cerrado region showed that 67% of the land surface (excluding non-Cerrado habitats) was in a disturbed or highly disturbed condition as a result of human activity (Mantovani and Pereira 1998). In the Pampas, less than 5% of the land was used for agriculture in 1890, but in high, mesic areas that figure is now greater than 50%. In the more arid and lowland areas of the Pampas, till- age agriculture represents less than 10% of the land use, but cattle grazing over seeded or nat- ural pastures is widespread (Leon et al. 1984). It is clear that similar rates of habitat loss have taken place elsewhere in Central and South America, from northern Mexico (Manzano-Fi- scher et al. 1999) to Argentina (Collar et al. 1992, Dinerstein et al. 1995, Tubaro and Gabelli 1999). It is distressing that conversion of native grasslands for agricultural purposes in South America has been "so utterly neglected as an international conservation issue" (Collar et al. 1992: 35). In Brazil, remnants of native grassland are now largely restricted to national parks (Col- lar et al. 1992). In Argentina, there is no national park protecting a representative sample of pam- pas (Burkart and Valle Ruiz 1994). Moreover, a recent attempt to create a national park in the Pampas failed because the landowner plowed and destroyed the grassland on his hacienda when he realized the government was consid- ering appropriating the area (P. Tubaro, pers. comm.). The most acutely imperiled grasslands in Central and South America are the Cerrado, 8 STUDIES IN AVIAN BIOLOGY TABLE 1. OBLIGATE GRASSLAND BIRDS OF NORTH AMERICA, MEXICO, AND THE CARIBBEAN NO. 19 Family Breeding distribution Arctic/ alpine Temperate Sub- tropical/ Mexico Caribbean Hawks Northern Harrier Swainson's Hawk Ferruginous Hawk Rough-legged Hawk Falcons Aplomado Falcon Partridge, grouse, Old World quail Rock Ptarmigan White-tailed Ptarmigan Sharp-tailed Grouse Greater Prairie-Chicken Lesser Prairie-Chicken New World quail Montezuma Quail Ocellated Quail Stone curlews Double-striped Thick-knee Plovers, lapwings American Golden-Plover Pacific Golden-Plover Mountain Plover Shorebirds Upland Sandpiper Eskimo Curlew* Bristle-thighed Curlew Long-billed Curlew Marbled Godwit Baird's Sandpiper Buff-breasted Sandpiper Gulls, jaegers Pomarine Jaeger Parasitic Jaeger Long-tailed Jaeger Owls Snowy Owl Burrowing Owl Long-eared Owl Short-eared Owl Larks Horned Lark Wrens Sedge Wren Pipits American Pipit Sprague's Pipit Emberizids Ruddy-breasted Seedeater Saffron Finch Grassland Yellow-Finch Accipitridae Circus cyaneus Buteo swainsoni Buteo regalis Buteo lagopus Falconidae Falco femoralis Phasianidae Lagopus mutus Lagopus leucurus Tympanuchus phasianellus Tympanuchus cupido Tympanuchus pallidicinctus Odontophoridae Cyrtonyx montezumae Cyrtonyx ocellatus Burhinidae Burhinus bistriatus Charadriidae Pluvialis dominica Pluvialis fulva Charadrius montanus Scolopacidae Bartramia longicauda Numenius borealis Numenius tahitiensis Numenius americanus Limosa fedoa Calidris bairdii Tryngites subruficollis Laridae Stercorarius pomarinus Stercorarius parasiticus Stercorarius longicaudus Strigidae Nyctea scandiaca Athene cunicularia Asio otus Asio fiammeus Alaudidae Eremophila alpestris Troglodytidae Cistothorus platensis Motacillidae Artthus rubescens Artthus spragueii Emberizidae Sporophila minuta Sicalis fiaveola Sicalis luteola INTRODUCTION--Vickery et al. 9 TABLE 1. CONTINUED Family Breeding distribution Arctic/ alpine Sub- tropicalJ Temperate Mexico Caribbean Cassin's Sparrow Bachman's Sparrow Botteri's Sparrow Striped Sparrow** Vesper Sparrow Lark Bunting Savannah Sparrow Grasshopper Sparrow Baird's Sparrow Henslow's Sparrow Le Conte's Sparrow Sierra Madre Sparrow** McCown's Longspur Lapland Longspur Smith's Longspur Chestnut-collared Longspur Snow Bunting McKay's Bunting Cardinals and allies Dickcissel Meadowlarks, blackbirds Bobolink Eastern Meadowlark Western Meadowlark Aimophila cassinii Aimophila aestivalis Aimophila botterii Oriturus superciliosus Pooecetes gramineus Calamospiza melanocorys Passerculus sandwichensis gmmodramus savannarum Ammodramus bairdii Ammodramus henslowii Ammodramus leconteii Xenospiza baileyi Calcarius mccowniœ Calcarius lapponicus Calcarius pictus Calcarius ornatus Plectrophenax nivalis Plectrophenax hyperboreus Cardinalidae Spiza americana Icteridae Dolichonyx oryzivorus Sturnella magna Sturnella neglecta 7 7 7 7 7 7 7 Note: This list was derived from numerous sources. including Bond 1971; Johnsgard 1981; Hayman et al. 1986; Raffaele 1989; Howell and Webb 1995: AOU 1998; and J. L. Dunn. pets. comm. * Possibly extinct. ** Autecology poorly known. chaco savannas, Pampas, and Beni savannas (Bolivia), and more regionally, the savannas near Veracruz and Tehuantepec, Mexico (Diner- stein et al. 1995). Although habitat loss is frequently viewed primarily as conversion to cropland or other uses, it also includes more subtle forms of deg- radation, among them unnatural grazing re- gimes, planting of exotic grasses, and succession to shrublands (Vickery et al. in press). In Pata- gonia, overgrazing by sheep has degraded tall- grass habitats (Fjeldsfi 1988), and in the western pampas of Argentina it is contributing to the spread of chafiar trees (Geoffroea decorticans; Anderson 1977). In North America, shortgrass prairie is adapted to intensive grazing by native herbivores, but contemporary cattle management emphasizes rotations that maintain moderate ground cover, which is less suitable for some rare species such as Mountain Plover (Knopf and Rupert 1999). THE IMPETUS FOR GRASSLAND BIRD AND HABITAT CONSERVATION Habitat loss and degradation have been the two most important factors influencing the de- cline of grassland birds in North and South America (Collar et al. 1992, Knopf 1994, Her- kert et al. 1996, Stotz et al. 1996, Vickery et al. in press). In South America, excessive hunting and illegal trapping have also contributed to some grassland bird declines (Bucher and Nores 1988, Collar et al. 1992, Fraga et al. 1998). In North America, most grassland bird popu- lations have been declining for half a century (Askins 1993, Peterjohn and Sauer 1999). Pop- ulations of at least 13 grassland species declined significantly between 1966 and 1996, whereas populations of only 3 species are known to have increased during that period (Peterjohn and Sauer 1999). There is additional concern be- cause tfiese declines have prevailed across much of the continent. It is unlikely that there is a single underlying cause of these declines; in- stead, multiple causes are probably responsible. It is clear, however, that these declines are not local, isolated phenomena (Peterjohn and Sauer 1999). Similar declines have taken place throughout South America, especially in lowland grasslands (Bucher and Nores 1988, Fjeldsfi 1988, Caval- 10 STUDIES IN AVIAN BIOLOGY TABLE 2. FACULTATIVE GRASSLAND BIRDS OF NORTH AMERICA, MEXICO, AND THE CARIBBEAN NO. 19 Breeding distribution Sub- Arctic/ tropical/ Family alpine Temperate Mexico Caribbean Herons Ardeidae American Bittern Botaurus lentiginosus / Cattle Egret Bubulcus ibis ,/ Storks Ciconiidae Jabiru Jabiru mycteria New World vultures Cathartidae Turkey Vulture Cathartes aura ,/ Lesser Yellow-headed Vulture Cathartes burrovianus Waterfowl Anatidae Greater White-fronted Goose Anser albifrons ,/ Emperor Goose Chen canagica ,/ Snow Goose Chen caerulescens ,/ Ross's Goose Chen rossii ,/ Canada Goose Branta canadensis ,/ ,/ Brant Branta bernicla ,/ Gadwall Anas strepera  American Wigeon Anas americana / Mallard Anas platyrhynchos ,/ Blue-winged Teal Anas discors ,/ Northern Shoveler Anas clypeata ,/ Northern Pintail Anas acuta ,/ Green-winged Teal Anas crecca ,/ ,/ Falcons Falconidae Crested Caracara Carcara plancus American Kestrel Falco sparverius ,/ Merlin Falco columbarius ,/ ,/ Gyrfalcon Falco rusticolus / Peregrine Falcon Falco peregrinus ,/ ,/ Prairie Falcon Falco mexicanus ,/ Partridge, grouse, Old World quail Phasianidae Gray Partridge* Perdix perdix ,/ Ring-necked Pheasant* Phasianus colchicus ,/ Willow Ptarmigan Lagopus lagopus ,/ ,/ New World quail Odontophoridae Scaled Quail Callipepla squamata Elegant Quail Callipepla douglasii Northern Bobwhite Colinus virginianus ,/ Black-throated Bobwhite Colinus nigrogularis Crested Bobwhite Colinus cristatus Rails Rallidae Yellow Rail Coturnicops noveboracensis ,/ Cranes Gruidae Sandhill Crane Grus canadensis ,/ ,/ Whooping Crane Grus americana / Plovers, lapwings Charadriidae Black-bellied Plover Pluvialis squatarola / Killdeer Charadrius vociferux ,/ Shorebirds Scolopacidae Lesser Yellowlegs Tringa fiavipes ,/ Willet Catoptrophorus semipalmatus ,/ Whimbrel Numenius phaeopus ,/ 4' INTRODUCTION--Vickery et al. 11 TABLE 2. CONTINUED Family Breeding distribution Arctic/ alpine Temperate Sub- tropical/ Mexico Caribbean Hudsonian Godwit Surfbird Red Knot Sanderling Semipalmated Sandpiper Western Sandpiper Least Sandpiper White-romped Sandpiper Pectoral Sandpiper Purple Sandpiper Rock Sandpiper Dunlin Short-billed Dowitcher Long-billed Dowitcher Common Snipe Wilson's Phalarope Gulls Franklin's Gull Doves Mourning Dove Common Ground-Dove Barn Owls Barn Owl Owls Striped Owl Goatsuckers Lesser Nighthawk Common Nighthawk Common Poorwill Tyrant flycatchers Say's Phoebe Ash-throated Flycatcher Cassin's Kingbird Western Kingbird Eastern Kingbird Scissor-tailed Flycatcher Fork-tailed Flycatcher Shrikes Loggerhead Shrike Northern Shrike Crows, jays Chihuahuan Raven Thrashes Eastern Bluebird Western Bluebird Mountain Bluebird Thrashers Bendire's Thrasher Wood-Warblers Common Yellowthroat Limosa haemastica Aphdza vit2qata Calidds canutus Calidds alba Calidds pusilia Calidds maud Calidds minutilla Calidds fuscicollis Calidds melanotos Calidds madtima Calidds ptilocnemis Calidds alpina Limnodromus gdseus Limnodromus scolopaceas Gallinago gallinago Phalaropus tricolor Laridae Larus pipixcan Columbidae Zenaida macroura Columbina passedna Tytonidae Tyro alba Strigidae Pseudoscops clamator Caprimulgidae Chordeiles acutipennis Chordeiles minor Phalaenoptilus nuttallii Tyrannidae Sayornis $aya Myiarchus cinerascens Tyrannus vocyerans Tyrannus verticalis Tyrannus Lyrannus Tyrannus fitficatus ryranna$ savana Laniidae Lanius ludovicianus Lanius excubitor Corvidae Corvus co'ptoleucus Turdidae Sialia sialis Sialia mexicana Sialia currucoides Mimidae Toxostoma bendirei Pamlidae Geothlypis talchas ,/ ½' ,/ 12 TABLE 2. CONTINUED STUDIES IN AVIAN BIOLOGY NO. 19 Family Breeding distribution Arctic/ alpine Temperate Sub- tropical/ Mexico Caribbean Emberizids Blue-black Grassquit Yellow-bellied Seedeater Yellow-faced Grassquit Canyon Towbee Rufous-winged Sparrow Rufous-crowned Sparrow Oaxaca Sparrow** Clay-colored Sparrow Worthen's Sparrow** Lark Sparrow Meadowlarks, blackbirds Red-winged Blackbird Brewer's Blackbird Shiny Cowbird Bronzed Cowbird Brown-headed Cowbird Finches Gray-crowned Rosy-Finch Black Rosy-Finch Brown-capped Rosy-Finch Emberizidae Volatinia jacarina Sporophila nigricollis Tiaris olivacea Pipilo fuscus Aimophila carpalis Aimophila ruficepx Aimophila notosticta Spizella pallida Spizella wortheni Chondeste$ grammacu$ Icteridae Agelaius phoeniceux Euphagus cyanocephalus Molothrus bonariensis Molothrus aeneus Molothru$ ater Fringillidae Leucosticte tephrocoti$ Leucosticte atrata Leucoxticte australis Note: This list was derived from numerous sources, including Bond 1971; Johnsgard 1981; Hayman et al. 1986; Raffaele 1989; Howell and Webb 1995; AOU 1998; and J. L. Dunn. pets. comm. * Introduced. ** Autoecology poorly known. canti 1999a, Tubaro and Gabelli 1999). Accord- ing to Wege and Long (1995), 12% of the Neo- tropic's threatened bird species live in grasslands and savannas. At least 34% of the grassland bird species rank as high conservation priorities, and 80% of the campos grassland birds are at risk (Storz et al. 1996). CONSERVATION STRATEGIES People involved in grassland bird conserva- tion efforts need to recognize the historical dy- namics under which these unique habitats evolved. Where feasible, management should in- corporate the ecological processes that have gen- erated and maintained these distinctive ecosys- tems. The timing, intensity, and seasonality of grazing, fire, and other disturbances on grassland conservation areas should mimic natural pro- cesses as closely as possible. This is important for many of the plants and animals that occur in these unique habitats. In North America, for ex- ample, intensive grazing by native herbivores such as prairie dogs (Cynomys spp.), bison (Bi- son bison), and pronghorn (Antilocapra ameri- cana) was one of the major ecological forces that shaped and maintained shortgrass prairies (Vickery et al. in press). Fires, ignited both nat- urally and by Native Americans, were primarily responsible for maintaining tallgrass prairies in the Midwest and native grasslands in the North- east. In Florida, lightning was the primary dis- turbance that helped maintain prairie habitat. Prescribed fires have generally been conducted in winter, however, whereas natural fires burn primarily in summer--and research has demon- strated that at least two species of grassland birds, Florida Grasshopper and Bachman's spar- rows, generally prolong their breeding activities after summer burns (Shriver et al. 1996). In cen- tral Brazil, Parker and Willis (1997) reported that several grassland birds shift their habitats every few years in response to local fires: tall- grass species (e.g., Sharp-tailed Grass-Tyrant [Culcivora caudacuta] and Bearded Tachuri [Polystictus pectoralis]) move to older grass- lands, whereas birds that prefer sparser cover (e.g., Coal-crested Finch [Charitospiza eucos- ma] and Campo Miner [Geobates poecilopte- rus]) shift to newly burned sites. Large or con- nected areas are needed to provide both types of habitats; small reserves protected from fire turn to scrub, whereas annually burned ranches sup- port few species (Parker and Willis 1997). It is especially important that small individual sites (< 500 ha) not be managed for the greatest diversity of grassland bird species. Management INTRODUCTIONVickery et al. 13 TABLE 3. PRELIMINARY LIST OF OBLIGATE GRASSLAND BIRDS OF SOUTH AMERICA Family Rheas Lesser Rhea Tinamous Red-winged Tinamou Huayco Tinamou Darwin's Nothura Spotted Nothura Lesser Nothura Dwarf Tinamou Waterfowl Andean Goose Ruddy-headed Goose Hawks SwainsoWs Hawk Falcons Carunculated Caracara Mountain Caracara White-throated Caracara Striated Caracara Aplomado Falcon Stone curlews Double-striped Thick-knee Plovers, lapwings Southern Lapwing Andean Lapwing Rufous-chested Plover Tawny-throated Dotterel Diademed Sandpiper-Plover Seedsnipes Rufous-bellied Seedsnipe White-bellied Seedsnipe Grey-breasted Seedsnipe Shorebirds Upland Sandpiper Eskimo Curlew Buff-breasted Sandpiper South American Snipe Puna Snipe Giant Snipe Andean Snipe Doves Blue-eyed Ground-Dove Black-winged Ground-Dove Golden-spotted Ground-Dove Owls Burrowing Owl Short-eared Owl Goatsuckers Least Nighthawk Lesser Nighthawk Band-winged Nightjar White-tailed Nightjar White-winged Nightjar Spot-tailed Nightjar Rheidae Rhea pennata Tinamidae Rhynchotus rufescens Rhynchotos maculicollis Nothura darwinii Nothura maculosa Nothura minor raoniscus nanus Anatidae Chloephaga melanoptera Chloephaga rubidiceps Accipitridae Buteo swainsoni Falconidae Phalcoboenus carunculatus Phalcoboenus megalopterus Phalcoboenus albogularis Phalcoboenus australis Falco .femoralis Burhinidae Burhinus bixtriatus Charadriidae Vanellus chilensis Vanellus resplendens Charadrius modestus Eudromias ruficollis Phegornis mitchellii Thinocoridae Attagis gayi Attagis malouinus Thinocorus orbignyianus Scolopacidae Bartramia longicauda Numenius borealis Tryngites subruficollis Gallinago paraguaiae Gallinago andina Gallinago undulata Gallinago jamesoni Columbidae Columbina cyanopis Metriopelia melanoptera Metriopelia aymara Strigidae Athene cunicularia Asio fiammeus Caprimulgidae Chordeiles pusillus Chordeiles acutipennis Caprimulgus longirostris Caprimulgus cayennensis Caprimulgus candicans Caprimulgus maculicaudus 14 TABLE 3. CONTINUED STUDIES IN AVIAN BIOLOGY NO. 19 Family Hummingbirds White-tailed Goldenthroat Tepui Goldenthroat Ecuadorian Hillstar Andean Hillstar White-sided Hillstar Black-breasted Hillstar Olivaceous Thornbill Blue-mantled Thornbill Bronze-tailed Thornbill Rainbow-bearded Thornbill Bearded Helmetcrest Hooded Visorbearer Hyacinth Visorbearer Horned Sungem Ovenbirds Campo Miner Common Miner Puna Miner Dark-winged Miner Creamy-rumped Miner Short-billed Miner Rufous-banded Miner Slender-billed Miner Cipo Canastero Austral Canastero Junin Canastero Scribble-tailed Canastero Straight-billed Reedhaunter Tapaculos Varzea Tapaculo Tyrant flycatchers Sharp-tailed Grass-Tyrant Bearded Tachuri Cock-tailed Tyrant Fork-tailed Flycatcher Larks Horned Lark Wrens Sedge Wren Merida Wren Pipits Correndera Pipit South Georgia Pipit Short-billed Pipit Hellmayr's Pipit Pararno Pipit Yellowish Pipit Chaco Pipit Ochre-breasted Pipit Emberizids Grasshopper Sparrow Grassland Sparrow Black-masked Finch Plumbeous Sierra-Finch Red-backed Sierra-Finch White-throated Sierra-Finch Trochilidae Polytmus guainumbi Polytmus milleri Oreotrochilus chimborazo Oreotrochilus estella Oreotrochilus leucopleurus Oreotrochilus melanogaster Chalcostigma olivaceum Chalcostigma stanleyi Chalcostigma heteropogon Chalcostigma herrani Oxypogon guerinii Augastes lumachellus Augastes scutatus Heliactin cornuta Furnariidae Geobates poecilopterus Geositta cunicularia Geositta punensis Geositta saxicolina Geositta isabellina Geositta antarctica Geositta rufipennis Geositta tenuirostri$ Asthenes luizae Asthenes anthoides Asthenes virgata Asthenes maculicauda Limnornis rectirostris Rhinocryptidae Scytalopus iraiensis Tyrannidae Culicivora caudacuta Polystictus pectoralis Alectrurus tricolor ryrannus savana Alaudidae Eremophila alpestris Troglodytidae Cistothorus platensis Cistothoru$ meridae Motacillidae Anthus correndera Anthus antarcticus Anthus furcatux Anthus hellmayri Anthus bogotensis Anthus lutescens Anthus chacoensis Anthus nattereri Emberizidae Ammodramus savannarum Ammodramus humerails Coryphaspiza melanotis Phrygilus unicolor Phrygilus dorsalis Phrygilus erythronotos INTRODUCTION--Vickery et al. 15 TABLE 3. CONTINUED Family Canary-winged Finch White-winged Diuca-Finch Short-tailed Finch Puna Yellow-Finch Bright-rumped Yellow-Finch Greater Yellow-Finch Patagonian Yellow-Finch Grassland Yellow-Finch Wedge-tailed Grass-Finch Duida Grass-Finch Lesser Grass-Finch Great Pampa-Finch Plumbeous Seedeater Capped Seedeater Ruddy-breasted Seedeater Tawny-bellied Seedeater Dark-throated Seedeater Marsh Seedeater Rufous-rumped Seedeater Chestnut Seedeater Narosky's Seedeater Black-bellied Seedeater Blue Finch Cardinals and allies Dickcissel Meadowlarks, blackbirds Bobolink Saffron-cowled Blackbird White-browed Blackbird Peruvian Meadowlark Red-breasted Blackbird Pampas Meadowlark Long-tailed Meadowlark Eastern Meadowlark Yellow-rumped Marshbird Melanodera melanodera Diuca speculifera Idiospar brachyurus Sicalis lutea Sicalis uropygialis Sicalis auriventris Sicalis lebruni Sicalis luteola Emberizoides herbicola Emberizoides duidae Emberizoides ypiranganus Embernagra platensis Sporophila plumbea Sporophila bouvreuil Sporophila minuta Sporophila hypoxantha Sporophila ruficollis Sporophila palustris Sporophila hypochroma Sporophila cinnamonea Sporophila zelichi Sporophila melanogaster Porphyrospiza caerulescens Cardinalidae Spiza americana Icteridae Dolichonyx oryzivorus Agelaius flayus Sturnella superciliaris Sturnella bellicosa Sturnella militaris Sturnella defilippii Sturnella loyca Sturnella magna Pseudoleistes guirahuro Note: This list was derived primarily from the following sources: Hayman et al. 1986; Ridgely and Tudor 1989; Storz et al. 1996; and R. S. Ridgely, pers. comm. for enhanced alpha diversity is neither necessary nor practical and is likely to be counterproduc- tive to regional conservation goals (Vickery et al. in press). It is important to recognize that certain sites are usually best suited to manage- ment for a particular subset of grassland birds. Sedge meadows, for example, are better suited to management for Sedge Wrens and Le Conte's Sparrows than to a full range of grassland spe- cies (Herkert et al. 1993, Sample and Mossman 1997, Vickery et al. in press). REGIONAL CONSERVATION PLANNING To be effective, grassland habitat conservation planning and action must be conducted within a large regional context. Although conservation action and management usually take place on a local scale at specific sites, cooperative manage- ment on a landscape or regional level makes it possible to address the complete range of habitat needs required by different species, including rare and endangered species, and to minimize the risks of stochastic catastrophic events. In Florida, extensive research on and management of the endangered Florida Grasshopper Sparrow have been site specific but have not yet incor- porated landscape planning or conservation ac- tion. Despite intensive site management, popu- lations of this endemic sparrow are declining, in part because of the absence of a broader geo- graphic framework (Shriver and Vickery 1999). Regional grassland habitat and bird manage- ment plans are developing in many parts of North America and are becoming established in parts of South America. These broad initiatives provide the best opportunities for grassland bird and ecosystem conservation. Partners in Flight, an international effort to 16 STUDIES IN AVIAN BIOLOGY TABLE 4. PRELIMARY LIST OF FACULTATIVE GRASSLAND BIRDS OF SOUTH AMERICA NO. 19 Family Rheas Greater Rhea Tinamous Small-billed Tinamou Ornate Tinamou Andean Tinamou Curve-billed Tinamou Elegant Crested-Tinamou Quebracho Crested-Tinamou Puna Tinamou Patagonian Tinamou Herons Whistling Heron Cattle Egret Ibis Plumbeous Ibis Buff-necked Ibis Black-faced Ibis Storks Wood Stork Maguari Stork Jabiru New World vultures Black Vulture Turkey Vulture Lesser Yellow-headed Vulture Andean Condor Waterfowl Upland Goose Ashy-headed Goose Hawks Pearl Kite White-tailed Kite Long-winged Harrier Northern Harrier Cinereus Harrier Savanna Hawk Harris's Hawk Black-chested Buzzard-Eagle Crowned Eagle White-tailed Hawk Variable Hawk Falcons Crested Caracara Yellow-headed Caracara Chimango Caracara Spot-winged Falconet Seriemas Red-legged Seriema Black-legged Seriema Stone curlews Peruvian Thick-knee Seedsnipes Least Seedsnipe Rheidae Rhea americana Tinamidae Crypturellus parvirostris Nothoprocta ornata Nothoprocta pentlandii Nothoprocta curvirostris Eudromia elegans Eudromia formosa Tinamotis pentlandii Tinamotis ingoufi Ardeidae Syrigma sibilatrix Bubulcus ibis Threskiornithidae Theristicus caerulescens Thetistitus caudatus Theristicus melanopis Ciconiidae Mvcteria americana Ciconia maguari Jabiru mycteria Cathartidae Coragyps atratus Cathartes aura Cathartes burrovianus Vultur gryphus Anatidae Chloephaga picta Chloephaga poliocephala Accipitridae Gampsonyx swainsonii Elanus leucurus Circus buffoni Circus cyaneus Circus cinereus Buteogallus meriodionalis Parabuteo unicinctus Geranoaetus melanoleucus Harpyhaliaetus coronatus Buteo albicaudatus Buteo polyosoma Falconidae Caracara plancus Milvago chimachima Milvago chimango Spiziapteryx circumcinctus Cariamidae Carlama cristata Chunga burmeisteri Burhinidae Burhinus supercilaris Thinocoridae Thinocorus rumicivorus INTRODUCTIONVickery et al. 17 TABLE 4. CONTINUED Family Shorebirds Hudsonian Godwit Baird's Sandpiper Fuegian Snipe Doves Picazuro Pigeon Spot-winged Pigeon Eared Dove Common Ground-Dove Plain-breasted Ground-Dove Ruddy Ground-Dove Buckley's Ground-Dove Picui Ground-Dove Bare-faced Ground-Dove Moreno's Ground-Dove Long-tailed Ground-Dove Scaly Dove Parrots Burrowing Parakeet Monk Parakeet Green-rumped Parrotlet Cuckoos Striped Cuckoo Smooth-billed Ani Groove-billed Ani Barn Owls Barn Owl Owls Striped Owl Goatsuckers Nacunda Nighthawk Scrub Nightjar Scissor-tailed Nightjar Hummingbirds Fiery-throated Hummingbird Green-tailed Goldenthroat Woodpeckers Andean Flicker Campo Flicker Ovenbirds Straight-billed Earthcreeper Rock Earthcreeper Scale-throated Earthcreeper Bar-winged Cinclodes Long-tailed Cinclodes Dark-bellied Cinclodes White-winged Cinclodes Rufous Hornero Pale-breasted Spinetail Lesser Canastero Cordilleran Canastero Streak-throated Canastero Streak-backed Canastero Puna Canastero Many-striped Canastero Hudson's Canastero Firewood-gatherer Scolopacidae Limosa haemastica Calidris bairdii Gallinago stricklandii Columbidae Columba picazuro Columba maculosa Zenaida auriculata Columbina passerina Columbina minuta Columbine talpacoti Columbine buckleyi Columbina picui Metriopelia ciciliae Metriopelia morenoi Uropelia campestris Scardafella squammata Psittacidae Cyanoliseus patagonus Myiopsitta rnonachus Forpus passerinus Cuculidae Tapera naevia Crotophaga ani Crotophaga sulcirostris Tytonidae Tyto alba Strigidae Rhinoptynx clametor Caprimulgidae Podager nacunda Caprimulgus anthonyi Hydropsalis brasiliana Trochilidae Panterpe insignis Polytmus theresiae Picidae Colaptes rupicola Colaptes campestris Furnariidae Upucerthia ruficauda Upucerthia andaecola Upucerthia dumetaria Cincloides fuscus Cincloides pabsti Cincloides patagonicus Cincloides atacamensis Furnarius rufus Synallaxis albescens Asthenes pyrrholeuca Asthenes modesta Asthenes humilis Asthenes wyatti Asthenes sclateri Asthenes fiarnmulata Asthenes hudsoni Anumbius annumbi 18 TABLE 4. CONTINUED STUDIES IN AVIAN BIOLOGY NO. 19 Family Tapaculos Collared Crescent-chest Tyrant flycatchers Plain-crested Elaenia Rufous-crowned Elaenia Lesser Elaenia Grey-backed Tachuri Rufous-sided Pygmy-Tyrant Grey Monjita Black-crowned Monjita White-rumped Monjita White Monjita Rusty-backed Monjita Black-and-white Monjita Chocolate-vented Tyrant Black-billed Shrike-Tyrant White-tailed Shrike-Tyrant Great Shrike-Tyrant Grey-bellied Shrike-Tyrant Lesser Shrike-Tyrant Spot-billed Ground-Tyrant Dark-faced Ground-Tyrant Cinnamon-bellied Ground-Tyrant Rufous-naped Ground-Tyrant Puna Ground-Tyrant White-browed Ground-Tyrant Plain-capped Ground-Tyrant Cinereous Ground-Tyrant White-fronted Ground-Tyrant Ochre-naped Ground-Tyrant Black-fronted Ground-Tyrant Austral Negrito Spectacled Tyrant Strange-tailed Tyrant Streamer-tailed Tyrant Cattle Tyrant Crows, jays White-necked Raven Emberizids Rufous-collared Sparrow Yellow-browed Sparrow Coal-crested Finch Many-colored Chaco-Finch Ash-breasted Sierra-Finch Carbonated Sierra-Finch Yellow-bridled Finch Long-tailed Reed-Finch Black-and-rufous Warbling-Finch Stripe-tailed Yellow-Finch Pale-throated Serra-Finch Blue-black Grassquit Grey Seedeater Variable Seedeater Caqueta Seedeater Wing-barred Seedeater Rusty-collared Seedeater Lesson's Seedeater Lined Seedeater Black-and-white Seedeater Rhinocryptidae Melanopareia torquata Tyrannidae Elaenia cristata Elaenia ruficeps Elaenia chiriquensis Polystictus superciliaris Euscarthmus rufomarginatus Xolmis cinerea Xolmis coronata Xolmis velata Xolmis irupero Xolmis rubetra Heteroxolmis dominicana Neoxolmis rufiventris Agriornis montana Agriornis andicola Agriornis livida Agriornis microptera Agriornis murina Muscisaxicola maculirostris Muscisaxicola macloviana Muscisaxicola capistrata Muscisaxicola ruffvertex Muscisaxicola juninensis Muscisaxicola albilora Muscisaxicola alpina Muscisaxicola cinerea Muscisaxicola albifrons Muscisaxicola fiavinucha Muscisaxicola frontalis Lessonia tufa Hymenops perspicillatus Alectrurus risora Gubernetes yetapa Machetornis rixosus Corvidae Corvus cryptoleucus Emberizidae Zonotrichia capensis Ammodramus aurifrons Charitospiza eucosma Saltatricula multicolor Phrygilus plebejus Phrygilus carbonarius Melanodera xanthogramma Donacospiza albifrons Poospiza nigrorufa Sicalis citrina Embernagra longicauda Volatinia jacarina Sporophila intermedia Sporophila corvina Sporophila murallae Sporophila americana Sporophila collaris Sporophila bouvronides Sporophila lineola Sporophila luctuosa INTRODUCTION--Vickery et al. 19 TABLE 4. CONTINUED Family Yellow-bellied Seedeater Double-collared Seedeater White-bellied Seedeater Chestnut-bellied Seedeater Chestnut-throated Seedeater Large-billed Seed-Finch Great-billed Seed-Finch Lesser Seed-Finch Band-tailed Seedeater Plain-colored Seedeater Yellow-faced Grassquit Black-faced Grassquit Meadowlarks, blackbirds Red-winged Blackbird Yellow-hooded Blackbird Brown-and-yellow Marshbird Chopi Blackbird Bay-winged Cowbird Screaming Cowbird Shiny Cowbird Bronzed Cowbird Sporophila nigricollis Sporophila caerulescens Sporophila leucoptera Sporophila castaneiventris Sporophila telasco Oryzoborus crassirostrix Oryzoborus maximiliana Oryzoborus angolensis Catamenia analis Catamenia inornata Tiaris olivacea Tiaris bicolor Icteridae Agelaius phoeniceus Agelaius icterocephalus Pseudoleistes virescenx Gnorimopsar chopi Molothrus badiux Molothrus rufoaxillaris Molothrus bonariensis Molothrus aeneus Note: This list was derived primarily from the following sources: Hayman et al. 1986: Ridgely and Tudor 1989: Storz et al. 1996; and R. S. Ridgely, pers. comm. protect and enhance North American bird pop- ulations, is organized at state, regional, national, and international levels and provides an excel- lent, flexible structure for facilitating regional conservation efforts (Finch and Stangel 1992). For example, a Northeast Grassland Bird Work- ing Group functions within the rubric of the Northeast Working Group. As a specialist group, the Northeast Grassland Bird Working Group fa- cilitates communication, inventory, and planning across a 13-state region from Maine to Virginia. In 1997 this group was involved in a seven-state inventory of grassland birds, emphasizing re- gionally rare species such as Upland Sandpiper and Henslow's Sparrow (Shriver et al. 1997). Because Partners in Flight has been instrumental TABLE 5. ESTIMATED HABITAT LOSS TO GRASSLAND ECOSYSTEMS IN THE UNITED STATES SINCE EUROPEAN SET- TLEMENT Esti mated Ecosystem loss (%) Reference Critically endangered ecosystems (> 98% habitat loss) a Tallgrass prairie east of Missouri River > 99 Sedge meadows, Wisconsin > 99 Black belt prairie, Alabama and Mississippi > 99 Sandplain grassland, Long Island, NY 99.9 Native prairie, Willamette Valley, OR 99.5 Palouse prairie, Montana, Idaho, Oregon, and Washington 99.9 California grasslands, all types 99 Ungrazed sagebrush steppe, Intermountain West > 99 Endangered ecosystems (80-98% habitat loss) Tallgrass prairie, all types combined 90 Grassland shrubsteppe, Washington and Oregon > 90 Shortgrass prairie, Montana 80-90 Shortgrass prairie, North Dakota 90 Coastal heathland, s. New England and Long Island, NY > 90 Sandplain grassland, New England > 90 Palmetto dry prairie, Florida 81 Noss et al. 1995 Reuter 1986 Noss et al. 1995 Niering 1992 Ingersoll and Wilson 1991 Noss et al. 1995 Kreissman 1991 West 1995 Madson 1990 Noss et al. 1995 Chadde 1992 Madson 1989 Noss et al. 1995 Noss et al. 1995 Shriver and Vickery 1999 a Classification of critically endangered and endangered ecosystems adapted from Noss et al. 1995. 20 STUDIES IN AVIAN BIOLOGY NO. 19 in bringing together multiple agencies, more than 30 collaborators and dozens of volunteers contributed to the grassland inventory, which censused nearly 1,100 sites (Shriver et al. 1997). More importantly, organizations and agencies in each of these states have become invested in the results of this regional effort. In New York, ma- jor breeding habitat for grassland birds has been included in the state's registry of important bird areas and has also received legislative protection (Wells 1998). In the midwestern United States, a multistate plan for grassland bird conservation has devel- oped a broad outline of the region's conservation priorities (Herkert et al. 1996). Within the re- gion, more detailed state plans have been de- veloped. In Wisconsin, for example, Sample and Mossman (1997) have produced a plan that de- scribes goals and organizing principles of grass- land bird management, including a detailed dis- cussion of overall management philosophy; they also identify management priorities for both grassland birds and their habitats within this broad geographic area. The plan supplies de- tailed habitat management guidelines and man- agement recommendations based on individual species' responses to specific management prac- tices and identifies specific landscapes, sites, and properties worthy of special management atten- tion. This type of specific targeting of conser- vation activities will undoubtedly result in on- the-ground management that is likely to benefit grassland birds in the target area. In Canada, conservation of prairie grassland habitat and birds has been gaining momentum through the actions of many organizations since 1990. The scope of these partnerships and inter- actions has grown, culminating in the formation of provincial implementation groups for the Prairie Conservation Action Plan (PCAP) and the formation of provincial (Manitoba) and re- gional Partners in Flight-Canada groups. PIF- Canada sets general priorities based on trends and geographic responsibility (based on propor- tion of range) as set forth by Dunn 1997. In most cases, Canadian prairie fragments in national and provincial parks, federal govern- ment bird sanctuaries, national wildlife areas (NWAs), military bases, Prairie Farm Rehabili- tation Administration (PFRA) holdings, and fed- eral and provincial crown grazing lands are se- cure. Examples of large blocks include Grass- lands National Park, Saskatchewan (90,000 ha); Last Mountain Lake NWA, Saskatchewan (15,000 ha); and Canadian Forces Base Suffield, Alberta (270,000 ha). Large holdings include PFRA pastures (75 million ha) and Saskatche- wan crown grazing lands (2.9 million ha). Because there is presently no federal endan- gered species legislation in Canada, complemen- tary provincial and federal legislation to desig- nate species is being developed, with an empha- sis on rewarding stewardship rather than punish- ing offenders. Efforts have centered around changing adverse government policy and work- ing with agriculture to find "Best Management Practices" for conserving remaining native prai- rie and other grassland habitats. For example, the recent abolition of grain-shipping subsidies based on the number of hectares under cultiva- tion has removed one incentive to plow native prairie. Most farmland in Canada is privately owned, and conservation funding is limited. Identifying options that make it worthwhile for landowners to maintain native prairie or use bird-friendly cropping methods has thus proven to be the most effective and economical approach to conserv- ing grassland habitats. Among such options are subsidy-based programs such as Agriculture Canada's Permanent Cover Program (PCP). In- stituted in 1989, the PCP has converted 450,000 ha in poor soil classes to grass cover for 10 or more years. The payment to landowners covers some of the cost of seeding, and the landowner may use the land for haying or grazing so long as it is not broken. A recent study showed that many grassland obligates use PCP sites (Mc- Master and Davis 1998). In Brazil, high-priority areas for biodiversity conservation in the Cerrado were identified in a 1998 workshop in which more than 200 scien- tists participated. The workshop was part of the Brazilian government's biome-level biodiversity program to establish biodiversity priorities in the country. Important criteria for designating sites included species richness, number of endemic species, presence of rare and/or endangered spe- cies, and sites of unique communities or key ar- eas for migratory species. Eighty-seven priority areas were identified, 20 of which were recom- mended for reserve status because of their im- portance for birds (Silva 1998a). Priorities for conservation action for each of these areas were then determined by cross-referencing biodiver- sity data with data on human encroachment and land-cover changes (Cavalcanti 1999b). In addition to creating new reserves in the Cerrado, new strategies must be adopted as soon as possible to minimize the impact of human activities on the biota of this region (Silva 1998b). The most pressing need is to provide the agricultural technology to help landowners in- crease productivity of lands already under cul- tivation. It is hoped that this will reduce the pressure on lands covered by natural vegetation. Macedo (1994) has suggested that by increasing productivity on lands already used for agricul- INTRODUCTION--Vickery et al. 21 ture in the Cerrado region, it would be possible to produce 100 millions tons of food annually, or enough to feed 250 million people. The sec- ond strategy is to establish legal mechanisms that would preclude the destruction of the bio- logical resources of the Cerrado; as an example, new agriculture projects in areas covered by nat- ural vegetation could be banned until their im- pacts on fauna and flora were rigorously as- sessed. HEMISPHERIC CONSERVATION PLANNING Since most grassland birds migrate between breeding and wintering areas, it is necessary to understand the habitat requirements and conser- vation needs in both these areas. In South Amer- ica, some grassland species breeding in Tierra del Fuego and Patagonia winter in the southern Pampas. This is the case for Upland Goose (Chloephaga picta), Ashy-headed Goose (C. po- liocephala), and the endangered continental race of Ruddy-headed Goose (C. rubidiceps). Other grassland species, such as seedeaters and some tyrant flycatchers, breed in the Pampas but win- ter in northern Argentina, Paraguay, and Brazil (Ridgely and Tudor 1989, Chesser 1994). Although some species of North American grassland birds are long-distance neotropical mi- grants, most species migrate relatively short dis- tances and winter primarily in the southern Unit- ed States and northern Mexico. This provides conservation opportunities for species wintering in North America and Mexico but also under- scores the need for coordinated research and conservation efforts across international borders (Hagan and Johnston 1992, Wilson and Sader 1993, Vickery et al. in press). The habitat requirements of many species wintering in Central and South America are poorly understood. Recently there have been en- couraging research and educational efforts in grassland habitats in Mexico (e.g., Colorado Bird Observatory 1996, Manzano-Fischer et al. 1999) and other parts of Central and South America. For example, the Canadian Wildlife Service's newly developed Latin American Pro- gram is working to train local avian biologists and build local capacity to study and protect mi- gratory and resident birds (Hyslop 1996). The U.S. Fish and Wildlife Service is undertaking similar collaborative efforts. Additionally, pri- vate nonprofit conservation organizations such as The Nature Conservancy and BirdLife Inter- national have also developed international bird conservation programs. There are few efforts, however, directed exclusively toward grassland bird and habitat protection. Widespread efforts by farmers in Venezuela to reduce Dickcissel crop damage (Basili and Temple 1999) and the use of pesticides in Argentina that has killed many Swainson's Hawks (Krapovickas and de Perez 1997) clearly demonstrate the need for ex- panded international grassland bird research and conservation. Changing agricultural practices in Argentina have profoundly reduced the amount of native grassland in that country, and the loss is seri- ously affecting populations of endemic grass- land birds such as the Pampas Meadowlark (Tu- baro and Gabelli 1999). This habitat change is likely to affect populations of nearctic breeders as well and may be particularly significant for long-distance migrants such as Swainson's Hawk, Eskimo Curlew, Upland Sandpiper, Buff- breasted Sandpiper (Tryngites subruficollis), and Bobolink, all of which winter in Argentina (O1- cog 1984). Similar agricultural changes else- where in Central and South America will un- doubtedly have consequences for both neotrop- ical and nearctic grassland breeders. The Western Hemisphere Shorebird Reserve Network (WHSRN), an international conserva- tion network focused specifically on shorebirds (Bildstein et al. 1991), may provide an excellent model for international grassland bird conser- vation efforts. WHSRN has successfully collab- orated with more than 120 other agencies, in- cluding the North American Waterfowl Manage- ment Plan and Partners in Flight, on internation- al wetland and shorebird conservation issues and has helped protect more than 3.6 million ha of habitat in 7 countries (J. Corven, pers. comm.). For example, joint efforts by the Suriname For- est Service, Canadian Wildlife Service, and WHSRN have helped protect critical wintering habitat for Semipalmated Sandpipers (Calidris pusilla) in Suriname (J. Corven, pers. comm.). Recognizing the rapid decline of many South American grassland birds, especially Sporophila seedeaters, Silva (1999) has suggested a system of reserves across South America that would protect a large majority of grassland endemics. Such planning, critical for the protection of en- demic neotropical species, could be coupled with efforts to protect nearctic migrants such as Swainson's Hawks and Dickcissels, and thus to develop a comprehensive system for grassland bird protection throughout the Western Hemi- sphere. Although international efforts, initiated largely by the American Bird Conservancy, in Argentina in 1995 stopped or minimized inci- dental Swainson's Hawk mortality that resulted from insecticide use on agricultural fields, the absence of an established international network meant that emergency measures were required (Anonymous 1996, Krapovickas and de Perez 1997). It is hoped that an established interna- tional grassland bird network would anticipate 22 STUDIES IN AVIAN BIOLOGY NO. 19 such a major crisis and thus minimize the need for such emergency actions. We hope that pub- lication of this volume will facilitate such a net- work. SEEKING COMMON GROUND The effective management of grassland land- scapes will require the involvement of a diverse group of natural resource professionals, includ- ing range managers, game and nongame biolo- gists, soil conservationists, agronomists, farm- ers, and ranchers (Vickery et al. in press). In many areas, grassland management has histori- cally emphasized soil conservation. To increase the likelihood of successfully conserving grass- land habitat, it will be important to combine the goals of avian habitat conservation with those of soil conservation and agriculture. Because the ecological and habitat requirements of many en- dangered grassland species in South America are poorly understood, it will be most difficult to achieve these disparate goals in South America. Although habitat loss is the main cause of grass- land bird declines in South America (Bucher and Nores 1988, Cavalcanti 1988), more subtle fac- tors such as competitive interactions, nest para- sitism, social facilitation, and failure to colonize new patches are probably also involved. These factors are probably stronger when populations are small and fragmented. The North American Waterfowl Management Plan (NAWMP), through Ducks Unlimited Can- ada's Prairie Care program, has established graz- ing systems on about 132,000 ha in the grass- land portion of Canada's three prairie provinces (Alberta, Manitoba, and Saskatchewan). Provin- cial agricultural extension services helped pro- ducers revamp grazing systems on many addi- tional hectares. Because these systems make grazing more economically viable, they keep the land under grass cover. Initial studies show that a greater variety of bird species, including many grassland obligates, use these sites than use con- tinuous-grazing (i.e., season-long) sites (Dale and McKeating 1996) and that avian productiv- ity is about the same as it was before the grazing systems were instituted (Prescott et al. 1998). The initial demonstration farms and agreements with cattle ranchers required a substantial input, but as the economic benefits became clear and neighboring cattle ranchers saw the results, the conservation management was voluntarily adopted on many more farms and ranches. NAWMP has proven to be a good partner in grassland bird conservation. The Canadian Wildlife Service initiated nongame evaluations of NAWMP in 1989 and was joined in this by provincial partners in 1993 (Dale and Mc- Keating 1996). GRASSLAND RESTORATION Because loss of native grassland habitat has been so extensive and has occurred over such a broad region, habitat restoration has become in- creasingly important for many regions and may be critical for the persistence of some rare and endangered species. For example, a recent land- scape analysis in Florida demonstrated that only 19% of the original prairie remains and that the configuration of remaining prairie is insufficient to maintain and enhance populations of the U.S. federally endangered Florida Grasshopper Spar- row (Shriver and Vickery 1999). The best option for the long-term viability of this rare taxon ap- pears to be major habitat restoration (Shriver and Vickery 1999). Although similar landscape anal- yses have not been undertaken in South Amer- ica, the sharp decline in Pampas Meadowlark populations in Argentina (Tubaro and Gabelli 1999) and the rapid destruction of grassland habitat in the Cerrado of central Brazil (Caval- canti 1999a) both suggest that some form of habitat restoration may be critical for the long- term survival of endemic grassland birds in South America. At least in the Pampas, habitat restoration should be possible to achieve in a relatively short time if land is left undisturbed (Leon and Oesterheld 1982, Leon et al. 1984). In North America, several grassland species have adapted to agricultural fields (Graber and Graber 1963, Knopf 1994) or to other artificial habitats such as airports and reclaimed surface mines (Melvin 1994, Jones and Vickery 1997). Because few native prairie or grassland rem- nants remain in most of midwestern and north- eastern North America, effective grassland bird conservation will require the protection and en- hancement of artificial grassland habitats. Re- claimed surface mines in West Virginia, Penn- sylvania, Ohio, and Indiana provide important habitat for Henslow's Sparrow and other grass- land birds, and airfields in northeastern North America support some of the largest New En- gland populations of several regionally threat- ened species, notably Upland Sandpiper and Grasshopper Sparrow (Jones and Vickery 1997). Protection and enhancement of these non-native habitats that serve as refugia for many grassland birds will be critical. Where feasible, however, efforts to restore native habitats should be a long-term objective. FUTURE RESEARCH From a hemispheric perspective, the most pressing needs are additional research and relat- ed conservation in Central and South America, where loss of habitat and population declines are becoming more acute. The number of endemic INTRODUCTION--Vickery et al. 23 species and families in the Neotropics, and the fact that this area provides habitat for wintering nearctic breeders, makes this the highest hemi- spheric priority for conservation research and action. As in North America, a better under- standing of the ecological effects of fire and grazing on South American obligate grassland birds and their habitats should be a high priority (Collar et al. 1992). Grassland bird conservation programs in the United States and elsewhere in the Western Hemisphere need to address both breeding and wintering ecology (Vickery et al. in press). Al- though the wintering ecology of most grassland birds is poorly known, there continues to be lit- tle research on the wintering habitat require- ments of many grassland bird species, as the paucity of papers on wintering ecology in this book clearly demonstrates (3, versus 23 for the breeding season). It is unclear whether habitat loss and degradation on the wintering grounds are primarily responsible for the population de- clines reported for many species. Winter survi- vorship may be critically important in the long- term declines of some grassland species (Herkert and Knopf 1998, Vickery et al. in press). Additionally, although there has been substan- tial research on some arctic-nesting birds, nota- bly waterfowl (e.g., Snow Goose [Chen caeru- lescens]; Ganter et al. 1996) and shorebirds (Charadriidae and Scolopacidae; e.g., Whitfield and Brade 1991), there has been little research on other grassland species, especially passedfies, that breed at high latitudes or altitudes. In par- ticular, there is essentially no research on the winter ecology of these species on temperate grasslands, although initial efforts are underway (E. Dunn, pers. comm.). Winter habitat use, pop- ulation dynamics, and survivorship of species such as Smith's Longspur and the rosy-finches (Leucosticte spp.) are largely unknown and mer- it careful study. Unlike in North America, most species of grassland birds in Central and South America are still poorly known, and information regard- ing their ranges, habitat preferences, and migra- tory movements are based on relatively few ob- servations and limited museum specimens. For instance, Silva (1995) found that approximately 70% of the Cerrado region has never been ade- quately sampled for birds. Well-sampled locali- ties are usually natural areas near major cities or national parks with easy access. This probably reflects the situation for most of the major grass- land regions in Latin America. The taxonomy for several Central and South American grass- land species should be re-evaluated, as they like- ly comprise two or more distinct phylogenetic species, each one indicating a region where con- servation actions need to be taken. Unfortunate- ly, funds for basic ornithological inventory and taxonomic studies in Central and South America are scarce and, when available, are directed at studies on forests rather than grasslands or other open habitats. Any international conservation project directed at Latin American grasslands must include support for both long-term studies on threatened bird populations and basic biolog- ical inventory and taxonomic studies. ACKNOWLEDGMENTS We especially thank B. Dale and R. S. Ridgely for their knowledge and insights of Canadian and South American grassland birds, respectively, and for their valuable contributions to this manuscript. We also thank R. A. Askins, R. Cannings, J. L. Dunn, P. W. Dunwiddie, A. L. Jones, J. E. Pierson, J. T. Rotenberry, and W. G. Shriver for helpful comments on earlier drafts of the manuscript. Many people shared their in- sights regarding lists for North and South American grassland birds, and we express our deepest gratitude to them: C. Bock, S. Davis, E. Dunn, J. L. Dunn, R. Fraga, E Handford, L. Igl, E Knopf, M. Koenen, C. Norment, J. Pierson, E Rabufetti, J. C. Reboreda, R. S. Ridgely, G. Shriver, and D. Storz. We thank V. May- nard for meticulous map preparation. We are grateful to the following institutions for their support: Center for Biological Conservation, Massachusetts Audubon Society and the Switzer Foundation (Vickery); Insti- tuto de Biolog/a y Medicina Experimental CONICET (Tubaro); Conselo Nacional de Desenvolvimento Cien- fffico e Tecno16gico (CNPq), Brazil (Silva); U.S. Geo- logical Survey, Biological Resources Division (Peter- john); Illinois Endangered Species Protection Board (Herkert); and Departamento de Zoologia, Universi- dade de Brasflia, CNPq, and Conservation Internation- al, Inc. (Cavalcanti). LITERATURE CITED AMERICAN ORNITHOLOGISTS' UNION. 1998. Check-list of North American birds. 7th ed. American Orni- thologists' Union, Washington, D.C. ANDERSON, D. L. 1977. Las causas de la invasirn de chafiar en el firea medanosa de pastizales e isletas de chaffan Pp. 11-13 in Limitacirn en la produccirn ganadera de San Luis debido alas lefiosas invasoras [no editor]. Gobierno de la Provincia de San Luis. Instituto Nacional de Teconologia Agropecuaria. San Luis, Argentina. ANONYMOUS. 1996. Hawk deaths spur action. P 14 in Bird Conservation (Fall Migration) [magazine of the American Bird Conservancy, Washington, D.C.]. ASKINS, R. A. 1993. Population trends in grassland, shrubland, and forest birds in eastern North Ameri- ca. Current Ornithology 11:1-34. BAS1L1, G. D., AND S. m. TEMPLE. 1999. Winter ecol- ogy, behavior, and conservation needs of Dickcissels in Venezuela. Studies in Avian Biology 19:289-299. BILDSTEIN, r. L., G. T BANCROFT, 1 J. DUGAN, D. H. GORDON, R. M. ERWIN, E. NOL, X. PAYNE, AND S. E. SENNER. 1991. Approaches to the conservation of coastal wetlands in the western hemisphere. Wilson Bulletin 103:218-254. 24 STUDIES IN AVIAN BIOLOGY NO. 19 BOND, J. 1971. Birds of the West Indies. Houghton Mifflin, Boston, MA. BUCHER, E. H., AND M. mORES. 1988. Present status of birds in steppes and savannas on northern and cen- tral Argentina. Pp. 71-79 in P. D. Goriup (editor). Ecology and conservation of grassland birds. ICBP Technical Publication no. 7. International Council for Bird Preservation, Cambridge, U.K. BURKART, R., AND L. DEL VALLE RUIZ. 1994. Las eas naturales protegidas del pals, datos, historia y eval- uaci6n. Pp. 22 64 in E1 sistema de areas naturales protegidas de la Argentina [no editor]. Administra- ci6n de Parques Nacionales, Buenos Aires, Argen- tina. CABRERA, m. L., AND m. WILLINK. 1980. Biogeograffa de Amdrica Latina. Organization of American States, Washington, D.C. CAVALCANTI, R. B. 1988. Conservation of birds in the cerrado of central Brazil. Pp. 59-66 in P. D. Goriup (editor). Ecology and conservation of grassland birds. ICBP Technical Publication no. 7. Internation- al Council for Bird Preservation, Cambridge, U.K. CAVALCANTI, R. B. 1999a. Bird species richness and conservation in the cerrado region of central Brazil. Studies in Avian Biology 19:244-249. CAVALCANTI, R. B. (COORDINATOR). 1999b. Executive summary. Workshop on priority actions for the con- servation of the biodiversity of the cerrado and pan- tanal, Brasflia, Brazil. Funda9ao Pc6 Natureza, Con- servation International, Funda9ao Biodiversitas, Universidade de Brasflia, Brasflia, Brazil. CHADDE, S. 1992. Decline of natural ecosystems in Montana. Unpublished report. U.S. Forest Service, Missoula, MT CHESSER, R. T. 1994. Migration in South America: an overview of the austral system. Bird Conservation International 4:91 107. COLLAR, N.J., L. P. GONZAGA, N. KRABBE, A. MAD- RONO NIETO, L. G. NARANJO, T. A. PARKER III, AND D.C. WEGE. 1992. Threatened birds of the Ameri- cas. Smithsonian Institution Press, Washington, D.C. COLORADO BIRD OBSERVATORY. 1996. Annual report. Colorado Bird Observatory, Brighton, CO. DALE, B.C., AND G. MCKEATING. 1996. Finding com- mon ground: the nongame evaluation of the North American Waterfowl Management Plan in Canada. Pp. 258-265 in J. T Ratti (editor). 7th international waterfowl symposium. Ducks Unlimited, Memphis, TN. DARWIN, C. 1876. Journal of researches into the natural history and geology of the countries visited during the voyage of HMS Beagle round the world. Murray, London, U.K. DINERSTEIN, E., D. M. OLSON, D. J. GRAHAM, A. L. WEBSTER, S. A. PRIMM, M.P. BOOKBINDER, AND a. LEBEC. 1995. A conservation assessment of the ter- restrial ecoregions of Latin America and the Carib- bean. World Wildlife Fund and World Bank, Wash- ington, D.C. DUNN, E. H. 1997. Setting priorities for conservation, research and monitoring of Canada's landbirds. Technical report no. 293, Canadian Wildlife Service, Ottawa, ON. DUNNING, J. B. 1993. Bachman's Sparrow (Aimophila aestivalis). In A. Poole, P. Stettenhein, and E Gill (editors). The birds of North America no. 38. Phil- adelphia Academy of Natural Sciences, Philadel- phia, PA, and American Ornithologists' Union, Washington, D.C. EITEN, G. 1972. The cerrado vegetation of Brazil. Bo- tanical Review 38:201-341. ENVIRONMENT CANADA. 1998. Terrestrial ecozones of Canada [map]. Hnp://wwwl.ec.gc.ca/--vignenes/ terr. html. FINCH, D. M., AND P. W. STANGEL (EDITORS). 1992. Sta- tus and management of neotropical migratory birds. USDA Forest Service Gem Tech. Rep. RM-229. USDA Forest Service Rocky Mountain Forest and Range Experiment Station, Fort Collins, CO. FJELDS,k, J. 1988. Status of birds of steppe habitats of the Andean zone and Patagonia. Pp. 81-95 in P. D. Goriup (editor). Ecology and conservation of grass- land birds. ICBP Technical Publication no. 7. Inter- national Council for Bird Preservation, Cambridge, U.K. FRAGA, R. M., H. CASAiqAS, AND G. PUGNALI. 1998. Natural history and conservation of the endangered Saffron-cowled Blackbird Xanthopsarfiavus in Ar- gentina. Bird Conservation International 8:255-267. CANTER, B., F. COOKE, AND P. MINEAU. 1996. Long- term vegetation changes in a Snow Goose nesting habitat. Canadian Journal of Zoology 74:965 969. GAYTON, D. V. 1991. Grazing pressure on Saskatche- wan rangelands. Rangelands 13:107-108. GOCHFELD, M. 1979. Brood parasite and host coevo- lution: interactions between Shiny Cowbirds and two species of meadowlarks. American Naturalist 113:855-870. GRABER, R. R., AND J. W. GRABER. 1963. A compara- tive study of bird populations in Illinois, 1906-1909 and 1956-1958. Illinois Natural History Survey Bul- letin 28:383-529. HAGAN, J. M., III, AND D. W. JOHNSTON (EDITORS). 1992. Ecology and conservation of neotropical mi- grant landbirds. Smithsonian Institution Press, Washington, D.C. HAYMAN, P., J. MARCHANT, AND r. PRATER. 1986. Shorebirds: an identification guide to the waders of the world. Houghton Mifflin, Boston, MA. HERKERT, J. R. 1991. Prairie birds of Illinois: popula- tion response to two centuries of habitat change. Il- linois Natural History Survey Bulletin 34:393-399. HERKERT, J. R. 1997. Bobolink Dolichonyx oryzivorus population decline in agricultural landscapes in the midwestern USA. Biological Conservation 80:107- 112. HERKERT, J. R. 1998. The influence of the CRP on Grasshopper Sparrow population trends in the mid- continental United States. Wildlife Society Bulletin 26:227-231. HERKERT, J. R., AND E L. KNOPF. 1998. Research needs for grassland bird conservation. Pp. 273-282 in J. M. Marzluff and R. Sallabanks (editors). Avian con- servation: research and management. Island Press, Washington, D.C. HERKERT, J. R., D. W. SAMPLE, AND R. E. WARNER. 1996. Management of midwestern grassland land- scapes for the conservation of migratory birds. Pp. 89-116 in F. R. Thompson III (editor). Managing midwestern landscapes for the conservation of neo- INTRODUCTION--Vickery et al. 25 tropical migratory birds. USDA Forest Service Gen. Tech. Rep. GTR-NC-187. USDA Forest Service North Central Forest Experimental Station, St. Paul, MN. HERKERT, J. R., R. E. SZAFONI, V. M. KLEEN, AND J. E. SCHWEGMAN. 1993. Habitat establishment, enhance- ment and management for forest and grassland birds. Natural Heritage Technical Publication no. 1. Illinois Department of Conservation. Springfield, IL. HowEt.t., S. N. G., AND S. WEBD. 1995. A guide to the birds of Mexico and northern Central America. Ox- ford University Press, Oxford, U.K. HYSLOP, C. 1996. From north to south: the Canadian Wildlife Service's Latin American program links two worlds. Pp. 12-13 in Bird Conservancy (Win- tering Grounds) [magazine of the American Bird Conservancy, Washington, D.C.]. IMBODEN, C. 1988. Foreword. P. vii in P. D. Goriup (editor). Ecology and conservation of grassland birds. ICBP Technical Publication no. 7. Internation- al Council for Bird Preservation, Cambridge, U.K. INGERSOLL, C. m., AND M. V. WILSON. 1991. Restora- tion plans of a western Oregon remnant prairie. Res- toration Plans and Management Notes 9:110-111. JOHNSGARD, ] m. 1981. The plovers, sandpipers, and snipes of the world. University of Nebraska Press, Lincoln, NE. JONES, m. L., AND P. D. VICKERY. 1997. Distribution and population status of grassland birds in Massa- chusetts. Pp. 187-199 in ED. Vickery and P. W. Dunwiddie (editors). Grasslands of northeastern North America: ecology and conservation of native and agricultural landscapes. Massachusetts Audubon Society, Lincoln, MA. KANTRUD, H. A. 1981. Grazing intensity effects on the breeding avifauna of North Dakota native grass- lands. Canadian Field-Naturalist 95:404-417. KNOPF, E L. 1988. Conservation of steppe birds in North America. Pp. 27-41 in ED. Goriup (editor). Ecology and conservation of grassland birds. ICBP Technical Publication no. 7. International Council for Bird Preservation, Cambridge, U.K. KNOPF, F. L. 1994. Avian assemblages on altered grass- lands. Studies in Avian Biology 15:247-257. KNOPF, E L., AND J. R. RUPERT. 1999. Use of cultivated fields by breeding Mountain Plovers in Colorado. Studies in Avian Biology 19:81-86. KRAPOVICKAS, $., AND J. m. C. DE PEREZ. 1997. Swain- son's Hawk in Argentina: international crisis and co- operation. World Birdwatch 19(4):12-15. KREISSMAN, B. 1991. California, an environmental at- las and guide. Bear Klaw Press, Davis, CA. LAUDER, T. B. 1991. Birds and the Conservation Re- serve Program: a retrospective study. M.S. thesis. University of Maine, Orono, ME. LEON, R. J. C., AND M. OESTERHELD. 1982. Envejeci- miento de pasturas en el norte de la depresi6n del Salado. Un enfoque sucesional. Revista de la Facul- tad de Agronomia 3:41-49. LEON, R. J. C., O. M. RUSCH, AND M. OESTERHELD. 1984. Pastizales pampeanos: impacto agropecuario. Phytocoenologia 12:201-218. MACEDO, J. 1994. Prospectives for the rational use of the Brazilian cerrados for food production. Anais da Academia Brasileira de Cincias 66:159-166. MADSON, C. 1989. Of wings and prairie grass. Nature Conservancy 1989(3):9-13. IADSON, C. 1990. On the Osage. Nature Conservancy 1990(3):7-15. MANTOVANI, J. E., AND A. PERMIRA. 1998. Estimativa da integridade da cobertura vegetal do Cerrado/Pan- tanal atrav6s de dados TM/Landsat. Grupo temitico de Geoprocessamento. Report. Workshop on biodi- versity conservation priorities for the Brazilian cer- rado and Pantanal, Brasœ1ia, Brazil. http:// www.bdt.org.br/bdt/workcerrado/relatorios/inpe. MANZANO-FISCHER, P., R. LIST, AND G. CEBALLOS. 1999. Grassland birds in prairie-dog towns in north- western Chihuahua, Mexico. Studies in Avian Bi- ology 19:263-271. McMASTER, D. G., AND S. K. DAVIS. 1998. Non-game evaluation of the Permanent Cover Program. Sas- katchewan Wetland Conservation Corporation, Re- gina, SK. McNic}oLL, M. K. 1988. Ecological and human influ- ences on Canadian populations of grassland birds. Pp. 1-12 in P. D. Goriup (editor). Ecology and con- servation of grassland birds. ICBP Technical Publi- cation no. 7. International Council for Bird Preser- vation, Cambridge, U.K. MELVIN, S. g. 1994. Military bases provide habitat for rare grassland birds. Natural Heritage News 4:3 [Massachusetts Division of Fisheries and Wildlife, Boston, MA]. MENGEL, R. g. 1970. The North American central plains as an isolating agent in bird speciation. Pp. 279-340 in W. Dort and J. K. Jones, Jr. (editors). Pleistocene and recent environments of the central great plains. University of Kansas Press, Lawrence, KS. NILRING, W. A. 1992. The New England forests. Res- toration Plans and Management Notes 10:24-28. Noss, R. E, E. t. LARoE, AND J. M. SCOTT. 1995. En- dangered ecosystems of the United States: a prelim- inary assessment of loss and degradation. Report no. 0611-R-01 (MF). U.S. Department of the Interior, National Biological Service, Washington, D.C. O'CONNOR, g. J., M. t. JONES, g. B. BOONE, AND t. B. LAUDER. 1999. Linking continental climate, land use, and land patterns with grassland bird distribution across the conterminous United States. Studies in Avian Biology 19:45-59. OLROG, C. C. 1984. Las aves Argentinas. Administra- cion de Parques Nacionales, Buenos Aires, Argen- tina. PARCER, T A., III, AND E. O. WILLIS. 1997. Notes on three tiny grassland flycatchers, with comments on the disappearance of South American fire-diversified savannas. Ornithological Monographs 48:549-555. PETERJOHN, B. O., AND J. R. SAUER. 1999. Population status of North American grassland birds from the North American Breeding Bird Survey 1966-1996. Studies in Avian Biology 19:27-44. PITELKA, E A., E Q. TOMICH, AND G. W. TREICHEL. 1955. Ecological relations of jaegers and owls as letoming predators near Barrow, Alaska. Ecological Monographs 25:85-117. PRESCOIT, D. R. C., B. C. DALE, AND R. D. DICKSON. 1998. Effects of timing and intensity of grazing on nest success of upland-nesting birds on the Univer- 26 STUDIES IN AVIAN BIOLOGY NO. 19 sity Ranch. North American Waterfowl Management Plan report 034. Land Stewardship Centre of Canada and Canadian Wildlife Service, Edmonton, AB. PRICE, J., S. DROEGE, AND A. PRICE. 1995. The summer atlas of North American birds. Academic Press, To- ronto, ON. RAFFAELE, H. A. 1989. A guide to the birds of Puerto Rico and the Virgin Islands. Princeton University Press, Princeton, NJ. REUTER, D. D. 1986. Sedge meadows of the upper midwest: a stewardship summary. Natural Areas Journal 6:2-34. REYNOLDS, R. E., T L. SHAFFER, J. R. SAUER, AND B. G. PETERJOHN. 1994. Conservation Reserve Pro- gram: benefit for grassland birds in the northern plains. Transactions of the North American Wildlife and Natural Resources Conference 59:328-336. RIDGELY, R. S., AND G. TUDOR. 1989. The birds of South America: the oscine passerines. University of Texas Press, Austin, TX. RISSER, P. G., E. C. BIRNEY, H. D. BLOCKER, S. W. MAY, W. J. PARTON, AND J. A. WIENS. 1981. The true prairie ecosystem. Vol. 16, United States/Inter- national Biological Program Synthesis Series. Hutchinson Ross, Stroudsburg, PA. RODENHOUSE, N. L., L. B. BEST, R. J. O'CONNOR, AND E. K. BOLLINGER. 1995. Effects of agricultural prac- tices and farmland structures. Pp. 269-293 in T E. Martin and D. M. Finch (editors). Ecology and man- agement of neotropical migratory birds. Oxford Uni- versity Press, Oxford, U.K. SAMPLE, D. W., AND M. J. MOSSMAN. 1997. Managing habitat for grassland birds: a guide for Wisconsin. Department of Natural Resources, Madison, WI. SAMSON, E, AND E KNOPF. 1994. Prairie conservation in North America. BioScience 44:418-421. SHR[VER, W. G., A. L. JONES, AND P. D. VICKERY. 1997. Northeast grassland bird survey. Report to the Na- tional Fish and Wildlife Foundation, project #96- 177. National Fish and Wildlife Foundation, Wash- ington, D.C. SHRIVER, W. G., AND P. D. VICKERY. 1999. Aerial as- sessment of potential Florida Grasshopper Sparrow habitat: conservation in a fragmented landscape. Florida Field Naturalist 27:1-9. SHR1VER, W. G., P. D. V1CKERY, AND S. A. HEDGES. 1996. Effects of summer burns on Florida Grass- hopper Sparrow. Florida Field Naturalist 24:68-73. SHRIVER, W. G., P. D. VICKERY, AND D. W. PERKINS. 1999. The effects of summer burns on breeding Flor- ida Grasshopper and Bachman's sparrows. Studies in Avian Biology 19:144-148. SILVA, J. M. C. 1995. Avian inventory of the cerrado region, South America: implications for biological conservation. Bird Conservation International 5:15- 28. SILVA, J. M. C. 1998a. Integrating biogeography and conservation: an example with birds and plants of the cerrado region. Anais da Academia Brasileira de Cincias 70:881-888. SILVA, J. M. C. 1998b. Grupo temitico das aves. Re- port. Workshop on biodiversity conservation priori- ties for the Brazilian cerrado and Pantanal, Brasflia, Brazil. Http://www.bdt.org.br/bdt/workcerrado/rela- torios/aves. SILVA, J. M. C. 1999. Seasonal movements and con- servation of seedeaters of the genus Sporophila in South America. Studies in Avian Biology 19:272- 280. SOmANO, A. 1991. Rio de la Plata grasslands. Pp. 367- 407 in R. T. Coupland (editor). Ecosystems of the world. Vol. 8A. Natural grasslands, introduction and western hemisphere. Elsevier, Amsterdam, Nether- lands. STATISTICS CANADA. 1997. Indicators and detailed sta- tistics. Government of Canada Catalogue no. 16- 200-XKE. Government of Canada, Ottawa, ON. STOTZ, D. E, J. W. FITZPATRICK, t. A. PARKER III, AND D. K. MOSKOVITS. 1996. Neotropical birds: ecology and conservation. University of Chicago Press, Chi- cago, IL. TUBARO, P. L., AND E M. GABELL[. 1999. The decline of the Pampas Meadowlark: difficulties of applying the IUCN criteria to neotropical grassland birds. Studies in Avian Bioloby 19:250-257. VICKERY, P. D., J. R. HERKERT, E L. KNOPF, J. RUTH, ANO C. E. KELLER. In press. Grassland birds: an overview of threats and recommended management strategies. In R. E. Bonney, Jr., D. N. Pashley, and R. Cooper (editors). Strategies for bird conservation: creating the Partners in Flight planning process. Cor- nell Laboratory of Ornithology, Ithaca, NY. WARNER, R. E. 1994. Agricultural land use and grass- land habitat in Illinois: future shock for midwestern birds. Conservation Biology 8:147-156. WEGE, D.C., ANO A. J. LONG. 1995. Key areas for threatened birds in the neotropics. BirdLife Inter- national, Cambridge, U.K. WELLS, J. V. 1998. Important bird areas in New York state. National Audubon Society of New York state, Albany, NY. WEST, N. E. 1995. Strategies for maintenance and re- pair of biotic community diversity on rangelands. Pp. 275-289 in R. Szaro (editor). Biodiversity in managed landscapes. Oxford University Press, Ox- ford, U.K. WHEELWRIGHT, N. T., AND R. A. MAUCK. 1998. Philo- parry, natal dispersal, and inbreeding avoidance in an island population of Savannah Sparrows. Ecology 79:755-767. WHEELWRIGHT, N. t., AND J. D. RISING. 1993. Savan- nah Sparrow (Passerculus $andwichensis). In A. Poole and E Gill (editors). The birds of North Amer- ica no. 45. Academy of Natural Sciences, Philadel- phia, PA, and American Ornithologists' Union, Washington, D.C. WHITFIELD, D. P., AND J. J. BRADE. 1991. The breeding behaviour of the Knot Calidrix canutux. Ibis 133: 246-255. WILSON, A. S. 1926. Lista de aves del sur de Santa Fe. Hornero 3:349-363. WILSON, M. H., AND S. A. SADER (EOITORS). 1993. Con- servation of neotropical migratory birds in Mexico. Miscellaneous Publication no. 727. Maine Agricul- tural and Forest Experiment Station, Orono, ME. Studies in Avian Biology No. 19:27-44, 1999. POPULATION STATUS OF NORTH AMERICAN GRASSLAND BIRDS FROM THE NORTH AMERICAN BREEDING BIRD SURVEY, 1966-1996 BRUCE G. PETERJOHN AND JOHN R. SAUER Abstract. We summarize population trends for grassland birds from 1966 to 1996 using data from the North American Breeding Bird Survey. Collectively, grassland birds showed the smallest per- centage of species that increased of any Breeding Bird Survey bird group, and population declines prevailed throughout most of North America. Although 3 grassland bird species experienced significant population increases between 1966 and 1996, 13 species declined significantly and 9 exhibited non- significant trend estimates. We summarize the temporal and geographic patterns of the trends for grassland bird species and discuss factors that have contributed to these trends. LA CONDICI(SN DE LA POBLACI(SN DE AVES DE PASTIZAL EN AMIRICA DEL NORTE UTILIZANDO EL BREEDING BIRD SURVEY DE NORTEAMIRICA, 1966-1996 Sinopsis. Resumimos las tendencias poblacionales para las aves de pastizal desde 1966 hasta 1996 utilizando datos del Breeding Bird Survey de Norteamrica. Colectivamente, las aves de pastizal tuvieron el menor porcentaje de especies que aumentaron entre todos los grupos de aves en el Breeding Bird Survey. Prevalecieron las disminuciones de poblaciones de estas aves en la mayorfa de Nortea- mfirica. Aunque 3 especies de aves de pastizal experimentaron importantes aumentos poblacionales entre 1966 y 1996, 13 especies disminuyeron significativamente y 9 manifestaron estimaciones de tendencias no significativas. Resumimos los patrones temporales y geogrficos de las tendencias para especies de aves de pastizal y analizamos los elementos que han contribuido a estas tendencias. Key Words: bird populations; Breeding Bird Survey; grassland birds. The status and distribution of grassland birds in North America have apparently undergone dra- matic changes in the past 200 yr. Settlement of the continent by Europeans had both positive and negative effects on grassland bird commu- nities. In eastern North America, the conversion from a forested to a largely agricultural land- scape enabled grassland species to increase pop- ulations and expand their distributions, primarily during the nineteenth century (Andrle and Car- roll 1988, Brewer et al. 1991, Peterjohn and Rice 1991). In contrast, the native grasslands of central and western North America suffered from settlement activities, particularly where the conversion to cultivated crops or overgrazing eliminated or severely altered these habitats (Bock and Bock 1988, Knopf 1988). Anecdotal evidence suggests that populations of most grassland birds have declined in North America during the twentieth century. Although a variety of factors have contributed to these de- clines, the continued degradation and destruction of native grassland habitats remain the most prominent factors across the continent (Mc- Nicholl 1988, Askins 1993, Knopf 1994). Changing agricultural land-use practices have also been detrimental, contributing to declines in species occupying non-native pastures and hay- fields (Bollinger et al. 1990, Askins 1993). In this paper, we use data from the North American Breeding Bird Survey (BBS; Robbins et al. 1986, Peterjohn and Sauer 1996) to de- scribe the geographic and temporal patterns in trends of grassland bird populations during the breeding seasons between 1966 and 1996. We evaluate regional patterns of observed species richness and mean trends for all grassland birds, document the percentage of grassland species with increasing trends, and compare this per- centage to other species groups of management and ecological interest. METHODS The BBS is a roadside survey of approximately 4,000 randomly located survey routes established along secondary roads in the continental United States and Canada (see Peterjohn 1994). Although route cov- erage varies temporally and geographically, more than 2,800 routes have been surveyed annually since 1980. Each route is 39.4 km long with 50 stops spaced at 0.8-kin intervals and is surveyed once annually by a single observer during the peak of the breeding season, primarily in June. The observer records all birds heard or seen within 0.4 km of each stop during a 3-min period. The BBS was started in 1966 in eastern North America, and by 1968 routes were established across the continental United States and southern Canada. Additional information on the history and methodol- ogy of the survey is provided by Robbins et al. 1986. STATISTICAL ANALYSES We used the total number of individuals of a species counted over the entire BBS route as a population in- 27 28 STUDIES IN AVIAN BIOLOGY NO. 19 dex. Most of our analyses of BBS data used the time series of population indices from the routes to estimate trend (a measure of population change over a pre- scribed interval, usually presented as percent change/ year) and relative abundance (mean count). We also summarized this information for regions. Because the sites were georeferenced by route starting points, we were able to map information such as trends, relative abundance, and species richness. Species richness maps We calculated the number of grassland species re- corded on each BBS route in 1966-1996. We then de- veloped contour maps of species richness, using the route species totals as input to smoothing procedures (Isaaks and Srivastava 1989, Cressie 1992). We used inverse distancing (Isaaks and Srivastava 1989) to smooth these data. In this procedure, abun- dance was estimated at a location as a distance-weight- ed average of counts from nearby sites. We used in- verse distancing to estimate abundances for uniformly spaced locations on a 21.4-km grid across the conti- nental United States and southern Canada and then used the Arc/Info Geographic Information System pro- gram to make a contour map from the estimated abun- dances (Environmental Systems Research Institute 1991). See Sauer et al. 1995 and 1997 for applications and discussions regarding mapping of survey data. Trend estimation We estimated trends from each route using the es- timating equation procedure in which a multiplicative trend is modeled (Link and Sauer 1994). As in earlier analyses, we incorporated observer effects in the mod- el to minimize bias associated with improved observer quality over time (Sauer et al. 1994). Maps of regional patterns of trend were also esti- mated using contouring. To accommodate the differ- ences in quality of information among routes, however, trend estimates were weighted by estimates of vari- ances of trends and relative abundances from individ- ual routes. This weighting was similar to that used in the estimation of the regional mean trends (e.g., Geiss- let and Sauer 1990, Link and Sauer 1994). We estimated regional trends as a weighted mean of the route-specific trends. Weights of abundance along routes, precision of trend estimates, and areas were used to accommodate inequities in data quality and regional variation in numbers of samples (Geissler and Sauer 1990). The areas of BBS physiographic strata in states and provinces were used in the weighting (see Butcher 1990 for a map of BBS physiographic strata). Bootstrapping was used to estimate variances of trends. Regional trends were estimated for the entire survey area of the BBS (hereafter called "continen- tal"), BBS regions (Eastern, Central, and Western; Bystrak 1981), states, and provinces. Regional trends were estimated for three time periods: 1966-1996, 1966-1979, and 1980-1996. Subinterval trend esti- mates were based on smaller samples of routes and were sometimes much less precise than long-term es- timates. Hence, long-term trends may not be accurately reflected in either or both of the subinterval estimates. Annual indices To evaluate nonlinear patterns of population change, we used the residual method for estimating annual in- dices of abundance (Sauer and Geissler 1990). In these analyses, a composite yearly index of abundance was estimated as a mean residual from the estimated re- gional trend. Nonlinear patterns in the indices were illustrated with LOESS smooths (James et al. 1990). Composite analysis of grassland birds Although several authors have developed lists of grassland bird species (Udvardy 1958, Mengel 1970), no consensus exists regarding the composition of this ecological grouping. In this paper we use the 25 spe- cies included in the grassland bird group of Peterjohn and Sauer 1993. This group is generally restricted to obligate grassland species, although the raptors occupy large territories that may include mixed grassland- shrubland communities, open areas, and other habitats. The group does not include species that regularly use nongrassland habitats during some seasons of the year or in sizable portions of their ranges, however. For each BBS route, we calculated the total number of grassland bird species and mean trend for the species group. Maps were made using inverse distancing to illustrate geographic patterns in distributions and trends. We estimated percentages of increasing species for other groupings of birds to compare with composite trends in grassland birds. These groups are defined in Peterjohn and Sauer 1993 and include groups based on breeding habitat (wetland, scrub/successional, wood- land, urban), migration form (short-distance migrant, neotropical migrant, permanent resident), nest type (cavity, open cup), and nest location (ground/low, midstory/canopy). For each group, we estimated trends for each species in the group over the surveyed area and determined the percentage of species with positive trend estimates using a procedure based on empirical- Bayes methods that incorporates the relative variances of the component trend estimates (Link and Sauer 1995). We used a z-test to evaluate the null hypothesis that the percentage of species with increasing trends did not differ from 50. RESULTS Grassland bird species richness in North America was greatest in the Great Plains, espe- cially in portions of North Dakota, Montana, and the adjacent prairie provinces of Canada (Fig. 1). Species richness was noticeably reduced east and west of the Great Plains, becoming most depauperate in the southeastern states. Trends for the entire group showed declines prevailing throughout most of the United States and south- ern Canada (Fig. 2). Areas with increasing pop- ulations of grassland birds were small and lo- cally distributed, although one of these areas (in northeastern Montana and northwestern North Dakota) corresponded with the northern Great Plains where species richness was greatest. Only 23% of grassland bird species--the smallest pro- portion of any BBS bird group--showed posi- BBS POPULATION STATUSPeterjohn and Sauer 29 limit 60-100 , 20 - 60 of 5-20 1 - 5 10tal FIGURE 1. Species richness map for grassland bird group, 1966-1996, expressed as the percentage of the total group. tive population trend estimates during 1966- 1996 (Fig. 3). The continental trend estimates for all grass- land birds (25 species) are summarized in Table 1. Although the entire grassland bird group gen- erally declined throughout North America be- tween 1966 and 1996, individual species showed a variety of temporal and geographic patterns in population trends. These trends are discussed in greater detail below. SPECIES WITH INCREASING POPULATION TRENDS BBS data indicate that only three grassland bird species experienced significant increases in their continental populations between 1966 and 1996. Few Ferruginous Hawks (Buteo regalis) were recorded along BBS routes prior to 1980, although their representation improved subse- quently. As is true for many raptors, this species is not well sampled by the BBS methodology and was recorded in small numbers throughout its range. Population increases after 1980 were largely responsible for the increasing trends shown over the entire survey period (Table 1). These increases were evident across most of the breeding range (Fig. 4). Upland Sandpipers (Bartramia longicauda) were more adequately surveyed by the BBS methodology than was the preceding species. Population increases were most evident from the Great Plains westward, whereas declines were concentrated from the Great Lakes into Minne- sota and Wisconsin (Fig. 5). The remnant, lo- cally distributed populations elsewhere in east- ern North America are currently poorly moni- tored by the BBS. Although some regional var- iability exists in the temporal patterns of trends in this species, the continental increases were most consistent between 1978 and 1992 (Fig. 6). Sedge Wrens (Cistothorus platensis) are op- FIGURE 2. Population trend map for grassland bird group, 1966-1996. The map presents areas of consis- tent population change, grouped into categories of de- clining (< -0.25% change per year), indeterminate (-0.25 to +0.25% change per year), and increasing (> 0.25% change per year) trends. portunistic breeders, apparently exhibiting little site fidelity (Burns 1982). This species' erratic seasonal movements may obfuscate population- trend estimates, and they should thus be viewed with caution. The long-term trend was generally positive, reflecting increases during the 1980- 1996 interval. In contrast, most trend estimates for the species were negative during the 1966- 1979 interval. Increases between 1966 and 1996 were most prevalent from the Dakotas into Man- itoba and portions of Minnesota, whereas de- clines prevailed eastward from Iowa and Wis- consin and in Saskatchewan (Fig. 7). 70  50 o 4o  30 lO 0 Gr We Su Wo Ur Ca Oc Sd Pr Nm Gn Mc All Species group FIGURE 3. Percentages of increasing species for all BBS bird groups, 1966-1996. Gr = grassland birds, We = wetland birds, Su = shrub and successional birds, Wo = woodland birds, Ur - urban birds, Ca = cavity nesters, Oc = open-cup nesters, Sd  short- distance migrants, Pr = permanent residents, Nm = neotropical migrants, Gn = ground- and low-nesting birds, Mc midstory- and canopy-nesting birds, All = all species. An asterisk (*) denotes a significant (P < 0.05) deviation from 50%. 30 STUDIES IN AVIAN BIOLOGY TABLE 1. BBS CONTINENTAL TREND ESTIMATES FOR GRASSLAND BIRDS NO. 19 1966-1996 1966-1979 1980-1996 Species Trend a pb N c RA d Trend P N Trend P N Northern Harrier -0.6 - 891 0.49 -1.4 - 397 -0.7 - 745 Ferruginous Hawk 5.2 *** 186 0.25 2.6 - 34 7.2 ** 170 Ring-necked Pheasant - 1.0 ** 1,206 7.30 -0.8 - 735 -0.6 - 1,060 Sharp-tailed Grouse 0.3 - 124 0,55 -4.8 - 53 1.8 - 105 Mountain Plover -2.7 ** 33 0.31 2.0 - 9 3.7 - 29 Upland Sandpiper 1.3 *** 581 2.22 3.0 ** 315 -0.9 - 486 Long-billed Curlew - 1.4 202 1.45 1.7 74 -2.0 - 155 Short-eared Owl -2.8 - 132 0.21 4.1 - 54 -0.8 - 90 Horned Lark - 1.3 *** 1,805 27.02 -0.4 1,064 -2.0 *** 1,559 Sedge Wren 2.2 ** 307 1.14 -3.3 ** 162 1.9 - 262 Sprague's Pipit -4.7 *** 108 1.41 -6.6 *** 51 -4.5 - 94 Dickcissel - 1.6 *** 783 16.29 -5.5 *** 559 0.4 - 706 Cassin's Sparrow -2.5 *** 203 16.31 0.4 96 -0.2 - 186 Vesper Sparrow -0.8 ** 1,462 7.84 1.9 *** 816 0.1 1,232 Lark Bunting -0.9 332 42.97 -4.0 ** 154 0.2 - 288 Savannah Sparrow -0.6 ** 1,477 8.40 0.2 - 810 -0.2 - 1,336 Baird's Sparrow - 1.6 - 115 1.87 -4.7 ** 52 - 1.1 95 Grasshopper Sparrow -3.6 *** 1,404 3.97 -4.6 *** 857 -2.1 *** 1,193 Henslow's Sparrow -8.8 *** 149 0.15 -6.0 ** 99 10.4 *** 80 Le Conte's Sparrow 1.4 - 154 0.73 -1.9 - 44 4.6 *** 138 McCown's Longspur 1.1 - 59 4.57 3.5 - 27 2.7 - 44 Chestnut-collared Longspur -0.1 145 9.27 1.6 - 76 1.1 - 125 Bobolink -1.6 *** 1,134 5.35 1.1 ** 761 -3.8 *** 1,026 Eastern Meadowlark -2.6 *** 1,92 l 20.29 - 1.4 *** 1,338 3.0 *** 1,761 Western Meadowlark -0.6 ** 1,480 44.48 - 1.4 ** 800 -0.3 - 1,348 Note: See American Ornithologists' Union 1983 for scientific nantes. a Average percent change per year. b ** 0.01 < p < 0.05, *** P < 0.01. c Number of BBS route on which each specie has been recorded. d Relative abundance; expressed as mean number of individuals per BBS route within the range of the species. SPECIES WITH NONSIGNIFICANT POPULATION TRENDS Nine grassland bird species had nonsignificant trend estimates, although for six of these species the trends were in a negative direction (Table 1). Sharp-tailed Grouse (Tympanuchus phasianel- lus) and Short-cared Owls (Asio flammeus) were poorly sampled by the BBS methodology, and estimates may not be representative of actual population trends (Table 1). Other species, in- cluding Le Conte's Sparrow (Ammodramus le- conteii) and McCown's Longspur (Calcarius mccownii), have restricted ranges or are other- wise infrequently encountered along BBS routes. Their population trends were generally imprecisely estimated and should also be viewed with caution. Four other species are summarized BB$ limi! Percent Change per Year Less than -0.25 -O.25 to +0.2r n +0,25 FIGURE 4. 1966-1996. BBI limit Percent Change per Year Le$s than -0,25 -0.25 tO +0.25 Greater than +0.25 Ferruginous Hawk population trend map, FIGURE 5. Upland Sandpiper population trend map, 1966-1996. BBS POPULATION STATUS---Peterjohn and Sauer 31 3 2.s . ': : ß '  1.5 i 1 0.5 066 70 74 718 YeaSt 2 86 90 94 FIGURE 6. Continental indices for Upland Sandpip- er, 1966-1996. below to illustrate the temporal and geographic patterns in their trends. Northern Harriers (Circus cyaneus) were widely encountered in relatively small numbers throughout their range. BBS data indicate that the most consistent declines occurred from the Great Plains westward through the intermoun- tain states, although increases were evident in portions of the Dakotas, Montana, and Wyoming (Fig. 8). Consistent increases were also evident from Wisconsin eastward to the maritime prov- inces and in the states bordering the Pacific coast. This species' continental population trends remained negative from 1966 to 1996 (Table 1). Long-billed Curlews (Numenius americanus) are not particularly conspicuous during the breeding season except when they vocalize at dawn (Fitzner 1978). Their relative abundance may be under-represented by the BBS method- ology (Redmond et al. 1981). Also, reports of nonbreeders along BBS routes may have ob- scured population trends in some areas. Despite these potential limitations, BBS data suggest that Long-billed Curlews declined throughout the western Great Plains but tended to increase west 'p' 'B$ limit ercent Change per Year FIGURE 8. Northern Harrier population trend map, 1966-1996. of the Rocky Mountains except in Utah (Fig. 9). No consistent temporal patterns were evident in these trends (Table 1). Lark Buntings (Calamospiza melanocorys) can be nomadic during the breeding season, and these short-term movements may obscure or ac- centuate long-term population trends (Stewart 1975, Andrews and Righter 1992). Along BBS routes, population declines predominated through- out most of the Lark Bunting's range (Fig. 10). Increasing populations were small and localized, except in Montana. Population declines during 1966-1979 were largely responsible for the long-term trends in this species, although the es- timates became more positive after 1980 (Table 1). As a result of their limited distribution on the northern Great Plains and historic declines in some populations (Stewart 1975), Baird's Spar- rows (Ammodramus bairdii) have received con- siderable attention. The BBS population trend map indicates that declines were prevalent in North Dakota and along the northern periphery of this species' range, whereas increases were evident elsewhere (Fig. 11). The trend estimates FIGURE 7. 1966-1996. Sedge Wren population trend map, FIGURE 9. Long-billed Curlew population trend map, 1966-1996. 32 STUDIES IN AVIAN BIOLOGY NO. 19 FIGURE 10. Lark Bunting population trend map, 1966-1996. 10 Percent Ch e per Year Less than -0.25 -0.25 to +0.25 er than +0.25 0 i . , 66 70 74 76 aSr2 86 90 94 Ye FIGURE 12. Continental indices for Ring-necked Pheasant, 1966-1996. were nonsignificant for the entire survey period, although a significant decline occurred during 1966-1979 (Table 1). SPECIES WITH DECLINING POPULATION TRENDS Most of the 13 species that experienced sig- nificant declines in their continental populations during 1966-1996 were widely distributed and well sampled by the BBS (Table 1). Mountain Plovers (Charadrius montanus), however, were recorded on a relatively small number of routes which precluded a detailed analysis of the spe- cies' population trends. Other species, including Sprague's Pipit (Antbus spragueii) and Hen- slow's Sparrow (Ammodramus henslowii), have relatively limited breeding distributions but ex- perienced consistent significant rangewide de- clines between 1966 and 1996. Cassin's Spar- rows (Aimophila cassinii) exhibit considerable annual fluctuations in abundance which produce imprecise trend estimates, but they have shown a declining tendency since the mid-1970s. Ex- amples of other temporal and geographic pat- terns in population trends shown by declining species are described below.' Populations of Ring-necked Pheasants (Phas- ianus colchicus) exhibited consistent trends dur- ing 1966-1979 and 1980-1996 (Table 1). The surveywide indices declined noticeably during the mid-1970s, followed by a slight recovery and then another decline during 1982-1985 (Fig. 12). Increasing populations were most evident in the Great Plains, whereas declines were wide- spread from the Rocky Mountains westward and from Wisconsin and Illinois east into New Eng- land (Fig. 13). BBS data indicate that Horned Lark (Eremo- phila alpestris) populations experienced wide- spread declines between 1966 and 1996 (Table 1, Fig. 14). Declining trends were prevalent in most regions of the continent, although local in- creases were evident from the western Great Lakes across the northern Great Plains and into the intermountain western states. The status and distribution of the Dickcissel (Spiza americana) have always been confound- ed by the species' irregular population move- ments (Emlen and Wiens 1965, Ewert and Can- tino 1967). These movements normally produce influxes near the northern periphery of the breeding range that are inversely correlated with BB$ limit Perten! Chanoe per Year Less Ihan -0.25 -0.25 to +0.25 n + 0.25 FIGURE 11. Baird's Sparrow population trend map, FIGURE 13. Ring-necked Pheasant population trend 1966-1996. map, 1966-1996. BBS POPULATION STATUS---Peterjohn and Sauer 33 FIGURE 14. Horned Lark populion trend map, 1966-1996. habitat suitability in the southern portion of the range (Fretwell 1986). Perhaps as a result of these irregular movements, the geographic pat- terns in the long-term trends of Dickcissels are not uniform (Fig. 15). Declines were most prev- alent across the northern half of the range and in central Texas. Increases predominated from northern Texas through Oklahoma into Kansas and from Arkansas and Louisiana east into Al- abama and Tennessee. The continental annual indices exhibited a distinct decline from 1966 through the late 1970s, followed by variable but fairly stable counts (Fig. 16A). Declines during the first 10 yr of the BBS were evident in the Eastern and Central BBS regions, but popula- tions in both regions were reasonably stable be- tween 1980 and 1996 (Figs. 16B and C). Vesper Sparrow (Pooecetes gramineus) pop- ulations showed consistent declines from Min- nesota, Wisconsin, and Indiana eastward and from Montana and South Dakota south to north- em New Mexico, northern Arizona, and Nevada (Fig. 17). Increasing populations predominated from Illinois across Iowa to Kansas and north- ward into North Dakota. The continental indices A. Continental 30  20- [ ' ß . o 66 70 74 76 82 86 90 94 Year B. Eastern 66 70 74 78 62 66 9'0 9 Year C. Central 25 t  15 1 20 O 66 70 74 78 2 6 90 94 Year FIGURE 16. BBS annual indices for Dickcissel pop- ulations, 1966-1996: A = continental, B = Eastern BBS region, C = Central BBS region. varied, but declines were most evident before the mid-1970s (Fig. 18). This temporal pattern reflected similar trends in the Central and West- ern BBS regions; populations in the Eastern BBS region declined throughout the survey pe- riod (Fig. 17). Savannah Sparrows (Passerculus sandwich- ensis) have expanded their breeding range dur- ing the twentieth century, most noticeably in the midwestern states where their breeding distri- bution has spread southward from the upper Great Lakes (Monroe et al. 1988, Peterjohn and Rice 1991). Despite this range expansion, pop- ulations declined from Ontario, Minnesota, and Iowa eastward between 1966 and 1996 (Fig. 19). limit per c;, t. Ct hh  2  .p;sr Year -0.25 to +0.55 JIBS limit Percent Chan0e per Year Less than 0.25 ater than + 0.25 -0.25 to +0.25 FIGURE 15. Dickcissel population trend map, 1966- FIGURE 17. Vesper Sparrow population trend map, 1996. 1966-1996. 34 STUDIES IN AVIAN BIOLOGY NO. 19 ! I 66 70 74 78 82 86 90 94 Year FIGURE 18. Continental indices for Vesper Sparrow, 1966-1996. A mosaic of increases and decreases was evident elsewhere, with increases most prevalent in the Rocky Mountains and northern Great Plains. Re- gional trends exhibited increases in the Western BBS region and more variation, including a fair- ly distinct decline during the late 1970s, in the Central BBS region (Figs. 20A and B). The con- tinental indices also showed the most marked declines in the late 1970s (Fig. 20C). Grasshopper Sparrows (Ammodramus savan- narum) showed some of the most consistent de- clines of any grassland bird. The declines pre- vailed throughout most of the Grasshopper Spar- row's range (Fig. 21), although some local in- creases were evident in the western states and elsewhere. These trends were fairly consistent from 1966 to 1996, with a slight moderation in the rate of decline in recent years. BBS data indicate that Bobolinks (Dolichonyx oryzivorus) have generally declined throughout their breeding range (Fig. 22). Areas with in- creasing populations were small and locally dis- tributed, most notably from North Dakota to western Ontario and in Pennsylvania. The con- tinental indices were fairly stable through the late 1970s, followed by a consistent decline A. Continental 10 F 66 7'0 74 78 8'2 86 90 94 Year B. Western 10  8 68 72 76 C. Central 12  o  8 8 ' 2 ifi o 80 84 88 92 96 Year 66 70 74 ' 7'8 8'2 8'6 90 94 Year FIGURE 20. BBS annual indices for Savannah Spar- row populations, 1966-1996: A = continental, B = Western BBS region, C = Central BBS region. (Fig. 23). The general population declines in 1980-1996 contrasted sharply with the popula- tion estimates for 1966-1979, when the conti- nental population significantly increased (Table 1). Eastern Meadowlarks (Sturnella magna) also exhibited consistent declines on the BBS. De- clining populations prevailed throughout most of the range except in the southwestern states (Fig. 24). The long-term trends were almost entirely negative, and most declines were significant. These declining trends prevailed during the 1966-1979 and 1980-1996 intervals (Table 1). ' ./Bs limit ercen! Change per Year FIGURE 19. Savannah Sparrow population trend map, 1966-1996. ater lhan +0.25 FIGURE 21. Grasshopper Sparrow population trend map, 1966-1996. BBS POPULATION STATUS--Peterjohn and Sauer 35 7 , BBS limit  Percent Change per Year Less than -0,25 -0.25 to +0.25 n +o.2s 06, 6 7i 0 , , , I 74 76 82 Year FIGURE 23. 1966-1996. FIGURE 22. Bobolink population trend map, 1966- 1996. Robbins et al. (1986) reported declines in Eastern Meadowlark populations associated with the severe winters of 1976-1978. These declines were most apparent in the Midwest, especially in Illinois, Indiana, Michigan, and Kentucky (Figs. 25A-D). Populations in Indiana recovered in 3 yr, but no substantial recovery was evident in the other states. Western Meadowlarks (S. neglecta) have un- dergone a range expansion in the twentieth cen- tury, spreading eastward into the Great Lakes region (Lanyon 1956, DeVos 1964). This expan- sion largely occurred before the start of the BBS, and Western Meadowlark populations generally declined between 1966 and 1996 (Table 1). These declines were evident throughout most of the Western Meadowlark's range, although in- creases occurred from southern California across the southwestern states to Texas and locally northward along the Rocky Mountains and Great Plains (Fig. 26). DISCUSSION Although BBS data indicate the trends of grassland bird populations, they do not identify the factors responsible for these trends. Some factors, such as habitat alteration, degradation, and destruction, may be common to many grass- land birds, whereas other factors may influence only certain species or may operate in only por- tions of a species' range. The factors believed to be responsible for the reported population trends are discussed below, although for many species the causes of their population trends have not been conclusively identified. SPECIES WITH INCREASING POPULATION TRENDS Ferruginous Hawk populations have experi- enced significant declines and range contractions in the past 100 yr, but these trends were most evident before 1960 (Houston and Bechard 1984, Schmutz 1984, Houston and Schmutz 66 90 94 Continental indices for Bobolink, 1999). The conversion of native grasslands to agricultural fields was largely responsible for these trends (Schmutz 1984). Contributing fac- tors included persecution, reductions in prey populations, fewer fires, and shortages of suit- able nest sites (Houston and Bechard 1984). Ferruginous Hawk population trends after 1960 have been less certain. Since most tillable lands had already been converted into agricul- tural fields, additional declines in response to habitat loss have been relatively small (Houston and Bechard 1984). The declines apparently have been reversed in portions of the species' range, where Ferruginous Hawks may have in- creased during the late 1980s (Harlow and Bloom 1989). These increases were reflected in the positive BBS trend estimates for 1980-1996 (Table 1). Short-term population increases in Ferruginous Hawks are not unexpected, as the birds are known to be fairly nomadic and local influxes have been documented in response to prey availability (Gilmer and Stewart 1983). Ad- ditional data from the BBS and other sources are needed to determine if the increases since 1980 reflect short-term fluctuations or a long-term re- versal of historic declines. BBS limit Percenx Change per Year Less than -0.25 ater than +0.25 -0.25 to +0.25 FIGURE 24. Eastern Meadowlark population trend map, 1966-1996. 36 STUDIES IN AVIAN BIOLOGY NO. 19 A. Illinois 60 5O 66 7'0 7'4 78 82 8 910 94 Year C. Michigan 25 5 0 B. Indiana o o eeee eel e ß . 66 70 74 76 82 86 90 94 Year D. Kentucky 80 ß 60 eee ß eee ß 66 70 74 78 62 86 90 94 66 70 74 78 82 86 90 94 Year Year FIGURE 25. BBS annual indices for Eastern Meadowlark populations, 1966-1996: A = Illinois, B = Indiana, C = Michigan, D = Kentucky. Upland Sandpiper populations suffered signif- icant declines in the late nineteenth and early twentieth centuries as a result of market hunting and habitat destruction (Bent 1929). Their num- bers recovered, however, when hunting ceased. More recent declines have been evident in the eastern portion of the species' range since the 1940s, and only small isolated populations re- main in most of this region (Peterjohn and Rice 1991, Carter 1992). BBS trend estimates were FIGURE 26. Western Meadowlark population trend map, 1966-1996. consistent with this pattern; there were declines in the Eastern BBS region, but relatively few Upland Sandpipers were recorded on those BBS routes, and the eastern regional trend estimates thus had little influence on the continental trend estimates. Increases in the Great Plains were largely re- sponsible for the positive continental trend es- timates for Upland Sandpipers. Factors that con- tributed to these positive trends merit additional study since the increases occurred in areas where most other grassland birds declined. Breeding Upland Sandpipers occupy much larg- er home ranges than other grassland birds (Mitchell 1967), so nesting pairs may be less affected by unfavorable agricultural practices in individual fields. These sandpipers also prefer to nest where the grass cover is of mixed short and medium heights (Kirsch and Higgins 1976, Ailes 1980). Hence, they tolerate moderate lev- els of grazing, especially in habitats where the grass cover may otherwise be too tall. This abil- ity to tolerate some disturbance may have al- lowed populations to increase in recent decades. Overgrazing, standing water, burning, and mow- ing, however, can still make breeding habitats unsuitable for this species (Mitchell 1967). BBS POPULATION STATUS--Peterjohn and Sauer 37 Factors on the Upland Sandpiper's South American winter range may also have influenced BBS population trends. Unfortunately, this spe- cies' winter ecology is poorly understood (White 1988). Although pesticide use and unfavorable agricultural practices remain threats, deforesta- tion and the subsequent creation of grasslands are probably beneficial for Upland Sandpipers. Some range extensions have been reported in deforested areas of South America (White 1988). The erratic intraseasonal movements of Sedge Wrens may not have greatly influenced the BBS trend estimates since these movements are most evident in late summer after breeding surveys have been completed (Meanley 1952, Schwilling 1982). Bedell (1996), for example, documented Sedge Wrens on some Nebraska BBS routes in August, even though the species had never been recorded during the June surveys. Although lack of site fidelity during the breeding season may contribute to considerable annual fluctuations in abundance of this species (Burns 1982), the BBS data may reasonably reflect population trends in June. How these June population trends relate to the entire Sedge Wren population, however, can- not be determined until the birds' seasonal movements are better understood. Despite fluctuations, some marked declines have been apparent in Sedge Wren populations in the twentieth century, especially in the north- eastern states and near the eastern Great Lakes (Brewer et al. 1991, Peterjohn and Rice 1991, Gibbs and Melvin 1992). These trends have con- tinued in recent decades, with declines along BBS routes most consistent in the eastern half of the species' range. Habitat loss appears to have been the most important factor contributing to these declines, although burning and over- grazing may have been important in some areas (Gibbs and Melvin 1992). In contrast, Sedge Wren population increases on the Great Plains during 1980-1996 were largely responsible for generally positive conti- nental trend estimates in 1980-1996 and 1966- 1996 (Table 1). These increases were most evi- dent in the 1990s. Since Sedge Wrens frequently occupy grasslands created by the Conservation Reserve Program (CRP; Johnson and Schwartz 1993), increased habitat availability may have contributed to these increases. Also, increased annual precipitation has improved wetland con- ditions in this portion of the Great Plains (U.S. Fish and Wildlife Service 1997), expanding the extent of damp grassland habitats favored by nesting Sedge Wrens. SPECIES WITH NONSIGNIFICANT POPULATION TRENDS For those species that are poorly sampled by the BBS, trend estimates may not represent the true status of their breeding populations. Any discussion of factors responsible for these re- ported trends would be speculative. Only those species adequately surveyed by the BBS are dis- cussed below. The BBS trends for Northern Hamer popu- lations were similar to those reported by Martin (1989), Serrentino and England (1989), and Sweet (1991). Martin (1989) indicated that sev- eral factors combine to prevent clear interpre- tation of population trends for this species. Its specialized predation on voles (Microtus spp.) produces extensive nomadic movements which, in concert with a fluctuating prey base, obscure distinguishing between actual trends and normal fluctuations. Precipitation extremes also influ- ence population levels since droughts or floods can affect habitat suitability and prey popula- tions. Severe winter weather and its effects on prey populations can also influence short-term fluctuations in Northern Harrier populations (Hamerstrom 1986). In the BBS data, these marked short-term fluctuations were apparent only in the annual indices for states and prov- inces and in physiographic strata. Where long- term declines were documented, as in portions of the northeastern and midwestern United States, habitat destruction and intensified agri- cultural use of remaining grasslands are believed to have been largely responsible (Serrentino and England 1989, Sweet 1991). Reforestation has also eliminated many suitable grasslands in the Northeast (Serrentino and England 1989). Breeding Long-billed Curlews are associated with shortgrass steppe communities on the west- ern Great Plains and with grasslands in the Great Basin. They prefer habitats that have been heavi- ly grazed where the vegetation is less than 10 cm high and the soils are moist (Knopf 1988). Populations were decimated by uncontrolled hunting in the nineteenth and early twentieth centuries, causing a noticeable contraction of the breeding range (Page and Gill 1994). The con- version of native grasslands to agricultural fields has not permitted a sustained population recov- ery (Page and Gill 1994), and the breeding range has experienced some local contractions since 1960 (McCallum et al. 1977, Renaud 1980). Since Long-billed Curlews prefer grazed habi- tats, major threats to their breeding populations are the continued conversion of grasslands into cultivated fields and the loss of wetlands, which may eliminate the moist soils preferred for feed- ing (Knopf 1988, Page and Gill 1994). These factors may be responsible for the general pop- ulation declines of Long-billed Curlews along BBS routes on the western Great Plains between 1966 and 1996. However, loss of coastal forag- ing habitats during the winter, exposure to toxic 38 STUDIES IN AVIAN BIOLOGY NO. 19 chemicals, and increased predation may also have contributed to the declines of some popu- lations (Page and Gill 1994). Factors that may have contributed to the apparent increase in pop- ulations along BBS routes in the Great Basin have not been identified. Breeding Lark Buntings are conspicuous oc- cupants of short- and mixed-grass communities of the Great Plains. Their nomadic movements are poorly understood, but fluctuations in pre- cipitation levels and the influence of these fluc- tuations on habitat conditions and food avail- ability are believed to be primarily responsible for these movements (Stewart 1975, Andrews and Righter 1992). Large influxes of Lark Bun- tings may appear for a single year or for several years in any portion of the species' range, only to disappear quickly when conditions change (Shane 1996). Regional populations may show cyclical fluctuations, and Shane (1996) has the- orized that a single population cycle may require several decades to complete. Until this species' population fluctuations are better understood, the biological significance of BBS trend estimates for Lark Burnings remains in question. Despite these fluctuations, the generally neg- ative trends for Lark Bunting populations be- tween 1966 and 1996 were associated with the destruction and degradation of their preferred grassland habitats (Andrews and Righter 1992). These buntings are also nomadic during the win- ter months (Shane and Seltman 1995), and changing habitat conditions in their winter range may have contributed to population trends. Since 1990, however, Lark Buntings have ben- efited from habitats created by the CRP on the Great Plains (Johnson and Igl 1995). These new habitats have allowed some populations to ex- pand and may have contributed to the more pos- itive population trend estimates since 1980. Baird's Sparrows have not been extensively studied, and the factors affecting their popula- tion trends are poorly understood. They frequent mixed-grass communities or, in more arid re- gions, the taller grasslands bordering wetlands, lakes, or other water sources. They prefer rela- tively undisturbed grasslands and avoid inten- sively grazed areas (Stewart 1975). Habitat de- struction and degradation during the breeding season have been associated with population de- clines in the past (Stewart 1975). Habitats cre- ated by the CRP may have benefited this species in the 1990s (Johnson and Igl 1995). Factors in- fluencing grassland habitats on the winter range in the southwestern United States and Mexico may also influence population trends of Baird's Sparrows; their winter distribution is poorly un- derstood (Phillips et al. 1964), however, and their habitat preferences and ecological require- ments are largely unknown. SPECIES WITH DECLINING POPULATION TRENDS Many of the factors important in the declines of grassland bird populations are common to most species. These factors are discussed first, followed by a summary of the species-specific factors that have contributed to declining trends. Habitat destruction The destruction of grassland habitats has been implicated in the declines of all grassland birds. The loss of native grasslands in North America has been dramatic, especially in the eastern half of the continent (Vickery 1996). Most of these grasslands disappeared before the twentieth cen- tury, but the conversion of shortgrass commu- nities into agricultural habitats has continued (Knopf 1988). The conversion of non-native pastures and hayfields into other agricultural habitats has also been prevalent during the past 50 yr (Herkert 1994). Until recently, few new grasslands had been created to compensate for habitats converted into agricultural fields or urban development or lost through forest regeneration. Reclamation of strip mines in portions of the northern Appala- chian Mountain region created extensive grass- lands in areas that were formerly forested, pro- ducing local increases in some grassland birds (Whirmore and Hall 1978). Reproductive suc- cess may be low in these habitats, however, so they may actually serve as population "sinks" for some species (Wray et al. 1982). Beginning in the 1980s, the CRP was initiated to reduce agricultural overproduction. Millions of hectares of cropland were converted to grass- lands or other perennial cover, with the greatest enrollment of area in the central United States (Young and Osborn 1990). Although the CRP lands represent a very small proportion of all agricultural lands in North America, creation of these habitats has benefited grasslands birds and even reversed the long-term declines of some regional populations (Johnson and Schwartz 1993, Johnson and Igl 1995). This reversal of some population trends in association with the creation of CRP lands demonstrates the impor- tance of habitat availability in influencing pop- ulation trends in grassland birds. Habitat fragmentation Although the effect of habitat fragmentation on woodland bird communities has been the subject of many studies, little information is available on its effects on grassland birds. Her- kert (1994) examined area relationships of grass- land birds in Illinois and noted that both area BBS POPULATION STATUS--Peterjohn and Sauer 39 and vegetative structure significantly influenced the composition of grassland bird communities. Five species were area sensitive in Illinois, and only Dickcissels were unaffected by this factor. In Maine, Vickery et al. (1994) found similar results, suggesting that large area requirements for grassland birds may be an important factor influencing habitat use and that fragmentation of remaining grassland habitats may play a signif- icant role in the population trends of some spe- cies. Habitat degradation Grazing has been implicated in the declines of some local grassland bird populations, but its impact on regional populations is difficult to as- sess. Some species, such as Grasshopper Spar- row, may benefit from light to moderate grazing in portions of their range but be adversely af- fected in other areas (Vickery 1996). Other spe- cies may be fairly sensitive to grazing across their range and may serve as indicators of over- all habitat quality (Bock and Webb 1984, Baker and Guthry 1990). Species that prefer shortgrass habitats, such as Long-billed Curlew and Horued Lark, may benefit from intensive graz- ing under some circumstances (Knopf 1988). Agricultural practices, particularly those as- sociated with hay cropping, have also been det- rimental to many grassland birds. Hayfields are being cropped more frequently and at earlier dates (Rodenhouse et al. 1993), which under many circumstances prohibits grassland birds from successfully raising young during the breeding season (Warner and Etter 1989, Bollin- ger et al. 1990). Since hayfields provide most of the remaining grassland habitats in easteru North America (Herkert 1994), this agricultural prac- tice may have serious adverse effects on the re- gional populations of grassland birds. Grazing, fire, and agricultural practices also influence the successional stages of grassland communities. Habitats that have not been dis- turbed for 5-10 yr are favored by a few species such as Upland Sandpiper, Grasshopper and Henslow's sparrows, and Eastern Meadowlark (Bollinger 1995), which are less numerous in or completely absent from younger grasslands. Hence, regular disturbance to grasslands may fa- vor generalist species such as Savannah Sparrow but may contribute to the declines of species fa- voring more mature grasslands (Bollinger 1995). Mortality from toxic chemicals Direct mortality of grassland birds from poi- soning by toxic chemicals such as Carbofuran and Fenthion has been reported in a few cases (Stone 1979, Deweese et al. 1983). The geo- graphic extent of this problem is poorly under- stood, however. Species that spend considerable time in agricultural fields, such as Ring-necked Pheasant and Horued Lark, may be most sus- ceptible to the toxic effects of these chemicals. Nest parasitism Brown-headed Cowbirds (Molothrus ater) are known to parasitize the nests of most grassland- nesting passerines (Friedmann 1963). Rates of nest parasitism are generally believed to be low, but there is much interspecific variability in par- asitism rates within sites (Hill 1976) as well as intraspecific variability across a species' range (Vickery 1996). Nest parasitism has been shown to have a negative impact on the recruitment of Dickcissels (Zimmerman 1983) and may have a similar effect on other species. Adverse winter weather Unusually severe winter weather is known to significantly reduce populations of some bird species (Robbins et al. 1986, Sauer and Droege 1990, Sauer et al. 1996). These reductions are normally short-term, with populations returuing to normal within several years following the re- turu to normal weather patterus. These weather conditions are most likely to affect species breeding in the northeru United States, although during exceptionally harsh winters, such as oc- curred during 1975-1977, harsh conditions can extend south to the Gulf Coast states. Among grassland birds, species that winter in the north- em United States, such as Ring-necked Pheasant and Easteru Meadowlark, appear to be most sus- ceptible to these conditions. Species-specific factors Habitat destruction, fragmentation, and deg- radation are common causes of declines of all grassland species, but additional factors may have contributed to the population trends evident between 1966 and 1996. Factors responsible for declines in Ring- necked Pheasant populations included intensi- fied agricultural land-use practices (resulting in reduced habitat availability), increased use of pesticides, and adverse weather conditions (Dahlgren 1988). Adverse weather normally produced short-term population fluctuations, as exemplified by declines during the mid-1970s in portions of the range, whereas the other factors were largely responsible for long-term declines. The loss of agricultural fields to reforestation and development contributed to Horned Lark population declines in easteru North America (Laughlin and Kibbe 1985, Buckelew and Hall 1994). Factors responsible for this species' de- cline elsewhere are poorly understood. Since Horued Larks are frequently associated with ag- 40 STUDIES IN AVIAN BIOLOGY NO. 19 ricultural fields throughout their range, their ex- tensive decline may indicate that some agricul- tural practices have contributed to these negative trends. For example, mortality of Horned Larks has been reported after exposure to certain pes- ticides (Beason 1995). In contrast, population in- creases in the intermountain region of the west- ern United States corresponded with the conver- sion of sagebrush (Artemisia) habitats to grass- lands (Knick and Rotenberry 1999). Several factors contributed to trends in Dick- cissel populations. This species is well adapted to residing in agricultural landscapes, inhabiting hayfields, pastures, weedy fallow fields, and weedy margins of ditches and roadsides. Con- version of these habitats to cultivated fields and more frequent mowing of hayfields, however, contributed to declines in some areas (Fretwell 1986). Brood parasitism by Brown-headed Cow- birds had a negative effect on Dickcissel recruit- ment (Zimmerman 1983), as did increased nest predation in certain habitats (Zimmerman 1984). Factors on the Dickcissel's tropical winter range are also believed to be important; the birds are viewed as pests in grain fields, and control op- erations at winter roosts have caused extensive mortality (Fretwell 1986, Basili and Temple 1999). Several factors contributed to declines in Ves- per Sparrow populations in eastern North Amer- ica. Loss of grassland habitat to reforestation and urbanization was a major factor, although "clean farming" practices such as the removal of hedgerows and more frequent mowing of hayfields also contributed to declines (Laughlin and Kibbe 1985, Brauning 1992). The Vesper Sparrow is one of the first species to occupy reclaimed strip mines, however, and it has ex- panded its range in heavily forested portions of West Virginia and surrounding states since 1960 (Whitmore and Hall 1978). The factors responsible for the trends in Sa- vannah Sparrow populations are poorly under- stood (Wheelwright and Rising 1993). In the eastern United States, reforestation, conversion of grasslands to cultivated crops, and more fre- quent mowing of hayfields contributed to de- clines (Laughlin and Kibbe 1985, Peterjohn and Rice 1991). Factors associated with population trends in western North America, where areas of increase and decline were interspersed, were not identified. Habitat destruction, fragmentation, and deg- radation have been the primary factors respon- sible for declines in Grasshopper Sparrow pop- ulations since 1966 (Vickery 1996). Because this species prefers relatively large grassland tracts (Herkert 1994, Vickery et al. 1994, Bol- linger 1995), it may be particularly susceptible to changes in habitat availability. Additionally, early mowing of hayfields can result in aban- donment of breeding territories and can contrib- ute to local annual fluctuations in abundance (Smith 1963). In addition to habitat destruction, the factor most frequently cited for declines in Bobolink populations is more frequent mowing of hay- fields (Bollinger et al. 1990). Many hayfields are cut in late May and at regular intervals through- out the summer, which does not provide Bobo- links with an opportunity to successfully raise a brood between mowing operations. Also, habitat preferences and other aspects of the species' winter biology are poorly understood, so factors on its South American winter range may also have contributed to population declines. Eastern Meadowlarks tend to winter farther north than most other grassland birds, which may explain their greater susceptibility to peri- odic severe winter weather, as shown in the BBS annual indices. Their population declines, how- ever, have generally been attributed to habitat destruction, more frequent mowing of hayfields, and similar factors affecting the populations of most grassland birds (Peterjohn and Rice 1991, Brauning 1992). Breeding meadowlarks are very sensitive to disturbance around the nest, either by people or livestock. Certain agricultural prac- tices, such as spring tillage (which can reduce nest success and increase adult mortality), are also detrimental to breeding populations (Lan- yon 1995). Factors responsible for declines in Western Meadowlark populations are believed to be sim- ilar to those described for Eastern Meadowlarks (Lanyon 1994). Extensive droughts in the 1930s may have contributed to this species' eastward range expansion into the Great Lakes area, but causes for this expansion and subsequent decline have never been fully explained (Lanyon 1956). CONCLUSIONS AND CONSERVATION IMPLICATIONS BBS data indicate negative population trends for most grassland bird species between 1966 and 1996. The decfines were fairly consistent throughout the survey period and, for many spe- cies, prevailed across most of the breeding range. A few exceptions existed, with Ferrugi- nous Hawks, Upland Sandpipers, and Sedge Wrens exhibiting significant increases. The gen- eral declines in grassland birds shown by the BBS, however, reflected similar trends reported in the decades prior to the 1960s. Although BBS data can be used to identify temporal and geographic patterns in population trends, the data do not identify the causes of these patterns. Other sources of information BBS POPULATION STATUS---Peterjohn and Sauer 41 must be used to establish the factors responsible for these trends. The common factors of habitat destruction, fragmentation, and degradation in- fluence population trends of most grassland bird species. Agricultural practices such as earlier and more frequent mowing of hayfields may also contribute to population declines in some species, whereas other factors may be important for individual species or in specific geographic areas. These factors affect each species differ- ently and produced the variety of geographic and temporal patterns in population trends evi- dent in the BBS estimates. Despite the prevalence of negative trend es- timates, the situation is not entirely bleak for grassland birds. The CRP has shown that dete- rioration of grassland habitats can be reversed over the short term, even on fairly large geo- graphic scales. Efforts to mitigate some of the other adverse factors discussed above can only help grassland birds. Grassland birds have evolved in relatively harsh and constantly changing habitats, requiring considerable adapt- ability in order to survive in this environment. With some assistance from humans, this adapt- ability may allow many of these species to re- cover if habitat availability and conditions im- prove. The reversal of population declines resulting from the habitats created by the CRP is just the first step toward an overall improvement in the status of grassland birds. The conservation of these species must receive greater priority, par- ticularly in the Great Plains where grassland bird communities are richest. Additional research is needed to better understand how these species respond to the factors that affect their population trends. This understanding may be crucial for the development of successful efforts to produce a long-term reversal of the general decline in grassland bird populations. ACKNOWLEDGEMENTS We thank the thousands of observers who have sur- veyed BBS routes. Our analyses would have been im- possible without the data they collected. We are grate- ful to K. Boone and K. Pardieck for preparing the fig- ures. Earlier drafts of this manuscript benefited from comments by D. Dawson, K. Pardieck, C. Robbins, S. Knick, ED. Vickery, and an anonymous reviewen LITERATURE CITED AlLES, I. W. 1980. Breeding biology and habitat use of the Upland Sandpiper in central Wisconsin. Passen- ger Pigeon 42:53-63. AMERICAN ORNITHOLOGISTS' UNION. 1983. Check-list of North American birds. 6th ed. American Orni- thologists' Union, Washington, D.C. ANDREWS, R., AND R. RIGHTER. 1992. Colorado birds. Denver Museum of Natural History, Denver, CO. ANDRLE, R. E, AND J. R. CARROLL (EDITORS). 1988. The atlas of breeding birds in New York state. Cornell University Press, Ithaca, NY. ASKINS, R. A. 1993. Population trends in grassland, shrubland, and forest birds in eastern North Ameri- ca. Current Ornithology 11:1-34. BAKER, D. L., AND E $. GUTHRY. 1990. Effects of con- tinuous grazing on habitat and density of ground- foraging birds in south Texas. Journal of Range Management 43:2-5. BASILI, G. D., AND $. A. TEMPLE. 1999. Winter ecol- ogy, behavior, and conservation needs of Dickcissels in Venezuela. Studies in Avian Biology 19:289-299. BEASON, R. C. 1995. Horned Lark (Eremophila alpes- tris). In A. Poole and E Gill (editors). The birds of North America, no. 195. Academy of Natural Sci- ences, Philadelphia, PA, and American Ornitholo- gists' Union, Washington, D.C. BEDELL, P. A. 1996. Evidence of dual breeding ranges for the Sedge Wren in the central Great Plains. Wil- son Bulletin 108:115-122. BENT, A. C. 1929. Life histories of North American shore birds. U.S. National Museum Bulletin 146:55- 69. Boche, C. E., AND J. H. BoChe. 1988. Grassland birds in southeastern Arizona: impacts of fire, grazing and alien vegetation. Pp. 43-58 in ED. Goriup (editor). Ecology and conservation of grassland birds. ICBP Technical Publication no. 7. International Council for Bird Preservation, Cambridge, U.K. BOCK, C. E., AND B. WEBB. 1984. Birds as grazing indicator species in southeastern Arizona. Journal of Wildlife Management 48:1045-1049. BOLLINGER, E. K. 1995. Successional changes and hab- itat selection in hayfield bird communities. Auk 112: 720-730. BOLLINGER, E. K., P. B. BOLLINGER, AND t. A. GAVIN. 1990. Effects of hay-cropping on eastern popula- tions of the Bobolink. Wildlife Society Bulletin 18: 142-150. BRAUNING, D. W. (EDITOR). 1992. Atlas of breeding birds in Pennsylvania. University of Pittsburgh Press, Pittsburgh, PA. BREWER, R., G. A. MCPEEK, AND R. J. ADAMS, JR. (ED- ITORS). 1991. The atlas of breeding birds of Michi- gan. Michigan State University Press, East Lansing, MI. BUCKELEW, A. R., JR., AND G. A. HALL. 1994. The West Virginia breeding bird atlas. University of Pittsburgh Press, Pittsburgh, PA. BURNS, J. T. 1982. Nests, territories and reproduction of Sedge Wrens (Cistothorus platensis). Wilson Bul- letin 94:338-349. BUTCHER, G. S. 1990. Audubon Christmas Bird Counts. Pp. 5-13 in J. R. Sauer and S. Droege (ed- itors). Survey designs and statistical methods for the estimation of avian population trends. U.S. Fish and Wildlife Service Biological Report 90(1). BYSTRAK, D. 1981. The North American Breeding Bird Survey. Studies in Avian Biology 6:34-41. CARTER, J. W. 1992. Upland Sandpiper. Pp. 235-252 in K. J. Schneider and D. M. Pence (editors). Mi- gratory nongame birds of management concern in the northeast. U.S. Fish and Wildlife Service, New- ton Corner, MA. 42 STUDIES IN AVIAN BIOLOGY NO. 19 CRESSIE, N. 1992. Statistics for spatial data. John Wiley and Sons, New York, NY. DAHLGREN, R. B. 1988. Distribution and abundance of the Ring-necked Pheasant in North America. Pp. 29-43 in D. L. Hallerr, W. R. Edwards, and G. V. Burger (editors). Pheasants: symptoms of wildlife problems on agricultural lands. Wildlife Society, Bloomington, IN. DEVOS, A. 1964. Range changes of birds in the Great Lakes region. American Midland Naturalist 71:489- 502. DEWELSE, L. R., L. C. McEwEN, L.g. SETFIMI, AND R. D. DEBLINGER. 1983. Effects on birds of Fenthion aerial application for mosquito control. Journal of Economic Entomology 76:906-911. EMLEN, J. T., AND J. A. WIENS. 1965. The Dickcissel invasion of 1964 in southern Wisconsin. Passenger Pigeon 27:51-59. ENVIRONMENTAL SYSTEMS RESEARCH INSTITUTE. 1991. Surface modeling with TIN. Environmental Systems Research Institute, Redlands, CA. EWERT, D. N., AND P. D. CANTINO. 1967. Summer ob- servations of Dickcissels and Western Meadowlarks in the Lansing area. Jack-Pine Warbler 45:128-130. FITZNER, J. $. 1978. The ecology and behavior of the Long-billed Curlew (Numenius americanus) in southeastern Washington. Ph.D. dissertation. Wash- ington State University, Pullman, WA. FRETWELL, S. D. 1986. Distribution and abundance of the Dickcissel. Current Ornithology 4:211-242. FRIEDMANN, H. 1963. Host relations of the parasitic cowbirds. U.S. National Museum Bulletin 233. GElSSEER, P. H., AND J. R. SANER. 1990. Topics in route-regression analysis. Pp. 54-57 in J. R. SaNer and S. Droege (editors). Survey designs and statis- tical methods for the estimation of avian population trends. U.S. Fish and Wildlife Service Biological Report 90(1). GIBBS, J.P., AND S. M. MELVIN. 1992. Sedge Wren. Pp. 191-209 in K. J. Schneider and D. M. Pence (edi- tors). Migratory nongame birds of management con- cern in the northeast. U.S. Fish and Wildlife Service, Newton Corner, MA. GILMER, D. S., AND R. E. STEWART. 1983. Ferruginous Hawk populations and habitat use in North Dakota. Journal of Wildlife Management 47:146-157. HAMERSTROM, E 1986. Harrier, hawk of the marsh: the hawk that is ruled by a mouse. Smithsonian Insti- tution Press, Washington, D.C. HARLOW, D. L., AND P. H. BLOOM. 1989. Buteos and the Golden Eagle. Pp. 102-110 in B. G. Pendleton (editor). Proceedings of the western raptor manage- ment symposium and workshop. National Wildlife Federation Scientific Technical Series no. 12. HERKERT, J. R. 1994. The effects of habitat fragmen- tation on midwestern bird communities. Ecological Applications 4:461-471. HILL, R. A. 1976. Host parasite relationships of the Brown-headed Cowbird in a prairie habitat of west central Kansas. Wilson Bulletin 88:555-565. HOUSTON, C. S., AND M. J. BEGHARD. 1984. Decline of the Ferruginous Hawk in Saskatchewan. American Birds 36:166-170. HOUSTON, C. S., AND J. K. SCHMUTZ. 1999. Changes in bird populations on Canadian grasslands. Studies in Avian Biology 19:87-94. ISAAKS, E. H., AND R. M. SRIVASTAVA. 1989. An intro- duction to applied geostatistics. Oxford University Press, New York, NY. JAMES, E C., C. E. MCCULLOCH, AND L. E. WOLFE. 1990. Methodological issues in the estimation of trends in bird populations with an example: the Pine Warbler. Pp. 84-97 in J. R. Sauer and S. Droege (editors). Survey designs and statistical methods for the estimation of arian population trends. U.S. Fish and Wildlife Service Biological Report 90(1). JOHNSON, D. H., AND L. D. IGL. 1995. Contributions of the Conservation Reserve Program to populations of breeding birds in North Dakota. Wilson Bulletin 107:709-718. JOHNSON, D. H., AND M. D. SCHWARTZ. 1993. The Con- servation Reserve Program and grassland birds. Conservation Biology 7:934-937. KIRSCH, L. M., AND K. E HIGGINS. 1976. Upland Sand- piper nesting and management in North Dakota. Wildlife Society Bulletin 4:16-20. KNICK, S. T., AND J. T. ROTENBERRY. 1999. Spatial dis- tribution of breeding passerine bird habitats in a shrubsteppe region of southwestern Idaho. Studies in Arian Biology 19:104-111. KNOPF, E L. 1988. Conservation of steppe birds in North America. Pp. 27-41 in P. D. Goriup (editor). Ecology and conservation of grassland birds. ICBP Technical Publication no. 7. International Council for Bird Preservation, Cambridge, U.K. KNOPF, E L. 1994. Arian assemblages on altered grass- lands. Studies in Arian Biology 15:247-257. LANYON, W. E. 1956. Ecological aspects of sympatric distribution of meadowlarks in the north-central states. Ecology 37:98-108. LANYON, W. E. 1994. Western Meadowlark (Sturnella neglecta). In A. Poole and E Gill (editors). The birds of North America, no. 104. Academy of Natural Sci- ences, Philadelphia, PA, and American Ornitholo- gists' Union, Washington, D.C. LANYON, W. E. 1995. Eastern Meadowlark (Sturnella magna). In A. Poole and E Gill (editors). The birds of North America, no. 160. Academy of Natural Sci- ences, Philadelphia, PA, and American Ornitholo- gists' Union, Washington, D.C. LAUGHLIN, S. B., AND D. P. KmBE (EIITORS). 1985. The atlas of breeding birds of Vermont. University Press of New England, Hanover, NH. LINK, W. A., AND J. R. SAUER. 1994. Estimating equa- tions estimates of trend. Bird Populations 2:23-32. LINK, W. A., AND J. R. SAUER. 1995. Estimation of empirical mixing distributions in summary analyses. Biometrics 51:810-821. MARTIN, J. W. 1989. Harriers and kites. Pp. 92-101 in B. G. Pendleton (editor). Proceedings of the western raptor management symposium and workshop. Na- tional Wildlife Federation Scientific Technical Series no. 12, Washington, D.C. MARTIN, S. G., AND T. A. GAVIN. 1995. Bobolink (Do- lichonyx oo;zivorous). In A. Poole and E Gill (edi- tors). The birds of North America, no. 176. Acade- my of Natural Sciences, Philadelphia, Pa., and American Ornithologists' Union, Washington, D.C. MCCALLUM, D. A., W. D. GRAUL, AND R. ZACCAGNINI. BBS POPULATION STATUS--Peterjohn and Sauer 43 1977. The status of the Long-billed Curlew in Col- orado. Auk 94:599-601. McNICHOLL, M. K. 1988. Ecological and human influ- ences on Canadian populations of grassland birds. Pp. 1-25 in ED. Goriup (editor). Ecology and con- servation of grassland birds. ICBP Technical Publi- cation no. 7. International Council for Bird Preser- vation, Cambridge, U.K. MEANLEY, B. 1952. Notes on the ecology of the Short- billed Marsh Wren in the lower Arkansas rice fields. Wilson Bulletin 64:22-25. MENGEL, R. M. 1970. The North American central plains as an isolating agent in bird speciation. Pp. 280-340 in W. Dort and J. K. Jones (editors). Pleis- tocene and recent environments of the central Great Plains. University of Kansas Press, Lawrence, KS. MITCHELL, G. J. 1967. The Upland Plover and its status in relation to environmental conditions and situa- tions, past and present. Blue Jay 25:58-63. MONROE, B. L., JR., A. L. STAMM, AND B. L. PALMER- BALL, JR. 1988. Annotated checklist of the birds of Kentucky. Kentucky Ornithological Society, Louis- ville, KY. PAGE, G. W., AND R. E. GILL, JR. 1994. Shorebirds in western North America: late 1800s to late 1900s. Studies in Avian Biology 15:147-160. PETERJOHN, B.G. 1994. The North American Breeding Bird Survey. Birding 26:386-398. PETERJOHN, B.G., AND D. L. RICE. 1991. The Ohio breeding bird atlas. Ohio Department of Natural Re- sources, Columbus, OH. PETERJOHN, B.G., AND J. R. SAUER. 1993. North Amer- ican Breeding Bird Survey annual summary 1990- 1991. Bird Populations 1:52-67. PETERJOHN, B.G., AND J. R. SAUER. 1996. The 1994 and 1995 summary of the North American Breeding Bird Survey. Bird Populations 3:48-66. PHILLIPS, A., J. MARSHALL, AND G. MONSON. 1964. The birds of Arizona. University of Arizona Press, Tuc- son, AZ. REDMOND, R. L., T. K. BICEK, AND D. A. JENNI. 1981. An evaluation of breeding season census techniques for Long-billed Curlews (Numenius americanus). Studies in Avian Biology 6:197-201. RENAUD, W. E. 1980. The Long-billed Curlew in Sas- katchewan: status and distribution. Blue Jay 38:221- 237. ROBBINS, C. S., D. BYSTRAK, AND P. H. GEISSLER. 1986. The Breeding Bird Survey: its first fifteen years 1965-1979. U.S. Fish and Wildlife Service Research Publication no. 157. RODENHOUSE, N. L., L. B. BEST, R. J. O'CONNOR, AND E. r. BOLLINGER. 1993. Effects of temperate agri- culture on neotropic migrant landbirds. Pp. 280-295 in D. M. Finch and P. W. Stangel (editors). Status and management of neotropical migratory birds. USDA Forest Service Gen. Tech. Rep. RM-229. USDA Forest Service Rocky Mountain Forest Sta- tion, Fort Collins, CO. SAUER, J. R., AND S. DROEGE. 1990. Recent population trends of the Eastern Bluebird. Wilson Bulletin 102: 239-252. SAUER, J. R., AND P. H. GEISSLER. 1990. Annual indices from route regression analyses. Pp. 58 62 in J. R. Sauer and S. Droege (editors). Survey designs and statistical methods for the estimation of avian pop- ulation trends. U.S. Fish and Wildlife Service Bio- logical Report 90(1). SAUER, J. R., J. E. HINES, G. GOUGH, I. THOMAS, AND B.G. PETERJOHN. 1997. The North American Breed- ing Bird Survey results and analysis. Ver. 96.3: www.mbr-pwrc.usgs.gov/bbs/bbs.html. U.S. Geo- logical Survey, Patuxent Wildlife Research Center, Laurel, MD. SAUER, J.e., G. W. PENDLETON, AND S. ORSILLO. 1995. Mapping of bird distributions from point count sur- veys. Pp. 151-160 in C. J. Ralph, J. R. Sauer, and S. Droege (editors). Monitoring bird populations by point counts. USDA Forest Service Gen. Tech. Rep. PSW-GTR-149. USDA Forest Service Pacific Southwest Research Station, Albany, CA. SAUER, J. R., G. W. PENDLETON, AND B. O. PETERJOHN. 1996. Evaluating causes of population change in North American insectivorous songbirds. Conser- vation Biology 10:465-478. SAUER, J. R., B.G. PETERJOHN, AND W. m. LINK. 1994. Observer differences in the North American Breed- ing Bird Survey. Auk 111:50-62. SCHMUTZ, J. K. 1984. Ferruginous and Swainson's hawk abundance and distribution in relation to land use in southeastern Alberta. Journal of Wildlife Management 48:1180-1187. SCHWILLING, M.D. 1982. Sedge Wrens nesting into September. Kansas Ornithological Society Bulletin 33:22-23. SERRENTINO, P., AND M. ENGLAND. 1989. Northern Har- rier Pp. 37-46 in B.G. Pendleton (editor). Proceed- ings of the northeast raptor management symposium and workshop. National Wildlife Federation Scien- tific Technical Series no. 13, Washington, D.C. SHANE, T. G. 1996. The Lark Bunting: in peril or mak- ing progress? Colorado Field Ornithologists' Journal 30:162-168. SHANE, r. G., AND S.S. SELTMAN. 1995. The historical development of wintering Lark Bunting populations north of the thirty-seventh parallel in Colorado and Kansas. Kansas Ornithological Society Bulletin 46: 36-39. SMITH, R. L. 1963. Some ecological notes on the Grasshopper Sparrow. Wilson Bulletin 75:159-165. STEWART, R. E. 1975. Breeding birds of North Dakota. Harrison Smith, Lund Press, Minneapolis, MN. STONE, W. B. 1979. Poisoning of wild birds by organ- ophosphate and carbamate pesticides. New York Fish and Game Journal 26:37-47. SWEET, M. J. 1991. Kites and Northern Harrier. Pp. 3 41 in B.G. Pendleton and D. L. Krahe (editors). Proceedings of the midwest raptor management symposium and workshop. National Wildlife Fed- eration Scientific Technical Series no. 15, Washing- ton, D.C. UDVARDY, M.D. F. 1958. Ecological and distributional analysis of North American birds. Condor 60:50- 66. U.S. FISH AND WILDLIFE SERVICE. 1997. Waterfowl: status and fall flight forecast, 1997. U.S. Fish and Wildlife Service, Office of Migratory Bird Manage- ment, Laurel, MD. VICKERV, P. D. 1996. Grasshopper Sparrow (Ammo- dramus savannarum). In A. Poole and F. Gill (edi- 44 STUDIES IN AVIAN BIOLOGY NO. 19 tors). The birds of North America, no. 239. Acade- my of Natural Sciences, Philadelphia, PA, and American Ornithologists' Union, Washington, D.C. VICKERY, P. D., M. L. HUNTER, JR., AND S. M. MELVIN. 1994. Effects of habitat area on the distribution of grassland birds in Maine. Conservation Biology 8: 1087-1097. WARNER, R. E., AND S. L. ETTER. 1989. Hay cutting and the survival of pheasants: a long-term perspec- tive. Journal of Wildlife Management 53:455-461. WHEELWRIGHT, N.m., AND J. D. RISING. 1993. Savan- nah Sparrow (Passerculus sandwichensis). In A. Poole and E Gill (editors). The birds of North Amer- ica, no. 45. Academy of Natural Sciences, Philadel- phia, PA, and American Ornithologists' Union, Washington, D.C. WHITE, R. P. 1988. Wintering grounds and migration patterns of the Upland Sandpiper. American Birds 42: 1247-1253. WmTMORE, R. C., ANt) G. A. HALL. 1978. The response of passefine species to a new resource: reclaimed strip mines in West Virginia. American Birds 32:6-9. WRAY, T, II, K. A. STRAIT, ANt) R. C. WHITMORE. 1982. Reproductive success of grassland sparrows on a reclaimed surface mine in West Virginia. Auk 99:157-164. YOUNG, E. C., AND C. T. OSBORN. 1990. Costs and benefits of the Conservation Reserve Program. Jour- nal of Soil and Water Conservation 45:370-373. ZIMMERMAN, J. L. 1983. Cowbird parasitism of Dick- cissels in different habitats and at different nest den- sities. Wilson Bulletin 95:7-22. ZIMMERMAN, J. L. 1984. Nest predation and its rela- tionship to habitat and nest density in Dickcissels. Condor 86:68-72. Studies in Avian Biology No. 19:45-59, 1999. LINKING CONTINENTAL CLIMATE, LAND USE, AND LAND PATTERNS WITH GRASSLAND BIRD DISTRIBUTION ACROSS THE CONTERMINOUS UNITED STATES RAYMOND J. O'CONNOR, MALCOLM T. JONES, RANDALL B. BOONE, AND T. BRUCE LAUBER Abstract. Associations of the abundance and temporal incidence of 17 grassland bird species with climate, weather, farm crops, and landscape metrics were determined for the conterminous United States using hierarchical models. We developed statistical models using two versions of classification and regression tree analysis in which the variation of each species' response variable (both as number of individuals [1973 1989] and as temporal incidence [1981-1990] per Breeding Bird Survey route) was recursively partitioned into statistically distinct chains of environmental determinants or associ- ations. The predictive power of these models was bimodal, yielding high R 2 values (above 38 percent) for one group of 12 species and low values (below 20 percent) for a second group of 5 (generaally scarce or restricted-range) species. The fit of the models was strongly correlated with the size of each species' range. Climate variables--long-term annual precipitation, January temperature, and July tem- perature appeared in many of the species models, often with strong effects (large R 2 values). January weather (annual deviation from long-term mean temperature) was also a consistent, though weaker, correlate. Sorghum (Sorghum vulgate) was the only strong crop correlate of most species abundances, but grain corn (Zea mays) and enrollment in the Conservation Reserve Program were consistent smaller contributors to most models. Wheat (Triticum aestivum) and durum wheat (T. durum) were other noteworthy variables, occurring in about half of the species models. The presence of soybeans (Glycine max) was a local modifier of abundance for almost all species. Considering only the leading variables for individual species, precipitation occurred in five species, grain corn in three, and durum wheat and sunflower (Helianthus sp.) in two each. The Conservation Reserve Program variable pre- empted grain corn for two species in the two years Conservation Reserve Program data were available. Other leading variables each appeared in only one species. A parallel analysis using remotely sensed land-use data to assess the relative roles of land-cover proportions and habitat patch attributes showed that grassland species were more strongly influenced by habitat patch variables, but less strongly influenced by land-cover proportions, than were nongrassland species. Grassland species' sensitivity to habitat patch variables appeared to be greater in wooded and cropland habitats than in habitats dominated by grass. EL ENLACE ENTRE EL CLIMA CONTINENTAL, EL USO DE TERRENO Y LOS PATRONES DE TERRENO CON LA DISTRIBUCION DE AVES DE PASTIZAL A TRVES DE LOS ESTADOS UNIDOS CONTIRMINOS Sinopsis. Se determinaron asociaciones de abundancia e incidencia temporal de 17 especies de aves de pastizal con el clima, el tiempo, las cosechas y las mediciones de paisaje para los Estados Unidos contrminos utilizando modelos jerirquicos. Elaboramos modelos estadfsticos aplicando dos versiones diferentes de amilisis de rboles de regresi6n y clasificaci6n. En ellos, la variaci6n de la variable respuesta de cada especie (tanto el nfimero de individuos [1973-1989] como la incidencia temporal [1981-1990] por ruta del Breeding Bird Survey) se divide recursivamente en cadenas de determinantes ambientales o asociaciones que difieren estadfsticamente. La capacidad de predicci6n de estos modelos fue bimodal, lo que produjo altos valores R 2 (mis de un 38 por ciento) para un grupo de 12 especies y bajos valores (menos de un 20 por ciento) para un segundo grupo de 5 especies (generalmente escasas o con una extensi6n restringida). La conformidad de los modelos se correlacion6 estrechamente con el tamafio de la extensi6n de cada especie. Las variables de clima--precipitaci6n anual a largo plazo, temperatura en enero y temperatura en julioaparecieron en muchos de los modelos de especie, a menudo con grandes efectos (altos valores de R2). E1 tiempo en enero (la desviaci6n anual de la temperatura promedio a largo plazo) fue tambifin un correlativo congruente, aunque de menor impor- tancia. E1 sorgo (Sorghum vulgate) fue el tinico correlativo de cosecha marcado para la abundancia de la mayorfa de las especies, pero el maiz (Zea mays) y la inscripci6n en el Programa de Reservas de Conservaci6n fueron factores menores siempre presentes que contribuyeron en la mayorfa de los modelos. E1 trigo (Triticum aestivum) y el Triticum durum fueron otras variables que cabe mencionar, que aparecieron en aproximadamente la mitad de los modelos de especie. La presencia de soya (Gly- cine max) fue un modificador local de abundancia para casi todas las especies. Tomando en cuenta solamente las variables principales para las especies individuales, hubo precipitaci6n en cinco especies, mafz en tres, y Triticum durum y girasol (Helianthus sp.) en dos cada uno. La variable del Programa de Reservas de Conservaci6n reemplaz6 la variable de mafz para dos especies durante los dos aftos en que habfa datos disponibles del Programa de Reservas de Conservaci6n. Otras variables principales aparecieron en s61o una especie cada una. Un amilisis paralelo utilizando datos de usos del territorio obtenidos por detecci6n remota para evaluar los papeles relativos de las proporciones de cobertura de 45 46 STUDIES IN AVIAN BIOLOGY NO. 19 terreno y las caracterfsticas de rodales de hfibitat demostr6 que las especies de pastizal fueron influidas en mayor grado pot las variables de rodales de hfibitat, pero que fueron infiuidas en menor grado pot las proporciones de cobertura de terreno, que las especies que no eran de pastizal. La sensibilidad de especies de pastizal alas variables de rodales de htbitat pareci6 set mts grande en hfibitats de firboles y de cosechas queen hSbitats dominados pot la hierba. Key Words: agriculture; area sensitivity; climate: grassland birds; landscape ecology; regression trees. Grassland birds have generally declined in the United States because of intensification of agri- culture in the Midwest (Askins 1993) and refor- estation and increased urbanization in the East (Witham and Hunter 1992, Litvaitis 1993). These declines have been particularly severe where the prairie has been fragmented and dis- turbed by farming, as in Illinois (Warner 1994). Farmland intensification has been aided by the development of new mechanical and chemical means of treating cropland and by economic support systems promoting their use (O'Connor and Shrubb 1986). In the United States these trends have been reflected in intensified corn (Zea mays) and soybean (Glycine max) produc- tion and in reductions in small-grain and forage crops, livestock, and pasture. Additionally, most hayfields are now intensively cultivated alfalfa (Medicago sativa) monocultures rather than mixed-species grasslands. The shift from peren- nial grassland to annually cultivated cropland is thought to be a major factor in the decline of several formerly common grassland bird species (Johnson and Schwartz 1993). Much of the information available on the hab- itat requirements of grassland birds originates in site-intensive studies and focuses on microhab- itat features. More spatially extensive studies, such as those by Johnson and Schwartz (1993), have used a regional set of sites and mesoscale habitat variables to characterize the correlates of favorable and unfavorable sites, and With (1994) has taken an explicitly landscape approach in studying the requirements of McCown's Long- spur (Calcarius mccownii). Another approach is that of Whitmore (1981), who compared his re- sults with those of Wiens (1973) to demonstrate that the habitat requirements of Grasshopper Sparrows (Ammodramus savannarum) are simi- lar in different parts of the country. Much less is known about the correlates of these species' distributions over large spatial ex- tents; the assumption is that the effects of mi- crohabitat or mesoscale correlates determine the larger distribution (Brown 1984). Distributions, and particularly continental distributions, are more likely to be controlled by hierarchies of controlling or constraining factors (Krebs 1985). Before effective conservation programs for grassland species can be developed, we need to identify controlling factors at spatial scales other than that of the microscale of the local habitat patch (Wiens 1981). In this paper we take a mac- roecological approach (Brown 1995) to assess the pattern of environmental correlates for 17 species of grassland birds in the conterminous United States (Table 1). We used a class of sta- tistical models known as classification and re- gression tree (CART) analysis that can handle hierarchical effects (see Rodenhouse et al. 1993). METHODS BIRD DATA The bird data we analyzed were from the Breeding Bird Survey (BBS) for the conterminous United States. The BBS is based on 40-km roadside surveys, each containing 50 stops at 0.8-km intervals. Approximately 2,000 BBS routes are distributed randomly in the con- terminous United States in 1-degree blocks of latitude and longitude by state. The number of routes per 1- degree block of latitude and longitude varies among states but is held constant in a state (Bystrak 1981). We used only "type one" routes (routes passing all quality-assurance checks) for the period 1973 through 1990. We used the total count (i.e., number of individ- uals) for each species and the incidence (i.e., propor- tion of years observed) for each species on each route. For crop analyses, we assigned each route to the coun- ty in which its starting coordinates lay, and in a spatial tessellation of a remotely sensed land-use analysis we assigned each route to the corresponding hexagon (see below). A variety of spatial autocorrelation analyses TABLE l. TOTAL VARIANCE ACCOUNTED FOR BY THE DECISION-TREE MODELS FOR INDIVIDUAL SPECIES Species % variance Western Meadowlark Dickcissel Horned Lark Eastern Meadowlark Ring-necked Pheasant Bobolink Savannah Sparrow Vesper Sparrow Grasshopper Sparrow Lark Bunting Upland Sandpiper Chestnut-collared Longspur Baird's Sparrow Gray Partridge Long-billed Curlew Henslow's Sparrow McCown's Longspur 76.1 71.8 64.8 64.6 62.5 62.3 59.6 59.0 52.1 51.6 41.3 38.9 15.3 11.7 11.1 3.7 3.3 CONTINENTAL BIRD DISTRIBUTIONS--O'Connor et al. 47 indicated that these assignments retained adequate spa- tial resolution for the purposes of our study. Data for 17 species were extracted for analysis on the basis of adequate data in our time period (Table 1). Range estimates were derived from maps of breed- ing densities prepared by the Patuxent Wildlife Re- search Center from BBS data (Sauer et al. 1997). Im- ages of each species' range were converted to raster coverages using a Geographic Information System. The proportion of North America sampled by the BBS that was occupied by the species was used as the range estimate for each species. Although this approach may underestimate the total range of some species, by ex- cluding the southern- and northernmost extents, it is spatially consistent with the abundance data for each species, and in our analyses an underestimate of range for a widespread species would be a conservative error. AGRICULTURE DATA Agriculture data for each county came from the pe- riodic Censuses of Agriculture (U.S. Department of Commerce, Bureau of the Census) and the annual Na- tional Agricultural Statistics Service (NASS) estimates (U.S. Department of Agriculture). We used the Cen- suses of Agriculture for 1974, 1978, 1982, and 1987 that contained summary statistics for thousands of ag- riculture variables for each county in the United States. Censuses of Agriculture include acreages of common crops (e.g., corn, cotton [Gossypium sp.], and hay) as well as of uncommon crops (e.g., mint [Mentha spp.] for oil, hops [Humulus lupulus], and kale [Brassica oleracea]). The NASS compiles annual estimates of agriculture for each county in the conterminous United States. Counties are grouped into crop-reporting dis- tricts by state and according to climate, cropping prac- tices, and other variables (U.S. Department of Agri- culture 1987). The NASS agricultural statistics include annual estimates of common crops, with total acreage planted, seeded, and harvested. Estimates of crops sown and harvested from 1972 to 1989 were included in our database. Thus, the NASS agricultural statistics provide data for years without direct Census of Agri- culture information. Data on the county acreage of land enrolled in the Conservation Reserve Program (CRP), a federal program initiated in 1986 which re- tires cropland from production, were obtained from the Natural Resources Conservation Service and were in- cluded as an additional cropping variable. WEATHER AND CLIMATE DATA The primary weather and climate data used in the crop analyses were the Climatic Division Data from World WeatherDisc, a commercial product from WeatherDisc Associates, Inc. (Seattle, Washington). The WeatherDisc data we used covered the period 1961 through 1988; data from mid-1988 through 1990 came from the National Climatic Data Center. We used mean January and July temperatures and mean annual precipitation as parsimonious representatives of bird- relevant weather. We computed 30-yr averages for 1961-1990 to index long-term weather (i.e., climatic conditions) and computed the deviations of the annual values from these means as measures of short-term weather conditions. Thus, we had six climate or weath- er variables for each spatial unit in our analyses. REMOTELY SENSED DATA For a subsidiary analysis, we used data from O'Connor et al.'s (1996) regression tree analysis of bird distribution in relation to remotely sensed data. O'Connor et al. (1996) used data from the Loveland et al. (1991) land-cover prototype, supplemented with an urban layer from the Digital Chart of the World (Danko 1992). This chart classifies each 1-km 2 pixel in the United States in 1990 in one of 159 (160 with the urban class) land-cover classes, doing so on the basis of the seasonal Advanced Very High Resolution Radiometry (AVHRR) profile for that point. O'Connor et al. (1996) adopted the U.S. Environmental Protec- tion Agency's (EPA) Environmental Monitoring and Assessment Program hexagonal grid (Overton et al. 1990, White et al. 1992) as a spatial framework for this analysis. Each hexagon is approximately 635 km 2, with a point-to-point (center-to-center) spacing of ap- proximately 27 km. All environmental correlates were determined as values typifying each hexagon, using only the 1,198 hexagons with BBS data satisfying our data quality criteria. Although this hexagon-based sampling averages the environmental data over a fixed area, the point-to-point spacing of 27 km across hexa- gons is acceptable given the length of each BBS route (40 km). O'Connor et al.'s (1996) approach captured spatial variation in landscape structure that might reflect hab- itat fragmentation and other land-use "stressors," do- ing so by calculating various metrics of spatial pattern developed under the rubric of "landscape ecology" (Turner and Gardner 1990). In this context, stressors were regarded as any measures, or metrics, reflecting negative impacts on species richness. Various land- scape metrics were calculated from the landscape pat- tern delineated with AVHRR imagery. The distribution of pixels in each hexagon was analyzed by treating contiguous pixels as "patches" for which metrics such as dominance, contagion, fractal dimensions, connec- tivity, and patch and edge characteristics could be cal- culated (O'Neill et al. 1988). Three metrics were de- termined for each land-cover class in each hexagon: the average size of patches of that class, the size of the largest patch of that class, and the largest value of the patch perimeter calculated for all patches of that class. Where a land-cover class was absent from the hexagon, the corresponding metric was set to zero. In addition, the average patch size in each hexagon, ir- respective of land-cover class, was computed. Four cli- matic variables were available from the analyses: long- term averages of January mean temperature, July mean temperature, and annual precipitation and an index of seasonality, which was computed as the within-pixel change between the January and July temperature val- ues (for further details see O'Connor et al. 1996). ANALYTICAL APPROACH We programmed a Fortran version of the decision- tree algorithm of Sonquist et al. (1973; Knowledge Seeker, version 2.0) to assess the association of our independent variable, the count of species on a BBS route, with a set of independent variables spanning cli- mate and cropping information. Counts of zero were fairly frequent, and consequently bird counts were first normalized by use of the random normal scores trans- 48 STUDIES IN AVIAN BIOLOGY NO. 19 formation (Bradley 1968). The decision-tree algorithm sorted the bird counts in the region on each indepen- dent variable in turn and determined the best threshold along this gradient that maximized the difference be- tween the dependent variable values in the two subsets. For example, in evaluating wheat (Triticum aestivum) as a splitting variable, the BBS routes were ordered from those in the area with the lowest wheat acreage to those in the area with the highest wheat acreage. The data set was split at the median wheat value into a low-wheat group and a high-wheat group, and the normalized bird counts in the two groups were tested for significant difference (P < 0.01) by means of a t- test. The remaining explanatory variables were then analyzed and similarly tested. If more than one vari- able resulted in a significant difference between spe- cies counts in high and low groups, the variable ex- plaining the greatest percentage of the variance in the set of routes was chosen and the routes were split into two subsamples at the threshold for that variable. The splitting process was repeated for each of the two groups, leading to the identification of four subsam- ples. The process was again repeated until no division of a group across any of the available variables re- sulted in a significant difference in average bird counts between subsamples. The final output was represented as a decision tree with a series of end-nodes whose values for species abundance were set by the chain of environmental conditions along the path back to the root node. This method identified the extent and pat- tern of correlation between dependent and independent variables, and in particular allowed for the occurrence of constraints and of contingent effects (Breiman et al. 1984). We summarized the output of the algorithm by com- puting the proportion of variance accounted for by a given model and by dividing this variance among the individual explanatory variables present in the final model (Clark and Pregibon 1992). To incorporate sam- pling variance in our estimates, we used a bootstrap- ping approach (Efron 1982) to select repeated random samples of the data set for analysis and reported the median percentage of variance in the data set explained by each variable over all bootstrap replicates. Prelim- inary analysis suggested that 60-plus bootstrap repli- cates were needed to stabilize the variance of these medians. Our final analyses were based on 100 boot- strap replicates. We analyzed data for each odd-numbered year from 1973 through 1989. Differences in results between years arose for two reasons: because some variables were mutually correlated and varied from year to year, or because the true association of a species with a var- iable changed substantially from year to year because of changes in cropping practices, weather, or other var- iables. Because of the computational complexity of the method, not every explanatory variable was consid- ered in the tree construction for every species. An ab- breviated screening analysis, based on 10 bootstrap samples of the data set, was performed first for each species for 1973, 1979, 1985, and 1989. This analysis was used to determine which variables were likely to be statistically significant in the final analysis. A full analysis, based on 100 bootstrap samples, was then performed on data from every other year using vari- ables that had been identified in the initial screening analysis. The final results considered 30 variables that could potentially explain the BBS counts in each year examined (Table 2). Twenty-two variables measured land use (percent of county land planted in a crop and CRP acreage); three measured climate (30-yr averages of annual precipitation and January and July temper- atures); three measured weather (deviation from 30-yr averages of annual precipitation and January and July temperatures); and two were geographic variables (lat- itude and longitude). Measurements of most explana- tory variables were available from 1973 through 1989, but occasionally a variable had to be omitted for a year in which its value was unreported. We also examined the environmental correlates, de- rived from the remotely sensed data, of temporal in- cidence for each grassland species using the regression tree modules of the S-plus statistical package (MathSoft Inc., Seattle, Washington). We used cross- validation techniques to optimize the fit of each re- gression tree (Clark and Pregibon 1992), an approach preferable to the bootstrap sampling we used in the crop analyses (Breiman et al. 1984). For these analyses incorporating landscape metrics, we report the percent mean deviance explained as a measure of the good- ness-of-fit equivalent to an R 2 value (S. Urquhart, pers. comm.). RESULTS GOODNESS-OF-FIT OF MODELS The percentage of variance explained by each species model ranged from 76.1% for Western Meadowlark (Sturnella neglecta) to 3.3% for McCown's Longspur (Table 1). The models fell into two groups: 12 species whose models ac- counted for 38% or more of the variance in abundance and 5 species whose models account- ed for less than 20% of the variance (Table 1). We were interested in determining whether the range in the variance explained by each of these 17 models might be a scale phenomenon (Table 1). Given the spatially extensive nature of variables such as climate and common crop acreages, a wide-ranging species might be ex- pected to adapt to one or more of these vari- ables, whereas a restricted-range species might simply incorporate the variation in these same variables across its range as a constant (Allen and Starr 1982). If this were the case, one would expect model fit to be correlated with range size across species. We tested this hypothesis by computing the Spearman rank correlation of model fit (as percent variance explained) with the proportion of the North American BBS area occupied by the species and found a strong cor- relation to support this explanation (Spearman rho = 0.733, P < 0.002). PREDICTOR VARIABLES The variables that appeared in most species models were mean annual precipitation (15 spe- CONTINENTAL BIRD DISTRIBUTIONS--O'Connor et al. 49 TABLE 2. NUMBER OF SPECIES SHOWING CORRELATION WITH INDIVIDUAL VARIABLES AND SUMMARY STATISTICS FOR CORRESPONDING SPECIES VARIANCES FOR EACH VARIABLE Species effect (% variance explained) Number Median Variable (+/- effect) of species Mean SD Minimum Maximum Years rank Mean annual precip. (-) 15 7.2 11.52 0.05 36.4 9 3.0 CRP (+) 10 6.1 6.05 0.80 19.6 2 4.5 January climate (-) 13 4.2 10.04 0.05 37.4 9 10.0 Sorghum (+) 12 4.0 8.23 0.05 29.9 9 7.0 Latitude (+) 13 3.9 7.06 0.05 26.0 9 7.0 July climate (m) 12 3.8 4.30 0.05 16.2 9 6.0 Longitude (+) 13 3.8 3.32 0.05 11.1 9 5.0 Durum wheat (+) 7 3.5 8.77 0.05 23.4 6 3.0 Grain corn (+) 11 2.9 2.90 0.05 8.4 9 5.0 Wheat (+) 8 2.9 3.56 0.05 9.4 4 4.5 Oats (+) 10 2.5 3.20 0.70 10.5 9 15.0 January weather (m) 11 1.7 1.32 0.05 5.4 9 5.0 All hay (+) 7 1.5 1.27 0.05 3.8 4 6.0 Soybeans (+) 13 1.5 1.61 0.05 5.1 9 10.0 Winter wheat (+) 12 1.4 1.29 0.05 4.0 9 9.0 Spring wheat (+) 8 1.4 3.06 0.05 8.9 5.5 6.0 Alfalfa (+) 7 1.3 0.56 0.70 2.3 4 11.0 Sunflowers (+) 5 1.2 1.29 0.05 3.2 5 14.0 Barley (m) 10 1.2 0.67 0.05 2.5 9 13.0 Other hay (m) 6 1.2 0.31 0.80 1.7 4 15.5 Deviation precip. (+) 10 1.0 0.42 0.05 1.7 9 10.5 July weather (+) 9 1.0 0.55 0.05 2.0 9 12.0 Tobacco (m) 1 0.9 0.90 0.90 0.9 9 23.0 Corn silage (+) 8 0.8 0.53 0.05 1.4 9 11.5 Cotton (-) 3 0.7 0.56 0.20 1.3 9 22.0 Beans (+) 1 0.6 0.60 0.60 0.6 8 17.0 Peanuts (-) 2 0.6 0.07 0.60 0.7 8.5 20.0 Potatoes (+) 4 0.5 0.32 0.05 0.8 4 17.5 Sugar beets (+) 3 0.3 0.49 0.05 0.9 2 9.0 Flaxseed (+) 5 0.2 0.42 0.05 1.0 2 10.0 Note: "Years" is the median of the number of years for which the variable was a correlate of the individual species (maximum 9 odd-numbered years. 1973 1989). Median rank was computed by ranking all variables in each species model as 1, 2, etc., by size and taking the median for each variable across species. Signs in parentheses indicate the dominant direction of the effect of the variable; "m" indicates mixed effects. "Climate" refers to 30-yr mean temperatures; "weather" refers to deviations about these means. cies, accounting on average for 7.2% of the var- iance), mean January temperature (13 species, average effect 4.2%), latitude and longitude (13 species each, with average effects of 3.9 and 3.8%, respectively), soybeans (13 species, av- erage effect 1.5%), and sorghum (Sorghum vul- gare; 12 species, average effect 4.0%; Table 2). Note that these are highly summarized estima- tors. The effects of each variable considered were estimated in each of the nine annual mod- els (alternate years from 1973 through 1989) computed for each species; the median of these annual effects for the variable was tabulated as a summary statistic for the species; and the spe- cies-specific medians were averaged across those species with non-zero medians as a sum- mary statistic of the influence of that variable. In calculating these averages, species with no correlation with the variable were omitted rather than treated as zeros. We omitted these species because the magnitude of effect is of most in- terest for species correlated with that variable, whereas the proportion of species associated with the variable could be summarized separate- ly. For individual species, both the median ef- fects and the effects in individual years were of- ten much higher (see below). It is important to remember that these effects are statistical cor- relates and may be directly responsible for the response or may have an indirect effect, the lat- ter occurring in the case of variables that may be highly correlated with an unmeasured vari- able (in sensu "surrogacy" of Breiman et al. 1984). The number of species correlated with a var- iable and the average size of the correlation ef- fect were themselves broadly correlated, but there were a few exceptions (Table 2). Soybeans (13 species), winter wheat (Triticum sp.; 12 spe- cies), and perhaps barley (Hordeum vulgare; 10 species) all appeared in more of the species models than was typical for their mean effects 50 STUDIES IN AVIAN BIOLOGY NO. 19 (Table 2). This was also true for the three weath- er variableseviations from long-term mean January and July temperatures and from long- term annual precipitation--but this was likely due to correlated responses by all the species to regional weather in individual years. Converse- ly, some variables had atypically few species correlates (Table 2), most notably the level of CRP enrollment in the county, a variable with strong effects on certain species (Lauber 1991, Johnson and Schwartz 1993). CRP data were available in our analyses only for 1987 and 1989, and the low representation of the CRP en- rollment variable surely reflects that fact. Durum wheat (T. durum; 7 species, average effect 3.5%), wheat (8 species, average effect 2.9%), and all hay (7 species, average effect 1.5%), however, were under-represented variables that lacked such obvious analysis bias by way of ex- planation. Some variables, such as tobacco (Ni- cotiana tabacum), beans (Leguminosae), and peanuts (Arachis hypogea), were only weakly correlated with just one or two species (Table 2). As a measure of the consistency of these as- sociations between species and crop or environ- mental variables, we tallied the number of years in which the correlation was significantly non- zero for each species (maximum of 9 odd-num- bered years, 1973-1989). In most cases, strongly correlated variables had the most consistent re- sults, occurring in all years for all species (ac- knowledging that the CRP variable could appear in at most 2 yr for each species). The exceptions were durum wheat, wheat, and all hay, all of which appeared only in four of the year-specific models and in fewer species models than might have been expected. Conversely, some variables (e.g., cotton [Gos- sypium] and corn silage) had weak effects but appeared consistently in the annual species mod- els (Table 2). The remaining cases with lower numbers of years with effects were all variables with weak overall effects and with eight or few- er species correlates. Most crops were positively associated with the abundance of the species with which they were correlated (Table 2); only peanuts and cotton, both minor influences, were consistently negative. January climate and mean annual precipitation had negative correlations with most species, but July climate had different effects with different species. Weather effects were likewise variable; warm summers and wet years favored most species, but January weather was more varied in its effects. As previously noted, the distribution of the variance explained by all models was bimodal. It was possible that the importance of some var- iables in well-fitting models was diluted by weak associations of those same variables with species with poor-fitting models. We therefore addressed the question of whether certain variables might not be consistently the most important variable (largest variance explained) across many spe- cies, irrespective of the size of the variance ex- plained by the variable. We ranked the variables in each species model based on the size of the contribution to explained variance to obtain the median ranks across all species (Table 2). This ranking revealed a consistent pattern: certain variables (e.g., mean annual precipitation and extent of durum wheat cultivation) were gener- ally the strongest predictors in individual grass- land bird species models; CRP, cultivation of wheat or grain corn, annual January tempera- ture, and longitude were usually the five stron- gest predictor variables for individual species. Spring wheat (Triticum sp.), hay, and sorghum production, July climate, and latitude were also fairly high ranking variables. Other variables, notably beans, potatoes (Solanum tuberosum), peanuts, cotton, and tobacco, were typically low-ranking predictors for most species (Table 2). Most crop correlates were again positive and most weather and climate correlates negative (Table 2). Examination of the major correlates (i.e., var- iables accounting for > 5% of the median var- iance explained) for each species showed clear patterns when summarized across species. Ag- riculture and climate variables had substantial effects in the breeding distribution of all 17 spe- cies, whereas geographic and weather variables had substantial effects in only 11 and 6 species, respectively (Table 3). Similar patterns were ob- served after grouping all predictor variables into three categories (agricultural, climatic, and geo- graphic [latitude/longitude]) and ranking them by total variance explained for each species (Ta- ble 4). Agriculture variables were dominant for eight species (Gray Partridge [Perdix perdix], Ring-necked Pheasant [Phasianus colchicus], Upland Sandpiper [Bartramia longicauda], Grasshopper Sparrow, Baird's Sparrow [Ammo- dramus bairdii], Henslow's Sparrow [A. henslo- wii], Chestnut-collared Longspur [ Calcarius or- natus], and Dickcissel [Spiza americana]); cli- mate variables for six species (Long-billed Cur- lew [Numenius americanus], Horned Lark [ Eremophila alpestris], Lark Bunting [ Calamos- piza melanocorys], McCown's Longspur, Bobo- link [Dolichonyx oryzivorous], and Western Meadowlark); and geographic variables for only one species (Savannah Sparrow [Ammodramus savannarum]). Two species (Vesper Sparrow [Pooecetes gramineus] and Eastern Meadowlark [Sturnella magna]) had models in which agri- culture and climatic variables accounted for sim- CONTINENTAL BIRD DISTRIBUTIONS--O'Connor et al. 51 TABLE 3. VARIABLES ACCOUNTING FOR AT LEAST 5'/ OF THE MEDIAN VARIANCE EXPLAINED FOR EACH SPECIES Species Predictor variables Gray Partridge Ring-necked Pheasant Upland Sandpiper Long-billed Curlew Horned Lark Vesper Sparrow Lark Bunting Savannah Sparrow Grasshopper Sparrow Baird's Sparrow Henslow's Sparrow McCown's Longspur Chestnut-collared Longspur Dickcissel Bobolink Eastern Meadowlark Western Meadowlark sunflower (+), CRP (+), durum wheat (+), January climate (-), all hay CRP (+), grain corn (+), wheat (+), latitude (+), mean annual precipita- tion sunflower (+), durum wheat (+), mean annual precipitation, sorghum, flax- seed, January weather, CRP, longitude mean annual precipitation (-), January climate (-), longitude (+), winter wheat (+), soybeans, spring wheat mean annual precipitation (-), CRP (+), wheat, soybeans, grain corn, win- ter wheat spring wheat (+), January climate (-), latitude (+), mean annual precipita- tion, oats mean annual precipitation (-), longitude (m), July climate, January climate, spring wheat, January weather latitude (+), oats (+), January climate, sorghum, July climate CRP (+), grain corn (+), wheat, all hay, January weather, cotton, July weather, sorghum, mean annual precipitation durum wheat (+), latitude (+), mean annual precipitation (-), soybeans, wheat grain corn (+), July climate (-), potatoes (+) mean annual precipitation (-), corn silage (-) durum wheat (+), longitude (+), mean annual precipitation (-), spring wheat, flaxseed, soybeans sorghum (+), longitude (+), July climate, January weather, soybeans, all hay January climate (-), longitude (-), oats July climate (+), January weather (m), grain corn, longitude, latitude, all hay, sorghum, mean annual precipitation mean annual precipitation (-), CRP (+), January weather (m) Note: Variables accounting for at least 10% of the median variance show direction of effect in parentheses; effects are positive (+), negative (-), or mixed (m). Data on the Conservation Reserve Program (CRP) were aaflable only for 1987 and 1989. "Climate" refers to 30 yr mean temperatures; "weather" refers to deviations about these means. TABLE 4. RELATIVE IMPORTANCE OF VARIABLE GROUPINGS (AGRICULTURE, CLIMATE, AND GEOGRAPHIC) FOR EACH SPECIES Rank of category based on mean % variance explained Species 1 N 2 N 3 N Dickcissel Agriculture (42.2%) 10 Climate (18.8%) 6 LatLong (10.8%) 2 Ring-necked Pheasant Agriculture (40.3%) 15 Climate (13.4%) 5 LatLong (8.9%) 2 Vesper Sparrow Agriculture (26.3%) 12 Climate (22.8%) 6 LatLong (9.9%) 2 Grasshopper Sparrow Agriculture (33.1%) 16 Climate (14.7%) 6 LatLong (4.0%) 2 Upland Sandpiper Agriculture (27.0%) 12 Climate (11.4%) 6 LatLong (2.9%) 2 Chestnut-collared Longspur Agriculture (28.4%) 7 LatLong (6.9%) I Climate (3.7%) 1 Baird's Sparrow Agriculture (11.9%) 3 LatLong (1.9%) 1 Climate (1.5%) 1 Gray Partridge Agriculture (9.8%) 9 Climate (1.8%) 1 LatLong (0.14%) 1 Henslow's Sparrow Agriculture (2.2%) 2 Climate (1.5%) 1 -- - Western Meadowlark Climate (50.3%) 6 Agriculture (17.6%) 9 LatLong (8.2%) 2 Horned Lark Climate (35.9%) 6 Agriculture (24.9%) 12 LatLong (4.1%) 2 Eastern Meadowlark Climate (29.9%) 7 Agriculture (27.1%) 17 LatLong (7.6%) 2 Bobolink Climate (42.4%) 5 Agriculture (11.2%) 8 LatLong (8.6%) 2 Lark Bunting Climate (29.1%) 5 LatLong (13.5%) I Agriculture (8.8%) 7 Long-billed Curlew Climate (6.0%) 2 Agriculture (2.9%) 5 LatLong (2.1%) 2 McCown's Longspur Climate (2.9%) I Agriculture (0.4%) 1 -- - Savannah Sparrow LatLong (26.7%) 2 Agriculture (19.3%) 9 Climate (13.7%) 6 Note: Total variance explained for each category, in parentheses, was calculated by summing across each category's variables. The number of variables that contributed to each category is given as N. 52 STUDIES IN AVIAN BIOLOGY TABLE 5. RELATIVE IMPORTANCE OF LAND-COVER AND PATCH VARIABLES IN DIFFERENT HABITATS NO. 19 Anderson land covers Frequency of correlations (negative) Land-cover Patch Total % patch Class Type variables variables variables variables 4 grass-dominated 7(1) 6(1) 13 46 1 cropland and pasture 4 8(1) 12 67 3 woodland/cropland 3 7(2) 10 70 9 mixed (decid./conif.) forest 3 4(2) 7 57 6 mixed grass/shrub rangeland 2(1) 2 4 50 7 deciduous forest 3 1 4 25 5 shrub-dominated rangeland 1 0 1 0 8 coniferous forest 0 1 1 100 14 urban 1 0 1 0 2 grassland/cropland 1 0 1 0 Totals 25(2) 29(6) 54 54 ilar amounts of the explained variation (Table 4). Of these three categories, geographic variables typically (11 of 17 species) accounted for the least amount of explained variation (Table 4). Among climate variables, it is interesting to note the lower median ranking of January weather (annual deviation from long-term mean temperature) than of January climate (mean tem- perature) values (median rank of 5 versus 10; Table 2). This was the result of a markedly bi- modal distribution of ranks for January climate. For six species (Gray Partridge, Long-billed Curlew, Vesper Sparrow, Lark Bunting, Savan- nah Sparrow, and Bobolink), January climate was among the top four most significant vari- ables (Table 3). For another group of seven spe- cies (Ring-necked Pheasant, Upland Sandpiper, Horned Lark, Grasshopper Sparrow, Dickcissel, and Eastern and Western meadowlarks), January climate appeared only at rank 10 or higher, in- dicating that the variable had only a peripheral or local effect in the decision trees involved. For all 13 of these species except Gray Partridge and Long-billed Curlew, however, January weather also appeared in the species models, as a sub- sidiary modifier to climate effects for four of the species above (Vesper Sparrow, Lark Bunting, Savannah Sparrow, and Bobolink) and as a var- iable dominant to climate for the last seven spe- cies. For species wintering in Central and South America, these correlations must be due to in- direct effects. When considering other weather and climate variables, annual precipitation had the strongest effects. For Horned Lark, Lark Bunting, and Western Meadowlark, long-term annual precip- itation was far more important (R 2 = 29.3, 15.0, and 32.0%, respectively) than January climate. July climate had a large effect only for Eastern Meadowlark (R 2 = 14.3%), though Horned Lark, Lark Bunting, and Western Meadowlark all had high-ranking contributions (range 4- 6%). Annual variation in July temperature had its strongest link with Grasshopper Sparrow (but only at R 2 = 3.2%), and variation in annual pre- cipitation was weakly linked (approximately 2%) with Upland Sandpiper, Dickcissel, and Eastern and Western meadowlarks. An analysis of satellite-derived land-use and land-pattern variables yielded regression tree models of the incidence of each species over a set of 1,198 BBS routes in the conterminous United States (Table 5). Incidence is the propor- tion of surveys on each BBS route in which a species was recorded between 1981 and 1990. For most species, incidence and abundance were well correlated (Wright 1991). The independent variables considered were climatic data, propor- tions of each land-cover class (of 160 classes) around each route, and various pattern metrics of patch size and edge characteristics of the landscape around each route. The data set is de- scribed in more detail in O'Connor et al. 1996. The frequency of occurrence of land-use and land-pattern variables that occurred in the re- gression tree models is summarized across all 17 species (Table 5). To avoid excessive detail, the summary collapses the 160 land-cover classes used in the analyses to the 14 classes of an An- derson et al. (1976) Level II classification. Thus, the incidence of seven species was correlated with the extent of one of the grass-dominated habitats that comprise the Anderson land-cover Class 4 (Table 5). Of these correlates, only one was negative. Similarly, six species models con- tained a statistical dependence on one of the patch attributes of habitat in land-cover Class 4, with five species more abundant and one less abundant in areas with larger patches of this cover type (Table 5). Some evidence suggests that habitat patch features may be significant in certain land clas- CONTINENTAL BIRD DISTRIBUTIONS--O'Connor et al. 53 Patch Compositional 60- 50-  40-  30- d z 20- 10- 0 2 1 o o Non-grassland 3o 1 10- Non-grassland o Grassland Grassland 10- 8- 6- 4- 2- 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 LEVEL LEVEL 7 8 9 FIGURE 1. Frequency distributions of the level (e.g., level I is the root node) in decision-tree models at which patch variables and proportion of land-class (i.e., compositional) variables act. Grassland bird species show a significant skew to the left for patch variables and a significant skew to the right for compositional variables. ses (Table 5). Overall, patch variables occurred more frequently in the models (53.7%) than did land-cover variables (Table 5). However, these patch correlates were more frequently influential in the Anderson Level II land classes not dom- inated by grass or rangeland, and particularly so in those classes with five or more correlations in total (Classes 1, 3, and 9). As expected, Class 4 (grass dominated) had the most correlations for grassland species; in this and the two rangeland classes, patch variables provided about half the predictors of species incidence (8 of 18 corre- lates). Patch variables were in the minority (1 of 4, or 25%) in Class 7 (deciduous forest). Among the four classes with most correlations, 46% of the correlations were patch related in Class 4, but 66% (19 of 29) were patch related in Classes 1, 3, and 9. The sample sizes were too small to obtain significant results even with the marked imbalance between land classes. We analyzed the significance of patch vari- ables for grassland species further, by comparing the relative influence of patch and land-propor- tion (compositional) variables in models for grassland and for all other species in the BBS (R. J. O'Connor, unpubl. data). We considered only those species with models involving either of these variable types. We plotted the frequency with which patch variables had their effect at the root level (level 1) of their trees, at the next level down (level 2), and so on (Fig. 1). Variables acting nearer the root of a regression tree have a more widespread, and usually stronger, influ- ence than do variables acting deeper in the tree. A comparison of patch variables for nongrass- land and grassland species showed a significant skew to the left for grassland species (Wilcoxon test, P = 0.012; Fig. 1). Similarly, a comparison of the location of action of associations involv- ing the proportion of land class present in the hexagon showed the reverse: grassland species had compositional variables acting farther away from the tree root than did other species (Wil- coxon test, P = 0.028; Fig. 1). These results con- 54 STUDIES IN AVIAN BIOLOGY NO. 19 firm the idea that patch variables were more crit- ical, and compositional variables less critical, in the distribution of grassland species than of oth- er species. DISCUSSION Grassland bird distributions at the spatially extensive scale of our analyses were markedly influenced first by crop distribution, second by climate, and third by habitat patch size and shape (Tables 2 and 4). CROP CORRELATES Areas of extensive cultivation of sorghum, wheat, and grain corn were generally favorable to grassland bird species. Correlations with these crops were among the largest contributors of variance explained in many species models and, for at least some species, explained much of the variance. Several other crops, either more re- gional in distribution or only locally grown, were likewise positively associated with the abundance of grassland species but contributed less to the overall variance explained and had low ranking in the individual regression trees. This last feature can be summarized as reflecting local modification of predictions, with these var- iables having only local effects in regions where abundance was set by constraints imposed by other crop or climate variables. Our results suggest that farmers' choices about cropping practices have implications for grassland species at several levels. First, it is clear that high enrollment in the CRP has major benefits for grassland birds. Despite our having only 2 yr of CRP data, many species were more abundant in CRP areas than in non-CRP areas, and several species, notably Gray Partridge and Ring-necked Pheasant, showed strong positive associations with CRP enrollment, which were consistent with previous analysis of BBS data with respect to the CRP (Lauber 1991). Lauber found that many species showed spatial associ- ations of density with CRP enrollment but that many of these correlations were apparent even prior to the advent of the CRP. However, he was able to show that for several species, including Ring-necked Pheasant and Western Meadow- lark, densities increased differentially in these areas with the advent of the CRP. A possible explanation for high densities in what later proved to be areas of high CRP enrollment may well be that these have long been areas of high soil erosion; enrollment by farmers in earlier "set-aside" programs to reduce erosion could have favored grassland species by the cessation of tillage operations inimical to the birds' suc- cess (Lauber 1991). Lauber's analyses of tem- poral trends are supported by the results of de- tailed studies of use of CRP fields in particular regions. Johnson and Schwartz (1993) found that several prairie species with restricted ranges (particularly Lark Bunting, Grasshopper and Baird's sparrows, Dickcissel, and Bobolink) were less abundant on annually tilled cropland than on CRP lands and that many of these spe- cies had previously been declining in the central United States. The only grassland species they found to be less abundant on CRP than on non- CRP lands were Vesper Sparrow and Chestnut- collared Longspur, both of which prefer sparse vegetation. A second conclusion to be drawn from our findings in relation to cropping is that the spa- tially extensive cultivation of certain crops, no- tably sorghum, grain corn, and wheat, may cre- ate agricultural environments conducive to breeding for grassland birds. We do not claim that these crops are necessarily favorable through cause and effect; instead, they may share with grassland birds environmental re- quirements we did not directly measure (e.g., to- pography, soil type). Alternatively, the cultiva- tion practices and associated land-management patterns may create conditions favorable to these birds. Long-billed Curlews apparently do well in wheat fields in Oklahoma that are subjected only to spraying (Shackford 1994), and Horned Larks have long been known to thrive in cultivated fields (Graber and Graber 1963). The benefits of particular crops need not accrue solely to breed- ing birds; certain crops modify the effects of snow cover in winter, permitting foraging to continue in fields where other vegetation would trap the snow in an impenetrable mass (Larsen et al. 1994). For some grassland birds, small grains may approximate a natural grassland, as shown by Warner (1994) for Ring-necked Pheasants; only Horned Larks, however, appear capable of persisting in a monoculture of cereal crops (Owens and Myres 1973). It is also pos- sible that cultivation maintains ephemeral con- ditions that some species prefer. Grasshopper Sparrows favor open grasslands providing open- ings and gaps through which the birds can move while foraging (Whirmore 1981), and open- planted crops may provide an adequate substi- tute. Chestnut-collared Longspurs, however, though needing open vegetation in which to for- age, will not nest in cultivated fields (Owens and Myres 1973). The third point to be drawn from our crop analyses is that some crops--among them soy- beans, oats (Avena), alfalfa, sunflowers (Helian- thus spp.), and barley--appeared only as local modifiers of species distributions already largely constrained by other factors. Some of these may be chance correlations; sunflowers, for example, CONTINENTAL BIRD DISTRIBUTIONS--O'Connor et al. 55 are grown mainly in the Dakotas, and their cor- relations with species with ranges centered in this region may be due to confounding effects. Barley and soybeans, however, consistently dis- played the same correlations from year to year (Table 2), so they may have a more ecological basis. Some of these small correlations probably reflect use of the crop as an adequate substitute for native habitat; thus the correlation of durum wheat with Baird's Sparrow (Table 3) may result from the species using this crop as a source of vegetative cover in what is otherwise an agri- cultural waste (Owens and Myres 1973). INFLUENCE OF CLIMATE The pattern of influence of climate variables in our analysis is of considerable interest given Root's (1988a) demonstration of the power of climatic limits to constrain winter bird distribu- tion. We found that in most grassland bird spe- cies there were significant associations between climate and weather variables and local breeding abundance. Breeding-season analyses of climate influences on bird populations are more likely to be mediated indirectly (e.g., by climate influence on productivity; Rotenberry et al. 1995). Curfie (1991) hypothesized that the latitudinal gradient in breeding-species richness was set largely by the corresponding productivity gradient. In con- trast, wintering-distribution limits appear to be very close to those set by a model of physiolog- ical limits to maximum daily metabolic rates for resident species (Root 1988a). Among the seven species for which a climate variable was the largest correlate of breeding distribution, only that for the Bobolink involved winter climate (Table 3). Because Bobolinks winter in South America, however, winter temperatures in the United States cannot be directly related to their abundance; whatever effects winter temperatures have on Bobolinks must therefore be indirect. Price (1995) reached the same conclusion from his model of the climate envelope of the Bobo- link. Our other major climate effects largely differ from those reported by Price (1995). Our results implied that Eastern Meadowlarks were more abundant in hot summer areas (Table 3), but Price (1995) found a complex pattern of re- sponses to temperature for this species, with quadratic functions of mixed sign describing de- pendencies on temperatures in summer and in the wettest month of the year. Where we found negative effects of annual precipitation for Horned Lark, Lark Bunting, and Western Mead- owlark, Price (1995) reported generally negative effects only for Lark Bunting. For Horned Lark he found positive effects of precipitation in spring and in the coldest month, and for Western Meadowlark his largest terms were positive con- tributions from precipitation in winter and in the hottest and driest month but with negative terms for other seasonal components of precipitation. Thus, the overall effect of precipitation on West- ern Meadowlarks depends on the distribution of seasonality of that precipitation. It is also worth noting here how different methods yield differ- ent answers to what is apparently the same gen- eral question of climate correlates. Price (1995) derived a single climate envelope for the entire range of each species, producing rules applica- ble over the entire range but expressed in strong- ly seasonal aspects of climate. Our study mod- eled regional abundance in terms of less season- al climate variables using CART. Walker (1990) has previously shown how the climate envelope and CART approaches yield complementary perspectives on the environmental correlates of species distribution and has pioneered their in- tegration into a common model. Root's (1988a, b) work focused on the cor- relation of wintering limits with midwinter cli- mate conditions. She has elsewhere identified the lack of knowledge about breeding limits as a critical gap in our understanding of environ- mental constraints on birds (Root 1993). We found that, for several species, breeding densi- ties were also correlated with midwinter cli- mates, with fewer breeding pairs where winters are cold (Table 2); however, winter climate was a major predictor only for the Bobolink, a neo- tropical migrant (Table 3). Whereas climatic constraint by January temperatures was clearly widespread across species (Table 2), year-to- year variation in January temperatures also ap- peared regularly and at some strength in species models. The bimodal distribution of ranks for January climate, in combination with consistent effects for January weather, revealed the exis- tence of one group of species with breeding dis- tributions sensitive to midwinter climate condi- tions (Gray Partridge, Long-billed Curlew, Ves- per Sparrow, Lark Bunting, Savannah Sparrow, and Bobolink) and a second group (Ring-necked Pheasant, Upland Sandpiper, Horned Lark, Grasshopper Sparrow, Dickcissel, and Eastern and Western meadowlarks) more sensitive to year-to-year variation in conditions. Dickcissels have long been known to be sensitive to annual variations in winter conditions, largely because poor winters may constrain the rate of progres- sion of the spring migration northward (Fretwell 1986). Our findings suggest that other grassland bird species may share a similar sensitivity. For resident species such as Gray Partridge and Ring-necked Pheasant, however, winter condi- tions can have direct effects; heavy precipitation 56 STUDIES IN AVIAN BIOLOGY NO. 19 as snow covers the ground, and food resources for overwintering become limiting (Riley 1995). Our restfits suggest that summer temperature limits are not as important for grassland birds as studies suggest they are for other groups of birds (Blake et al. 1992, O'Connor 1992). Similarly, Price (1995) found only 7 of 23 species to be correlated either with summer temperatures or with temperatures in the hottest month, generally with negative effects evident. The link with drought identified by Blake et al. (1992) is con- gruent with the importance of precipitation as- sociations found here. July temperatures typi- cally may be associated with drought, such that our analyses attribute variations in density to variations in precipitation rather than to temper- ature. Drought effects appear to be rather short- lived in grassland birds, with most species re- covering within 1 yr (George et al. 1992). It is tempting to suggest that the absence of summer temperature links is because grassland birds re- distribute themselves each year in line with the prevailing distribution of weather-controlled re- sources, as argued previously of Dickcissels by Fretwell (1986). If this were the case, one might expect a pattern of strong correlations with the weather variables considered here instead of the lack of correlations we found (Table 2). There- fore, we consider whether breeding distributions might not instead be constrained in a climate envelope by the distribution of habitat. INFLUENCE OF PATCH VARIABLES Our analyses lend considerable support to ear- lier research that suggested that grassland birds are particularly sensitive to habitat fragmenta- tion. Area sensitivity is well established. Samson (1980) and Johnson and Temple (1986) conclud- ed that small fragments of grasslands cannot support species that need interior habitats, and Vickery et al. (1994) and Herkert (1994) have shown through site-specific studies that grass- land birds are more likely to occur on large patches of grassland habitat than on small ones. Vickery et al. (1994) demonstrated area sensitiv- ity for Upland Sandpipers, Vesper, Savannah, and Grasshopper sparrows, Bobolinks, and East- ern Meadowlarks. Our results expand these stud- ies in two significant ways. First, we found that grassland bird species as a class are more influ- enced by habitat patch variables, and less influ- enced by land-use proportions in an area, than are other bird species (Fig. 1). Patch variables are present significantly higher in the regression tree models for grassland species than for other species, thus ensuring that they affect propor- tionately more of the survey area and that they constrain distributions more than does landscape composition. Thus, grassland species are differ- entially susceptible to habitat fragmentation. Second, we raise the possibility that patch variables are more influential in habitats less dominated by grass (Table 5). A differential to- ward stronger association of incidence and patch variables in nongrassland habitats implies that patch configuration or size issues have become more important in the cropland and wooded ar- eas that have replaced native prairies. To our knowledge, no one has previously suggested that species might be more acutely selective in less favored habitats than in preferred habitats, though it is a logical outcome of more general phenomena such as the habitat hierarchies of Brown (1969) and Fretwell and Lucas (1969). One might expect that birds using secondary habitats would be more selective as to which parts of these habitats they use if they are unable to settle in their preferred habitats. In their study in Maine, Vickery et al. (1994) noted that the grassland species they studied may have favored grassland-barrens rather than hayfields and pas- tures simply because grassland-barrens were the principal source of large expanses of grassland habitat. If grassland habitats are generally becoming scarcer in North America, and particularly in ag- ricultural areas (Askins 1993, Warner 1994), birds are likely to use near-equivalent patches of cropland and other nonnatural habitat (Litvaitis 1993, Vickery et al. 1994) but to require larger areas of such before settling there. Thus, With (1994) suggests that the natural habitat of McCown's Longspur in native short-grass prai- ries has now become a mosaic of pastures var- iably grazed by cattle and fragmented by agri- cultural activities and human development, and that the species may treat heavily grazed pas- tures as near approximations of the original hab- itat. In cropland-pasture, one would expect larg- er patches to be favored over smaller ones. Sim- ilarly, Warner (1994) found that the diversity of grassland species was highest on those study sites closest to grassland. Warner also demon- strated that Ring-necked Pheasant nests hatched more successfully the greater the amount of grassland (whether strip cover, forage crops, or small grains) surrounding the nest. These studies indicate how the increased influence of patch variables in secondary habitats might arise, un- der the assumption that these habitats are being used by populations displaced from preferred grassland habitats. A more alarmist interpretation is also possi- ble: some of the scarcer grassland species may display greater apparent selectivity simply be- cause there are now so few individuals remain- ing that they fill only the better components of CONTINENTAL BIRD DISTRIBUTIONS--O'Connor et al. 57 the available habitats (O'Connor 1981, Vickery et al. 1994). Whatever the processes underlying the pattern, our findings endorse the need to pre- serve remaining large plots of grassland habitats and to consolidate smaller patches in manage- ment efforts. The role of the CRP may be critical in this regard. A CAUTIONARY NOTE An important reservation about the findings here, not only for crop variables but for climate and habitat patch analyses, is that the CART models we used, despite their sophistication, re- turn only estimates of correlations. Therefore, our conclusions are subject to the normal cave- ats of correlation analysis, in particular that cor- relation does not ensure causation. Some con- clusions are likely to be stronger than first ap- parent with a correlation analysis. Our emphasis on patch variables and on CRP effects are each based on analyses with very different biases than in the site-specific studies of Vickery et al. (1994) and Herkert (1994) for area sensitivity and of Johnson and Schwartz (1993) for the CRP. Hence, our arrival at a similar assessment of the importance of these variables for grass- land bird species lends strength to all the studies; different sources of bias are unlikely to yield similar conclusions in the absence of a real eco- logical effect. The broad spatial extent of our analyses and their replication across multiple years provide a robust overview of the correlates of grassland bird distribution that has hitherto been unavailable. Our results highlight particular patterns of correlation as deserving of further attention and raise some important new ques- tions about constraints on the distribution of grassland bird species. ACKNOWLEDGMENTS We thank C. Hunsaker, S. Timmons, and B. Jackson (Oak Ridge National Laboratory) for providing land- scape metrics and G. Nielsen (U.S. EPA, Corvallis, Oregon) for data from the Historical Climate Network climate databank. We thank D. White for help in using the climate data. We thank S. Moulton for secretarial support. R. J. O'Connor was supported by a National Research Council Senior Research Fellowship at the U.S. EPA, Corvallis, Oregon, while on sabbatical leave from the University of Maine 1993 1994. We ac- knowledge the following financial support for this work: U.S. EPA Cooperative Agreements CR 818843- 01-0 and CR 823806-01-0 (RJOC, MTJ) and U.S. De- partment of Agriculture Forest Service Cooperative Agreement PNW 93-0462 (RJOC and MTJ). Addi- tional financial support came from a consortium of ag- rochemical corporations: American Cyanamid, CIBA- Geigy, E. I. Dupont de Nemours and Company, FMC Corporation, Mobay, and Rh6ne-Poulenc. LITERATURE CITED ALLEN, r. F. H, AND r. B. STARR. 1982. Hierarchy: perspectives for ecological complexity. University of Chicago Press, Chicago, IL. ANDERSON, J. R., E. E. HARDY, J. r. ROACH, AND R. E. WITMER. 1976. A land use and land cover classifi- cation system for use with remote sensor data. U.S. Geological Survey Professional Paper 964. ASKINS, R. A. 1993. Population trends in grassland, shrubland, and forest birds in eastern North Ameri- ca. Current Ornithology 11:1 34. BLAKE, J. O., O. J. NIEMI, AND J. M. HANOWSKI. 1992. Drought and annual variation in bird populations. Pp. 419-430 in J. M. Hagan III and D. W. Johnston (editors). Ecology and conservation of neotropical migrant landbirds. Smithsonian Institution Press, Washington, D.C. BRADLEY, J. V. 1968. Distribution-free statistical tests. Prentice-Hall, Englewood Cliffs, NJ. BREIMAN, L., J. H. FRIEDMAN, R. A. OLSHEN, AND C. J. STONE. 1984. Classification and regression trees. Wadsworth, Belmont, CA. BROWN, J. H. 1984. On the relationship between abun- dance and distribution of species. American Natu- ralist 124:255-279. BROWN, J. H. 1995. Macroecology. University of Chi- cago Press, Chicago, IL. BROWN, J. L. 1969. The buffer effect and productivity in tit populations. American Naturalist 103:347-354. BYSTRAK, D. 1981. The North American Breeding Bird Survey. Studies in Avian Biology 6:34-41. CLARK, L. A., AND D. PREGIBON. 1992. Tree-based models. Pp. 377-419 in J. M. Chambers and T. J. Hastie (editors). Statistical models in S. Wadsworth & Brooks/Cole Advanced Books & Software, Pacif- ic Grove, CA. CURRIE, D. J. 1991. Energy and large-scale patterns of animal and plant species richness. American Natu- ralist 137:27-49. DANKO, D, M. 1992. The digital chart of the world. Geolnfo Systems 2:29 36. EFRON, B. 1982. The jackknife, the bootstrap and other resampling plans. Society for Industrial and Applied Mathematics, Philadelphia, PA. FRETWELL, S. D. 1986. Distribution and abundance of the Dickcissel. Current Ornithology 4:211-242. FRETWELL, S. D., AND H. L. LUCAS. 1969. On territorial behaviour and other factors influencing habitat dis- tribution in birds. I. Theoretical development. Acta Biotheoretica 19:16-36. GEORGE, T L., A. C. FOWLER, R. L. KNIGHT, AND L. C. MCEWEN. 1992. Impacts of a severe drought on grassland birds in western North Dakota. Ecological Applications 2:275 284. GRABER, R. R., AND J. W. GRABER. 1963. A compara- tive study of bird populations in Illinois, 1906-1909 and 1956-1958. Illinois Natural History Survey Bul- letin 28:383-528. HERKERT, J. R. 1994. The effects of habitat fragmen- tation on midwestern grassland bird communities. Ecological Applications 4:461 471. JOHNSON, D. G., AND M.D. SCHWARTZ. 1993. The Con- servation Reserve Program and grassland birds. Conservation Biology 7:934 937. JOHNSON, R. G., AND S. h. TEMPLE. 1986. Assessing 58 STUDIES IN AVIAN BIOLOGY NO. 19 habitat quality for bird nesting in fragmented tall- grass prairies. Pp. 245-249 in J. Verner, M. L. Mor- rison, and C. J. Ralph (editors). Wildlife 2000: mod- eling habitat relationships of terrestrial vertebrates. University of Wisconsin Press, Madison, WI. KREBS, C. J. 1985. Ecology: the experimental analysis of distribution and abundance. 3d ed. Harper and Row, New York, NY. LARSEN, D. T., P. L. CROOKSTON, AND L. D. FLAKE. 1994. Factors associated with Ring-necked Pheasant use of winter food plots. Wildlife Society Bulletin 22:620-626. LAUBER, T. B. 1991. Birds and the Conservation Re- serve Program: a retrospective study. M.S. thesis. University of Maine, Orono, ME. LITVAITIS, J. A. 1993. Response of early successional vertebrates to historic changes in land use. Conser- vation Biology 7:866-873. LOVELAND, T. R., J. W. MERCHANT, D. J. OHLEN, AND J. E BROWN. 1991. Development of a land-cover characteristics database for the conterminous U.S. Photogrammetric Engineering and Remote Sensing 57:1453-1463. O'CONNOR, R. J. 1981. Habitat correlates of bird dis- tribution in British census plots. Studies in Avian Biology 6:533-537. O'CONNOR, R. J. 1992. Population variation in relation to migrancy status in some North American birds. Pp. 64-74 in J. M. Hagan III and D. W. Johnston (editors). Ecology and conservation of neotropical migrant landbirds. Smithsonian Institution Press, Washington, D.C. O'CONNOR, R. J., M. T. JONES, D. WHITE, C. HUNSAK- ER, m. LOVELAND, B. JONES, AND E. PRESTON. 1996. Spatial partitioning of the environmental correlates of arian biodiversity in the lower United States. Bio- diversity Letters 3:97-110. O'CONNOR, R. J., AND M. SHRUBB. 1986. Farming and birds. Cambridge University Press, Cambridge, U.K. O'NEILL, R. V., J. R. KRUMMEL, R. H. GARDNER, G. SUGIHARA, B. JACKSON, D. L. DEANGELIS, B. T. MILNE, M. G. TURNER, B. ZYGMUNT, S. W. CHRIS- TENSEN, V. H. DALE, AND R. L. GRAHAM. 1988. In- dices of landscape pattern. Landscape Ecology 1: 153-162. OVERTON, W. S., D. WHITE, AND D. L. STEVENS. 1990. Design report for EMAP: environmental monitoring and assessment program, EPA/600/3-91/053. U.S. Environmental Protection Agency, Office of Re- search and Development, Washington, D.C. OWENS, R. h., AND M. T. MYRES. 1973. Effects of ag- riculture upon populations of native passefine birds of an Alberta fescue grassland. Canadian Journal of Zoology 51:697-713. PRICE, J. T. 1995. Potential impacts of global climate change on the summer distributions of some North American grassland birds. Ph.D. thesis. Wayne State University, Detroit, MI. RILEY, T. Z. 1995. Association of the Conservation Re- serve Program with Ring-necked Pheasant survey counts in Iowa. Wildlife Society Bulletin 23:386- 390. RODENHOUSE, N. L., I_,. B. BEST, R. J. O'CONNOR, AND E. K. BOLLINGER. 1993. Effects of temperate agri- culture on neotropical migrant landbirds. Pp. 280- 295 in D. M. Finch and P. W. Stangel (editors). Sta- tus and management of neotropical migratory birds. USDA Forest Service Gen. Tech. Rep. RM-229. USDA Forest Service Rocky Mountain Forestry and Range Experimental Station, Fort Collins, CO. ROOT, T. L. 1988a. Energy constraints on arian distri- butions and abundances. Ecology 69:330-339. ROOT, T. L. 1988b. Atlas of wintering North American birds: an analysis of Christmas Bird Count data. University of Chicago Press, Chicago, IL. ROOT, T. L. 1993. Effects of global climate change on North American birds and their communities. Pp. 280-292 in E M. Kareiva, J. G. Kingsolver, and R. B. Huey (editors). Biotic interactions and global change. Sinaner Associates, Sunderland, MA. ROTENBERRY, J. T., R. J. COOPER, J. M. WUNDERLE, AND K. G. SMITH. 1995. When and where are populations limited. 9 The roles of insect outbreaks, fire, and other natural perturbations. Pp. 55-84 in T. E. Martin and D. M. Finch (editors). Ecology and management of neotropical migratory birds: a synthesis and review of critical issues. Oxford University Press, Oxford, U.K. SAMSON, EB. 1980. Island biogeography and the con- servation of prairie birds. Proceedings of the North American Prairie Conference 7:293-305. SAUER, J. R., J. E. HINES, G. GOUGH, I. THOMAS, AND B. G. PETERJOHN. 1997. The North American Breed- ing Bird Survey results and analysis. Ver. 96.4: www.mbr. nbs.gov/bbs/bbs.html. Patuxent Wildlife Research Center, Laurel, MD. SHACKFORD, J. S. 1994. Nesting of Long-billed Cur- lews on cultivated fields. Bulletin of the Oklahoma Ornithological Society 27:17-20. SONQUIST, J. A., E. L. BAKER, AND J. N. MORGAN. 1973. Searching for structure. Survey Research Cen- ter, University of Michigan, Ann Arbor, MI. TURNER, M. G., AND R. H. GARDNER. 1990. Quantita- tive methods in landscape ecology: the analysis and interpretation of landscape heterogeneity. Springer Verlag, New York, NY. U.S. DEPARTMENT OF AGRICULTURE. 1987. Agricultural statistics. U.S. Government Printing Office, Wash- ington, D.C. VICKERY, P. D., M. L. HUNTER, JR., AND S. M. MELVIN. 1994. Effects of habitat area on the distribution of grassland birds in Maine. Conservation Biology 8: 1087-1097. WALKER, P. A. 1990. Modeling wildlife distributions using a geographic information system: kangaroos in relation to climate. Journal of Biogeography 17: 279-289. WARNER, R. E. 1994. Agricultural land use and grass- land habitat in Illinois: future shock for midwestern birds? Conservation Biology 8:147-156. WHITE, D., J. KIMMERLING, AND W. S. OVERTON. 1992. Cartographic and geometric components of a global design for environmental monitoring. Cartography and Geographic Information Systems 19:5-22. WHITMORE, R. C. 1981. Structural characteristics of Grasshopper Sparrow habitat. Journal of Wildlife Management 45:811-814. WIENS, J. A. 1973. Interterritorial habitat variation in Grasshopper and Savannah sparrows. Ecology 54: 877-884. CONTINENTAL BIRD DISTRIBUTIONS--O'Connor et al. 59 WIENS, J. A. 1981. Scale problems in avian censusing. Studies in Avian Biology 6:513-521. WITH, K. A. 1994. The hazards of nesting near shrubs for a grassland bird, the McCown's Longspur. Con- dor 96:1009-1019. WITHAM, J. W., AND M. L. HUNTER. 1992. Population trends of neotropical migrant landbirds in northern coastal New England. Pp. 85-95 in J. M. Hagan III and D. W. Johnston (editors). Ecology and conser- vation of neotropical migrant landbirds. Smithsonian Institution Press, Washington, D.C. WRIGHT, D. H. 1991. Correlations between incidence and abundance are expected by chance. Journal of Biogeography 18:463-466. Studies in Avian Biology No. 19:60-71, 1999. HISTORY OF GRASSLAND BIRDS IN EASTERN NORTH AMERICA ROBERT A. ASKINS Abstract. Until recently the severe decline in the populations of many species of grassland birds in eastern North America has aroused relatively little concern or conservation action. This response appears to be rooted in the perception that grassland birds invaded the East Coast from western grasslands after European settlers cleared the forest. Detailed historical accounts and analysis of pollen deposits, however, show that open grasslands existed on the East Coast of North America at the time of European settlement. Extensive grasslands resulted from burning and agricultural clearing by Native Americans. Natural disturbances, such as wildfire and beaver (Castor canadensis) activity, produced grasslands even before Native Americans cleared the forest. The presence of specialized grassland birds in Pleistocene deposits and in the earliest ornithological collections from eastern North America, and the existence of distinctive eastern populations of the Greater Prairie-Chicken (Tympanuchus cupido), Henslow's Sparrow (Ammodramus henslowii), and Savannah Sparrow (Passerculus sand- wichensis), indicate that grassland birds are an ancient component of biological diversity on the heavily forested East Coast of North America. LA HISTORIA DE LAS AVES DE PASTIZAL EN EL ESTE DE NORTEAMIRICA Sinopsis. Hasta hace poco tiempo, la declinaci6n severa de las poblaciones de muchas especies de aves de pastizal en el este de Amdrica del Norte ha causado poco interds o acci6n de conservaci6n. Esta respuesta se basa en la percepci6n de que las aves de pastizal invadieron la costa este de Amdrica del Norte de los pastizales occidentales despuds de la tala del bosque por los colonos europeos. Sin embargo, relatos hist6ricos detaliados y analisis de dep6sitos de polen indican que pastizales abiertos existfan en la costa este de Amdrica del Norte en la dpoca de la colonizaci6n europea. Quemas y desbrozos agrfcolos de los indfgenas norteamericanos se tradujeron en pastizales extensivos. Disturbios naturales, como los hechos por incendios y castores (Castor canadensis), produjeron pastizales antes de que los indfgenas talaran el bosque. La presencia de aves de pastizal especialistas en los dep6sitos pleistocenos yen las colecciones ornitol6gicas mas antiguas del este de Am6rica del Norte, y la existencia de poblaciones orientales distintivas de Tympanuchus cupido, el Gorri6n de Henslow (Am- modramus henslowii), y el Gorri6n Sabanero (Passerculus sandwichensis), indican que las aves de pastizal forman una panre antigua de la diversidad bio16gica en la arbolada costa este de Amdrica del Norte. Key Words: beaver; bird populations; disturbance; eastern grasslands; grassland birds. Many grassland bird species were common or even abundant along the East Coast through most of the nineteenth century, but their num- bers diminished noticeably between the late nineteenth and mid-twentieth centuries. Griscom (1949) described the Upland Sandpiper (Bartra- mia longicauda), Bobolink (Dolichonyx oryzi- vorus), Eastern Meadowlark (Sturnella magna), and Grasshopper Sparrow (Ammodramus savan- narum) as formerly common but declining in Massachusetts. Forbush (1925:449) mourned the virtual disappearance of the Upland Sandpiper from New England: "our children's children may never see an Upland Plover in the sky or hear its rich notes on the summer air. Its cries are among the most pleasing and remarkable sounds of rural life." Although the decline of grassland birds was obvious to any careful observer, it is only in re- cent decades that we have been able to calculate the precise rate and extent of these population changes. The best evidence for this comes from the Breeding Bird Survey, a system of roadside routes scattered throughout the United States and southern Canada where birds are counted each year (Peterjohn 1994). The results for all of the survey routes east of the Mississippi River indicate that since 1966, when the surveys be- gan, the abundance of 14 of the 19 species of grassland and savanna birds in eastern North America has declined significantly (Table 1). Some have shown rapid population changes. Be- tween 1966 and 1994, for example, Grasshopper Sparrows decreased at a rate of 6% per year, whereas the annual rates of decline were 3% for Vesper Sparrows (Pooecetes gramineus), 9% for Henslow's Sparrows (Ammodramus henslowii), and 3% for Eastern Meadowlarks. In contrast, only 2 of 40 species of forest-dwelling migra- tory birds--a group that has received consider- able attention from conservationistsdecreased at a rate of more than 2% per year during ap- proximately the same period (Askins 1993). Another indication that grassland birds in the eastern United States are in trouble comes from state lists of endangered and threatened species 60 HISTORY OF EASTERN GRASSLAND BIRDSAskins 61 TABLE 1. POPULATION TRENDS OF GRASSLAND AND SAVANNA SPECIALISTS IN NORTH AMERICA EAST OF THE MISSISSIPPI RIVER BETWEEN 1966 AND 1994 % change per Statistical Number of Species year significance a routes Grassland species b Northern Harrier (Circus cyaneus) Ring-necked Pheasant (Phasianus colchicus) Northern Bobwhite ( Colinus virginianus) Upland Sandpiper (Bartramia longicauda) Horned Lark (Eremophila alpestris) Dickcissel (Spiza americana) Vesper Sparrow (Pooecetes gramineus) Lark Sparrow ( Chondestes grammacus) Savannah Sparrow (Passerculus sandwichensis) Grasshopper Sparrow (Ammodramus savannarum) Henslow's Sparrow (A. henslowii) Bobolink (Dolichonyx oryzivorus) Eastern Meadowlark (Sturnella magna) Western Meadowlark (S. neglecta) Savanna species c Common Ground-Dove (Columbina passerina) Red-headed Woodpecker (Melanerpes erythrocephalus) Eastern Bluebird (Sialia sialis) Loggerhead Shrike (Lanius ludovicianus) American Goldfinch ( Carduelis tristis) +2.1 * 227 -2.2 - 522 3.3 *** 880 + 1.2 201 0.0 - 642 -4.3 *** 269 3.3 *** 650 -6.9 * 39 -1.7 *** 727 -5.9 *** 730 -9.3 *** 137 1.4 *** 745 -3.4 *** 1,330 7.2 *** 174 -3.4 *** 127 -2.1 *** 667 +2.2 *** 1,185 -4.3 *** 379 1.1 *** 1,349 Note: Data are from the Breeding Bird Survey database. U.S. Fish and Wildlife Service (Sauer et al. 1995), Habitat classifications are based on DeGraaf and Rudis 1986. DeGraaf et al. 1991, and Askins 1993. a * p < 0.05. *** P < 0.001. b Species dependent on open habitats dominated by grass and forbs, with little woody vegetation. c Species found primarily in open grassland with scattered trees or shrubs. (Vickery 1992). Of the 40 species listed as en- dangered, threatened, or of special concern in three or more northeastern states, 13 are grass- land or savanna specialists and only 3 are forest specialists. For example, Upland Sandpiper, Northern Harrier (Circus cyaneus), Loggerhead Shrike (Lanius ludovicianus), and Grasshopper, Henslow's, and Vesper sparrows are listed in all or most of the New England states (Vickery 1992). The populations of many of these species have declined in other parts of the eastern Unit- ed States, both in heavily forested areas along the East Coast and in the more agricultural Mid- west (Herkert 1991, Bollinger and Gavin 1992). What Mayfield (1988) called the "quiet de- cline" of grassland birds has attracted surpris- ingly little attention or concern from most gov- ernment wildlife agencies and conservation or- ganizations. This response is rooted in the wide- ly held view that before Europeans cleared the land, unbroken forest stretched from the Atlantic to the Great Plains, leading to the general im- pression that grassland species invaded the east- ern states from western savannas and prairies af- ter the clearing of the forest for agriculture. For example, Whitcomb (1987) argued that this in- vasion of the eastern "neosavanna" created by agriculture has been a "failed experiment for many of these species," which are now declin- ing. The implication is that this is a return to ecosystems more similar to those before Euro- pean settlement and therefore should not be a cause for concern. According to Whitcomb, these species could survive only with active management to preserve grassland "in a region where [grassland] is inappropriate as an equilib- rium community" (Whitcomb 1987:165). Many historians and botanists have depicted the landscape of the ancient East Coast of North America as carpeted with forest, a forest so con- tinuous that a squirrel could travel from the At- lantic Ocean to the Mississippi River without touching the ground (Day 1953). Clearly forests in eastern North America were extensive, and in some areas they were essentially unbroken (Sic- cama 1971, Lorimer 1977, Bormann and Likens 1979, Runkle 1990, Seischab and Orwig 1991). Since the early 1900s, however, some botanists have argued that the forest was not always con- tinuous; in coastal areas and even in some inland areas, it was interrupted by scrubland, barrens, glades, and even, in places, prairielike grass- lands (Day 1953). If this is true, then grassland birds would have had a place in the landscape before European settlement. 62 STUDIES IN AVIAN BIOLOGY NO. 19 WERE THERE GRASSLANDS ON THE EAST COAST BEFORE EUROPEAN SETTLEMENT? When Conrad (1935) visited Long Island's Hempstead Plains in the 1930s, much of the area was little bluestem (Schizachyrium scoparius) prairie, a yellow-green grassland dotted with the small, bright green hemispheres of wild indigo (Baptista tinctoria). In May the prairie was blue with the blossoms of birdfoot violets (Viola pe- data). As Conrad pointed out, this grassland on the New York coast was remarkably similar to the tallgrass prairies of Iowa and Nebraska. Moreover, the Hempstead Plains had a rich com- munity of grassland birds: Upland Sandpipers, Bobolinks, and Vesper and Grasshopper spar- rows were all common there in the 1920s (Bull 1974). European travelers described the Hempstead Plains as treeless in the 1600s (Harper 1911), so this grassland was not a product of European agriculture. The Hempstead Plains were char- acterized by thin soil resting on a porous foun- dation of quartz and granite pebbles (Conrad 1935, Cain et al. 1937), features that, in com- bination with periodic fires, appeared to favor the growth of grasses and herbs rather than trees and shrubs. The Plains once covered more than 20,000 ha, and for many years the area was used primarily for grazing sheep and racing horses (Svenson 1936, Stalter and Lamont 1987). Large areas of grassland remained in the 1930s, but after World War II these open areas were subdivided for housing or plowed for truck farms. Today only a few acres of this prairie survive: an 8-ha parcel belonging to Nassau County Community Col- lege and managed by The Nature Conservancy, and a 19-ha parcel managed as a nature preserve by Nassau County (Antenen et al. 1994). The smaller preserve has been maintained with con- trolled burning. Although the Hempstead Plains may have been one of the largest and most distinctive grasslands on the East Coast, it was not the only one. Another grassland, the Montauk Downs, covered approximately 2,400 ha of eastern Long Island (Taylor 1923), and several large grass- lands, called "glades," characterized a plateau in the Allegheny Mountains of western Penn- sylvania (Whimey 1994). Also, in the 1600s a savanna where occasional large oaks (Quercus) broke an expanse of tall, wiry grass (probably bluestem) stretched for 24 km along the Quin- nipiac River north of New Haven, Connecticut (Olmsted 1937). After decades of overgrazing, this area became the almost desertlike North Ha- ven Sand Plains, and subsequently most of the area was developed. Blueberry barrens, which are open expanses covered with lowbush blue- berry (Vaccinium angustifolium) shrubs and grasses, still cover large areas in eastern Maine, where they are maintained by burning for blue- berry production. Some of the largest East Coast populations of Upland Sandpipers, Vesper Spar- rows, and other species of grassland birds breed on these barrens (Vickery et al. 1994). Many of these grasslands may have resulted from the activities of Native Americans before European settlement. Early explorers and colo- nists frequently encountered open landscapes created by firewood harvesting, agricultural clearing, and burning to enhance hunting. For example, Giovanni da Verrazano described the area around Narragansett Bay (Rhode Island) in 1524 as open plains, without forests or trees, for many leagues inland (Day 1953). Samuel de Champlain and John Smith reported extensive areas of cleared land along the New England coast before Europeans colonized the area (Whitney 1994). Moreover, an early settler in Salem, Massachusetts, described "open plains, in some places five hundred acres...not much troublesome for to cleere for the plough to goe in" (Day 1953:331). These clearings were not restricted to coastal areas; accounts of early set- tlers indicate that river valleys had been cleared by Native Americans for farming and hunting (Patterson and Sassaman 1988). Early assessments that Native American ag- riculture had relatively little effect on the land- scape were based on population estimates after European settlement, but population densities were much higher before contact with Europe- ans triggered massive epidemics that killed a large proportion of the people in most tribes (Crosby 1972, Cronon 1983, Denevan 1992, Whitney 1994). As Kulikoff (1986:29) wrote re- garding the Chesapeake Bay area, "though En- glish settlers did not find a wilderness, they did create one"; extensive agricultural clearings re- verted to forest as Native American populations declined. Pilgrims traveling through the area near Warren, Massachusetts, in 1621 "saw the remains of so many once occupied villages and such extensive formerly cultivated fields that they concluded thousands of people must have lived there before the plague" (Russell 1980: 24). Maps, drawings, and written accounts of the landscape around Native American settlements in the southeastern United States before Euro- pean settlement provide evidence of extensive clearings created by farming and of "parklands" maintained by controlled burning (Hammett 1992). In New York and southern New England, relatively high population densities combined with slash-and-burn agriculture (Whitney 1994) HISTORY OF EASTERN GRASSLAND BIRDSAskins 63 would have resulted in extensive areas of cleared land in the form of both active and aban- doned fields. This would have produced a "mo- saic of forests and fields in varying stages of succession" (Patterson and Sassaman 1988: 115). Another view is that Eastern tribes used large permanent agricultural fields from which tree stumps had been removed rather than tem- porary fields cut out of the forest for slash-and- burn agriculture (Doolittle 1992). This perma- nent farmland would have had to be rested oc- casionally, however, producing "weed-covered" fallow fields of the sort seen by Champlain near the site of Boston in 1605 (Doolittle 1992). Re- gardless of whether Native Americans used slash-and-burn or permanent-field agriculture, their activities would have produced open hab- itats (abandoned or fallow fields) that could have been used by grassland birds. There also is good evidence that the Native Americans of the East Coast burned large areas to create open woodlands and grassland for hunting. For example, Roger Williams wrote in the 1640s that Native Americans in New En- gland "burnt up all the underwoods in the Coun- trey, once or twice a yeare and therefore as No- ble men in England possessed great Parkes ... onely for their game" (Williams 1963:47). In 1818, B. Trumbull reported that the Native Americans of Connecticut "so often burned the country, to take deer and other wild game, that in many of the plain dry parts of it, there was but little small timben Where the lands were thus burned there grew bent grass, or as some called it, thatch, two, three and four feet high" (Olmsted 1937:266). Native Americans in New York not only burned the woods each autumn to create a more open understory but also burned plains and meadows to improve hunting (Whit- ney 1994). Although fires were probably infrequent in most forests that were remote from Native American settlements (Russell 1983), fire and other disturbances near settlements provided ex- tensive habitat for early successional species, in- cluding potential habitat for grassland birds. An analysis of charcoal deposits in the sediments of 11 lakes in New England demonstrated that be- fore European settlement, fires were frequent in densely populated coastal areas but infrequent in inland and northern areas (Patterson and Sassa- man 1988). Moreover, Winne's (1988) analysis of pollen and charcoal in lake sediments showed that the area around Pineo Pond in eastern Maine has been characterized by frequent, mod- erate fires and scrubby, fire-adapted vegetation for at least 900 yr. Today this area is dominated by blueberry barrens that are maintained by con- trolled burning. These support a diversity of breeding grassland birds (Vickery et al. 1994). The extent of forest clearing by Native Amer- icans in the northeastern United States probably paled in comparison with the extensive agricul- tural fields created by the Moundbuilders, who lived along the Mississippi River and its tribu- taries in much of what is now the southeastern and midwestern United States. Moundbuilding cultures existed in the lower Mississippi River Valley as early as 1500 B.c. The early Mound- building cultures probably depended on a mix- ture of hunting, gathering of nuts, fishing, and small-scale farming based on native plants such as sunflower (Helianthus) and marsh elder (Iva frutescens; Shaffer 1992). Later, during the Mis- sissippian Period, which lasted from A.O. 700 until the early 1700s, large-scale agriculture sup- ported a dense population living in closely spaced villages. Corn (Zea mays) and beans (Le- guminosae) from Mexico replaced indigenous crops, and large areas were cleared for farming (Shaffer 1992). The largest population center was Cahokia, located on the Mississippi River near its confluence with the Missouri River (Shaffer 1992). This center covered 800 ha, with 160 ha enclosed in a wooden palisade. The site was dotted with as many as 120 earthen mounds, the largest of which rose to 30 m and covered more than 6 ha. The mounds supported wooden buildings, and the area below the mounds was densely packed with rectangular thatched-roof houses where an estimated 15,000-38,000 peo- ple lived. Cahokia and the many villages and towns around it were supported by farming the American Bottom, a 324-km 2 strip of rich allu- vial soil in the floodplain along the eastern bank of the Mississippi Riven In A.O. 1000 there were 50 villages and 8 other large or medium-sized centers within 40 km of Cahokia. There are no historical accounts of Cahokia because it was abandoned after A.O. 1200 (Shaf- fer 1992). Early Spanish visitors visited similar sites that were still occupied in the 1500s, how- ever. A chronicler of Hernando de Soto's expe- dition (1539-1542) described a Moundbuilder town along the Mississippi River as being "in an open field, that for a quarter of a league over was all inhabited; and at the distance of from half a league to a league off were many other large towns, in which was a good quantity of maize, beans, walnuts [Juglans], and dried Ameixas [persimmons]" (Bourne 1904:149). The Moundbuilding centers were abandoned long before Europeans settled the southeastern United States or Mississippi River Valley. This culture may have been destroyed by Old World diseases that swept inland from European out- posts on the Florida and Gulf Coasts (Crosby 64 STUDIES IN AVIAN BIOLOGY NO. 19 1986). Before the collapse of this agricultural society, however, there were extensive areas of open fields in many parts of the Southeast, es- pecially the lower Mississippi River Valley. GRASSLANDS BEFORE NATIVE AMERICAN AGRICULTURE The patterns of Native American land use ob- served in the 1600s began to emerge about 2,000 yr ago (Smith 1989, 1995). Many species of grassland birds may have colonized cultivated areas after the initiation of Native American, rather than European, agriculture, and their cur- rent decline represents a return to conditions be- fore humans began to substantially modify the vegetation of eastern North America. Undoubtedly, many apparently "natural" open grasslands of eastern North America were the product of human activities. For example, historical accounts and analysis of the pollen re- cord indicate that the extensive heathlands and sandplain grasslands on Nantucket Island, Mas- sachusetts, resulted from the clearing of oak for- est and grazing of sheep after Europeans settled that island (Dunwiddie 1989). Some of the open habitats on the East Coast may predate disturbance by Native Americans or Europeans, however. Smaller shrubby and grassy openings in the eastern forest result from dam-building by beavers (Castor canadensis). After beavers exhaust the food supply around a pond, they move to another area. When the abandoned pond drains, the pond bed often be- comes a "beaver meadow," a patch of shrubby vegetation or grassland. This meadow eventually is overgrown with young forest, and after 10- 30 yr beavers may recolonize the site and initiate another cycle (Remillard et al. 1987). Although beaver meadows are largely restrict- ed to flood plains, their total area was probably extensive before beavers were extirpated in most parts of eastern North America (Naiman et al. 1988). In Ontario's Algonquin Provincial Park, where beavers are protected, there is a high den- sity of beaver ponds and meadows (Coles and Orme 1983). After beavers became reestablished at Quabbin Reservoir in Massachusetts in 1952, the population grew rapidly until the density reached 0.8 colony per kilometer of stream (Howard and Larson 1985). The impact of a dense beaver population can be considerable. In the Adirondack Mountains of New York, beaver dams created patches of disturbance that covered an average area of 7 ha, with a maximum area of 12 ha (Remillard et al. 1987). Bela Hubbard, who surveyed land in Michigan before European settlement, reported that one-fifth of the area within 19 km of what is now Detroit was cov- ered with "marshy tracts or prairies which had their origin in the work of the beaver" (Whitney 1994:304). Coles and Orme (1983:99) argued that ancient forests in England must have been "moth-holed with clearings wherever beaver were present." These "grassy meadows of relict pools" were also an important feature of the pre- settlement landscape of eastern North America. Although beaver meadows are generally too small to accommodate many species of special- ized grassland birds, some other species (e.g., Eastern Meadowlark and Savannah Sparrow) occur in patches of grassland of similar size (5- 10 ha; Herkert 1994, Vickery et al. 1994). Many regions of the East were subject to dis- turbances that created large openings in the for- est (Runkle 1990). Grassland, savanna, and grassy scrub were probably created by large fires, particularly fires that burned following hurricanes or tornadoes that periodically leveled forests. These catastrophic disturbances were probably most frequent in low-lying sandy areas on the coastal plain. In many regions farther in- land, fires, windstorms, and other disturbances were infrequent, and consequently the forest canopy was almost continuous, with few large openings (Lorimer 1977). For example, the northern hardwood forest of western New York and of the White Mountains of New Hampshire probably formed an almost unbroken canopy (Bormann and Likens 1979, Seischab and Orwig 1991). Large grassy openings may have oc- curred in some of the river valleys in the interior, however. John Winthrop (Hosmer 1959:85) de- scribed how one of the first European expedi- tions to the White Mountains passed through "many thousands of acres of rich meadow" as it paddled birch-bark canoes up the Saco River in what is now Maine. A LONGER VIEW: PLEISTOCENE STEPPE AND SAVANNA When continental glaciers covered much of Canada and the northern United States, the re- gions immediately south of the glaciers were dominated by a spruce (Picea) parkland, a grassy savanna with scattered spruce trees (Webb 1988). Samples of pollen from lake sed- iments deposited 18,000-12,000 yr ago show that this savanna stretched westward from the Atlantic Coast to the Great Plains. Most species of deciduous trees, and presumably the closed forests where these species grow today, were re- stricted to the extreme southeastern United States (Webb 1988). Thus, in eastern North America there was a gradient from savanna in the north to dense forest in the south. Vegetation zones shifted and changed as the glaciers retreated northward beginning about 12,000 yr ago. Spruce parkland largely disap- HISTORY OF EASTERN GRASSLAND BIRDS--Askins 65 peared, and a new gradient, from eastern forest to western prairie, gradually formed (Webb 1988). Before this transition occurred, however, the spruce parkland was occupied by a diversity of large open-country mammals; caribou (Ran- gifer mrandus), mastodons (Mammut american- um), and long-nosed peccaries (Mylohyus nasu- tus) are frequently found in fossil deposits from the time of the spruce parkland (Kurt6n and An- derson 1980). One of the best samples of spruce parkland animals comes from a site called New Paris No. 4 in Pennsylvania (Guilday et al. 1964). Ap- proximately 11,000 yr ago, a deep sinkhole act- ed like a pitfall trap, collecting the skeletons of more than 2,700 animals that fell into the crev- ice and died. The mixture of small mammals and skeletal remains of Sharp-tailed Grouse (Tym- panuchus phasianellus) at this site suggests that there was extensive grassland habitat in this re- gion. A better picture of the birdlife of the postgla- cial period comes from another site, the caves of Natural Chimneys in Virginia, where skele- tons were deposited at about the same time as at the New Paris No. 4 site. The skeletons at Natural Chimneys were deposited in owl pellets, so both small mammals and birds were well rep- resented (Guilday 1962). Although remains of these animals may have accumulated over a long period while the vegetation was changing, they provide a glimpse of the bird community of the spruce parkland. The bones of Sharp-tailed Grouse, Northern Bobwhite (Colinus virgini- anus), Upland Sandpiper, Red-headed Wood- pecker (Melanerpes erythrocephalus), Black- billed Magpie (Pica pica), and Brown-headed Cowbird (Molothrus ater), along with other grassland vertebrates such as thirteen-lined ground squirrel (Spermophilus tridecemlinea- tus), point to a landscape with large amounts of open savanna or grassland. The remains of woodland species such as Red-bellied Wood- pecker (Melanerpes carolinus), Eastern Wood- Pewee (Contopus virens), and Red-breasted Nut- hatch (Sitm canadensis) in the same deposits suggest either that woodland and savanna were found in the area at the same time or that wood- land invaded and replaced the savanna while the bones accumulated at Natural Chimneys. In ei- ther case, it is clear that grassland birds occurred in eastern North America before the spruce parkland receded and disappeared. A site farther south, at Duck River, Tennessee, revealed that in the late Pleistocene typical co- niferous forest species (Northern Hawk-Owl [Surnia ulula], Boreal Owl [Aegolius funereus], Northern Saw-whet Owl [A. acadicus], Gray Jay [Perisoreus canadensis], and Pine Grosbeak [Pinicola enucleator]) lived alongside grassland species such as Sharp-tailed Grouse, Greater Prairie-Chicken (Tympanuchus cupido), Horned Lark (Eremophila alpestris), meadowlark (Stur- nella sp.), pocket gopher (Geomys spp.), and thirteen-lined ground squirrel (Parmalee and Klippel 1982). A key question is whether grassland birds could have survived the northward spread of closed-canopy forest over eastern North Ameri- ca after the continental glaciers melted. This is an issue not only for the current warm intergla- cial period but also for previous interglacial pe- riods. In all previous interglacial periods and at the beginning of the current postglacial period, large browsers such as mastodons and giant ground sloths (Megalonyx jeffersonii and other species) may have created and maintained open- ings in the forest in much the same way as Af- rican elephants (Loxodonm africana) maintain open savannas in East Africa today (Dublin et al. 1990). European ecologists have recognized that giant herbivores, particularly the extinct rel- atives of elephants, probably opened the forest, creating glades, parklike woods, or even savan- nas in areas that would otherwise be dominated by dense forest (Andersson and Appelquist 1990, Puchkov 1992). Such openings would have supported a variety of animal and plant species that depend on grassy habitats. Through most of the past I million yr, as forests retreated and advanced in response to the shrinking and growing of glacial ice sheets, woodland habitats may have been modified and opened by giant herbivores. Only in the present interglacial did mastodons, ground sloths, and other giants dis- appear from North America, perhaps as a result of the invasion of the continent by people who had already developed efficient tools and strat- egies for hunting large animals (Martin and Klein 1984). Human activities such as burning and agricultural clearing subsequently may have substituted for giant herbivores in creating a mo- saic of forest and openings (Andersson and Ap- pelquist 1990), permitting open-country species to persist in eastern woodlands. THE ORIGIN OF EASTERN GRASSLAND BIRDS The common impression that many species of grassland birds spread eastward from the prairies of the Midwest to the newly cleared farmland of the East Coast is substantiated by several well- documented examples of range expansion. For example, the prairie subspecies of Horned Lark (Eremophila alpestris praticola) spread east- ward from Illinois and Wisconsin, reaching Michigan and Ontario in the 1870s, New York in the 1880s, New England by 1891, and Penn- 66 STUDIES IN AVIAN BIOLOGY NO. 19 sylvania and Maryland by 1910 (Forbush 1927, Thomas 1951, Hurley and Franks 1976). The Dickcissel (Spiza americana) spread eastward from the tallgrass prairies in the early 1800s, but its range contracted after 1850, and it eventually disappeared as a regular breeding bird along the East Coast (Hurley and Franks 1976). Western Meadowlarks (Sturnella neglecta) expanded east into Wisconsin and Michigan after 1900 (Lan- yon 1956), and Lark Sparrows (Chondestes grammacus) spread eastward from the prairies into agricultural areas in the Ohio Valley, West Virginia, and western Maryland (Brooks 1938). Although these eastward range expansions were well documented, there is no similar evi- dence for invasion of the East by the species that are most abundant and widespread in eastern grasslands. Upland Sandpipers, Grasshopper Sparrows, Bobolinks, Eastern Meadowlarks, and other common grassland birds were reported by the earliest ornithologists who systematically documented the distribution of birds on the east- ern coast of North America. Alexander Wilson's American Ornithology (originally published be- tween 1808 and 1814; Brewer 1839) and John James Audubon's Ornithological Biography (Audubon 1831-1849) were published more than 100 yr after most of the eastern seaboard had been cleared, so it is possible that grassland birds colonized the meadows and pastures cre- ated by Europeans long before their occurrence was initially documented. Some seventeenth- century European observers, such as John Jos- selyn (Lindholt 1988) and William Wood (Vaughan 1977), described gamebirds and the more conspicuous songbirds, but only a few spe- cies are recognizable because descriptions are sketchy and the names of British birds were fre- quently used for North American species. Mark Catesby's Natural History of Carolina, Florida, and the Bahama Islands, completed in 1747, in- cludes descriptions and paintings of many spe- cies of eastern birds, including two species of grassland songbirds, Eastern Meadowlark and Bobolink (Feduccia 1985). It is not surprising, however, that there are relatively few descrip- tions of grassland songbirds from this period. Many grassland birds are small, inconspicuous, and dull colored, so they could have been over- looked by early observers. Significantly, the East Coast populations of three species of grassland birds were distinctive enough from western populations to be consid- ered separate subspecies. This suggests that these populations have existed in isolation in the East for many thousands of years, perhaps since unbroken grasslands reached from the Great Plains to the Atlantic during the last glacial pe- riod. The eastern Henslow's Sparrow (Ammo- dramus henslowii susurrans) has a breeding range restricted to central New York and south- ern New England south to Virginia, eastern West Virginia, and North Carolina (Smith 1968). It is darker than the western subspecies, with a deep- er bill and more buff on the underparts and more yellow in the wing (Smith 1968). The "Ipswich" Sparrow (Passerculus sand- wichensis princeps), a subspecies of the Savan- nah Sparrow, is also restricted to the East Coast (Wheelwright and Rising 1993). This population is so distinctive that it was considered a separate species until 1973. After reclassifying it as a subspecies of Savannah Sparrow, the American Ornithologists' Union (1957) recommended that it be designated by its vernacular name in quotes ("Ipswich" Sparrow). Vernacular names are only officially recognized for particularly dis- tinctive subspecies (Stobo and McLaren 1975). The "Ipswich" Sparrow was not described until 1868, when C. J. Maynard collected one in the coastal dunes at Ipswich, Massachusetts (El- liott 1968). Its breeding range on Sable Island, a 32-km-long island about 135 km off the coast of Nova Scotia, was not discovered until 1884. "Ipswich" Sparrows spend both summers and winters in extremely open habitats. During the breeding season they are virtually restricted to low shrubby vegetation and stands of marram grass (Ammophila breviligulata) on Sable Island (Stobo and McLaren 1975). In winter they occur primarily in a narrow zone of dunes near Atlan- tic beaches from Nova Scotia to Florida, with the highest densities on relatively undeveloped barrier islands and sandy peninsulas between New Jersey and Virginia (Stobo and McLaren 1971). The "Ipswich" Sparrow is adapted to living in the dunes and sandy scrub adjacent to the ocean. It is paler gray than other Savannah Spar- row subspecies, so it tends to be well camou- flaged in the light-colored dune and beach areas where it lives (Stobo and McLaren 1975). Also, it averages 9% larger than other eastern subspe- cies of Savannah Sparrow, which might be an adaptation to feeding on the exceptionally large seeds of dune grasses such as marram grass and sea oats (Uniola paniculata; Stobo and McLaren 1975). Finally, unlike other Savannah Sparrow subspecies, the "Ipswich" Sparrow has a short tail, making it more similar to Grasshopper Sparrow, Seaside Sparrow (Ammodramus mari- timus), and other sparrows that live in open hab- itats with few tall shrubs or trees. These specific adaptations indicate that grassy habitats must have existed along the outer beaches of the East Coast for a long time. The now-extinct Heath Hen (Tympanuchus cupido cupido) was the eastern subspecies of the HISTORY OF EASTERN GRASSLAND BIRDS--Askins 67 Greater Prairie-Chicken. During the early years of European settlement, the Heath Hen was common or even abundant in open grasslands and scrublands on Long Island and around Bos- ton, and it ranged along the coast from southern Maine as far south as Virginia (Gross 1932). Be- cause it was an important game species, it was described by many early settlers. In the 1600s, William Wood, Thomas Morton, and other ob- servers wrote that the Heath Hen was common in eastern Massachusetts (Forbush 1927, Gross 1932, Vaughan 1977). Heath Hens inhabited sandy scrub-oak plains, pine (Pinus) barrens, blueberry barrens, and other open habitats (For- bush 1927, Johnsgard 1983). In the nineteenth century they were common in the open grass- land of the Hempstead Plains on Long Island (Bull 1974). The abundance of Heath Hens at the time of European settlement, and the recognition of the Heath Hen and eastern populations of two other species of grassland birds as distinct subspecies, suggest that grassland birds inhabited the East Coast long before Europeans arrived or even be- fore Native Americans started clearing the land for farming. This is consistent with the evidence on grassland plants. Several plant species are re- stricted to eastern grasslands (Mehrhoff 1997; E Dunwiddie, pets. comm.), suggesting that they evolved in isolation from the grasslands of the Great Plains. Bushy rockrose (Helianthernurn durnosurn) is found from Massachusetts to Long Island; sandplain agalinis (Agalinis acuta) from Massachusetts to Maryland; and sickle-leaved golden aster (Pityopsis [Chrysopsis] falcata) from Massachusetts to New Jersey (Gleason and Cronquist 1991). In addition, a subspecies of northern blazing star (Liatris scariosa var. no- vae-angliae) is found only in eastern grasslands. Because of the paucity of historical records of small birds in the 1600s and 1700s, it is likely that only carefully dated skeletal remains could provide definitive evidence of the occurrence of most species of grassland birds before European settlement. There is strong evidence, however, that extensive grasslands and savannas occurred in eastern North America at the time of Euro- pean settlement. We also know that some grass- land species were found in the spruce parkland of postglacial times, about 11,000 yr ago, and that distinctive eastern subspecies evolved in three grassland species. Therefore, it is reason- able to conclude that many open-country species are native to the region, not recent invaders from the western prairies. During the eighteenth and nineteenth centuries these species probably be- came much more abundant than they had been before Europeans cleared the land, but subse- quently they may have declined far below the level of abundance characteristic of the preset- tlement landscape. Many of these species are now in danger of regional extinction, and they deserve the same attention from conservationists as birds associated with eastern forests, marshes, and lakes. CONSERVATION OF GRASSLAND BIRDS Many of the original grasslands, such as bea- ver meadows and recently burned areas, were ephemeral. Other areas may have been disturbed frequently enough to create stable grasslands; the Hempstead Plains in New York and some of the barrens in eastern Maine are obvious can- didates. Temporary grasslands are created much less frequently today because beavers are less abundant and fires are controlled, and most of the areas that may have been stable grasslands have been developed for agriculture or housing. The blueberry barrens of eastern Maine are an exception; these open habitats have been main- tained in a seminatural state by controlled burn- ing to sustain commercial blueberry production (Vickery et al. 1994). With the exception of Maine's blueberry bar- rens and a few other areas remaining as semi- natural open habitat for centuries, present-day habitats used by grassland birds along the East Coast are highly artificial. Populations of grass- land species have diminished primarily because much of the farmland in the Northeast and parts of the Southeast has been abandoned and has reverted to forest, and because the remaining farmland is now managed more intensively for agricultural production (Hart 1968, Askins 1993). For example, hayfields have become less suitable as nesting habitat for Eastern Mead- owlarks, Bobolinks, and some other grassland species because they are mowed earlier in the summer, before the end of the nesting season, and because they are rotated more frequently (Bollinger and Gavin 1992). In southern New England, most of the remaining populations of Grasshopper Sparrows and Upland Sandpipers are found in extensive mowed areas at airports and military airfields (Veit and Petersen 1993, Bevier 1994, Melvin 1994). The farmland once used by these species has either disappeared or become unsuitable for nesting. Regional populations of grassland birds can be maintained with proper management of arti- ficial grasslands such as fallow farmland and the mowed areas near airport runways. The Conser- vation Reserve Program (CRP), which pays farmers to take land out of production in order to manage it for conservation of soil and wildlife (Dunn et al. 1993), could potentially benefit grassland birds in the East as it already has in some western prairie regions (Johnson and 68 STUDIES IN AVIAN BIOLOGY NO. 19 Schwartz 1993, Johnson and Igl 1995). How- ever, most of the CRP land is concentrated in the northcentral United States (Rodenhouse et al. 1995), and abandoned farmlands in the East quickly become wooded, so a better approach might be to compensate farmers who use less intensive farming methods to create traditional hay meadows and other types of farmland that once sustained grassland birds. This approach has been successful in preserving open-country species in the Netherlands (Beintema 1988). Relatively simple changes in airport manage- ment (e.g., removing woody vegetation and changing mowing schedules to avoid the nesting season) have sustained or improved habitat for grassland birds at Westover Air Reserve Base in Massachusetts (Melvin 1994), Bradley Interna- tional Airport in Connecticut (Crossman 1989), and Floyd Bennett Field, a former naval air base on Long Island, New York (Lent and Litwin 1989). Habitat management at Westover resulted in substantial increases in the abundance of Grasshopper Sparrows and Upland Sandpipers between 1987 and 1994 (Melvin 1994). Even when grassland birds are absent from an area, it should be possible to create habitat that will attract them. Probably because eastern grassland birds have always depended on patch- es of ephemeral habitat, they have a remarkable ability to find and colonize remote sites, even sites far from other bird populations of the same species. When a field in Lincoln, Massachusetts, was managed to maintain tall grass for nesting Bobolinks, in 1995 it attracted a breeding pair of Henslow's Sparrow, a species that has almost disappeared from Massachusetts, with no known breeding records in the preceding 20 yr (Ells 1995). Similarly, when abandoned strip mines in heavily forested areas of West Virginia were re- stored and seeded with grass, they were colo- nized by Horned Larks, Eastern Meadowlarks, and Savannah, Vesper, and Grasshopper spar- rows (Whitmore and Hall 1978). These new grasslands were extremely isolated from other grasslands supporting grassland birds, but they still attracted breeding populations of several species. Expending scarce resources to maintain meadows, fallow fields, and airfields may seem unwise to many conservationists who are accus- tomed to protecting forests and wilderness areas. Yet many species of birds, insects, plants, and other organisms depend on these grassland hab- itats. Artificial habitats are critical for many of these species because people have destroyed most of the native grassland habitat, including most of the midwestern tallgrass prairies where these species may have once been most abun- dant (Bollinger et al. 1990). People have not only destroyed natural grasslands directly, but they also have interrupted or suppressed many of the natural processes of disturbance, such as fires and beaver activity, that once created the early successional habitats that grassland species need. In the near term, artificial grasslands rep- resent our best hope for maintaining grassland species. These species are an important, and probably ancient, component of biological di- versity along the East Coast of North America. ACKNOWLEDGMENTS I have gained insights about the questions addressed in this paper in discussions with P. D, Vickery, W. A. Niering, G. D. Dreyer, and R. M. DeGraaf. P. D. Vick- ery, J. R. Herkert, and an anonymous reviewer made helpful suggestions for improving the manuscript. My research was supported by funds from the USDA For- est Service, Northeastern Forest Experiment Station. An earlier version of this paper was published in the book Grasslands of Northeastern North America: Ecology and Conservation of Native and Agricultural Landscapes (P. D. Vickery and P. W. Dunwiddie [ed- itors]. 1997. Massachusetts Audubon Society, Lincoln, MA). This paper is also being revised for a chapter in a book by the author that will be published by Yale University Press. LITERATURE CITED AMERICAN ORNITHOLOGISTS' UNION. 1957. Check-list of North American birds. 5th ed. American Orni- thologists' Union, Washington, D.C. ANDERSSON, L., AND T. APPELQUIST. 1990. The influ- ence of the Pleistocene megafauna on the hemoral and the boreonemoral ecosystem: a hypothesis with implications for nature conservation strategy. Svensk Botanisk Tidskrift 84:355-368. ANTENEN, S., M. JORDAN, K. MOTIVANS, J. B. WASHA, AND R. ZAREMBA. 1994. Hempstead Plains fire man- agement plan, Nassau County, Long Island, New York. Report to Long Island Chapter of The Nature Conservancy, Cold Spring Harbor, NY. ASKINS, R. A. 1993. Population trends in grassland, shrubland, and forest birds in eastern North Ameri- ca. Pp. 1-34 in D. M. Power (editor). Current or- nithology, vol. 11. Plenum Publishing, New York, NY. AUDUBON, J. J. 1831-1849. Ornithological biography, or an account of the birds of the United States of America. Vol. 1, J. Dodson, Philadelphia, PA; vols. 2-5, A. and C. Black, Edinburgh, Scotland. BEINTEMA, A. J. 1988. Conservation of grassland bird communities in the Netherlands. Pp. 105-111 in P. D. Goriup (editor). Ecology and conservation of grassland birds. ICBP Technical Publication no. 7. International Council for Bird Preservation, Cam- bridge, U.K. BEVIER, L. R. (EDITOR). 1994. The atlas of breeding birds of Connecticut. State Geological and Natural History Survey of Connecticut Bulletin 113. BOLLINGER, E. K., P. B. BOLLINGER, AND T A. GAVIN. 1990. Effects of hay-cropping on eastern popula- tions of the Bobolink. Wildlife Society Bulletin 18: 142-150. HISTORY OF EASTERN GRASSLAND BIRDSAskins 69 BOLLINGER, E. K., AND T. A. GAVIN. 1992. Eastern Bobolink populations: ecology and conservation in an agricultural landscape. Pp. 497-506 in J. M. Ha- gan III and D. W. Johnston (editors). Ecology and conservation of neotropical migrant landbirds. Smithsonian Institution Press, Washington, D.C. BORMANN, E H., AND G. E. LIKENS. 1979. Catastrophic disturbance and the steady state in northern hard- wood forests. American Scientist 67:660-669. BOURNE, E.G. 1904. Narratives of the career of Her- nando de Soto. Vol. 1. A. S. Barnes and Co., New York, NY. BREWER, T. M. 1839. Wilson's American ornithology. Charles L. Cornish, New York, NY. BROOKS, M. 1938. The eastern Lark Sparrow in the upper Ohio valley. Cardinal 4:181 200. BULL, J. 1974. Birds of New York state. Doubleday, Garden City, NY. CAIN, S. A., M. NELSON, AND W. MCLEAN. 1937. An- dropogonetum Hempsteadi: a Long Island grassland vegetation type. American Midland Naturalist 18: 334-350. COLES, J. m., AND B. J. ORME. 1983. Homo sapiens or Castor fiber? Antiquity 57:95-102. CONRAD, H. S. 1935. The plant associations of central Long Island: a study in descriptive plant sociology. American Midland Naturalist 16:433-516. CRONON, W. 1983. Changes in the land: Indians, col- onists, and the ecology of New England. Hill and Wang, New York, NY. CROSBY, A. W., JR. 1972. The Columbian exchange: biological and cultural consequences of 1492. Greenwood Press, Westport, CT CROSBY, A. W., JR. 1986. Ecological imperialism: the biological expansion of Europe, 900-1900. Cam- bridge University Press, New York, NY. CROSSMAN, T. I. 1989. Habitat use of Grasshopper and Savannah sparrows at Bradley International Airport and management recommendations. M.S. thesis. University of Connecticut, Storrs, CT DAY, G. m. 1953. The Indian as an ecological factor in the northeastern forest. Ecology 34:329-346. DEGRAAF, R. M., AND D. D. RUDIS. 1986. New En- gland wildlife: habitat, natural history, and distri- bution. USDA Forest Service Gen. Tech. Rep. NE- 108. USDA Forest Service, Northeastern Forest Ex- periment Station, Broomall, PA. DEGRAAF, R. M., V. E. SCOTt, R. H. HAMRE, L. ERNST, AND S. H. ANDERSON. 1991. Forest and rangeland birds of the United States: natural history and habitat use. Agricultural handbook 688, USDA Forest Ser- vice, Washington, D.C. DENEVAN, W. m. 1992. The pristine myth: the land- scape of the Americas in 1492. Annals of the As- sociation of American Geographers 82:369-385. DOOLITTLE, W. E. 1992. Agriculture in North America on the eve of contact: a reassessment. Annals of the Association of American Geographers 82:386-401. DUBLIN, H. T., A. R. E. SINCLAIR, AND J. MCGLADE. 1990. Elephants and fire as causes of multiple stable states in the Serengeti-Mara woodlands. Journal of Animal Ecology 59:1147 1164. DUNN, C. P., E STEARNS, G. R. GUNTENSPERGEN, AND D. m. SHARPE. 1993. Ecological benefits of the Con- servation Reserve Program. Conservation Biology 7:132 139. DUNWIDDIE, P. W. 1989. Forest and heath: the shaping of vegetation on Nantucket Island. Journal of Forest History 33:126-133. ELLIOTT, J. J. 1968. Passerculus princeps (Maynard). Ipswich Sparrow. Pp. 657-675 in A. C. Bent (edi- tor). Life histories of North American grosbeaks, buntings, towhees, finches, sparrows, and allies. U.S. National Museum Bulletin 237, pt. 2. ELLS, S. F. 1995. Breeding Henslow's Sparrows in Lin- coln, Massachusetts, 1994. Bird Observer 23:113 115. FEDUCCIA, A. 1985. Catesby's birds of colonial Amer- ica. University of North Carolina Press, Chapel Hill, NC. FORBUSH, E. H. 1925. Birds of Massachusetts and other New England states. Pt. 1. Massachusetts Depart- ment of Agriculture, Boston, MA. FORBUSH, E. H. 1927. Birds of Massachusetts and other New England states. Pt. 2. Massachusetts Depart- ment of Agriculture, Bostom MA. GLEASON, H. A., AND A. CRONQUIST. 1991. Manual of vascular plants of northeastern United States and ad- jacent Canada. 2d ed. New York Botanical Garden, New York, NY. GRISCOM, L. 1949. The birds of Concord. Harvard Uni- versity Press, Cambridge, MA. GROSS, A. O. 1932. Heath Hen. Pp. 264-280 in A. C. Bent (editor). Life histories of North American gal- linaceous birds. U.S. National Museum Bulletin 162. GUlLDAY, J. E. 1962. The Pleistocene local fauna of the Natural Chimneys, Augusta County, Virginia. Annals of the Carnegie Museum 36:87 122. GUlLDAY, J. E., 19. S. MARTIN, AND A.D. MCGRADY. 1964. New Paris No. 4: a Pleistocene cave deposit on Bedford County, Pennsylvania. Bulletin of the National Speleological Society 26:121-194. HAMMESS, J. E. 1992. The shapes of adaptation: his- torical ecology of anthropogenic landscapes in the southeastern United States. Landscape Ecology 7: 121-135. HARPER, R. M. 1911. The Hempstead Plains: a natural prairie on Long Island. Bulletin of the American Geographic Society 43:351-360. HART, J. E 1968. Loss and abandonment of cleared land in the eastern United States. Annals of the As- sociation of American Geographers 58:417 440. HERKERT, J. R. 1991. Prairie birds of Illinois: popula- tion response to two centuries of habitat change. Il- linois Natural History Survey Bulletin 34:393-399. HERKERT, J. R. 1994. The effect of habitat fragmen- tation on midwestern grassland bird communities. Ecological Applications 4:461-471. HOSMER, J. K. 1959. Winthrop's journal, "History of New England", 1630-1649. Barnes and Noble, New York, NY. HOWARD, R. J., AND J. S. LARSON. 1985. A stream hab- itat classification system for beaver. Journal of Wild- life Management 49:19-25. HURLEY, R. J., AND E. C. FRANKS. 1976. Changes in the breeding ranges of two grassland birds. Auk 93: 108 115. JOHNSGARD, P. A. 1983. The grouse of the world. Uni- versity of Nebraska Press, Lincoln, NE. 70 STUDIES IN AVIAN BIOLOGY NO. 19 JOHNSON, D. H., AND L. D. IGL. 1995. Contributions of the Conservation Reserve Program to populations of breeding birds in North Dakota. Wilson Bulletin 107:709-718. JOHNSON, D. H., AND M.D. SCHWARTZ. 1993. The Con- servation Reserve Program and grassland birds. Conservation Biology 7:934-937. KULIKOFF, m. 1986. Tobacco and slaves: the develop- ment of southern cultures in the Chesapeake, 1680- 1800. University of North Carolina Press, Chapel Hill, NC. KURTIN, g., AND E. ANDERSON. 1980. Pleistocene mammals of North America. Columbia University Press, New York, NY. CANYON, W. E. 1956. Ecological aspects of the sym- pattic distribution of meadowlarks in the north-cen- tral states. Ecology 37:98-108. LENT, R. m., AND T. S. LITWIN. 1989. Bird-habitat re- lationships as a guide to ecologically-based manage- ment at Floyd Bennett Field, Gateway National Rec- reation Area. Part 1. Baseline study. Seatuck Re- search Program, Cornell Laboratory of Ornithology, Islip, NY. LINDHOLT, P. J. 1988. John Josselyn, colonial traveler: a critical edition of account of two voyages to New- England. University Press of New England, Hano- ver, NH. LORIMER, C. O. 1977. The presettlement forest and nat- ural disturbance cycle of northeastern Maine. Ecol- ogy 58:139-148. MARTIN, P.S., AND R. G. KLEIN (EDITORS). 1984. Qua- ternary extinctions: a prehistoric revolution. Univer- sity of Arizona Press, Tucson, AZ. MAYFIELD, H. E. 1988. Changes in bird life at the west- ern end of Lake Erie. Part 1. American Birds 42: 393-398. MEHRHOFF, L. J. 1997. Thoughts on the biogeography of grassland plants in New England. Pp. 15-23 in P. D. Vickery and E W. Dunwiddie (editors). Grass- lands of northeastern North America: ecology and conservation of native and agricultural landscapes. Massachusetts Audubon Society, Lincoln, MA. MELVIN, S. 1994. Military bases provide habitat for rare grassland birds. Massachusetts Division of Fish- edes and Wildlife Natural Heritage News 4:3. NAIMAN, R. J., J. M. MELILLO, AND J. E. HOBBIE. 1988. Ecosystem alteration of boreal forest streams by bea- ver (Castor canadensis). Ecology 67:1254-1269. OLMSTED, C. E. 1937. Vegetation of certain sand plains of Connecticut. Botanical Gazette 99:209-300. PARMALEE, P W., AND W. E. KLIPPEL. 1982. Evidence of a boreal avifauna in middle Tennessee during the late Pleistocene. Auk 99:365-368. PAlTERSON, W. h., III, AND K. E. SASSAMAN. 1988. Indian fires in the prehistory of New England. Pp. 107-135 in G. E Nichols (editor). Holocene human ecology in northeastern North America. Plenum Publishing, New York, NY. PETERJOHN, B. 1994. The North American Breeding Bird Survey. Birding 26:386-398. PUCHKOV, P. V. 1992. Uncompensated Wuermian ex- tinctions. Part 2. Transformation of the environment by giant herbivores. Vestnik Zoologii 0(1):58-66. REMICLARD, M. M., G. K. GRUENDLING, AND D. J. BOG- UCKI. 1987. Disturbance by beaver (Castorcanaden- sis Kuhl) and increased landscape heterogeneity. Pp. 103-122 in M. G. Turner (editor). Landscape het- erogeneity and disturbance. Springer-Verlag, New York, NY. RODENHOUSE, N. C., C. B. BEST, R. J. O'CONNOR, AND E. K. BOLLINGER. 1995. Effects of agricultural prac- tices and farmland structures. Pp. 269-293 in T. E. Martin and D. M. Finch (editors). Ecology and man- agement of neotropical migratory birds: a synthesis and review of critical issues. Oxford University Press, New York, NY. RUNKLE, J. R. 1990. Gap dynamics in an Ohio Acer- Fagus forest and speculations on the geography of disturbance. Canadian Journal of Forest Research 20:632-641. RUSSELL, E. W. B. 1983. Indian-set fires in the forests of the northeastern United States. Ecology 64:78- 88. RUSSELL, H. S. 1980. Indian New England before the Mayflower. University Press of New England, Han- over, NH. SAUER, J. R., g. G. PETERJOHN, S. SCHWARTZ, AND J. E. HINES. 1995. The grassland bird home page. Ver. 95.1: www.mbr. nbs.gov/bbs/grass/grass.htm. U.S. Geological Survey, Patuxent Wildlife Research Cen- ter, Laurel, MD. SEISCHAB, E K., AND D. ORWIG. 1991. Catastrophic dis- turbances in the presettlement forests of western New York. Bulletin of the Torrey Botanical Club 114:330-335. SHAFFER, L. N. 1992. Native Americans before 1492: the moundbuilding centers of the eastern woodlands. M. E. Sharpe, Armonk, NY. SCCAMA, T G. 1971. Presettlement and present forest vegetation in northern Vermont with special refer- ence to Chittenden County. American Midland Nat- uralist 85:153-172. SMITH, B. D. 1989. Origins of agriculture in eastern North America. Science 246:1566-1571. SMITH, g. D. 1995. The origins of agriculture in the Americas. Evolutionary Anthropology 3:174-184. SMITH, W. P. 1968. Eastern Henslow's Sparrow. Pp. 776-778 in A. C. Bent (editor). Life histories of North American grosbeaks, buntings, towbees, finches, sparrows, and allies. U.S. National Museum Bulletin 237, pt. 2. STALTER, R., AND E. E. CAMONT. 1987. Vegetation of the Hempstead Plains, Mitchell Field, Long Island, New York. Bulletin of the Torrey Botanical Club 114:330-335. STOBO, W. r., AND I. A. MCLAREN. 1971. Late winter distribution of the Ipswich Sparrow. American Birds 25:941-944. STOBO, W. T, AND I. A. MCLAREN. 1975. The Ipswich Sparrow. Nova Scotian Institute of Science, Halifax, NS. SVENSON, H. K. 1936. The early vegetation of Long Island. Brooklyn Botanic Garden Journal 25:207- 227. TAYLOR, N. 1923. The vegetation of Long Island, part 1. The vegetation of Montauk: a study of grassland and forest. Brooklyn Botanic Garden Memoirs vol. 2. THOMAS, E. S. 195l. Distribution of Ohio animals. Ohio Journal of Science 51:153-167. VAUGHAN, m. T. 1977. New England's prospect (by HISTORY OF EASTERN GRASSLAND BIRDSAskins 71 William Wood). University of Massachusetts Press, Amherst, MA. VEIT, R. R., AND W. R. PETERSEN. 1993. Birds of Mas- sachusetts. Massachusetts Audubon Society, Lin- coln, MA. VICKERY, P. D. 1992. A regional analysis of endan- gered, threatened, and special concern birds in the northeastern United States. Transactions of the Northeast Section of the Wildlife Society 48:1-10. VICKERY, P. D., M. L. HUNTER, JR., AND S. M. MELVIN. 1994. Effects of habitat area on the distribution of grassland birds in Maine. Conservation Biology 8: 1087-1097. WEBB, T., III. 1988. Eastern North America. Pp. 385 414 in B. Huntley and T. Webb III (editors). Vege- tation history. Kluwer Academic Publishers, Hingh- am, MA. WHEELWRIGHT, N. T., AND J. D. RISING. 1993. Savan- nah Sparrow (Passerculus sandwichensis). In A. Poole and F. Gill (editors). The birds of North Amer- ica, no. 45. Academy of Natural Sciences, Philadel- phi& PA, and American Ornithologists' Union, Washington, D.C. WHITCOMB, R. F. 1987. North American forests and grassland: biotic conservation. Pp. 163-176 in D. A. Saunders, G. W. Arnold, A. A. Burbidge, and A. J. M. Hopkins (editors). Nature conservation: the role of remnants of native vegetation. Surrey Beatty & Sons, Chipping Norton, New South Wales, Austra- lia. WHITMORE, R. C., AND G. A. HALL. 1978. The response of passerine species to a new resource: reclaimed surface mines in West Virginia. American Birds 32: 6-9. WHITNEY, G. G. 1994. From coastal wilderness to fruit- ed plain. Cambridge University Press, New York, NY. WILLIAMS, R. 1963. The complete writings of Roger Williams. Vol. 2. Russell and Russell, New York, NY. WINNE, J. C. 1988. History of vegetation and fire on the Pineo Ridge blueberry barrens in Washington County, Maine. M.S. thesis. University of Maine, Orono, ME. Studies in Avian Biology No. 19:72-80, 1999. GRASSLAND BIRD CONSERVATION IN NORTHEASTERN NORTH AMERICA JEFFREY V. WELLS AND KENNETH V. ROSENBERG Abstract. As a first step in the development of a conservation plan for grassland birds in the north- eastern United States, we prioritized species based on the percent of estimated total breeding popu- lation in states and provinces throughout North America. As expected, most species had only a small percent of their total breeding population in the Northeast. We estimated that 82 percent of all Savan- nah Sparrows (Passerculus sandwichensis), 47 percent of Vesper Sparrows (Pooecetes gramineus), and 37 percent of Bobolinks (Dolichonyx oryzivoroux) breed in Canada. An estimated 60 percent of North American Grasshopper Sparrows (Ammodramus ,'avannarum) breed in Kansas, Nebraska, North Dakota, and South Dakota. The grassland species we consider to be most at risk, Henslow's Sparrow (A. henslowii), has a relatively restricted breeding range with most of the population (more than 50 percent) in Ohio, Michigan, and Wisconsin but with a substantial percent (more than 20 percent) in the Northeast as well. This highlights the responsibility of different regions to the global, long-term persistence of species. A species-level analysis, however, does not consider regional genetic variability that may include taxonomic recognition below the species level. We considered this factor by repeating the above process for each subspecies of eastern grassland bird. We found that 100 percent of Eastern Henslow's Sparrows (A. h. susurrans) breed in the northeastern United States. Similarly, the Northeast supports 12 percent of the breeding Eastern Grasshopper Sparrows (A. s. pratensis) but only 4 percent of the total breeding population (all subspecies combined). This perspective counters the suggestion that the northeastern United States is unimportant for grassland birds. A regional grassland bird con- servation plan should include (1) standardized inventory and monitoring, particularly for Henslow's Sparrow; (2) identification of key nesting sites supporting high diversity and abundance of grassland birds and development of plans for management or acquisition of these sites where appropriate; (3) completion of a preliminary population viability analysis for the least abundant species to assess the relative importance of different sites; and (4) development of guidelines for how private landowners and public land managers can manage grasslands to benefit grassland birds. CONSERVACION DE AVES DE PASTIZAL EN EL NORESTE DE AMI2RICA DEL NORTE Sinopsis. Como primer paso en el desarrollo de un plan de conservaci6n para aves de pastizal en el noreste de los Estados Unidos, pusimos en orden de las especies a partir de su porcentaje de la tasa total de la poblaci6n reproductiva en estados y provincias en toda Amdrica del Norte. Como esperamos, la mayorfa de las especies tenfa s61o un pequefio porcentaje del total de su poblaci6n reproductiva en el noreste. Estimamos en un 82 por ciento los Gorriones Sabaneros (Passerc'ulus sandwichensis), en un 47 pot ciento los Gorriones Coliblancos (Pooecetes gramineus) yen un 37 pot ciento los Tordos Arroceros (Dolichonyx oryzivorous) que se reproducen en Canada. Un ndmero aproximado al 60 pot ciento de los Gorriones Chapulines (Ammodramus savannarum) de Amdrica del Norte se reproducen en Kansas, Nebraska, Dakota del Norte y Dakota del Sur. La especie de pastizal que consideramos con mas riesgo, el Gorri6n de Henslow (A. henslowii), tiene una extensi6n reproductiva relativamente reducida, con la mayorfa de la poblaci6n (mas de un 50 pot ciento) en Ohio, Michigan y Wisconsin, pero tambidn con una presencia importante (mas de un 20 por ciento) en el noreste. Esto subraya la responsabilidad de varias regiones ante la sobrevivencia mundial de especies a largo plazo. Sin em- bargo, un analisis a nivel de especie no considera la variaci6n gendtica regional que puede incluir un reconocimiento taxon6mico menor al de especie. Consideramos este elemento al repetir el proceso anterior para cada subespecie de ave de pastizal del este. Descubrimos que 100 por ciento de los Gorriones de Henslow del Este (A. h. susurrans) se reproducen en el noreste de los Estados Unidos. De igual manera el noreste mantiene un 12 pot ciento de los Gorriones Chapulines del Noreste (A. x. pratensis) en reproducci6n, pero tiene s61o un 4 por ciento de la poblaci6n reproductiva total (todas las subespecies combinadas). Esta perspectiva contradice la idea sugerida de que el noreste de los Estados Unidos no es importante para las aves de pastizal. Un plan de conservaci6n regional para aves de pastizal debe incluir (1) inventario y medici6n uniformes, especialmente para los Gorriones de Henslow; (2) identificaci6n de lugares de reproducci6n que mantienen mucha diversidad y abun- dancia de aves de pastizal, y desarrollo de planes para la administraci6n o adquisici6n de estos lugares donde sea apropiado; (3) realizaci6n de un analisis preliminar de la viabilidad de la poblaci6n para las especies menos abundantes, con el fin de determinar la importancia relativa de varios lugares; y (4) desarrollo de sugerencias para duerios particulares y administradores de terrenos pdblicos en re- laci6n al manejo de pastizales para beneficiar alas aves de pastizal. Key Words: bird conservation; conservation priorities; grassland birds; Henslow's Sparrow; north- eastern United States; subspecies. 72 GRASSLAND BIRD CONSERVATION--Wells and Rosenberg 73 Assessing regional species and habitat priorities is critical to the conservation planning process and is central to the North American songbird management plan under development through the Partners In Flight coalition (Finch and Stan- gel 1993). Bird species may be prioritized for conservation consideration based on several global and regional criteria (Hunter et al. 1993, Carter et al. in press) or on the proportion of their total breeding population supported in a particular region (Rosenberg and Wells 1995, in press). This latter approach, which provides a global perspective on the relative responsibility of each region to the overall conservation of each species, has been applied to all neotropical migratory landbirds breeding in the northeastern United States (Rosenberg and Wells 1995, in press). That analysis highlighted a potential con- flict between long-term planning for species with high proportions of their total breeding population in the region and local concern for species showing significant population declines. Bird species of grasslands and other early-suc- cessional habitats are at the center of this di- chotomy. In the northeastern United States, grassland and shrubland birds have been identified as the habitat-community groups showing the most widespread and persistent declines in abundance (Witham and Hunter 1992, Askins 1993). Ini- tially, however, these declines were treated as a matter of little importance because of a percep- tion that there had been little grassland and shrubland habitat in the Northeast prior to Eu- ropean colonization. More recently, careful re- views of the available evidence have shown that open grassland and shrubland habitats composed a significant proportion of the pre-European ar- rival landscape (Marks 1983, Askins 1993). One of the most compelling pieces of evidence for the existence--and rapid destruction-of such habitats is the evolution of a distinct taxonomic form of the Greater Prairie Chicken (Tympanu- chus cupido) in the eastern United States: the Heath Hen (T. c. cupido), which became extinct on Martha's Vineyard, Massachusetts, in 1932 (AOU 1957). Concern for the conservation of declining grassland birds has resulted in legal designation of one or more species on virtually every north- eastern state's threatened and endangered spe- cies list (Vickery 1992). For example, Upland Sandpiper (Bartramia longicauda) and Grass- hopper Sparrow (Ammodramus savannarum) are listed by at least seven states, and Vesper Spar- row (Pooecetes gramineus) and Henslow's Spar- row (Ammodramus henslowii) are listed by at least five states. In addition, Henslow's Sparrow was recently assessed as a potential candidate for listing by the U.S. Fish and Wildlife Service (USFWS; Pruitt 1996). In this paper we assess the conservation status of grassland bird species in the northeastern United States from a continental perspective. We first consider the importance of the Northeast to the total breeding population of each species; we then compare population trends in the Northeast with those in other regions of the United States and Canada. Finally, because of the potential distinctiveness of eastern populations of certain grassland birds, we discuss how consideration of subspecies affects our current view of grassland bird conservation in the Northeast. METHODS We considered the following eight species of eastern grassland birds: Upland Sandpiper, Horned Lark (Er- emophila alpestris), Savannah Sparrow (Pasxerculux sandwichensis), Grasshopper Sparrow, Henslow's Sparrow, Vesper Sparrow, Bobolink (Dolichonyx o- zivorus), and Eastern Meadowlark (Sturnella magna). As a measure of potential genetic diversity, we con- sidered the subspecies described in the 1957 American Ornithologists' Union's check-list (AOU 1957), which was the most recent reference that systematically re- viewed subspecies in all North American species north of Mexico. We estimated the proportion of the total population of each species breeding in each U.S. state and Ca- nadian province using the following procedure. First, using ranges described in Peterson 1980 and 1990, we estimated the proportion of each state occupied and then multiplied that proportion by the area of each state. We then multiplied the range area occupied in each state and province by the relative abundance cal- culated for that area based on USFWS Breeding Bird Survey (BBS) data from 1966 to 1994 (Sauer et al. 1996). These state and province values were then summed to get an index of the total breeding popula- tion size. Note that this index represents a species' total breeding range size as well as the species' relative abundance across its breeding range; it is not, however, an accurate estimate of the total number of individuals in the species' breeding population. For each species we then divided the value for each state or province by the total population size index to get the estimated percent of the total population breed- ing in each state and province. These percentages were then summed to give an estimate of the percent of the total population breeding in the Northeast (here de- fined as USFWS Region-5 [Maine, Vermont, New Hampshire, Massachusetts, Connecticut, Rhode Island, New York, New Jersey, Pennsylvania, Maryland, Del- aware, Virginia, and West Virginia]), other USFWS regions, and Canada (Table 1). For species with sub- species described in the American Ornithologists' Union's 1957 check-list, we also determined what pro- portion of the relevant subspecies breed in the North- east. Finally, to assess geographic patterns in popula- tion trends, we compared percent of population in USFWS regions and Canada against the 1966-1994 BBS trend (Sauer et al. 1996). 74 STUDIES IN AVIAN BIOLOGY NO. 19 TABLE 1. PERCENT OF TOTAL BREEDING POPULATION OF EIGHT GRASSLAND BIRD SPECIES IN THE NORTHEASTERN UNITED STATES USFVS REGION-5), OTHER USFYVS REGIONS, AND CANADA Species Region NE SE MW GP SW CAN Upland Sandpiper 0.3 a 0.0 5.6 85.0 0.0 9.1 Horned Lark 0.1 0.1 6.0 47.5 13.7 19.4 Savannah Sparrow 1.6 0.0 9.1 5.1 0.0 81.7 Grasshopper Sparrow 3.6 2.2 19.8 70.0 4.1 0.2 Henslow's Sparrow 21.3 0.0 b 78.6 0.0 0.0 c 0.0 Vesper Sparrow 0.6 0.0 15.1 31.7 0.0 46.6 Bobolink 13.7 0.0 32.6 17.0 0.0 36.5 Eastern Meadowlark 5.4 31.9 23.3 3.6 35.1 0.6 Note: NE - Northeast (Connecticut, Delaware, Maine, Mm'yland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island. Vermont, Virginia. West Virginia): SE = Southeast (Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi. North Carolina, South Carolina. Tennessee); MW - Midwest (Illinois, Indiana, Iowa, Michigan, Minnesota, Missouri. Ohio. Wisconsin); GP - Great Plains (Colorado, Kansas. Montana, Nebraska, North Dakota, South Dakota, Utah, Wyoming); SW = Southwest (Arizona, New Mexico. Oklahoma, Texas); CAN = Canada. a Estimate may be low because BBS abundance was not calculated for most northeastern states. b Small, recently discovered population in North Carolina. c Large, recently discovered population in Oklahoma. RESULTS AND CONSERVATION IMPLICATIONS Of the eight species considered at the species level, Henslow's Sparrow had the highest per- cent of its breeding population in the Northeast, with 21.3%, followed by Bobolink with 13.7% and Eastern Meadowlark with 5.4% (Table 1). None of the other species had more than 5% of their estimated population in the region. Most species had a high percent of their total breeding population in a single region. For example, 85% of Upland Sandpipers and 70% of Grasshopper Sparrows were estimated to breed in the Great Plains region (USFWS Region-6). Similarly, the Midwest (USFWS Region-3) supported an esti- mated 78.6% of the world's breeding Henslow's Sparrows. Canada had 81.7% of breeding Sa- vannah Sparrows and 46.6% of breeding Vesper Sparrows (Table 1). When we considered subspecies, the impor- tance of the Northeast to certain grassland birds was highlighted (Table 2). For example, al- though the Northeast held 1.6% of all breeding Savannah Sparrows, it supported 55% of the eastern subspecies, P.s. savanna. Similarly, the Northeast held 3.6% of breeding Grasshopper Sparrows but 11.5% of the eastern subspecies, A. s. pratensis. The most dramatic example was Henslow's Sparrow, for which the Northeast supported 21.3% of the entire breeding popula- tion but 100% of the remaining populations of the described eastern subspecies, A. h. susur- tans. All eight grassland species under consider- ation, except Upland Sandpiper, showed a sig- nificant negative trend based on continental BBS data (Sauer et al. 1996). Henslow's Sparrow, with the lowest population size index, also had the greatest negative trend (Fig. 1). If the percent of total breeding population in each region is compared with the trends for those regions, a geographic pattern of the con- servation status of each species emerges (Fig. 2). These patterns show where conservation efforts will have the greatest overall influence on each species and help put the conservation status of grassland birds in the Northeast in a larger per- spective. We summarize these patterns in the following species accounts. TABLE 2. PERCENT OF TOTAL BREEDING POPULATION OF NINE GRASSLAND BIRD SPECIES IN THE NORTHEASTERN UNITED STATES, CONSIDERING GEOGRAPHICALLY VARIABLE SUBSPECIES Species Eastern subspecies All subspecies Heath Hen 100.0 (T. c. cupido) ? Upland Sandpiper no subspecies 0.3 Horned Lark 2.1 (E. a. praticola) O. 1 Savannah Sparrow 55.2 (P.s. savanna) 1.6 Grasshopper Sparrow 11.5 (A. s. pratensis) 3.6 Henslow's Sparrow 100.0 (A. h. susurrans) 21.3 Vesper Sparrow 3.1 (P. g. gramineus) 0.6 Bobolink no subspecies 13.7 Eastern Meadowlark 12.9 (S. m. magna) 5.4 GRASSLAND BIRD CONSERVATION--Wells and Rosenberg 75 Upland Vesper Sandpiper Sparrow Savannah ß Sparrow ßß ß ßHorned Bobolink Lark ß Eastern ß Meadowlark Grasshopper Sparrow ß Henslow's Sparrow .01 .1 1 10 100 Total breeding population index (x 1) FIGURE 1. Total breeding population index versus 28-year population trend in eight species of northeast- ern grassland birds. Savannah Sparrow Vesper' Sparro Horne Lark Upland 7 ' -  Grasshopper Sparrow OW'S parrow Eastem ark FIGURE 3. Schematic showing general areas of highest abundance for eight species of northeastern grassland birds. UPLAND SANDPIPER The largest breeding populations of Upland Sandpipers occur in the Great Plains states of South Dakota (34%), North Dakota (19%), Ne- braska (15%), and Kansas (11%; Fig. 3). In the Great Plains, the species shows an increasing population trend, indicating that the overall pop- ulation is doing well where the species is most abundant (Fig. 2). The northeastern United States supports a small percentage of the total breeding population of this species, and no sub- species have been described. The long-term pop- ulation trend for the region appears to be stable, although there is clear evidence for declines in Upland Sandpipers in portions of the Northeast (Andrle and Carroll 1988, Brauning 1992, Foss 1994). Conservation of Upland Sandpipers in the Northeast will have little impact on the glob- al population status, but as this species is an in- dicator of intact and diverse grassland commu- Upland Sandpiper Horned Lark 3 SINCA N ß Grasshopper Sparrow Henslow's Sparrow o Savannah Sparrow 0 ßGP CAN Vesper Sparrow Bobolink Eastern Meadowlark - Percent of total breeding population FIGURE 2. Percent of total breeding population for U.S. Fish and Wildlife Service (USFWS) regions and Canada (see Table 1 for abbreviations) versus 28-year population trend for eight species of northeastern grassland birds. 76 STUDIES IN AVIAN BIOLOGY NO. 19 nities, local conservation efforts undoubtedly will continue. HORNED LARK The largest percentage of the total Horned Lark breeding population occurs in the western Great Plains states and provinces, including Montana (13%), Saskatchewan (12%), Colorado (10%), and North Dakota (7%; Fig. 3). Percent- ages for this species are probably exaggerated somewhat because far-northern populations in Canada and Alaska are not censused by the BBS and therefore are not included in our analyses. The northeastern United States currently sup- ports less than 1% of the total breeding popu- lation of this species and only 2% of the eastern subspecies, E. a. praticola. Horned Lark popu- lations are stable in areas where they are most abundant but are decreasing over much of the remainder of the breeding range (Fig. 2). Partic- ularly large declines are seen in northeastern states such as New York (-5.7%) and Pennsyl- vania (-9.7%) as well as at the southern limit of the eastern breeding range in Kentucky (-9.1%) and Tennessee (-7.9%). By itself the Horned Lark may not be considered a high pri- ority species in the region, as conservation ef- forts will have minimal impact on the species as a whole. To conserve the rich genetic diversity evident in this highly variable species, however, a continentwide strategy for stabilizing popula- tions is desirable. SAVANNAH SPARROW The largest proportion of all Savannah Spar- rows breed in the Canadian provinces from Al- berta to Quebec (Fig. 3). The northeastern Unit- ed States supports less than 2% of the total breeding population of this species but 55% of the eastern subspecies, P.s. savanna. Savannah Sparrows are increasing in the western Canadian provinces but decreasing in the eastern provinc- es (except Newfoundland), resulting in a stable trend overall in Canada (Fig. 2). Steepest de- clines are occurring in the Northeast, including most of the range of the eastern subspecies. Be- cause the northeastern United States supports a large percentage of the population of this de- clining subspecies, its status should be elevated to that of moderate conservation concern in the Northeast. Fortunately, both the global popula- tion and local densities of Savannah Sparrows are very high, and the species is more general- ized in its habitat selection than are most other grassland species in the region (Wheelwright and Rising 1993). GRASSHOPPER SPARROW Grasshopper Sparrow breeding populations are largest in the Great Plains states of Kansas (19%), South Dakota (17%), North Dakota (13%), and Nebraska (12%; Fig. 3). The north- eastern United States supports only 3.6% of the total breeding population of this species but 11.5% of the eastern subspecies, A. s. pratensis. Population declines are evident throughout the range of this species, except in the Southwest (Fig. 2). Declines are particularly steep in some northeastern states, especially New York (- 10.2%), New Jersey (- 10.2%), and West Vir- ginia (-12.5%), but an increasing population trend in the Piedmont of Virginia (+4.9%) caused the regionwide population decline to be less steep. Because Grasshopper Sparrows are in need of a continentwide conservation strategy, we think efforts to stabilize or enhance local populations in the Northeast are justified. HENSLOW' S SPARROW Henslow's Sparrow has the smallest total breeding range of any of the species we consid- ered, with virtually the entire global population concentrated in the Midwest and Northeast (Fig. 3). Henslow's Sparrow also shows the steepest declines of any grassland bird, both in the Northeast and throughout its range (Fig. 2). Per- centages for these regions may be exaggerated because this species is too rare to be detected on BBS routes in certain states, and these are there- fore not included in our trend analysis. For ex- ample, fairly large numbers have been found breeding in northeastern Oklahoma since the early 1990s, though the species has only occa- sionally been registered on BBS routes in that state (Pruitt 1996). We consider this species to rank first in con- servation priority among grassland birds in the Northeast, and efforts to stabilize or enhance re- gional populations should be made in coordi- nation with an overall strategy to protect this species. VESPER SPARROW The Vesper Sparrow's breeding distribution overlaps broadly with that of the Horned Lark (Fig. 3), with the largest percentage of the Ves- per Sparrow's total breeding population in Al- berta (18%), Montana (16%), and Saskatchewan (15%). Populations in these areas, and in the re- mainder of the species' western range, are stable or increasing (Fig. 2). In contrast, small popu- lations in the northeastern states are declining precipitously. These populations represent 3% of the eastern subspecies, P. g. gramineus, which is declining throughout its range. As with Horned Lark and other widespread species, con- GRASSLAND BIRD CONSERVATION--Wells and Rosenberg 77 servation efforts in the Northeast will have little impact on the total breeding population of Ves- per Sparrows; however, concern for the eastern subspecies, if warranted, could elevate this spe- cies' conservation status in the region. BOBOLINK Unlike most of the other species considered here, the Bobolink is fairly evenly distributed over a broad area; no single state or province supports more than about 10% of the total breed- ing population. The largest numbers occur in a belt from North Dakota and Minnesota east to southern Quebec (Fig. 3). Unlike populations of the other species we considered, Bobolink pop- ulations have been stable in most northeastern states (Fig. 2); the largest declines have occurred in the Midwest (especially Indiana [-8.7%] and Illinois [- 10.7 % ]) and in Quebec (- 4.1%). The Northeast has a relatively large percent of the total breeding population (13.7%); only Hen- slow's Sparrow has more of its total breeding population in this region (21.3%). The lack of declining populations and lack of geographic variation evident in this species, however, sug- gest that the Bobolink is only a moderate con- servation priority in the region. EASTERN MEADOWLARK The largest breeding populations of Eastern Meadowlarks occur in the southern Great Plains states, from Missouri to Texas (Fig. 3). Popula- tions in these states show a weak negative pop- ulation trend (Fig. 2). Throughout the remainder of the species' range, declines are much steeper, with declines of more than 8% per year in some northeastern states. All states with significant de- clining trends are in the range of the eastern sub- species, S. m. magna. Conservation status of the Eastern Meadowlark in the Northeast is similar to that of the Grasshopper Sparrow, in that al- though regional efforts will have little impact on the global population, they may be coordinated with a rangewide strategy to conserve the spe- cies. Similarly, recognition that the eastern sub- species is particularly vulnerable elevates its conservation priority in the region. DISCUSSION Our analyses confirm that the northeastern United States does not support a large propor- tion of the total breeding population of most grassland species. Henslow's Sparrow is the only species for which a substantial proportion of the total breeding population is restricted to the Northeast. Clearly, Henslow's Sparrow is the highest priority grassland species in the region based on this criterion, and any regional conser- vation plan should focus on stabilizing or en- hancing populations of this species. Indeed, Henslow's Sparrow ranked first in regional con- cern, even when compared with all nongrassland birds (Rosenberg and Wells 1995, in press). Therefore, in a regional grassland bird initiative, states such as Pennsylvania and New York, with high proportions of the total breeding popula- tion, are most responsible for plans to protect and manage Henslow's Sparrows. Recent status assessments of this species recommended that comprehensive inventory and monitoring pro- grams be undertaken throughout the breeding range, especially since the ability to monitor Henslow's Sparrows through BBS data will be- come increasingly problematic as the species de- clines (Smith 1992, Pruitt 1996). Should other grassland species be given low priority for conservation in the Northeast? Even though conservation actions in this region will have little effect on the long-term continental persistence of these species, several factors ar- gue for continued concern for regional popula- tions. First, most of the declining grassland spe- cies are found in habitats that support unique assemblages of plants, invertebrates, and other nonavian vertebrates. For example, northern blazing star (Liatris scariosa var. novae-angliae) is a rare grassland perennial found only in the northeastern United States; it occurs in habitats that also support grassland birds (Vickery 1996). Additionally, for several geographically vari- able species, eastern populations represent de- scribed subspecies, and in nearly every case the eastern subspecies are exhibiting the most pre- cipitous declines. Even for widespread species such as Savannah Sparrow and Eastern Mead- owlark, significant genetic diversity may be rep- resented in the Northeast and is therefore worthy of protection. Changes in species-level taxonomy, reflecting modern knowledge of genetic variation, contin- ue to have profound effects on conservation pri- orities. A recent example is the elevation of Bicknell's Thrush (Catharus bicknelli) to full species (Ouellet 1993, AOU 1995), changing its status in the Northeast from a marginal popula- tion of the widespread Gray-cheeked Thrush (C. minimus) to being among the highest priority landbird species in the region (Rosenberg and Wells 1995, in press). Similarly, the recognition of Salt-marsh Sharp-tailed Sparrow (Ammodra- mus caudacutus) as a separate species (AOU 1995) has made it one of the Northeast's highest priorities in terms of importance of the region to the global population of the species. Important- ly, these examples serve to focus attention on the restricted habitats of these "new" species, in these cases stunted mountaintop forests and coastal salt marshes, respectively. Although 78 STUDIES IN AVIAN BIOLOGY NO. 19 modern studies of geographic variation are lack- ing for most North American bird species (Zink and Remsen 1986), we feel that assessing con- servation priorities based on even a dated as- sessment of morphological distinctiveness (i.e., the 1957 American Ornithologists' Union's check-list) is preferable to ignoring potential re- gional genetic diversity in northeastern grass- land species. BIOGEOGRAPHIC PERSPECTIVE ON GRASSLAND BIRD CONSERVATION Our analyses have identified areas of North America where the largest breeding populations of each grassland species are concentrated (Fig. 3). Each of these areas has a high responsibility for the overall conservation of a particular spe- cies or suite of species, and conservation efforts outside these areas should be coordinated with efforts in the core of each species' range. Inter- estingly, species that are similar in overall breeding distribution also tend to be similar in their local distribution and habitat affinities in the Northeast. For example, Vesper Sparrows and Horned Larks, which both reach their high- est abundance in the western Great Plains, have similar breeding distributions in the northeastern United States based on atlas-block occurrence (Rosenberg and Wells 1995). Both species also tend to occupy dry, sparsely vegetated sites. These distributions are different, however, from those of Bobolinks and Savannah Sparrows, which are also similar to each other in continen- tal and regional breeding distribution as well as in habitat preference. A third example is that of Upland Sandpipers and Grasshopper Sparrows, whose highest abundances are concentrated in the prairie states from North Dakota to Kansas. These two species require larger habitat areas than other species in the Northeast (Vickery et al. 1994), and they tend to share a similar dis- tribution throughout the region. None of these grassland-species clusters, however, was strongly tied to any geographic portions of the region, such as particular phys- iographic areas (Rosenberg and Wells 1995). This means there is no clearly defined physio- graphic region in the Northeast where manage- ment plans could be developed for grassland birds as a whole. Instead, what is required is a larger, regionwide initiative that would set dif- ferent goals for different species in each area. DEVELOPING A CONSERVATION PLAN FOR THE NORTHEAST The first step in developing a management plan for grassland species in the Northeast is to carry out a comprehensive inventory so we know where the species of interest occur and how many occur at each site. Fortunately, in New England many of the sites where grassland birds occur have been identified and surveyed (Jones and Vickery 1995), and monitoring ef- forts are well underway. Inventories are also be- ing undertaken in other northeastern states, in- cluding New York and Pennsylvania, through the National Audubon Society Important Bird Areas programs in coordination with the North- eastern Grassland Bird Working Group of Part- ners In Flight. Estimates of many other demographic param- eters for each population (fecundity, mortality, mean population growth rates, etc.) are also needed to understand the factors affecting abun- dance trends and to model extinction probabili- ties (Boyce 1992, Burgman et al. 1993). There are, however, first approximations of these pa- rmeters available from other studies (Wells 1995) that allow at least preliminary consider- ation of the importance of different sites to the extinction risks of grassland species. In Maine, a population viability analysis (PVA) that was carried out for Grasshopper Sparrows yielded useful management recommendations (Wells 1995). Carrying out a preliminary PVA for the New England grassland bird species of concern would help identify those sites of highest im- portance for the long-term persistence of grass- land species in the Northeast. These sites could then be targeted for action, whether it be acqui- sition, easements, management agreements, or other options. Another logical step in developing a regional plan for grassland birds would be to identify sites that support multiple grassland species and to see how many of these sites are protected. This is the basic concept of GAP analysis as applied to a limited bird community (Scott and Csuti 1991). For example, consider the number of sites in New England (excluding Vermont) where all of the following species breed: Upland Sandpiper, Vesper Sparrow, and Grasshopper Sparrow. There are perhaps six sites where all three species are known to breed (P. Vickery, pers. comm.). Only one of these is a protected wildlife preserve (Kennebunk Plains Wildlife Management Area, Maine). Four of the other sites are military or municipal airports, and one is a privately owned parcel. The airports could be managed quite easily and effectively for grassland birds with little added expense or modification of airport management plans, as has been shown at Westover Air Force Base in Massachusetts (Melvin 1994, Jones and Vickery 1995). Clearly, however, the long-term persis- tence of any grassland species in New England will require preservation and management at more than these six sites. GRASSLAND BIRD CONSERVATION--Wells and Rosenberg 79 For Henslow's Sparrow, which we have iden- tified as one of the species of most immediate concern in the Northeast, conservation plans are not fully developed. The first priority for devel- oping a plan is to find out where the birds occur and how many are at each site. Therefore, a re- gional cooperative effort, largely involving New York and Pennsylvania, must be undertaken to identify and survey sites for Henslow's Sparrow. The majority of the most important sites have been identified through the site inventory pro- grams of the National Audubon Society Impor- tant Bird Areas programs in New York and Pennsylvania. These sites, and others throughout New England, are being inventoried and moni- tored using a standardized methodology coor- dinated through the Northeastern Grassland Bird Working Group of Partners In Flight. One aspect of the conservation of all species of grassland birds in the Northeast is the impor- tance of military installations and of commercial and municipal airports. Many of the largest con- centrations of grassland birds occur at these sites throughout the region. In New England, coop- erative management between airport or site man- agers and wildlife managers has been successful in increasing grassland bird populations (Melvin 1994). The importance of these sites should be assessed for the entire region, and coordination of management efforts among sites (particularly those in the same organizations, i.e., naval bases, air-force bases, etc.) should be encouraged. Finally, we note that the long-term persistence of grassland bird species in the Northeast is un- likely if these species are restricted to a few iso- lated, publicly managed sites. Private landown- ers and public land managers must he provided with guidelines for management practices that are beneficial to grassland birds, and they must be encouraged to implement them. ACKNOWLEDGMENTS This paper is partly based on a report submitted to USFWS Region-5; we are grateful to D. Pence and the USFWS for funding this project. For providing BBS data and advice, we thank B. Peterjohn and J. Sauer. We benefited from discussions with members of the Northeast Grassland Bird Working Group, in particular P. D. Vickery and D. Brauning, who also provided helpful comments on the manuscript. LITERATURE CITED AMERICAN ORNITHOLOGISTS' UNION. 1957. Check-list of North American birds. 5th ed. American Orni- thologists' Union, Washington, D.C. AMERICAN ORNITHOLOGISTS' UNION. 1995. Fortieth sup- plement to the American Ornithologists' Union check-list of North American birds. Auk 112:819- 830. ANDRLE, R. E, AND J. R. CARROLL. 1988. The atlas of breeding birds in New York state. Cornell University Press, Ithaca, NY. ASKINS, R. A. 1993. Population trends in grassland, shrubland, and forest birds in eastern North Ameri- ca. Current Ornithology 11:1-34. BOYCE, M. S. 1992. Population viability analysis. An- nual Review of Ecology and Systematics 23:481- 506. BRAUNINC, D. W. 1992. Atlas of breeding birds in Pennsylvania. University of Pittsburgh Press, Pitts- burgh, PA. BURGMAN, M. A., S. FERSON, AND H. R. AK(AKAYA. 1993. Risk assessment in conservation biology. Chapman & Hall, New York, NY. CARTER, M. E, W. C. HUNTER, D. N. PASHLEY, J. M. BRADLEY, J. PRICE, AND G. S. BUTCHER. In press. Setting landbird conservation priorities for states and physiographic regions within the United States. In R. E. Bonney, Jr., D. N. Pashley, and R. Cooper (editors). Strategies for bird conservation: creating the Partners in Flight planning process. Cornell Lab- oratory of Ornithology, Ithaca, NY. FINCH, D. M., AND P. W. STANGEL (EDITORS). 1993. Sta- tus and management of neotropical migratory birds. USDA Forest Service Gen. Tech. Rep. RM-229. USDA Forest Service, Fort Collins, CO. Foss, C. R. 1994. Atlas of breeding birds in New Hampshire. Audubon Society of New Hampshire, Concord, NH. HUNTER, W. C., M. E CARTER, D. N. PASHLEY, AND m. BARlcER. 1993. The Partners In Flight prioritization scheme. Pp. 109-119 in D. M. Finch and P. W. Stan- gel (editors). Status and management of neotropical migratory birds. USDA Forest Service Gen. Tech. Rep. RM-229. USDA Forest Service, Fort Collins, CO. JONES, A. L., AND P. D. VICKERY. 1995. Distribution and population status of grassland birds in Massa- chusetts. Bird Observer 23:896. MARlCS, P. L. 1983. On the origin of the field plants of the northeastern United States. American Naturalist 122:210-228. MELVIN, S. M. 1994. Military bases provide habitat for rare grassland birds. Natural Heritage News, Mas- sachusetts Division of Fisheries and Wildlife 4:3. OUELLET, H. 1993. Bicknell's Thrush.' taxonomic status and distribution. Wilson Bulletin 105:545-572. PETERSON, R. T. 1980. A field guide to the birds. Houghton Mifflin, Boston, MA. PETERSON, R. T 1990. A field guide to western birds. Houghton Mifflin, Boston, MA. PRUITT, L. 1996. Henslow's Sparrow status assessment. U.S. Fish and Wildlife Service, Bloomington Eco- logical Services Field Office, Bloomington, IN. ROSENBERG, m. V., AND J. V. WELLS. 1995. Final report: importance of geographic areas to neotropical mi- grant birds in the northeast. U.S. Fish and Wildlife Service, Region 5, Hadley, MA. ROSENBERG, m. V., AND J. V. WELLS. In press. Global perspectives on neotropical migrant bird conserva- tion in the northeast U.S. In R. E. Bonney, Jr., D. N. Pashley, and R. Cooper (editors). Strategies for bird conservation: creating the Partners in Flight planning process. Cornell Laboratory of Ornitholo- gy, Ithaca, NY. 80 STUDIES IN AVIAN BIOLOGY NO. 19 SAUER, J. R., B. G. PETERJOHN, S. SCHWARTZ, AND J. E. HINES. 1996. The North American Breeding Bird Survey home page. Ver. 95.1: www.mbr. nbs.gov/ bbs/bbs.html. U.S. Geological Survey, Patuxent Wildlife Research Center, Laurel, MD. SCOTt, J. g., AND g. CSUTI. 1991. GAP analysis: blue- print for proactive conservation. FOCUS on Renew- able Natural Resources 17:1-2. SMITH, C. R. 1992. Henslow's Sparrow, Ammodramus henslowii. Pp. 315-330 in K. J. Schneider and D. M. Pence (editors). Migratory nongame birds of management concern in the northeast. U.S. Fish and Wildlife Service, Newton, MA. VICKERY, P. D. 1992. A regional analysis of endan- gered, threatened, and special-concern birds in the northeastern United States. Transactions of the Northeast Section of the Wildlife Society 48:1-10. VICKERY, P. D. 1996. Effects of prescribed fire on the reproductive ecology of northern blazing star (Lia- tris scariosa var. novae-angliae). Report to Maine Natural Areas Program, Department of Conserva- tion, Augusta, ME. VICKERY, P. D., M. L. HUNTER, JR., AND S. M. MELVIN. 1994. Effects of habitat area on the distribution of grassland birds in Maine. Conservation Biology 8: 1087-1097. WELLS, J. V. 1995. Investigations into the distribution and abundance of species. Ph.D. dissertation. Cor- nell University, Ithaca, NY. WHEELWRIGHT, N. t., AND J. D. RISING. 1993. Savan- nah Sparrow (Passerculus sandwichensis). In A. Poole and E Gill (editors). The birds of North Amer- ica, no. 45. Academy of Natural Sciences, Philadel- phia, PA, and American Ornithologists' Union, Washington, D.C. WITHAM, J. W., AND g. L. HUNTER, JR. 1992. Popu- lation trends of neotropical migrant landbirds in northern coastal New England. Pp. 8545 in J. M. Hagan III and D. W. Johnston (editors). Ecology and conservation of neotropical migrant landbirds. Smithsonian Institution Press, Washington, D.C. ZINK, R. M., AND J. V. REMSEN. 1986. Evolutionary processes and patterns of geographic variation in birds. Current Ornithology 4:1-69. Studies in Avian Biology No. 19:81-86, 1999. USE OF CULTIVATED FIELDS BY BREEDING MOUNTAIN PLOVERS IN COLORADO FRITZ L. KNOPF AND JEFFERY R. RUPERT Abstract. Populations of breeding Mountain Plovers (Charadrius montanus) in North America de- clined an average of 3.7 percent per year from 1966 through 1993, resulting in a 63 percent total decline during that period. This decline led to listing the species as a Candidate Species under the federal Endangered Species Act. Mountain Plovers have been observed nesting on cultivated fields, but nest loss may be high on these sites. During the 1994 breeding season we surveyed Mountain Plover use of contiguous cultivated and native-prairie sites in Weld County, Colorado. Birds used both sites equally in April. The cultivated field was planted in early May, which probably destroyed nests and resulted in plovers reinitiating courtship and renesting. No resurgence of courtship was observed on native prairie during the same period. Observations of Mountain Plovers with radio transmitters during the 1994 breeding seasons revealed that some of the birds that lost nests or chicks on native prairie moved to the recently cultivated field to forage. Two of three Mountain Plovers that hatched eggs within 2 kilometers of the cultivated field moved chicks onto that field until the chicks fledged. We conclude that cultivated fields provide acceptable, and locally valuable, feeding habitat for Moun- tain Plovers. Because Mountain Plovers have also been reported to nest on plowed ground from Nebraska to Oklahoma, however, and because 31.9 percent of native habitats in the southwestern Great Plains have been cultivated, we also conclude that mechanical working of fields during the nest and early chick phases may contribute to the 3.7 percent annual rate of decline of this species. Four management options are suggested to improve Mountain Plover recruitment on and near cultivated lands. EL USO DE CAMPOS CULTIVADOS EN COLORADO POR PARTE DE LOS CHORLITOS LLANEROS EN REPRODUCCION Sinopsis. Las poblaciones en reproducci6n de Chorlitos Llaneros (Charadriux montanus) en Am6rica del Norte disminuyeron en un promedio de 3,7 por ciento por afio desde 1966 hasta fines de 1993, 1o que se tradujo en una disminuci6n total de un 63 por ciento durante aquel perfodo. Esta disminuci6n produjo la clasificaci6n de la especie como Especie Candidata estipulada en la ley federal Endangered Species Act. Se han observado Chorlitos Llaneros haciendo sus nidos en campos cultivados, pero la p6rdida de nidos puede ser alta en estos sitios. Durante la estaci6n de reproducci6n en 1994 censamos el uso por parte del Chorlito Llanero de dos clases de sitios 11aneros contiguos, cultivados y nativos, en el Condado de Weld, Colorado. En abril las aves usaron ambos sitios con la misma frecuencia. Se sembr6 el campo cultivado a principios de mayo, lo cual probablemente destruy6 los nidos e indujo a los chorlitos a reiniciar el cortejo y a hacer los nidos nuevamente. No se observ6 ningfn resurgi- miento de cortejo en la 11anura nativa durante el mismo perfodo. Las observaciones de Chorlitos Llaneros con radiotransmisores durante las estaciones de reproducci6n de 1994 revelaron que algunas de las aves que perdieron sus nidos o sus pollos en la 11anura nativa se mudaron al campo recin cultivado para forrajear. Dos de tres Chorlitos Llaneros que criaron pollos dentro de 2 ki16metros del campo cultivado trasladaron sus pollos all hasta que volaron. Concluimos que los campos cultivados proveen un hfibitat alimenticio aceptable y 1ocalmente valioso para los Chorlitos Llaneros. Sin em- bargo, dado que la informaci6n da cuenta de que los Chorlitos Llaneros hacen sus nidos en terreno arado desde Nebraska a Oklahoma, y como un 31,9 por ciento de los hfibitats nativos en el suroeste de la Gran Llanura han sido cultivados, concluimos asimismo que la labranza mecfinica de los campos durante las fases del nido y de los pollos nuevos puede explicar en gran medida la tasa anual de disminuci6n de 3,7 por ciento en esta especie. Se sugieren cuatro opciones para mejorar el restable- cimiento del Chorlito Llanero en terrenos cultivados y cerca de los mismos. Key Wordx: Charadrius montanus; Colorado; Mountain Plover. The Great Plains grasslands are the most endan- gered ecosystem in North America (Samson and Knopf 1994). As a group, grassland birds have shown the most universal and most severe de- clines of all native bird species, including neo- tropical migrants (Knopf 1994). Breeding pop- ulations of Mountain Plovers (Charadrius mon- tanus) declined 63 percent from 1966 to 1993, 81 despite what appeared to be normal rates of pro- ductivity in native habitats (Miller and Knopf 1993) and high adult survival (Knopf and Ru- pert 1995). Because of this decline, the U.S. Fish and Wildlife Service has listed the species as a Candidate Species for Threatened or Endan- gered status under the federal Endangered Spe- cies Act. 82 STUDIES IN AVIAN BIOLOGY NO. 19 Mountain Plovers nest across the western Great Plains and eastern Colorado Plateau re- gion, with a core breeding area in Weld County, Colorado (Graul and Webster 1976). The species nests in areas of shortgrass prairie historically grazed by native herbivores and currently man- aged as rangeland for domestic herbivores or as dryland (non-irrigated) farms. The breeding biology of Mountain Plovers is best known from studies at the Pawnee National Grassland in northcentral Colorado. Nests are usually located in areas of native shortgrass prai- rie dominated by blue grama (Bouteloua graci- lis) and buffalo grass (Buchlo' dactyloides; Graul 1975) with the area around nests being 30% or more bare ground (Knopf and Miller 1994). Chicks leave the nest shortly after hatch- ing and often move more than 1 km from the nest site (Knopf and Rupert 1996). Chicks raised on these grasslands generally use disturbed sites (e.g., areas that have suffered locally severe overgrazing, roadsides), especially where some forbs have invaded (Graul 1975). Breeding Mountain Plovers forage, and oc- casionally nest, on cultivated fields near native shortgrass-prairie landscapes. Since the early 1990s, nesting on such fields has been relatively common in areas along the eastern boundary of the shortgrass-prairie region, from Texas to Wy- oming (Shackford 1991; J. Shackford, pers. comm.). During our ongoing studies of Moun- tain Plovers in Colorado, we conducted a peri- odic survey of plover use of a cultivated field contiguous to native-prairie habitat on the Paw- nee National Grassland. In this paper we docu- ment relative use of native versus cultivated sites; describe habitats used for nesting and brood-rearing on native prairie; and document movements of birds that indicate that this spe- cies readily uses cultivated fields during the nesting and brood-rearing periods of the repro- ductive cycle. METHODS We studied Mountain Plovers on the Pawnee Na- tional Grassland, a 780-km 2 shortgrass prairie in Weld County, Colorado, during the 1992-1994 breeding sea- sons. Graul (1973) summarized the physiography, veg- etation, and climate of this region. In 1993 many cultivated fields within 2 km of our study area were left fallow during the April-July breeding season. In 1994 one field contiguous to the study area was left fallow. We implemented a 20-point survey of Mountain Plovers along the fenceline sepa- rating the native prairie and cultivated field, with sur- vey points 0.15 km apart. In 1994 we conducted 19 replications, from 20 April through 13 June. All sur- veys were conducted at sunrise. From each survey point we counted the number of adult Mountain Plo- vers we saw and/or heard on each side of the survey line. We were confident that the birds were equally visible on both sides of the survey line. Trends in the use of the two sites were compared using univariate repeated-measures analyses of variance and paired t- tests. In 1994 we captured 26 adult birds at nests on the Pawnee National Grassland and fitted them with radio transmitters before their eggs hatched. Birds were cap- tured with a leg snare or swing-door box trap (Knopf and Rupert 1996). We relocated each adult almost dai- ly from the time its chicks hatched until the adult left the study area. Because the landowner denied us ac- cess to the adjacent cultivated field, we did not conduct any nest searches there, nor did we capture and fit adults with transmitters on that land. We determined the relative coverage of grass versus bare ground around nests and brood-rearing sites (at distances of 10, 25, and 50 m in each of the cardinal directions) for 11 adult Mountain Plovers that fledged chicks in 1993 or 1994 on the Pawnee National Grass- land. Twelve 0.5-m 2 plots were photographed at each nest and brood-rearing site after Knopf and Miller 1994, except that in our study we took photographs in all four cardinal directions. A clear dot-grid was placed over each photograph to determine the percentages of area in grass or bare ground. We also recorded fre- quencies of cow manure piles and prickly pear cactus ( Opuntia spp.). RESULTS POPULATION SURVEYS Mountain Plovers were easily detected from distances up to 150 m away. Individuals occa- sionally could be seen from two adjacent survey points, in which case they were recorded only for the first point. Plovers moved freely back and forth between the cultivated field and native prairie. We frequently watched individual birds walk from one side of the fenceline to the other during a survey. Repeated-measures analysis of variance revealed no difference in the number of Mountain Plovers detected on native and culti- vated sites throughout the survey season (20 April-13 June; F = 1.35, df = 16, P = 0.16). The number of Mountain Plovers using na- tive-prairie (, = 12.5 _+ 4.24) and cultivated (, = 12.5 --- 4.12) sites (t < 0.01, P > 0.99) and the pattern of use (F = 0.79, df = 6, P = 0.58) were similar in the first 7 of the 19 surveys we conducted (Fig. 1). The number of birds peaked on the third survey but then declined markedly through the seventh survey. Between 3 and 6 May the cultivated field was chemically treated for weeds and planted. In the eighth survey, Mountain Plover detections on native prairie re- mained low, whereas detections on the cultivat- ed field peaked sharply (Fig. 1). Birds at this time were seen only foraging on native prairie but were seen mostly advertising territories and courting on the cultivated field (see Knopf 1996b for review of behaviors). MOUNTAIN PLOVERS IN COLORADO--Knopf and Rupert 83 20 ÷ 15 ÷ 10  Cultivated field ..... -A .... Prairie ß .?'  lay lun FIGURE 1. Number of adult Mountain Plovers de- tected on paired plots at a native-prairie and cultivated- field interface in Weld County, Colorado, in 1994. Plotted values are 3-point running means; statistics presented in text were calculated on raw data. The ar- row indicates the date when the cultivated field was planted. In the remaining 11 surveys, numbers of Mountain Plovers detected on native prairie re- mained relatively constant (Fig. 1). Numbers on the cultivated field peaked for a third time in the fifteenth survey, 36 d into the survey period (May 25); no courtship behaviors were seen at this time. The number of plovers detected was greater (t = 5.47, df = 6, P = 0.002) on culti- vated (, = 13.9 ñ 4.30) than on native-prairie (, = 6.1 -+ 1.95) sites in the last seven surveys, when all courtship behavior had ended on the cultivated field. MOVEMENTS TO CULTIVATED FIELDS After losing a nest or chicks, adult Mountain Plovers sometimes stayed in the study vicinity. After losing chicks to predators in mid-July 1994, 2 of 17 adults with transmitters moved to forage on recently cultivated fields. These indi- viduals remained on the cultivated fields 3-5 d as part of a loose flock that varied from 35 to 55 individuals. We often saw Mountain Plovers with small chicks on cultivated fields near our study site. In 1994 we monitored the movements of three adults with transmitters that had nested on the Pawnee National Grassland within 2 km of a cultivated field. Two of those adults moved their chicks to a cultivated field within 2 d of hatch- ing, and these families stayed on the field until the chicks fledged. BROOD HABITATS ON NATIVE-PRAIRIE GRASSLANDS In the Pawnee National Grassland, habitats used for brood-rearing contained more bare ground and less grass cover than did habitats surrounding nests (Table 1). Occurrences of forbs (F = 1.80, P = 0.18), prickly pear (F = 0.01, P = 0.94), and cow manure (F = 0.14, P = 0.71) were similar between nest and brood- rearing habitats. Many Mountain Plovers nesting in the Paw- nee National Grassland moved broods to the vi- cinity of cattle-watering tanks, which were gen- erally devoid of vegetation for more than 20 m around the tank. To determine if birds were sim- ply attracted to bare ground or if the presence of cattle enhanced the attractiveness of a site, we surveyed for Mountain Plovers at 56 stock tanks and compared plover presence to cattle use. Mountain Plovers occurred at 11 of 28 stock tanks with cattle but were absent at the 28 stock tanks where cattle were absent (X  = 29.29, df = 3, P > 0.0001). This survey indi- cated that Mountain Plovers are strongly attract- ed either to cattle or, as with cultivated fields, to recent site disturbance. DISCUSSION USE OF CULTIVATED FIELDS The first peak in Mountain Plover numbers was similar between the cultivated field and na- tive prairie. Detectability of birds began to drop in both areas as individuals began incubating eggs. Birds were confirmed nesting at this time, and six nests were found on the native-prairie side of the survey line. The cultivated field was private land, and although we were not granted permission to survey for nests, we used repro- ductive behaviors to track breeding phenology and were confident that birds were also nesting on the cultivated field. A second peak in Mountain Plover numbers on the cultivated field occurred immediately af- TABLE l. MEAN (+ SE) PERCENT COVER OF GRASS VERSUS BARE GROUND IN MOUNTAIN PLOVER NEST HABITAT AND BROOD-REARING HABITAT, WELD COUNTY, COLORADO, 1993--1994 Grass Bare ground Plot  ZdF P X ZdF P Nest habitat 87 +- 1.6 9 -+ 1.0 2.4 0.001 2.6 0.002 Brood habitat 84 _+ 1.4 15 +- 1.3 Note: Comparisons are for 11 adults that successfully raised chicks to fledging. All data are from native-prairie habitats only. 84 STUDIES IN AVIAN BIOLOGY NO. 19 ter the field was sprayed for weeds and then planted to millet (Setaria). The machinery work- ing the field probably destroyed most nests, as adults immediately began courting again. Num- bers of birds detected again dropped rapidly, as clutches presumably were completed and birds began incubating. Mountain Plover numbers on the cultivated field peaked a third time in late May, but no courtship behaviors were observed at this time. Instead, this third peak was associated with the hatching of the original nests on the Pawnee Na- tional Grassland and the movement of broods to the cultivated field to forage. One adult with a transmitter moved from the grassland to the cul- tivated field and remained there until its two chicks fledged. Numbers of Mountain Plovers using the cul- tivated field began to decline steadily in early June (Fig. 1). This decline was likely a response to the rapidly growing millet crop which made the site less acceptable as plover habitat. Moun- tain Plovers require very short vegetation, which facilitates predator detection (Graul 1975). At this time, plovers with broods moved back to the Pawnee National Grassland. The fate of most nests on the cultivated field was uncertain; how- ever, a single nest we observed daily was aban- doned when the millet reached a height of about 20 cm. Adult Mountain Plovers nesting in Pawnee National Grassland that lost all their eggs or chicks to predation sometimes moved to culti- vated fields to forage, where loose flocks of 50- 100 birds were common. These flocks also in- cluded some adults that had moved their chicks to the cultivated fields (Knopf and Rupert 1996). Generally, however, flocks stayed at a specific field only for a few days after it had been cul- tivated or planted and then moved again; this pattern is seen regularly among wintering flocks of Mountain Plovers in California (Knopf and Rupert 1995). In our study, adults with chicks either stayed on the same cultivated field until the chicks could fly or moved back to native prairie when the cropland vegetation became too tall. MOUNTAIN PLOVERS AS BARE-GROUND ASSOCIATES Mountain Plovers have evolved, as have other shortgrass-prairie birds, in an intensively grazed ecosystem dominated by bison (Bison bison), prairie dogs (Cynomys spp.), and pronghorn (An- tilocapra americana; Knopf 1996a). In taller, mixed-grass prairies, Mountain Plovers are as- sociated primarily with the intensive grazing found in prairie-dog towns (Knowles et al. 1982, Olson-Edge and Edge 1987). On the Pawnee National Grassland, Mountain Plovers select both nest and brood-rearing sites that have more bare ground than do surrounding areas. We do not think, however, that Mountain Plovers choose to nest specifically in areas of approxi- mately 30% bare ground or to raise chicks in areas of approximately 15% bare ground. Rath- er, these percentages represent the average max- imum bare ground currently available to birds on the Pawnee National Grassland. Mountain Plovers regularly use cultivated fields on migration and in winter, as well as in the breeding season (Grinnell and Miller 1944, Laymon et al. 1986, Knopf and Rupert 1995). Knopf and Miller (1994) first concluded that Mountain Plovers are associated with bare ground, based on data collected at the nest site. Prior to that time, this species had been de- scribed as a prairie associate of blue grama and buffalo grass landscapes. During three breeding seasons (1992-1994) in Colorado, however, we found no nests in buffalo grass (N = 147). This grass reproduces asexually by sending out sto- lons and has a tendency to mat, thus precluding a bare-ground component for situating nests. Breeding Mountain Plovers in Colorado use cultivated lands where range-management prac- tices strive to protect soils and provide relatively uniform landscapes of grass cover. Most grazing prescriptions on public lands use some variation of the allotment approach to regulate stocking densities and herbage removal, thus favoring ho- mogenous grass cover across broad landscapes (Knopf 1996c). Standardized grazing of allot- ments precludes areas of excessive grass/soil disturbances characteristic of native ungulate and rodent herbivores---disturbances to which Mountain Plovers have evolved. Using allot- ments contrasts with grazing by bison; this na- tive grazer preferentially forages on black-tailed prairie dog (Cynomys ludovicianus) towns (Krueger 1986), thus maximizing grazing pres- sure at some sites while leaving others only lightly grazed. These intensively grazed sites provide specific habitats used by other grassland birds (Knopf 1996a). MANAGING FOR MOUNTAIN PLOVERS ON OR NEAR PLOWED GROUND Mountain Plovers in Colorado appear to be equally attracted to cultivated fields and grazed native prairie. Most cultivated fields, however, are usually planted to a late-season crop or are recultivated every 4-6 wk to control weeds. These activities certainly destroy some nests and chicks, which use crypsis to avoid detection (Sordahl 1991). Mountain Plovers have been documented nesting on plowed fields in Texas, Oklahoma, Kansas, Colorado, Nebraska, New MOUNTAIN PLOVERS IN COLORADO--Knopf and Rupert 85 TABLE 2. RELATIVE EXTENT OF CROPLAND VERSUS NATIVE RANGELAND IN THE PRIMARY BREEDING RANGE OF MOUNTAIN PLOVERS IN THE SHORTGRASS-PRAIRIE REGION OF THE SOUTHWESTERN GREAT PLAINS Cropland Rangeland Cropland (ha) (ha) (%) Colorado 2,760,763 6,225,134 30.7 Kansas 329,387 93,150 78.0 Nebraska 112,351 59,454 65.4 New Mexico 344,291 631,695 35.3 Oklahoma 128,223 262,845 32.8 Texas 514,310 823,527 38.4 Wyoming 167,873 1,215,203 12.1 Totals 4,357,198 9,311,008 31.9 Note: Data are from the Natural Resource Inventory, U.S. Department of Agriculture, 1994 Mexico, and Wyoming (J. Shackford, pers. comm.). Because more than 30% of native hab- itats used by Mountain Plovers have been con- verted to cropland in this region (Table 2), we hypothesize that reduced productivity as a result of tillage may explain part of the 3.7% annual rate of decline of this species continentally from 1966 to 1993. It seems likely that cultivated fields represent regional reproductive "sinks" for nesting Mountain Plovers (Pulliam 1988). In view of the fact that agricultural practices may play a large role in the decline of Mountain Plovers, we offer four management options to reduce nest and chick losses on cultivated fields. 1. Encourage farmers to prepare and plant fields used by Mountain Plovers in a short win- dow of time in May and June. Fields are cur- rently prepared weeks or even months in ad- vance of planting. 2. If weed control is necessary during the pe- riod 1 May-15 July, encourage chemical rather than physical treatments on fields used by Mountain Plovers. 3. Mandate seeding of native grasses only and allow grazing of lands registered in the Con- servation Reserve Program (CRP). Current prac- tices often result in tame (introduced) cool-sea- son grasses being planted on the western plains and preclude grazing in an ecosystem that evolved with intensive grazing pressure. Grazing on CRP lands will increase the amount of habitat suitable for grassland species and will also pro- vide additional economic incentives to enroll in the CRP. 4. Management of publicly owned (or pri- vate) grazing allotments adjacent to cultivated fields could be changed to make them more at- tractive to Mountain Plovers during the period when the birds select nest sites. Highly inten- sive, long-term grazing of contiguous native grasslands should enhance nesting habitat. In ad- dition, Mountain Plovers tend to select grass- lands that are occupied by cattle or other her- bivores. Cattle generally are not moved onto the Pawnee National Grassland until late May, de- pending on growth of the warm-season grasses. Moving cattle onto pastures in early May should further enhance the attractiveness of native-prai- rie sites over cultivated lands. Mountain Plovers are also attracted to recently burned grasslands (Knopf and Rupert 1995). Winter or early spring burning could be used to make native rangelands more attractive than cultivated lands for breed- ing Mountain Plovers. ACKNOWLEDGMENTS We thank the U.S. Forest Service for financial as- sistance and personnel at the Rocky Mountain National Forest, Arapaho-Roosevelt National Forest, and Paw- nee National Grassland for technical assistance. Nat- ural Resource Inventory data (U.S. Department of Ag- riculture) were provided courtesy of A. Allen and T Osborn. We are especially grateful to L. Mullen for his interest and administrative support. F. C. Knopf pro- vided assistance in the field. E. K. Bollinger, J. R. Her- kert, and P. D. Vickery reviewed the manuscript. LITERATURE CITED GRAUL, W. D. 1973. Adaptive aspects of the Mountain Plover social system. Living Bird 12:69-94. GRAUL, W. D. 1975. Breeding biology of the Mountain Plover. Wilson Bulletin 87:6-31. GRAUL, W. D., AND L. E. WEBSTER. 1976. Breeding status of the Mountain Plover. Condor 78:265-267. GRINNELL, J., AND A. H. MILLER. 1944. The distribution of the birds of California. Pacific Coast Avifauna 27. KNOPF, E k. 1994. Arian assemblages on altered grass- lands. Studies in Avian Biology 15:247 257. KNOPF, E L. 1996a. Prairie legacies--birds. Pp. 135 148 in E B. Samson and E L. Knopf (editors). Prai- rie conservation: preserving North America's most endangered ecosystem. Island Press, Covelo, CA. KNOPF, E L. 1996b. Mountain Plover (Charadrius tnontanu). In A. Poole and E Gill (editors). The birds of North America, no. 211. Academy of Nat- ural Sciences, Philadelphia, PA, and American Or- nithologists' Union, Washington, D.C. KNOPF, E L. 1996c. Grazing nongame bird habitats. Pp. 51 58 in P. R. Krausmann (editor). Rangeland wildlife. Society for Range Management, Denver, CO. KNOPF, F. L., AND B. J. MILLER. 1994. Charadrius mon- tanus montane, grassland, or bare-ground plover? Auk 11:504-506. KNOPF, F. L., AND J. R. RUPERT. 1995. Habits and hab- itats of Mountain Plovers in winter. Condor 97:743- 751. KNOPF, F. L., AND J. R. RUPERT. 1996. Reproduction and moveinents of Mountain Plovers breeding in Colorado. Wilson Bulletin 108:28-35. KNOWLES, C. J., C. J. STONER, AND S. P. GIEB. 1982. Selective use of black-tailed prairie dog towns by Mountain Plovers. Condor 84:71-74. KRUEGER, K. 1986. Feeding relationships among bison, 86 STUDIES IN AVIAN BIOLOGY NO. 19 pronghorn, and prairie dogs: an experimental anal- ysis. Ecology 67:760-770. LAYMON, P., J. MARCHANT, AND T. PRATER. 1986. Shorebirds. Houghton Mifflin, Boston, MA. MILLER, B. J., AND F. L. KNOPF. 1993. Growth and survival of Mountain Plovers. Journal of Field Or- nithology 64:500-506 and 65:193. OLSON-EDGE, S. L., AND W. D. EDGE. 1987. Density and distribution of the Mountain Plover on the Charles M. Russell National Wildlife Refuge. Prairie Naturalist 19:233-238. PULLIAM, H. R. 1988. Sources, sinks, and population regulation. American Naturalist 132:652-661. SAMSON, EB., AND E L. KNOPF. 1994. Prairie conser- vation in North America. BioScience 44:418-421. SUACKFORD, J. S. 1991. Breeding ecology of the Moun- tain Plover in Oklahoma. Bulletin of the Oklahoma Ornithological Society 24:9-13. SORDAHL, r. 1991. Antipredator behavior of Mountain Plover chicks. Prairie Naturalist 23:109-115. Studies in Avian Biology No. 19:87-94, 1999. CHANGES IN BIRD POPULATIONS ON CANADIAN GRASSLANDS C. STUART HOUSTON AND JOSEF K. SCHMUTZ Abstract. Before the Canadian prairies were settled in the 1880s, the grassland birds of that region were catalogued in the 1820s, 1850s, and 1870s by John Richardson, Thomas Blakiston, and Elliott Coues, respectively. Ernest Thompson Seton recorded changes in southern Manitoba during the first 10 years of settlement. Tree plantings on the open plains made nest sites available for several species, but human encroachment was harmful to other, especially larger, species. Breeding Bird Surveys on the Canadian prairies between 1966 and 1994 documented steep declines of Sprague's Pipit (Anthus spragueii; 7.3 percent per year) and Loggerhead Shrike (Lanius ludovicianus; 5.6 percent per year) and less severe (but still more than 2.0 percent per year) declines of Northern Harrier (Circus cyaneus), Killdeer (Charadrius vociferus), Burrowing Owl (Athene cunicularia), Short-eared Owl (Asio fiam- meus), and Western Meadowlark (Sturnella neglecta). Swainson's Hawk (Buteo swainsoni) and Fer- ruginous Hawk (B. regalis) have shown significantly reduced productivity, coincident with sharp declines in their main prey, Richardson's ground squirrel (Spermophilus richardsonii). Both hawk species are now showing evidence of population declines as well. Introduced trees in deserted farm- steads are dying from neglect, drought, herbicides, and bulldozers, offering fewer nesting sites and less protective cover. Until recently, the Canadian government has encouraged plowing native prairie and substituting grain crops or crested wheatgrass (Agropyron cristaturn) in its place. Marked increases in red fox (Vulpesfulva) numbers (which may have contributed to decreased numbers of Richardson's ground squirrels) and increased use of pesticides, fertilizers, and other chemicals have coincided with bird declines. Declines in numbers of small grassland birds and in numbers and productivity of three grassland raptors seem disproportionate to these factors and may be more severe than in the western United States. CAMBIOS EN LAS POBLACIONES DE AVES EN LOS PRADOS DE CANADA Sinopsis. Antes de la colonizaci6n de los prados canadienses en los aftos de la dficada de 1880, las aves de pastizal de esa regi6n habfan sido catalogadas en las dficadas de 1820, 1850 y 1870 por John Richardson, Thomas Blakiston y Elliott Coues, respectivamente. Ernest Thompson Seton anot6 cam- bios en el sur de Manitoba durante los primeros 10 aftos de colonizaci6n. La siembra de firboles en las 11anuras descampadas proporcion6 sitios de nidos para algunas especies, pero la invasi6n humana fue perjudicial para otras especies, especialmente las ms grandes. Los Breeding Bird Surveys en las 11anuras canadienses entre 1966 y 1994 documentaron disminuciones precipitadas de la Bisbita de Sprague (Antbus spragueii; 7,3 por ciento por afio) y de Lanio Americano (Lanius ludovicianus; 5,6 por ciento por afio) y disminuciones menos severas (pero de todos modos de mils de un 2,0 por ciento por afio) del Gavilrn Rastrero (Circus cyaneus), del Chorlito Tildfo (Charadrius vociferus), del Bfiho Llanero (Athene cunicularia), del Btiho Orejicorto (Asiofiammeux) y del Pradero Occidental (Sturnella neglecta). El Aguililla de Swainson (Buteo swainsoni) y el Aguililla Real (B. regalis) han experimen- tado una importante reducci6n en la productividad, coincidente con una drfistica disminuci6n de su presa principal, la ardilla terrestre de Richardson (Spermophilux richardsonii). Asimismo, ambas es- pecies de halc6n ahora muestran indicios de disminuciones poblacionales. Los firboles introducidos en cortijos abandonados estrin muriendo por el descuido, por la sequfa, por los herbicidas y por los bulldozers, y ofrecen menos sitios de nidos y menos cobertura protectora. Hasta hace poco, el gobierno canadiense ha favorecido el arado de la pradera nativa y, en su lugar, ha dado preferencia a la susti- tuci6n de cosechas de granos o de Agropyron cristaturn. Aumentos marcados de los mimeros de los zorros rojos (Vulpesfulva) (que pueden haber contribuido a los nfimeros redicidos de ardillas terrestres de Richardson) y el aumento del uso de pesticidas, de abonos y de otras sustancias qufmicas han coincidido con las disminuciones de las aves. Las disminuciones de los mimeros de aves pequefias de pastizal y de los mimeros y la productividad de tres aves rapaces de pastizal parecen desproporcionadas para estos factores y pueden ser mils severas queen el oeste de los Estados Unidos. Key Words: agriculture; Alberta; Canadian prairies; grassland birds; Manitoba; presettlement; Sas- katchewan. There is widespread concern about the decline in grassland birds throughout the Canadian prai- ries, a profoundly altered habitat. "Between plowing and overgrazing, it is perhaps the most extensively altered biome on the planet" (Gay- ton 1990:25). With historical information that extends back to the 1820s (Richardson and Swainson 1832, Blakiston 1861-1863, Coues 1878), no locality in North America can surpass the presettlement inventory available for Sas- katchewan. In this paper we summarize for se- lected species 177 yr of observations in southern Saskatchewan, describe population trends for se- lected species since 1966, and suggest possible 87 88 STUDIES IN AVIAN BIOLOGY NO. 19 / / / 52ON 120øN 120%V 116*W 112%V  108w 104½W I 100øw 96'W / 100 20km ALBERTA SASKATCHEWAN M MONTANA I t\ MINNESOTA I NORTH DAKOTA 112 ø .... øW 1;4 W 100øw k 9øw FIGURE l. Map of Canadian prairie grasslands and of cities and villages mentioned in text. Dark shading represents open grassland, lighter shading represents moist grassland with some aspen copses. (Map by K. Bigelow, Department of Geography, University of Saskatchewan, Saskatoon, SK.) links between species declines and widespread ecosystem change. We stress some of the major changes in bird populations resulting from the extirpation of bison (Bison bison), the conver- sion of native grassland to agricultural fields, the cessation of regular prairie fires, the regrowth of quaking aspen (Populus tremuloides) from dor- mant roots (Maini 1960), the planting of trees in shelterbelts, and the declines in trees associated with deserted farmsteads, herbicides, bulldozers, and drought. Some bird species, among them Mourning Dove (Zenaida macroura), Western Kingbird (Tyrannus verticalis), Black-billed Magpie (Pica pica), American Crow (Corvus brachyrhyn- chos), Tree Swallow (Tachycineta bicolor), Barn Swallow (HiTundo rustica), and Mountain Blue- bird (Sialia currucoides), adapted to human set- tlement and the tree planting that followed it and increased in numbers (Houston 1977a, b, 1979, 1986; Houston and Houston 1988, 1997). Pop- ulations of most other species in the mixed prai- rie ecosystem of the prairie ecozone (Padbury and Acton 1994) remained relatively stable until about 1970, although satisfactory monitoring of numbers by Breeding Bird Surveys (BBSs) has been available only since 1966. In the early 1970s, about the same time as agriculture be- came more technological with higher chemical inputs (Goldsborough 1993), declines in popu- lations and productivity of several grassland spe- cies became evident (Downes and Collins 1996). For Burrowing Owls (Athene cunicularia; Hous- ton et al. 1996) and Loggerhead Shrikes (Lanius ludovicianus; Peterjohn and Sauer 1995), evi- dence of decline has been universal and consis- tent across the Canadian prairies. HISTORICAL PERSPECTIVE THE NINETEENTH CENTURY In the 1820s, bison and recurrent fires main- tained open grassland north to Carlton House, Saskatchewan (52o52 ' N, 106032 ' W; Fig. 1) on the North Saskatchewan River (Houston 1977a). This was the site of intensive natural-history cat- aloguing by two Scotsmen, surgeon-naturalist Dr. John Richardson and naturalist Thomas Drummond (Richardson 1823, 1829, 1836; Sa- bine 1823; Richardson and Swainson 1832; Kir- by 1837; Hooker 1840). On his first visit to Carl- ton House, in May 1820, Richardson noted that the interface between mixed forest and grassland had exceptional diversity (Houston and Street 1959). BIRD POPULATIONS ON CANADIAN GRASSLANDS--Houston and Schmutz 89 Three decades later, from October 1857 until June 1858, the birdlife at Carlton House was as- sessed by English surveyor Thomas Wright Bla- kiston (Blakiston 1861-1863, Houston and Street 1959). Blakiston found the American Crow so uncommon he could not collect a spec- imen of it, whereas the Common Raven (Corvus corax) was numerous and nested commonly on open grasslands. In 1873 and 1874, further stud- ies prior to the advent of farming were carried out along the Manitoba-United States boundary by American surgeon-naturalist Elliott Coues (1878), who reported that Tree Swallows were rare and that Barn Swallows nested sparingly on cliff faces, separate from the more common Cliff Swallows (Hitundo pyrrhonota). Red-winged Blackbird (Agelaius phoeniceus) was the least common blackbird, and the American Crow was still uncommon. Upland Sandpipers (Bartramia longicauda) were numerous, and both Whoop- ing Cranes (Grus americana) and Sandhill Cranes (G. canadensis) were sparingly but quite evenly distributed. Coues commented that be- tween Pembina Mountain (present-day Snow- flake) and Turtle Mountain, Manitoba (Fig. 1), Baird's Sparrow (Ammodramus bairdii) was the "most abundant and characteristic species...in some places outnumbering all the other birds to- gether" (Coues 1873:695-696). Coues also not- ed that in the same area Chestnut-collared Long- spurs (Calcarius ornatus) occurred "in profu- sion" (Coues 1878:579). West of where the Mis- souri Coteau crosses the 49th parallel, near the present boundary between North Dakota and Montana, shortgrass prairie predominated and McCown's Longspurs (C. mccownii) became abundant as Chestnut-collared Longspurs de- clined. When Walter Raine, a lithographer by trade and an oologist by avocation, visited new ranching territory at Rush Lake, Saskatchewan (Fig. 1), in 1891, McCown's Longspur was the most common small bird on the elevated prairies (Raine 1892, Houston 1981). In 1873 and 1874, Chestnut-collared Longspurs and Sprague's Pip- its (Anthus spragueii) were also abundant (Coues 1873); at the eastern crossing of the Sou- ris River loop, Sprague's Pipits were "so nu- merous that the air seemed full of them" (Coues 1878:560). By 1882, naturalist and well-known author Ernest E. Thompson (later known as Ernest Thompson Seton) had made careful observations in southwestern Manitoba and adjacent Sas- katchewan (Thompson 1890); settlers were still thinly scattered but the Passenger Pigeon (Ec- topistes migratorius) had all but vanished. Ten years later, Seton noted the influx of Mourning Doves, the westward advance of the Greater Prairie-Chicken (Tympanuchus cupido) and Eastern Bluebird (Sialia sialis) with settlement, and the virtual disappearance of Upland Sand- pipers and Sprague's Pipit as native prairie was plowed for agriculture (Houston 1980). Seton also noted declines in Chestnut-collared Long- spurs and Swainson's Hawks (Buteo swainsoni) and increases in Western Meadowlarks (Sturne- lla neglecta) and Horned Larks (Eremophila al- pestris) during the first 10 yr of settlement (Houston 1980). Because they provided food, large birds were often hunted by farmers. Canada Geese (Branta canadensis) soon became less common. Greater Sandhill Cranes (Grus canadensis tabida) and Whooping Cranes were also shot for food. Some large birds, however, did not disappear until the end of the nineteenth century. At Rush Lake in 1891, Turkey Vultures (Cathartes aura), present since the days of the bison, were still common on the open prairie, as were a few Common Ra- vens (Raine 1892). THE TWENTIETH CENTURY In Saskatchewan, Black-billed Magpies (Pica pica) retreated from the plains in the late 1800s and by the early twentieth century were restrict- ed to the Cypress Hills in southwestern Sas- katchewan (Houston 1977a; Fig. 1). From 1904 through 1910, magpies disappeared even from nearby Maple Creek and Eastend (Fig. 1), but in the 1920s they increased and spread out. They reappeared in small numbers to the north and east at Unity and Sheho in 1926, Percival in 1929, and Nipawin in 1930, and they were com- mon at Wauchope in 1939 and Yorkton in 1951 (Fig. 1). They became city residents in Saska- toon (Fig. 1) beginning in the late 1960s and have increased throughout the province since then (Houston 1977a). Between 1885 and 1903, Mourning Doves spread out from the Qu'Appelle River valley onto the newly settled plains near Indian Head (Houston 1986; Fig. 1). Spreading northeastward from river valleys onto the plains as domestic trees reached about 6 m in height, the Western Kingbird has served as a useful indicator species for tree growth. For example, at the Harley Ranson farm at Tyvan, Saskatchewan (Fig. 1), where trees were planted in 1903, Western Kingbirds first nested in 1924; at the Stewart Houston farm 8 km to the west, where trees were planted in 1917, the kingbirds took up residence in 1937 (Houston 1979). Red- tailed Hawks (Buteo jamaicensis), once quite uncommon even in migration, extended their range southward onto former prairie areas as as- pen (Populus spp.) copses, locally known as "bluffs," grew up from dormant roots once prai- rie fires came under control about 1910 (Hous- ton and Bechard 1983). 90 STUDIES IN AVIAN BIOLOGY NO. 19 Grasslands and trees Native grasslands were quickly plowed fol- lowing settlement. By 1911 there were 95,013 farms in Saskatchewan comprising an area of 113,843 km 2, of which 48,076 km 2 (42%) were plowed (Archibold and Wilson 1980). The num- ber of farms peaked at 142,391 in 1936 and dropped to 60,840 by 1991, yet total farm area increased to 268,738 km 2 in 1991, a year when 134,624 km 2 (50%) were in crops, 57,143 km 2 (21%) in summerfallow, 10,759 km 2 (4%) in tame pasture, and 66,198 km 2 (25%) were un- designated (Saskatchewan Agriculture and Food 1994). As a result of these changes, remaining grasslands have became smaller, more frag- mented, and of poorer quality. Between 1976 and 1986, 8% of grassland was lost in Alberta, 8% in Manitoba, and 6% in Saskatchewan (We- 11icome and Haug 1995). By 1978, nearly 18% (69,243 of 385,832 ha) of the Prairie Farm Re- habilitation Administration (PFRA) pastures in the mixed grassland ecoregion had been seeded (Agriculture and Agri-Food Canada 1996). A 1972 study of 16 townships in Saskatche- wan, together representing more than 1% of Sas- katchewan grassland, showed 269 occupied farmsteads and 244 abandoned farmsteads, 80% of the latter still with tree cover (Smith 1973). Insecticides, herbicides, and fertilizer Chemicals were used sparingly by Canadian farmers prior to 1947. Use of 2,4-D, one of the phenoxy herbicides, began at this time, and by 1966 high volumes of it were in use (Goldsbo- rough 1993). Other herbicides appeared in 1965 and were being used in high volumes by 1980. Peregrine Falcons (Falco peregrinus) were rare and local breeders along the Frenchman River valley and Battle Creek in the grasslands of extreme southwestern Saskatchewan until at least 1917 (Bechard 1981, 1982) but persisted until the 1950s in Alberta valleys such as that of the Red Deer River (e.g., Taverner 1919; Fig. 1). In 1975, following widespread use of DDT, a survey of historical Peregrine Falcon breeding sites in southern Alberta showed that none were occupied (Fyfe et al. 1976). With extensive use of dieldrin in the late 1950s and early 1960s, Merlins (Falco colum- barius) declined moderately in Alberta and al- most disappeared from the grassland near Kin- dersley, Saskatchewan, for 10 yr (Hodson 1976, Houston and Schmidt 1981); by 1995, Merlin numbers had returned to preinsecticide levels in Saskatchewan (Houston and Hodson 1997). Mammalian predators Red foxes (Vulpes fulva) were extremely rare in southern Saskatchewan until 1965 at Luseland and 1966 at Kyle (Fig. 1); they quickly became common in the 1970s (Jordheim 1995, Finley 1996). Coyotes (Canis latrans) have increased in the same area since the late 1980s (Finley 1996). It is likely that increasing populations of foxes on the Canadian prairies have had a det- rimental effect on populations of grassland birds. In North Dakota, Sovada et al. (1995) found that duck nesting success averaged 32% where coyotes were the principal canid but fell to 17% where foxes were the principal canid. Breeding Bird Surveys BBSs began in Manitoba, Saskatchewan, and Alberta in 1966. Since that time, populations of many grassland species in the southern portions of the Canadian prairie provinces have shown significant negative trends. For example, BBS data for grassland portions of Alberta, Saskatch- ewan, and southwestern Manitoba indicate sig- nificant (P < 0.05) population declines between 1966 and 1994 for Sprague's Pipit (-7.3% per annum), Loggerhead Shrike (-5.2%), Killdeer (Charadrius vociferus; -3.3%), Short-eared Owl (Asio fiammeus; -2.9%), Western Mead- owlark (-2.1%), and Burrowing Owl (-2.0%) (Downes and Collins 1996). Grassland species that have shown nonsignificant (P > 0.05) pop- ulation declines during the same period include Lark Bunting (Calamospiza melanocorys; -11.2% per annum), McCown's Longspur (-9.0%), Lark Sparrow (Chondestes gramma- cus; -3.2%), Chestnut-collared Longspur (-2.2%), and Horned Lark (-1.1%) (Downes and Collins 1996). All of these species, with the exception of the shrike, are ground-nesters that are likely to be extremely vulnerable to preda- tion. Knopf (1994:25), referring to both the United States and Canada, noted that "grassland birds have shown steeper, more consistent and more geographically widespread declines than any other behavioral or ecological guild of North American species." Burrowing Owls Burrowing Owls, which in the 1830s extend- ed north at least to Carlton House, Saskatchewan (Houston and Street 1959), have declined steadi- ly in range and abundance in the three prairie provinces since the late 1970s (Houston et al. 1996). The harmful effects of the insecticide Carbofuran on this species were first demon- strated in 1986 (James and Fox 1987), but the decline has continued even after restrictions on using this chemical within 250 m of owl colo- nies were implemented. This decline is due in part to habitat loss and fragmentation and to in- creases in predator populations (Wellicome and Haug 1995). On the plains near Regina, Sas- BIRD POPULATIONS ON CANADIAN GRASSLANDS Houston and Schmutz 91 4.5- >. 2.5- 2.0 ß . Year FIGURE 2. Productivity (young per successful nest produced to banding age) of Ferruginous Hawks in Saskatchewan, 1969-1996. Curved lines represent the 95% confidence interval about the linear regression (r = -0.63, P < 0.001) katchewan, the Burrowing Owl population de- clined steadily from 76 pairs in 1987 to 29 pairs in 1992 and 9 pairs in 1994 (Warnock and James 1997). During this period, the percentage of suc- cessful pairs dropped from 72 to 45%, and the number of young produced per nest attempt dropped from 3.1 to 1.8 (James et al. 1997). In large PFRA pasturals in the Kindersley, Sas- katchewan, region (Fig. 1), Burrowing Owls were last seen breeding at Antelope Park in 1980, Heart's Hill in 1985, Mantario in 1986, Newcombe in 1990, Eagle Lake in 1993, Kin- dersley-Elna in 1993, and Progress in 1994. A single pair persisted at Mariposa in 1996. At Kindersley-Elna Pasture (63.5 km 2) there were 18 pairs of Burrowing Owls in 1991, 9 pairs in 1992, 2 pairs in 1993, and none thereafter (Houston et al. 1996). The Committee on the Status of Endangered Wildlife in Canada ele- vated the Burrowing Owl's status from threat- ened to endangered in 1995 (Wellicome and Haug 1995). Swainsoh's and Ferruginous hawks In Saskatchewan, Swainson's Hawk produc- tivity averaged 2.09 young per successful nest from 1964 through 1987 (N = 985 successful nests; S. Houston, unpubl. data). Productivity then declined sharply, averaging 1.63 young per successful nest from !988 through 1994 and dropping as low as 1.27 young per successful nest in 1993 (N = 602 successful nests; Houston and Schmutz 1995). Near Hanna, Alberta (Fig. 1), Swainson's Hawk productivity 'fell from a long-time mean of 2.03 to 1.14 young per successful nest in 1993 (N = 1,170; Houston and Schmutz 1995), ß 3.0 14'  N Z 1975 1985 1990 1995 Year FIGURE 3. Population (nests per 100 km 2) and pro- ductivity (young per successful nest) of Ferruginous Hawks near Hanna, Alberta, 1975-1996. rose to 1.69 in 1994, and then dropped to 1.22 in 1995 and 1.40 in 1996 (J. Schmutz, unpubl. data). We know of no mechanism by which short-acting pesticide (monocrotophos) poison- ing on this species' wintering grounds in Argen- tina (Goldstein et al. 1996) should have detri- mental effects on brood size 6 mo later. Ferruginous Hawks have disappeared from nearly half of their presettlement territory in Saskatchewan (Houston and Bechard 1984), and their productivity has declined. Between 1969 and 1987, the number of young fledged per suc- cessful nest averaged 3.01 (N = 369 successful nests; S. Houston, unpubl. data). Since 1988 this number has remained below 2.82 and has aver- aged 2.63 (N = 488 successful nests; S. Hous- ton, unpubl. data). Overall productivity has de- clined in recent years (r = 0.63, P < 0.001; S. Houston, unpubl. data; Fig. 2). The species has declined even more severely in Alberta. The number of nests in the Hanna study area dropped from a peak of 14 per 100 km 2 in 1989 to 7 and 6 per 100 km 2 in 1995 and 1996, respectively, and productivity declined from 3.2 young per successful nest in 1986 to 2.1 in 1995 and 1996 (J. Schmutz, unpubl. data; Fig. 3). The nesting period of the Ferruginous Hawk coincides with the peak abundance of young Ri- chardson's ground squirrels (Spermophilus ri- chardsonii), which are the main prey of Ferru- ginous Hawks in Saskatchewan and Alberta (Schmutz et al. 1980). The decline in productiv- ity of both Ferruginous and SwainsoWs hawks is probably the result of the sharp decline in this rodent since 1987. This decrease began at Kin- dersley, Saskatchewan, and extended west to Mamario, Saskatchewan, and then Hanna, berta (Houston and Schmutz 1995; S. Houston and J. Schmutz, unpubl. data). One possible ex- planation for the steep decline in ground squir- rels is the substantial increase in the red fox pop- ulation. 92 STUDIES IN AVIAN BIOLOGY NO. 19 Nest sites for both Ferruginous and Swain- son's hawks have also been lost. South of the aspen parkland belt, the remaining planted trees, largely in shelterbelts of deserted farms, are dy- ing from neglect, drought, and herbicide sprays and are razed by bulldozers as farmers try to increase the amount of land they have in pro- duction. DISCUSSION The declines of small grassland birds in Ca- nada since the mid-1960s and the earlier decline of the Upland Sandpiper more than a century ago can be explained, at least in part, by an ever- diminishing and ever-more-fragmented amount of native grassland. This pertains particularly to Sprague's Pipit and Chestnut-collared Longspur, species that prefer native grassland over seeded pasture, hayland, and cropland (S. K. Davis, un- publ. data). The decline in Sprague's Pipit may have been hastened by overgrazing (Dale 1984, Sutter 1996), but Horned Larks and Chestnut- collared Longspurs, which usually respond fa- vorably to grazing (Owens and Myres 1973), have also declined (Dale 1984). In Saskatche- wan, numbers of Baird's Sparrows correlated positively with grass/sedge (Carex) cover and negatively with bare ground cover (Sutter et al. 1995). Clay-colored Sparrows (Spizella pallida) were also detected more often in native pasture than in any seeded pasture with crested wheat- grass (Agropyron cristatum; Davis and Duncan 1999). Fragmentation of prairie can also have adverse effects, including an increase in Brown- headed Cowbird (Molothrus ater) parasitism (Davis and Sealy in press). The Canadian government has provided mon- etary incentives that have encouraged grain rath- er than cattle production and hence has encour- aged the breaking of marginal lands (Fulton et al. 1989; Anderson et al. 1991; Riemer 1993, 1995). The "Crow's Nest Pass Rate," for ex- ample, which was in effect from 1897 until 1995, subsidized grain but not cattle shipments to ports, and the grain quota system, abandoned only in 1996, allowed sales based on cultivated acreage--the more land a farmer plowed, the more grain he was allowed to sell. The govern- ment has paid farmers when grain (but not cat- tie) prices fell below a 5-yr average and has sub- sidized crop insurance to underwrite the risk of growing grain. These are but a few examples of the government policies that have favored grain production over cattle production in the twenti- eth century. Additionally, the development of larger farm machinery and of mechanical rock- pickers has allowed hilly and rocky pastures to be broken, large sloughs to be drained, and shel- terbelts and streamside vegetation to be removed (Anderson et al. 1991). Increasing farm debt, which averaged $89,000 per farm by 1985, has added financial pressures to this mix (Anderson et al. 1991). Although major government assis- tance programs were discontinued in 1995 and 1996, which should reduce pressure to convert pastures into cropland, such changes are so re- cent that beneficial effects will probably not be apparent or measurable for at least a few years. Concern is not restricted to small grassland birds and three raptor species. Waterfowl nest success in the prairie pothole region has de- clined at a significant annual rate of about 0.5%, even on islands without mammalian predators and in study areas where large predators have been removed or excluded by fencing (Beau- champ et al. 1996). Increased predation by mammals, especially red foxes, striped skunks (Mephitis mephitis; Pasitschniak-Arts and Mess- ier 1995), and Franklin's ground squirrels (Sar- geant et al. 1987), has contributed to duck de- clines, with as yet undetermined effects on ground-nesting passedfies. The loss of prairie alone cannot explain the severity of the Burrowing Owl decline in the Canadian prairies and northwestern North Da- kota (R. K. Murphy, pers. comm.), which con- trasts with slowly declining numbers in south- western North Dakota and increasing numbers in southwestern Idaho from the 1970s through the 1990s (K. Steenhof, pers. comm.). Carbo- furan may have contributed to Burrowing Owl declines at one stage, but its use is now prohib- ited within 250 m of a Burrowing Owl nest. The declines in productivity among Swainson's and Ferruginous hawks, which have continued for 9 yr in Saskatchewan and Alberta, appear to differ from anything described in the United States. The causes of these declines, though probably associated with acute declines in prey species, are still not well understood. Unprecedented industrialization of farming has occurred, partially overlapping with the de- clines in grassland birds described here. Such associations, while intriguing, may be partially coincidental, but further scrutiny is indicated. We do not know the cause of the recent but widespread declines in productivity, largely or entirely restricted to Canada, among three grass- land raptor species. Careful study of these spe- cies must continue as we search for answers. ACKNOWLEDGMENTS M. Fulton, Saskatchewan Institute of Pedology, pro- vided valuable insight and commentary; G. Riemer, B. Bristol, and G. Wrubleski provided unpublished data about the PFRA and other government projects; and K. Bigelow, University of Saskatchewan, and D. S. Houston assisted with graphics. J. Harris, D. Francis, BIRD POPULATIONS ON CANADIAN GRASSLANDS--Houston and Schmutz 93 and D. Zazelenchuk found many Swainson's and Fer- ruginous hawk nests in Saskatchewan, and D. G. Mill- er and M. A. Gerard were among the many people who helped with the Saskatchewan banding operation. P. D. Vickery, B. Dale, and two anonymous reviewers pro- vided helpful criticism. LITERATURE CITED AGRICULTURE AND AGRI-FOoD CANADA. 1996. Sensitiv- ity analysis of economic impacts of proposed en- dangered species legislation and community pasture grazing. Internal report, Land Management and Di- versification Service, Prairie Farm Rehabilitation Adnfinistration, Regina, SK. ANDERSON, D. W., C. J. ROPPEL, AND R. M. GRY. 1991. Sustainability in Canadian agriculture. Sci- ence Council of Canada, Ottawa, ON. ARCHIBOLD, O. W., AND M. B. WILSON. 1980. The nat- ural vegetation of Saskatchewan prior to agricultural settlement. Canadian Journal of Botany 58:2031- 2042. BEAUCHAMP, W. D., T. D. NUDDS, AND R. G. CLARK. 1996. Duck nest success declines with and without predator management. Journal of Wildlife Manage- ment 60:258-264. BECHARD, M. J. 1981. Historic nest records of the Per- egrine Falcon in southern Saskatchewan and south- ern Manitoba. Blue Jay 39:182-183. BECHARD, M. J. 1982. Further evidence for a historic population of Peregrine Falcons in southern Sas- katchewan. Blue Jay 40:125. BLAKISTON, T 1861-1863. On birds collected and ob- served in the interior of British North America. Ibis 3:314-320; 4:3-10; 5:39-87; 5:121-155. COUES, E. 1873. Notes on two little-known birds of the United States. American Naturalist 7:695-697. COUES, E. 1878. Field-notes on birds observed in Da- kota and Montana along the forty-ninth parallel dur- ing the seasons of 1873 and 1874. Bulletin of the U.S. Geological and Geographical Survey of the Territories 4(3):545-661. DALE, B.C. 1984. Birds of grazed and ungrazed grass- lands in Saskatchewan. Blue Jay 42:102-104. DAVIS, S. K., AND D.C. DUNCAN. 1999. Grassland songbird abundance in native and crested wheatgrass pastures of southern Saskatchewan. Studies in Avian Biology 19:211-218. DAVIS, S. K., AND S. G. SEALY. In press. Cowbird par- asitism and nest predation in fragmented grasslands of southwestern Manitoba. In J. N. Smith, T L. Cook, S. I. Rothstein, S. G. Sealy, and S. K. Rob- inson (editors). Ecology and management of cow- birds. University of Texas Press, Austin, TX. DOWNES, C., AND B. t. COLLINS. 1996. The Canadian Breeding Bird Survey, 1966 1994. Canadian Wild- life Service Progress Notes no. 210. FINLEY, K. 1996. The red fox invasion and other changes in wildlife populations in west-central Sas- katchewan since the 1960s. Blue Jay 54:206-210. FULTON, g., K. ROSAASEN, AND A. SCHMITZ. 1989. Ca- nadian agricultural policy and prairie agriculture. Economic Council of Canada, Ottawa, ON. FYFE, R. W., S. A. TEMPLE, AND r. J. CADE. 1976. The 1975 North American Peregrine Falcon survey. Ca- nadian Field-Naturalist 90:228-273. GAYTON, D. 1990. The wheatgrass mechanism: science and imagination in the western Canadian landscape. Fifth House, Saskatoon, SK. GOLDSBOROUGH, L. G. 1993. Studies on the impact of agricultural and forestry herbicides on non-target aquatic plant communities. Pp. 49-57 in G. L. Hol- royd, H. L. Dickson, M. Regnier, and H. C. Smith (editors). Prairie conservation and endangered spe- cies workshop, Brandon, 1992. Provincial Museum of Alberta Occasional Papers no. 19. GOLDSTEIN, g. L., B. WOODBRIDGE, M. E. ZACCAGNINI, AND S. B. CANEVELLI. 1996. An assessment of mor- tality of Swainson's Hawks on wintering grounds in Argentina. Journal of Raptor Research 30:106-107. HODSON, K. A. 1976. The ecology of Richardson's Merlins on the Canadian prairies. M. S. thesis. Uni- versity of British Columbia, Vancouver, BC. HOOKER, W. J. 1840. Flora boreali-americana. William G. Bohn, London, U.K. HOUSTON, C. S. 1977a. Changing patterns of corvidae on the prairies. Blue Jay 35:149-155. HOUSTON, C. S. 1977b. The prairie bluebird trail. Na- ture Canada 6(2):3-9. HOUSTON, C. S. 1979. The spread of the Western King- bird across the prairies. Blue Jay 37:149-157. HOUSTON, C. S. 1980. Introduction. Pp. 5-15 in Ernest Thompson Seton in Manitoba [no editor]. Manitoba Naturalists' Society, Winnipeg, MB. HOUSTON, C. S. 1981. An assessment of Walter Raine and his Saskatchewan records. Blue Jay 39:168-181. HOUSTON, C. S. 1986. Mourning Dove numbers ex- plode on the Canadian prairies. American Birds 40: 52-54. HOUSTON, C. S., AND M. J. BECHARD. 1983. Trees and the Red-tailed Hawk in southern Saskatchewan. Blue Jay 41:99-109. HOUSTON, C. S., AND M. J. BECHARD. 1984. Decline of the Ferruginous Hawk in Saskatchewan. American Birds 38:166-170. HOUSTON, C. S., D. G. HJERTAAS, R. L. SCOTT, AND P. C. JAMES. 1996. Experience with Burrowing Owl nest-boxes in Saskatchewan, with comment on de- creasing range. Blue Jay 54:136-140. HOUSTON, C. S., AND K. A. HODSON. 1997. Resurgence of breeding Merlins, Falco colurnbarius richardso- nil, in Saskatchewan grasslands. Canadian Field- Naturalist 111:243-248. HOUSTON, C. S., AND M. I. HOUSTON. 1988. Tree Swal- low banding near Saskatoon, Saskatchewan. North American Bird-Bander 12:103-108. HOUSTON, C. S., AND M. I. HOUSTON. 1997. Saskatch- ewan bird species which increased with settlement. Blue Jay 55(2):90-96. HOUSTON, C. S., AND A. SCHMIDT. 1981. History of Richardson's Merlin in Saskatchewan. Blue Jay 39: 3037. HOUSTON, C. S., AND J. K. SCHMUTZ. 1995. Declining reproduction among Swainson's Hawks in prairie Canada. Journal of Raptor Research 29:198401. HOUSTON, C. S., AND M. G. STREET. 1959. The birds of the Saskatchewan River. Saskatchewan Natural History Society, Regina, SK. JAMES, E C., T. J. ETHIER, AND M. K. TOUTLOFF. 1997. Parameters of a declining Burrowing Owl popula- 94 STUDIES IN AVIAN BIOLOGY NO. 19 tion in Saskatchewan. Raptor Research Report 9:34- 37. JAMES, P. C., AND G. A. Fox. 1987. Effects of some insecticides on productivity of Burrowing Owls. Blue Jay 45:65-71. JORDHEIM, S. 1995. Foxes in south-central Saskatche- wan. Blue Jay 43:232-233. KIRBY, W. 1837. Fauna boreali-americana. Vol. 4. The insects. Josiah Fletcher, Norwich, U.K. KNOPF, E 1994. Avian assemblages on altered grass- lands. Studies in Avian Biology 15:247-257. MAINI, J. S. 1960. Invasion of grassland by Populus tremuloides in the northern Great Plains. Ph.D. dis- sertation. University of Saskatchewan, Saskatoon, SK. OWENS, R. A., AND M. T. MYRES. 1973. Effects of ag- riculture upon populations of native passefine birds of an Alberta fescue grassland. Canadian Journal of Zoology 51:697-713. PADBURY, G. A., AND D. E ACTON. 1994. Ecoregions of Saskatchewan. Central Survey and Mapping Agency, Regina, SK. PASITSCHNIAK-ARTS, M., AND E MESSIER. 1995. Pred- ator identification at simulated waterfowl nests using inconspicuous hair catchers and wax-filled eggs. Ca- nadian Journal of Zoology 73:984-990. PETERJOHN, B. G., AND J. R. SAUER. 1995. Population trends of the Loggerhead Shrike from the North American Breeding Bird Survey. Proceedings of the Western Foundation of Vertebrate Zoology 6:117- 121. RAINE, W. 1892. Bird nesting in north-west Canada. Hunter Rose, Toronto, ON. RICHARDSON, J. 1823. Botanical appendix. Pp. 729- 768 in J. Franklin. Narrative of a journey to the shores of the Polar Sea in the years 1819, 20, 21, and 22. John Murray, London, U.K. RICHARDSON, J. 1829. Fauna boreali-americana. Vol. 1. The mammals. John Murray, London, U.K. RICHARDSON, J. 1836. Fauna boreali-americana. Vol. 3. The fish. Richard Bentley, London, U.K. RICHARDSON, J., AND W. SWAINSON. 1832. Fauna bo- reali-americana. Vol. 2. The birds. John Murray, London, U.K. RIEMER, G. 1993. Agricultural policy impacts on land use decision making and options for reform. Pp. 123 132 in P. Rakowski and R. Massey (editors). Proceedings of the first national Wildlife Habitat workshop. Wildlife Habitat Canada, Ottawa, ON. RIEMER, G. 1995. The effects of GATT on Canadian environmental and farm policy. Pp. 129-135 in R. Gray, T Becker, and A. Schmitz (editors). World agriculture in a post-GATT environment: new rules, new strategies. University Extension Press, Univer- sity of Saskatchewan, Saskatoon, SK. SABINE, J. 1823. Zoological appendix. Pp. 647-703 in J. Franklin. Narrative of a journey to the shores of the Polar Sea in the years 1819, 20, 21, and 22. John Murray, London, U.K. SARGEANT, A. B., M. A. SOVADA, AND R. J. GREEN- WOOD. 1987. Responses of three prairie ground squirrel species, Spermophilus franklinii, S. richard- sonii, and S. tridecemlineatus, to duck eggs. Cana- dian Field-Naturalist 101:95-97. SASKATCHEWAN AGRICULTURE AND FOOD. 1994. Agri- cultural statistics 1993. Saskatchewan Agriculture and Food, Regina, SK. SCHMUTZ, J. K., S. M. SCHMUTZ, AND D. A. BOAG. 1980. Coexistence of three species of hawks (Buteo spp.) in the prairie-parkland ecotone. Canadian Jour- nal of Zoology 58:1075-1089. SMITH, A. R. 1973. The value of abandoned farmsteads as a wildlife resource. Department of Natural Re- sources, Saskatoon, SK. SOVADA, M. A., A. B. SARGEANT, AND J. W. GRIER. 1995. Differential effects of coyotes and red foxes on duck nest success. Journal of Wildlife Manage- ment 59:1-9. SUTTER, G. C. 1996. Habitat selection and prairie drought in relation to grassland bird community structure and the nesting ecology of Sprague's Pipit, Anthus spragueii. Ph.D. dissertation. University of Regina, Regina, SK. SUTTER, G. C., t. TROUPE, AND M. FORBES. 1995. Abundance of Baird's Sparrows, Ammodramux bair- diœ in native prairie and introduced vegetation. Ecoscience 2:344 348. TAVERNER, P. A. 1919. The birds of the Red Deer Riv- er, Alberta. Auk 36:1-21,248-265. THOMPSON, E. E. 1890. The birds of Manitoba. Pro- ceedings of the U.S. National Museum 13:457-643. WARNOCK, R. a., AND P. C. JAMES. 1997. Habitat frag- mentation and Burrowing Owls (Speotyto cunicular- ia) in Saskatchewan. Pp. 477-486 in J. R. Duncan, D. H. Johnson, and T H. Nicholls (editors). Biology and conservation of owls of the northern hemi- sphere. USDA Forest Service Gem Tech Rep. NC- 190. USDA Forest Service North Central Experi- ment Station, St. Paul, MN. WELLICOME, T. L., AND E. A. HAUG. 1995. Second up- date on status report on the Burrowing Owl, Speo- tyta cunicularia. Committee on the Status of Endan- gered Species in Canada, Ottawa, ON. Studies in Avian Biology No. 19:95-103, 1999. MULTISCALE HABITAT ASSOCIATIONS OF THE SAGE SPARROW: IMPLICATIONS FOR CONSERVATION BIOLOGY JOHN T. ROTENBERRY AND STEVEN T. KNICK Abstract. General patterns of habitat association of common shrubsteppe passerine birds, as indicated by correlations of population abundance with plot-level habitat variables, are well known. We dem- onstrated that for Sage Sparrows (Amphispiza bello these population correlations were consistent with the behavior of individual birds as they selected patches of shrubs on which to forage. Furthermore, individuals appeared to track local-scale variation in habitat variables that changed annually. Despite these linkages, the ability of plot-level correlations to predict abundances across sites or years was weak. We also examined landscape-level correlates of species distributions. For Sage Sparrows, pres- ence and persistence at a sampling point were related to landscape attributes such as shrubland frag- ment size and configuration. We concluded that failure of plot-level correlations to predict changes in abundance at broader scales stemmed both from failure to include landscape-level attributes and from failure to consider an intrinsic decoupling of population density from local habitat details. We discuss the implications of these results for conservation studies. ASOCIACIONES DE HBITAT MULTIESCALA DEL GORRI(SN DE ARTEMISIA: IMPLICACIONES PARA LA BIOLOG[A CONSERVACIONISTA Sinopsis. Son bien conocidos los esquemas generales de asociaci6n de hfibitat de aves paseriformes comunes en las estepas arbustivas. Estos fueron indicados a trav6s de las correlaciones de abundancia de poblaci6n con variables de hfibitat a nivel de parcela. Demostramos que para los Gorriones de Artemisia (Amphispiza belli), estas correlaciones de poblaci6n fueron concordantes con el comporta- miento de aves individuales cuando escogfan parcelas de arbustos donde forrajear. Adems, las aves individuales pareclan adoptar cambios anuales en las variables de hbitat a nivel local. A pesar de estas conexiones, la habilidad de predecir abundancias entre sitios o aftos usando correlaciones a nivel de parcela fue leve. Examinamos tambi6n los correlativos de distribuci6n de especies a nivel de paisaje. Para los Gorriones de Artemisia, se relacionaron su presencia y persistencia en un sitio de muestreo a caracterfsticas de paisaje tales como tamafio y configuraci6n del fragmento de matorral. Inferimos que la deficiencia de las correlaciones a nivel de parcela para predecir los cambios en abundancia en escalas amplias, radicaba en la falta de inclusi6n de atributos a nivel de paisaje yen no considerar una separaci6n intrfnsica entre la densidad de poblaci6n y los detalles del hbitat local. Analizamos las implicaciones de estos resultados para los estudios conservacionistas. Key Words: Amphispiza belli; Artemisia tridentata; bird-habitat relationships; Great Basin; individual behavior; landscape; sagebrush; Sage Sparrow; shrubsteppe. Research into the relationships between the abundance of vertebrates and certain features of their habitat, both physical and biotic, has been a cornerstone of modem ecology (Rotenberry 1981). Once such relationships are established, they can be applied both to theoretical and prac- tical questions. Many theoretical models that seek to explain adaptive variation in animal be- havior include as one of their essential elements the relationship between the number of individ- uals in a habitat and various aspects of environ- mental "quality," quality presumably relating to the fitness of the individual within that habitat. Such information is of increasing importance to wildlife managers concerned with preserving adequate numbers of individuals or species in an environment increasingly disrupted and frag- mented by humans. Indeed, the conservation value of identifying animal-habitat relationships cannot be underestimated. In the case of song- birds, their populations are usually managed by manipulating features of their habitat rather than by directly manipulating numbers. If we can identify habitat attributes that directly or indi- rectly influence bird-population numbers-- through the provision of food, shelter, nest sites, or protection from predators--then we can at- tempt to alter these attributes to achieve a de- sired conservation goal. The use of information about bird-habitat re- lationships rests on certain assumptions, how- ever, and it is an empirical examination of those assumptions that we examine here. This review is not exhaustive but instead relies heavily on our own research on Great Basin shrubsteppe birds, conducted over a span of almost 20 yr. Because of the hierarchical nature of the pro- cesses involved (e.g., Allen and Starr 1982), we discuss three levels of investigation in our re- search. Presumably, the process of habitat selec- tion that results in associations between a spe- cies and its habitat is an evolutionarily derived mechanism that ensures that individuals seek out and remain in the particular habitats to which 95 96 STUDIES IN AVIAN BIOLOGY NO. 19 they are adapted. Thus, we expect successful in- dividuals to reflect a phenotype that has been molded by and remains suitable for the habitat in which we find them. Expression of this phe- notype may be morphological or behavioral or, more likely, both. We further expect that the pat- terns of habitat selection by individuals are re- flected in habitat occupancy by populations. It is this emergent property of individual behavior that we assume is responsible for producing cor- relations between bird densities and habitat var- iables, and it provides the rationalization for in- terpreting these bird-habitat correlations in an adaptationist framework. It is also clear that the associations between a species and habitat variables recorded on any particular plot of ground can be influenced by the nature of the surrounding landscape (O'Neill et al. 1988). In most cases, these landscape-level effects are manifest through processes related to habitat fragmentation and its effects on popula- tion dynamics (e.g., Rolstad 1991, Porneluzi et al. 1993). Increasingly, however, it is recognized that it is often the structure of an entire land- scape mosaic that may be important to birds, not just the size and shape of individual fragments (e.g., Bolger et al. 1991, 1996; Pearson 1993; Knick and Rotenberry 1995). Thus, it is reason- able to expect an interaction between local- and landscape-level attributes in determining ob- served bird-habitat relationships. A second major assumption is that the eco- logical associations we observe are stable and consistent through time and space--that patterns detected at one time and place can be general- ized to other times and places. It is often as- sumed that natural selection for some sort of op- timal habitat response is strong and continuous, and thus that populations are generally at or near equilibrium with respect to the resources with which any set of habitat variables is associated (e.g., Cody 1981, 1985). We know that environments vary through time, however, and this can be especially true in arid regions. For example, one can easily doc- ument substantial fluctuation in the physical en- vironment in the form of annual variation in pre- cipitation. In arid lands these fluctuations can in turn drive enormous annual changes in primary and secondary productivity, and the difference between a dry and a wet year can be substantial (Noy-Meir 1973, Rotenberry and Wiens 1980, Cody 1981, Fuentes and Campusano 1985). This annual variation can influence the reproductive success of bird species in these ecosystems (Ro- tenberry and Wiens 1989, 1991). Likewise, abundance of bird species may also fluctuate substantially, both from year to year as well as from site to site within years (Wiens and Roten- berry 1981a, Knick and Rotenberry 1995). Of primary interest is whether these changes in bird abundance are associated with changes in habi- tat. In other words, are population numbers cou- pled to environmental variation, and do fluctu- ations in animal numbers represent a "tracking" of changes in habitat? Clearly, this is an important question for both scientists and conservationists to consider: are individuals and populations consistent and pre- dictable in their habitat associations through time and space? Do populations track environ- mental variation in a consistent fashion at the spatial and temporal scales over which habitat relationships are traditionally determined? There may be a variety of reasons why species abun- dances might not be associated with changes in habitat or its associated resources (see below). If so, population densities may become "decou- pied" from habitat parameters that might oth- erwise influence changes in local population siz- es. If this is the case, what are the implications for populations, and how do we go about study- ing them? Our studies of birds in shrubsteppe habitats of the northern Great Basin can shed some light on these issues. STUDY AREA AND SPECIES Our research was conducted in arid shrubsteppe habitat of western North America, primarily in the northern Great Basin and Snake River Plains. This shrubsteppe is dominated by sagebrush (Artemisia [pri- marily big sagebrush (A. tridentam)]), saltbush (Atri- plex), rabbitbrush (Chsothamnus [particularly gray rabbitbrush (C. nauseosus) and green rabbitbrush (C. viscidifiora)]), and greasewood (Sarcobatus) among the shrubs and by bluegrass (Poa), wheatgrass (Agro- pyron), rescue (Vulpia [= Festuca]), and brome (Bro- mus) among the grasses. In this paper we restrict our analyses to habitat re- lationships of the Sage Sparrow (Emberizidae: Amp- hispiza belli nevadensis), a common and widespread inhabitant of shrubsteppe. In the Great Basin this spe- cies is found mainly in association with sagebrush. In- dividuals weigh about 20 g and, where present, den- sities range from about 15 to 180 individuals per square kilometer (Wiens and Rotenberry 198 la). This subspecies is migratory, wintering in arid shrublands from central Nevada through northern Mexico (Martin and Carlson 1998). PLOT-LEVEL ASSOCIATIONS The most common method of establishing bird-habitat correlations is to census a series of representative plots or transects, usually ranging in area from 5 to 50 ha (Wiens and Rotenberry 1981b, Rotenberry 1982). A variety of habitat variables, both physical and biotic, are scored on the same plots or transects. Habitat relationships then are estimated using correlations between species abundance and environmental variables, HABITAT ASSOCIATIONS OF SAGE SPARROWSRotenberry and Knick 97 frequently employing a variety of bivariate and/ or multivariate approaches (Wiens and Roten- berry 1981b). As an example, we surveyed 14 "original" sites scattered throughout the northern Great Ba- sin of southeastern Oregon and northern Nevada and selected to represent an array of common shrubsteppe habitats (Wiens and Rotenberry 1981a). At each site we censused birds and mea- sured habitat features for 3 successive years. Birds were surveyed along 600-m Emlen-type transects (Emlen 1977). Percent coverage of each shrub species was determined from 10 100- m transects arrayed perpendicular to the bird transects and then reduced to independent axes using principal components analysis. We derived relationships of Sage Sparrows to habitat vari- ables using both bivariate and multiple correla- tions. Although we also measured a variety of other habitat attributes, we discuss below only those variables associated with shrub coverage because they yielded the strongest patterns with widespread shrubsteppe bird species (Wiens and Rotenberry 198 la). Sage Sparrow abundance was highly correlat- ed with sagebrush coverage (r = 0.61, df = 40, P < 0.001; Wiens and Rotenberry 1981a). Ad- ditionally, substantial variation in the distribu- tion of Sage Sparrows was statistically explained by a multiple regression of abundance on shrub- species components (R 2 = 0.70, N = 42, P < 0.001; Rotenberry 1986). The pattern of signif- icance of regression coefficients again implicat- ed sagebrush as the dominant covariate. A sim- ilar association was shown by Dobler (1994) us- ing a different approach. Examining 55 10-ha transects scattered throughout southeastern Washington, Dobler noted that transects with Sage Sparrows had significantly higher coverage of sagebrush than those without. Taken together, these observations, based on plot-level analyses, lead to the conclusion that Sage Sparrow popu- lation levels in shrubsteppe habitat are strongly associated with sagebrush coverage. INDIVIDUAL-LEVEL ASSOCIATIONS The strength of population-level patterns in Sage Sparrows led us to investigate the behavior of individual birds, to see if individuals acted in a manner consistent with those patterns (Wiens 1985, Rotenberry and Wiens 1998). We assumed that population-level correlations reflected the aggregate response of individuals to habitat var- iation. We observed individual birds in a study area and quantified their behavior throughout a range of habitat variations, which we also quan- tified. We changed methods of measuring habitat variables to reflect the fact that we changed the scale of our focus from 600-m transects to the TABLE 1. PATCH SELECTION BY SAGE SPARROWS Patch component Selection 1: size large*** II: % sage vs. % green sage*** rabbitbrush III: shape compact, densely foliated*** IV: % sage vs. % gray sage** rabbitbrush Note: Patch components are independent axes of variation in patch attri- butes determined from principal components analysis of 900 randomly selected patches. Selection denotes direction and significance of differ- ence between randomly selected and bird-selected (N - 181) patches. See Rotenberry and Wiens 1998 for details. ** P < 0.01, *** P < 0.001. few square meters in the vicinity of an individ- ual bird. Here we emphasized the attributes of individual or small clusters of shrubs ("patch- es"), most of which were less than 2 m in can- opy diameter. As before, we concentrated on both floristic and physiognomic variables. We then asked if birds used these patches in a non- random fashion. For each of 3 yr, we randomly selected 300 patches in an 800- X 300-m sampling area to characterize the structure and composition of patches available to foraging birds. A patch was defined as a more or less contiguous association of living and/or standing dead shrub material, distinctly set off from neighboring patches and usually consisting of one or a few closely im- bricated shrubs. We measured variables relating to the size, shape, and shrub-species composi- tion of each patch. During mornings, we fol- lowed individual Sage Sparrows and marked the patches in which they foraged. During after- noons, we returned to the plot and measured the same physical and compositional attributes of bird-selected patches that we had measured on randomly selected patches. We summarized independent patterns of co- variation of attributes of randomly selected patches using principal components analysis. We scored bird-selected patches on those compo- nents and then compared those average scores to the random ones using one-way analysis of variance (ANOVA) and multiple analysis of var- iance (MANOVA). Sage Sparrows did indeed use habitat nonran- domly (Table 1; Rotenberry and Wiens 1998), and the pattern of patch use by individuals was generally consistent with the population patterns noted before: individuals used sagebrush much more often than either green or gray rabbitbrush, were associated with larger shrubs, and were seen much more frequently in compact and densely, rather than sparsely, foliated shrubs. Furthermore, because we conducted this study 98 STUDIES IN AVIAN BIOLOGY NO. 19 over a 3-yr period during which precipitation (and hence patch attributes) varied considerably, we documented that patterns of use by birds changed in concert with variation in the patch variables we measured. For Sage Sparrows, the average scores on patch components not only varied significantly among years (MANOVA: F = 11.11; df = 4, 1076; P < 0.001), but these scores were also significantly correlated with changes in random patch components (r = 0.93, df -- 6, P < 0.001). In other words, not only did individual birds use features of the habitat non- randomly, they also tracked changes in those features from one year to the next. PROJECTING BIRD-HABITAT RELATIONSHIPS THROUGH SPACE AND TIME Results from the individually based studies, when combined with the strong patterns of cor- relations noted at the plot level, led us to expect that we would see variation in avian population numbers that closely matched temporal and spa- tial variation in habitat parameters. We expected that populations would be consistent in their ex- pression of the detailed habitat associations de- scribed by the various correlations and that we could take those correlations, combine them with values of habitat variables at any given site or time, and accurately predict bird numbers. To assess temporal consistency, we continued to census birds and measure habitat variables on a representative subset of five of our original sample sites for 4 yr or longer (Rotenberry 1986). We then ran the new habitat measure- ments through the previously derived multiple regression model (see above; Rotenberry 1986) to generate predicted bird abundances. We com- pared predicted with observed abundances; if the correlation was high, we inferred consisten- cy in the expression of the details of habitat re- lationships through time. For Sage Sparrows, the correlation between predicted and observed abundances was essen- tially nonexistent (r = -0.07, df = 18, P > 0.75; Rotenberry 1986). What was previously the best-fitting model (Sage Sparrow abundance and shrub-species principal components), with an R 2 of 70%, now explained less than 1% of the var- iation in Sage Sparrow abundances. This poor fit was not the result of some peculiarity of the sites selected for continued sampling, as a cross-val- idation correlation between predicted and ob- served abundances during the initial sampling period was 0.80 (df = 13). To assess the degree of spatial consistency in habitat association, we selected four additional shrubsteppe sites that were sampled during the original time period but were not included in the original analysis. Two of these were located in the same general area as the original sites, and two were located in similar habitat but about 500 km north in southeastern Washington (Roten- berry 1986). As before, we applied the original multiple regression model to the habitat mea- surements for the new sites to generate expected bird abundances which could then be compared to observed abundances. Because the sample size was small, however, correlation coefficients were too weak to detect a good relationship; in- stead, we used a t-test (each value predicted from the regression analysis had an associated standard error). If observed abundances were close to predicted abundances, a t-test would not be significant; if the t-test was significant, there would be a serious discrepancy between ob- served and predicted abundances. Results from these tests were inconsistent. The model predicted accurately for Sage Spar- rows at both of the distant sites (i.e., the prob- ability that the observed abundance at a site was sampled from the population estimated by the multiple regression was > 0.05) but failed at both of the near sites (the same probability was < 0.05; Rotenberry 1986). Finally, we wanted to know what happens when we intentionally modify the environment in a quasi-experimental design: do Sage Spar- rows respond in ways that are consistent with their previous habitat responses and thus are pre- dictable from the original correlations? We were afforded an excellent opportunity to address this question by assessing the effects of a large-scale habitat manipulation "experiment" conducted by the Oregon State Land Board. One of our original plots (Guano Valley) was included in an approximately 75-km 2 area that was sprayed with a broad-spectrum herbicide in an attempt to eradicate sagebrush. The area was subse- quently chained and reseeded with non-native grasses (mainly crested wheatgrass [Agropyron desertorum]) to make it more suitable for cattle- grazing. The herbicide treatment came at the end of our initial 3-yr monitoring period, so we con- tinued to visit the site to survey habitat and birds for an additional 3 yr postspray (Wiens and Ro- tenberry 1985). The treatment had an immediate effect: sagebrush coverage was much reduced and there was a substantial change in the struc- ture of the habitat (Wiens and Rotenberry 1985). Even after 4 yr the site had not recovered; sage- brush coverage remained low, whereas green rabbitbrush and grasses had increased consider- ably, filling in much of the previously bare areas between shrubs. How well did Sage Sparrows respond to this manipulation given their previous habitat rela- tionships? Not well at all, at least in terms of HABITAT ASSOCIATIONS OF SAGE SPARROWS--Rotenberry and Knick 99 200 150 100 50 0 ---observed 1977 1976 1979 1960 1981 1982 1983 Year FIGURE 1. Observed abundance of Sage Sparrows before and after major habitat modification (herbicide, chaining, reseeding) at Guano Valley, Oregon. Pre- dicted values (-+ 1 SE) from multiple regression of Sage Sparrow abundance on shrub-species components (R 2 = 0.70, N = 42, P < 0.001; Rotenberry 1986). Hatched area denotes year in which treatment was ap- plied. '5 0.6  0.,4 . 0.2 o.o lOO 300 ha) analysis included average size of shrubland or grassland patches, propor- tion of landscape in shrubland or grassland, and spatial similarity (related to fractal dimension; Palmer 1988). We used logistic regression of Sage Sparrow presence/absence to develop a habitat-selection model based on both local and landscape variables (e.g., Manly et al. 1993). The probability of the presence of Sage Spar- rows at a sampling point increased with increas- ing spatial similarity of sites (i.e., decreased hab- itat heterogeneity over the landscape scale), in- creasing shrubland patch size, and increasing lo- cal coverage of sagebrush and shadscale (Atriplex confertifolia; Fig. 2). Standardized es- timates of regression coefficients implied that landscape features were more important in pre- dicting presence than was coverage of individual shrub species, and that sagebrush was more im- portant than shadscale (Knick and Rotenberry 1995). Furthermore, it was clear that landscape features interacted with local habitat variables to alter the expression of probability of occurrence. For example, the form of the relationship be- tween percent cover of sagebrush and probabil- ity of occupancy by Sage Sparrows, although always positive (hence consistent with both plot- 100 STUDIES IN AVIAN BIOLOGY NO. 19 1.0  Small shrub patch o_ 0.2 _/' 0.0  i   20 40 60 80 100 Sagebrush cover (%) FIGURE 3. Relationship between probability of oc- cupancy by Sage Sparrows and local sagebrush cover for large (105 m 2) and small (10 2 m 2) average shrub patch sizes in a landscape. These curves are slices through the surface in Fig. 2. and individual-level results), changed shape as a function of shrub patch size (Fig. 3). Thus, dif- ferences in the structure of landscapes in which local plots are embedded may produce differ- ences in apparent bird-habitat relationships de- rived from those local plots. As we sampled a subset of 66 point counts for 4 consecutive years, we were also able to assess habitat correlat&s of temporal persistence. We classified points into three categories based on Sage Sparrow occurrence rates: unoccupied (never present), marginal (present 1-2 yr), or oc- cupied (present 3-4 yr). We then contrasted sites with different occupancy rates using canonical discriminant analysis applied to the set of local and landscape habitat variables. There were significant differences among habitat attributes associated with temporal per- sistence (MANOVA: F = 4.97; df = 2, 63; P < 0.001), and the first canonical axis (the only one statistically significant at P < 0.05) explained 63% of the variation in occupancy rates (Fig. 4). Persistence increased with increasing local cover of sagebrush and shadscale, increasing propor- tion of the landscape in shrubland and the min- imum size of shrubland patches, and decreasing local cover of Russian thistle (Salsola iberica [ = kali]), a species associated with severe habitat disturbance. As before, landscape variables were as important as local variables in determining habitat associations. SUMMARY AND CONCLUSIONS Although the behavior of individual Sage Sparrows was generally consistent with bird- habitat associations derived from plot-based cor- 16 14 12 u_ 6 4 2 0 R 2 = 0.63 never present present 1-2 yrs present 3-4 yrs -4 -2 0 2 4 6 Canonical Discriminant Analysis increasing sagebrush, shadscale, shrub patch size decreasing disturbance (Russian thistle) FIGURE 4. Histogram of canonical scores for sites where Sage Sparrows were observed in 0, 1-2, or 3- 4 yr at 66 sites surveyed in 3 consecutive years. In- terpretation of canonical discriminant axis is based on correlations of original variables with canonical scores. relations of population abundances, those pop- ulation-level relationships did not match habitat variation projected through space or time, or via habitat manipulation. From these observations and subsequent analyses, we draw two general conclusions. First, although there appears to be a basic component to habitat association in Sage Spar- rows (i.e., they are rarely found in areas in the Great Basin lacking sagebrush), once this coarse habitat preference is expressed, fluctuations in density become largely decoupled from the de- tails of habitat variation, both among sites and years. Thus, although there exist individual be- havioral responses that are generally appropriate to environmental variation, these do not translate readily into strong patterns linking variations in population density with variation in habitat. Such linkages may be eroded because this spe- cies is migratory; it breeds in the Great Basin, where we study it, but winters in the south- western United States and northern Mexico, up to 2,000 km farther south (Martin and Carlson 1998). We and others (e.g., Pulliam and Parker 1979, Rotenberry and Wiens 1980, Wiens and Rotenberry 1981b, Dunning and Brown 1982, Dunning 1986) have proposed that sizes of shrubsteppe passedfie populations are most like- ly regulated during winter. We assume that most mortality for postfledging and adult shrubsteppe birds occurs during migration and in winter, as is the case for many migrant passedfnes (e.g., Sherry and Holmes 1995). Thus, a breeding ter- ritory may become empty (and hence influence HABITAT ASSOCIATIONS OF SAGE SPARROWSRotenberry and Knick 101 estimates of population density) not because of its intrinsic character but because its previous owner perished 1,500 km to the south. If breed- ing-bird densities are not near saturation, as seems to be the case in our system (Rotenberry and Wiens 1980, Wiens and Rotenberry 1981a, Wiens 1985), then otherwise suitable territories may remain untilled for several years. The su- perposition of strong site tenacity for returning breeders, even in the face of substantial habitat alteration (Wiens and Rotenberry 1985, Wiens et al. 1986), further exacerbates this decoupling. Our second conclusion is that landscape-level features are as important as local-level features in determining patterns of local occupancy and abundance, as well as temporal persistence, of Sage Sparrows and other shrubland species (Knick and Rotenberry 1995, 1999). Perhaps more importantly, differences in the structure of landscapes in which local study plots are em- bedded may produce differences in apparent bird-habitat relationships derived from local plots. This may account not only for the inability to project plot-level relationships through time and space (because relevant habitat [landscape] variables have been omitted in deriving those relationships) but also for discrepancies in rela- tionships observed between different studies (e.g., Wiens and Rotenberry 1981a, Petersen and Best 1987) or in the same study over a large geographical area (e.g., Collins 1983). The foregoing observations carry important implications for how we study bird populations in a conservation context. Many birds breeding in temperate North America are migratory and thus manifest many of the same traits as Sage Sparrows. Because most bird species in North America migrate, studies of these species con- cern "open" ecological systems. Conditions far beyond the local plot's boundaries influence these birds on their breeding grounds. Instead of finding breeding populations in equilibrium at carrying capacity, we might expect to find a pat- tern of habitat occupancy not well explained in terms of local biotic conditions. The conserva- tion and management implications are clear: at- tempts to frame management policies based on assessing the effects of various treatments (e.g., herbicide application in our case), or any other scheme that alters habitat structure and compo- sition, will require much more than a single be- fore-and-after survey to determine those effects. Because systems such as shrubsteppe may be variable even when unaltered, it will take more than a single survey to determine the "normal" state of such systems prior to treatment (Wiens 1981). Clearly, natural-resource management programs will require a long-term perspective. Constraints imposed by the immediate need for management decisions prompted by political considerations may favor short-term studies or experiments as being better than none at all. In any system characterized by any of the attributes we have discussed, however, short-term studies may yield incomplete and even misleading re- sults. Thus, we strongly support the caveat brought forth by Petersen and Best (1999): in- adequate study design and duration can lead to inaccurate conclusions and misdirected conser- vation efforts. Although our principal argument is for the ne- cessity of considering landscape-level influences on habitat associations, we do not mean to imply that research focused on individual breeding birds should be secondary. Although individu- ally based patterns of habitat associations may be too variable to predict bird abundances effec- tively, they are nonetheless key to understanding how organisms are adapted and how they inter- relate with other species. On the breeding ground we can still study foraging behavior, mate selection, predator avoidance, reproductive success, and a host of other features that con- tribute to the fitness of individuals and popula- tions. Although populations of these individuals may not be in equilibrium with respect to habitat and its resources, this does not mean natural se- lection has stopped shaping other adaptations of these species. Perhaps most importantly, it is at the local level that we are most likely to deter- mine the biological mechanisms that produce landscape-level associations (e.g., nest preda- tion, nest parasitization; Johnson and Temple 1990). FUTURE DIRECTIONS The analyses presented above suggest three major avenues for future research. The first is to test the ability of the large-scale logistic regres- sion and discriminant models to predict occur- rences of Sage Sparrows through time and space. Although we are optimistic that the pres- ent models accurately capture the appropriate level at which habitat associations are most like- ly to be repeatable (Rotenberry 1986), we were previously optimistic about the abilities of the plot-level relationships to be projected as well. Our second goal is to document more carefully the apparent geographical variation in local hab- itat associations. We propose to do this by ex- amining a variety of data sets containing vege- tation and bird-abundance data collected by sev- eral different investigators throughout the Inter- mountain West. Finally, we are undertaking a study to examine potential mechanisms acting at the level of the individual that ultimately may be responsible for both plot- and landscape-level patterns in abundance. Our initial focus is on 102 STUDIES IN AVIAN BIOLOGY NO. 19 reproductive success as a function of local and landscape habitat features. We believe that by combining these three lines of investigation, we may approach a better understanding of the re- lationships among individual behavior, habitat selection, and population dynamics in Sage Sparrows. ACKNOWLEDGMENTS Our most recent studies have been funded by the Challenge Cost Share Program of the U.S. Bureau of Land Management in cooperation with the University of California at Riverside and by the U.S. Fish and Wildlife Service and U.S. National Biological Service (now Biological Resources Division, U.S. Geological Survey). We developed the vegetation map and Geo- graphic Information System coverages under the Glob- al Climate Change programs of the U.S. Bureau of Land Management and U.S. National Biological Ser- vice. Administrative support was provided by S. G. Coloff, M. R. Fuller, M. N. Kochert, and S. Sather- Blair. We thank J. B. Dunning, P. D. Vickery, and an anon- ymous reviewer for their insightful comments. JTR thanks J. A. Wiens for introducing him to shrubsteppe birds and for collaborating in the collection and anal- ysis of much of the data presented here. LITERATURE CITED ALLEN, T H. F., AND T B. STARR. 1982. Hierarchy: perspectives for ecological complexity. University of Chicago Press, Chicago, IL. BOLGER, D. T, A. C. ALBERTS, AND m. m. SOULS. 1991. Occurrence patterns of bird species in habitat frag- ments: sampling, extinction, and nested species sub- sets. American Naturalist 37:155-166. BOLGER, D. T., T A. SCOTT, AND J. T ROTENBERRY. 1996. Breeding bird abundance in an urbanizing landscape in coastal southern California. Conserva- tion Biology 11(2):1-16. CODY, m. L. 1981. Habitat selection in birds: the roles of vegetation structure, competition, and productiv- ity. BioScience 31:107-113. CODY, m. L. (EDITOR). 1985. Habitat selection in birds. Academic Press, New York, NY. COLLINS, S. L. 1983. Geographical variation in habitat structure of the Black-throated Green Warbler (Den- droica virenx). Auk 100:382-389. DOBLER, E C. 1994. Washington State shrubsteppe ecosystem studies with emphasis on the relationship between nongame birds and shrub and grass cover densities. Pp. 149-161 in S. B. Monsen and S. G. Kitchen (compilers). Proceedings: ecology and man- agement of annual rangelands. USDA Forest Service Gen. Tech. Rep. INT-GTR 313. USDA Forest Ser- vice Intermountain Research Station, Ogden, UT DUNNING, J. B. 1986. Shrub-steppe bird assemblages revisited: implications for community theory. Amer- ican Naturalist 128:82-98. DUNNING, J. B., AND J. H. BROWN. 1982. Summer rain- fall and winter sparrow densities: a test of the food limitation hypothesis. Auk 99:123-129. EMLEN, J. T. 1977. Estimating breeding season bird densities from transect counts. Auk 94:455-468. FUENTES, E. R., AND C. CAMPUSANO. 1985. Pest out- breaks and rainfall in the semi-arid region of Chile. Journal of Arid Environments 8:67-72. JOHNSON, R. G., AND S. A. TEMPLE. 1990. Nest pre- dation and brood parasitism of tallgrass prairie birds. Journal of Wildlife Management 54:106-111. KNICK, S. T., AND J. T. ROTENBERRY. 1995. Landscape characteristics of fragmented shrubsteppe habitats and breeding passerinc birds. Conservation Biology 9:1059-1071. KNICK, S. T., AND J. T. ROTENBERRY. 1999. Spatial dis- tribution of breeding passerinc bird habitats in a shrubsteppe region of southwestern Idaho. Studies in Avian Biology 19:104-111. KNICK, S. T., J. T. ROTENBERRY, AND T. J. ZARRIELLO. 1997. Supervised classification of Landsat thematic mapper imagery in a semi-arid rangeland by non- parametric discriminant analysis. Photogrammetric Engineering and Remote Sensing 63:79-86. MANLY, B., L. B. McDONALD, AND D. THOMAS. 1993. Resource selection by animals. Chapman and Hall, London, U.K. MARTIN, J. W., AND B. A. CARLSON. 1998. Sage Spar- row (Amphispiza bellO. In A. Poole and E Gill (ed- itors). The birds of North America, no. 326. Acad- emy of Natural Sciences, Philadelphia, PA, and American Ornithologists' Union, Washington, D.C. NoY-MEIR, I. 1973. Desert ecosystems: environment and producers. Annual Review of Ecology and Sys- tematics 4:25-51. O'NEILL, R. V., B. T. MILNE, M. G. TURNER, AND R. H. GARDNER. 1988. Resource utilization scales and landscape patterns. Landscape Ecology 2:63-69. PALMER, M. W. 1988. Fractal geometry: a tool for de- scribing spatial patterns of plant communities. Ve- getatio 75:91-102. PEARSON, S. M. 1993. The spatial extent and relative influence of landscape-level factors on wintering bird populations. Landscape Ecology 8:3-18. PETERSEN, K. L, AND L. B. BEST. 1987. Effects of pre- scribed burning on nongame birds in a sagebrush community. Wildlife Society Bulletin 15:317-325. PETERSEN, K. L., AND L. B. BEST. 1999. Design and duration of perturbation experiments: implications for data interpretation. Studies in Avian Biology 19: 230-236. PORNELUZI, P., J. C. BEDNARZ, L. J. GOODRICH, N. ZA- WADA, AND J. HOOVER. 1993. Reproductive perfor- mance of territorial Ovenbirds occupying forest fragments and contiguous forest in Pennsylvania. Conservation Biology 7:618 622. PULLIAM, H. R., AND T. A. PARKER III. 1979. Popula- tion regulation of sparrows. Fortschriftt ffir Zoologic 25:137-147. ROESTAD, J. 1991. Consequences of forest fragmenta- tion for the dynamics of bird populations: concep- tual issues and the evidence. Biological Journal of the Linnean Society 42:149-163. ROTENBERRY, J. T. 1981. Why measure bird habitat? Pp. 29-32 in D. Capen (editor). The use of multi- variate statistics in studies of wildlife habitat. USDA Forest Service Gen. Tech. Rep. RM-87. USDA For- est Service Rocky Mountain Research Station, Ft. Collins, CO. ROTENBERRY, J. T. 1982. Censusing birds in shrubsteppe HABITAT ASSOCIATIONS OF SAGE SPARROWS--Rotenberry and Knick 103 habitat. Pp. 307-309 in D. E. Davis (editor). Hand- book of census methods for terrestrial vertebrates. CRC Press, Boca Raton, FL. ROTENBERRY, J. T. 1986. Habitat relationships of shrubsteppe birds: even "good" models cannot pre- dict the future. Pp. 217-221 in J. Verner, M. L. Mor- rison, and C. J. Ralph (editors). Wildlife 2000: mod- eling habitat relationships of terrestrial vertebrates. University of Wisconsin Press, Madison, WI. ROTENBERRY, J. T., AND J. m. WIENS. 1980. Temporal variation in habitat structure and shrubsteppe bird dynamics. Oecologia 47:1-9. ROTENBERRY, J. T., AND J. m. WreNS. 1989. Reproduc- tive biology of shrubsteppe passerine birds: geo- graphical and temporal variation in clutch size, brood size, and fledging success. Condor 91:1-14. ROTENBERRY, J. T., AND J. m. WIENS. 1991. Weather and reproductive variation in shrubsteppe sparrows: a hierarchical analysis. Ecology 72:1325-1335. ROTENBERRY, J. T., AND J. m. WIENS. 1998. Foraging patch selection by shrubsteppe sparrows. Ecology 79:1160-1173. SHERRY, T. W., AND R. T. HOLMES. 1995. Summer ver- sus winter limitation of populations: what are the issues and what is the evidence. Pp. 85-120 in T. E. Martin and D. M. Finch (editors). Ecology and man- agement of neotropical migratory birds. Oxford Uni- versity Press, Oxford, U.K. WreNS, J. A. 1981. Single-sample surveys of commu- nities: are the revealed patterns real? American Nat- uralist 117:90-98. WreNS, J. A. 1985. Habitat selection in variable envi- ronments: shrub-steppe birds. Pp. 227-251 in M. L. Cody (editor). Habitat selection in birds. Academic Press, New York, NY. WreNS, J. A., AND J. T. ROTENBERRY. 1981a. Habitat associations and community structure of birds in shrubsteppe environments. Ecological Monographs 51:21-41. WIENS, J. m., AND J. T. ROTENBERRY. 198 lb. Censusing and the evaluation of arian habitat occupancy. Stud- ies in Avian Biology 6:522-531. WIENS, J. m., AND J. T. ROTENBERRY. 1985. Response of breeding passefine birds to rangeland alteration in a North American shrubsteppe locality. Journal of Applied Ecology 22:655-668. WIENS, J. A., J. T. ROTENBERRY, AND g. VAN HORNE. 1986. A lesson in the limitations of field experi- ments: shrubsteppe birds and habitat alteration. Ecology 67:365-376. Studies in Avian Biology No. 19:104-111, 1999. SPATIAL DISTRIBUTION OF BREEDING PASSERINE BIRD HABITATS IN A SHRUB STEPPE REGION OF SOUTHWESTERN IDAHO STEVEN T. KNICK AND JOHN T. ROTENBERRY Abstract. We mapped the spatial distribution of a habitat index for Sage Sparrows (Amphispiza bellO, Brewer's Sparrows (Spizella breweri), Horned Larks (Erernophila alpestris), and Western Meadowlarks (Sturnella neglecta) in shrubsteppe habitats of southwestern Idaho. Landscape-level habitat associa- tions of breeding passerinc birds were determined from presence or absence at 119 randomly located points surveyed each year from 1992 through 1995. We developed a multivariate description of habitats used by each species from variables derived from coverages in a Geographical Information System. Habitat variables were number of shrub, agriculture, and hydrography cells, mean and standard de- viation of shrub patch size, habitat richness, and a measure of spatial heterogeneity in a 1-kilometer radius around each survey point. We ranked each 50-meter cell in a Geographical Information System map by the generalized squared distance in multivariate space between values for habitat variables in the cell and the mean habitat vector for each species. We then generated a map of habitat probabilities of each species for a 200,000-hectare region in southwestern Idaho. In a verification survey at 39 sites, we correctly predicted presence or absence at approximately 80 percent of the sites for Sage and Brewer's sparrows and Western Meadowlarks but at only 36 percent of the sites for Horned Larks. Spatial distribution of habitats for breeding passerinc birds was strongly related to distribution of large shrub patches. Because fire is rapidly converting shrublands to exotic annual grasslands in this region, we expect shrubland-obligate species to decline because of habitat loss and grassland species to be- come more predominant unless management practices change. DISTRIBUCION ESPACIAL DE HABITATS DE AVES PASERIFORMES EN ß REPRODUCCION EN UNA REGION DE ESTEPA ARBUSTIVA DEL SUROESTE DE IDAHO Sinopsis. Delineamos mapas de distribuci6n espacial de un fndice de habitat para el Gorri6n de Artemisia (Amphispiza bellO, el Gorri6n de Brewer (Spizella brewen'), la Alondra Cornuda (Erernophi- la alpestris) y el Pradero Occidental (Stumella neglecta) en habitats de estepa arbustiva en el suroeste de Idaho. Las relaciones de habitat a escala de paisaje de los paseriformes reproductores se determi- naron a partir de la presencia o ausencia observada en 119 puntos escogidos al azar, cada afio entre 1992 y 1995. Desarrollamos una descripci6n multivariante de los habitats utilizados pot cada especie a partit de variables obtenidas de los datos de cobertura disponibles en un Sistema Geografico de Informaci6n. Las variables de habitat fueron: nfimero de cddas arbustivas, agrfcolas, e hidrograficas; media y desviaci6n tfpica del tamafio de rodales con matorral; riqueza del habitat; y una medici6n de la heterogeneidad espacial dentro de un radio de un ki16metro desde cada punto de censo. Ordenamos cada pixel de 50 metros de un mapa del Sistema Geografico de Informaci6n de acuerdo con la distancia cuadrada generalizada entre el valor correspondiente alas variables de habitat en el pixel y el valor medio correspondiente al vector habitat de cada especie. Luego produjimos un mapa de probabilidades de habitat para cada especie para una regi6n de 200.000 hectareas en el suroeste de Idaho. Censamos 39 sitios para verificar la presencia o ausencia de cada especie. Se predijeron correctamente en apro- ximadamente un 80 pot ciento de los sitios para los Gorriones de Artemisia y de Brewer y para los Praderos Occidentales; sin embargo, s61o se predijo un 36 pot ciento de los sitios para las Alondras Cornudas. La distribuci6n espacial de habitats para paseriformes reproductores se relacion6 estrecha- mente con la distribuci6n de rodales grandes de matorral. Como en esta regi6n el fuego esta convir- tiendo rapidamente los matorrales a pastizales ex6ticos anuales, pensmnos que, pot la pdrdida de habitat, disminuiran las especies dependientes del matorral. Pot otto lado, a menos que cambien las practicas de manejo, esperamos que las especies dependientes de los pastizales se hagan mas predo- minantes. Key Words: Amphispiza belli; Eremophila alpestris; exotic annual grassland; Geographical Infor- mation System; habitat-selection model; shrubsteppe; Spizella breweri; Sturnella neglecta. Interpreting the environment from a species' perspective is an important focus in studies of animal-habitat associations. Descriptions of hab- itat associations are well known for many spe- cies, and numerous statistical techniques exist to develop models of resource selection (e.g., John- son 1980; Alldredge and Ratti 1986, 1992; Man- ly et al. 1993). Recently, Geographical Infor- mation System (GIS) technology has advanced the capability to map the spatial distribution of single and multiple environmental variables. Just as realistic models have been developed to rep- 104 DISTRIBUTION OF BIRD HABITATS--Knick and Rotenberry 105 resent habitat use in a nonspatial context (Vemer et al. 1986), it is now possible to describe the spatial distribution of elements in the landscape as perceived by an animal by mapping appro- priate variables or indexes. It is then possible to manage landscapes for species based on con- cepts from theoretical biogeography, which have important implications for conservation biology (Urban and Shugart 1986, Temple and Cary 1988, Burkey 1989, Hansson and Angelstam 1991, Opdam 1991, Danielson 1992). For ex- ample, using maps of habitat distributions, man- agers can identify regions with high probability of use, maintain areas of sufficient size to con- tain viable populations, or identify habitat cor- ridors for dispersal. We studied four species of breeding passer- .ines in a shrubsteppe region in southwestern Ida- ho. Two of the species, Sage Sparrow (Arnphis- piza belli) and Brewer's Sparrow (Spizella brew- eri), are shrubland obligates, and two, Horned Lark (Erernophila alpestris) and Western Mead- owlark (Sturnella neglecta), are grassland spe- cies. Our first objective was to develop a re- source-selection model for each species by com- bining field surveys for species presence with landscape variables derived from a classified GIS map. We then mapped the selection func- tion for the entire study area to determine the spatial distribution of habitats potentially used by each species. Our study involved several assumptions. We assumed that the scale and selection of environ- mental variables we measured were relevant to the species (Wiens 1989) and that our multivar- iate statistical model appropriately described the species-habitat associations (Rotenberry 1986, Rotenberry and Knick 1999). We also assumed that the relative probabilities of habitat config- uration derived from the statistical model rep- resented the probability that a species would fill available habitats (e.g., Fretwell 1972, Van Home 1983, Hobbs and Hanley 1990, Vickery et al. 1992). STUDY AREA We conducted our study from 1992 through 1995 in a 200,000-ha region of shrubsteppe habitat in south- western Idaho that included portions of the Snake Riv- er Birds of Prey National Conservation Area (116 ø W, 43 ø N). The primary management mandate of this area, which was designated as a national conservation area in 1994 (U.S. Public Law 103-64; 4 August 1994), is to maintain and conserve the high densities of nesting raptors and their prey (U.S. Department of the Interior 1979). Multiple uses, including livestock grazing and military training, are permitted if compatible with rap- tor-conservation management (Kochert and Pellant 1986, U.S. Department of the Interior 1995). Wildfires are rapidly converting once-large expanses of big sagebrush (Artemisia tridentata), winterfat (Kraschenninikovia lanata), shadscale (Atriplex con- fertifolia), and other salt shrub (Atriplex spp.) com- munities in the Snake River plains into regions dom- inated by exotics such as cheatgrass (Brornus tecto- rum) or Russian thistle (Salsola kali; Whisenant 1990). As a result of numerous fires since 1980, the native shrub communities now are highly fragmented or have been converted to grassland (U.S. Department of the Interior 1996, Knick and Rotenberry 1997; Fig. 1). The highly flammable annual grasses increase fire fre- quency and reduce the potential for shrub reestablish- ment. More than 30% of the 76,910 ha that burned between 1980 and 1992 has reburned two to four times; human activities were responsible for 72% of the fire ignitions (U.S. Department of the Interior 1995). METHODS FIELD SURVEYS We conducted unlimited-radius point-count surveys (Ralph et al. 1995) each year from 1992 through 1995 to determine habitat associations of breeding passerine birds. We established 119 sites by selecting random coordinates throughout the study area in an attempt to sample all habitats in proportion to their available dis- tribution. Minimum distance between sites was more than 400 m, and we considered the sites spatially in- dependent because 97% of all observations (N = 5,757) were estimated from a distance of less than 200 m. Final coordinates of each site were determined by Global Positioning System to an accuracy of less than 5 m (August et al. 1994). Order of sampling in each year was randomly determined. Sites were sampled once each year between 0500 and 1000 on days with no precipitation and winds less than 12 km/hr. After waiting 2-5 min to reduce the disturbance from our arrival, we recorded all individ- uals we saw or heard at each site during a 5-min pe- riod. Sampling periods were 4 May-25 June 1992; 10 May-23 June 1993; 10 April-24 June 1994; and 3-29 May 1995. One observer participated in all 4 yr, one in 2 yr, and five in 1 yr. All observers participated in a 1- to 3-wk training before beginning surveys. GIS COVERAGES Our base coverage in the GIS was a vegetation map classified from Landsat thematic mapper satellite im- agery (Knick et al. 1997). Resolution of the vegetation map was 50 m (resampled from 27-m cells in the orig- inal thematic mapper satellite image), and the study area contained 1,752,340 cells. After identifying water and agriculture areas, we classified each 50-m cell in the habitat map into one of five categories: sagebrush, winterfat, shadscale, disturbance (dominated by Rus- sian thistle), or grasslands (including both cheatgrass and native grasses). Accuracy in separating shrub cells (>5% ground cover of shrub) from nonshrub cells was 80%; classification accuracy was 64% for classification of the five individual habitat classes (plus agriculture and water categories; Knick et al. 1997). Gross habitat characteristics such as percent shrub cover did not change during our study. We derived variables that described both the com- position and spatial heterogeneity of the landscape (Li 106 STUDIES IN AVIAN BIOLOGY NO. 19 0 6 10 16 I(belUterl Shrub FIGURE 1. Locations in a 200,000-ha shrubsteppe region of southwestern Idaho of (A) areas that burned between 1980 and 1992 and (B) shrub patches and grassland in 1994. Burned areas were digitized from U.S. Bureau of Land Management fire boundaries on 1:24:000 quadrangle maps. Shrub and grasslands were classified from Landsat thematic mapper satellite im- agery. and Reynolds 1994; Table 1). We developed an index of habitat diversity from the number of habitats in a 1-km radius of each cell in the vegetation map. We also determined the number of shrub cells and the av- erage size and standard deviation for shrub patches. We used the ratio of number of edges between shrub and grass cells to total the number of edges in the 1- km radius as an estimate of landscape heterogeneity. Habitat diversity was the only variable derived from the full habitat classification, which contained the low- er classification accuracy. For the other landscape var- iables we used the binary shrub/nonshrub classifica- tion. We created a coverage of all agriculture fields from a composite of a 1979 vegetation map of the Snake River Birds of Prey National Conservation Area (U.S. Department of the Interior 1979), 1993 U.S. Bureau of Reclamation agriculture maps, and our classified sat- ellite imagery. The composite agriculture map includ- ed areas of both actively growing and fallow vegeta- tion. We also created a coverage of hydrography, to include wetlands, lakes, and ephemeral or permanent streams and rivers, from U.S. Geological Survey 1: 100,000 digital line graphs. We used the number of agriculture and hydrography cells in the 1-km radius of each cell in our analyses. HABITAT ASSOCIATIONS We determined habitat associations for each species based on the mean values for the habitat variables at all sites where a species was detected. This multivar- iate habitat mean and the associated covariance matrix were then used to develop habitat maps for each spe- cies. We used a site in the analysis if a species was detected in any of the 4 yr of sampling. By including sites where a species was present in only 1 yr, we determined the optimistic range of habitat associations in our analysis because yearly variation in temporal persistence was related to habitat characteristics (Knick and Rotenberry 1995, Rotenberry and Knick 1999). Because of the relative uhiquity of Horned Larks, we defined presence of this species as more than four birds at a site. Six of the seven variables were transformed to best approximate a normal distribution as determined by Kolmogorov-Smirnov tests. Distributions of all vari- ables remained significantly different from normal (P < 0.05), but statistical power remained high because of large sample sizes in the GIS map. Although nor- mality tests failed, we proceeded without meeting this assumption using the best approximations to a normal distribution to minimize the potential bias. We trans- formed average patch size and standard deviation by log10(x + 1), number of shrub cells by x ø.4, landscape heterogeneity by x ø'8, habitat diversity by x 2, and hy- drography by x ø'9. The distribution of the agriculture variable could not be improved and was not trans- formed. HABITAT MAPS We created maps of habitat distributions for each species by first determining the generalized squared distance (or Mahalanobis distance) between the land- scape variables in each cell of the GIS map and the multivariate mean of those variables associated with a species (Clark et al. 1993, Knick and Dyer 1997). The generalized squared distance was then converted to a Chi-squared probability distribution with P (number of variables) - 1 = 6 degrees of freedom. Therefore, we simply rescaled the generalized squared distance, a di- mensionless statistic, to a probability distribution be- DISTRIBUTION OF BIRD HABITATS Knick and Rotenberry 107 TABLE 1. LANDSCAPE-LEVEL VARIABLES USED IN AN ANALYSIS OF HABITAT SELECTION BY SAGE AND BREWER'S SPARROWS, HORNED LARKS, AND WESTERN MEADOWLARKS IN A SHRUBSTEPEE REGION IN SOUTHWESTERN IDAHO Variable Description Agriculture Habitat richness Hydrography Shrubs Mean shrub patch Shrub patch variance Habitat patchiness Number of 50-m agriculture cells Number of different habitats Number of cells containing wetlands, lakes, or permanent or ephemeral streams and rivers Number of 50-m cells classified in shrubland category Mean patch size of all shrubland patches in the 1-km radius Standard deviation of all shrubland patches in the l-km radius Number of edges between shrubland and grassland cells Note: All variables were determined for a 1-km radius around each 50-m ceil in a GIS map of the study area. tween 0 and 1. As such, we determined the probability that the variables at a cell were similar to the multi- variate mean that described habitats associated with sites where a species was detected. VERIFICATION SURVEYS We conducted verification surveys in 1995 at an ad- ditional 39 sites located at random coordinates throughout the study area. Sites were classified into predicted absence or presence for each species based on a X2 probability of 0.5 for the cutpoint. RESULTS At 119 points surveyed annually from 1992 through 1995, we observed Sage Sparrows at 36 points, Brewer's Sparrows at 83, Horned Larks at 102, and Western Meadowlarks at 96. Habi- tats associated with Sage and Brewer's sparrows included large shrub patches and relatively low- er amounts of edge between shrubland and grasslands than at habitats associated with Horned Larks and Western Meadowlarks (Table 2). Maps of habitat probabilities for each cell re- flected the strong association of Sage and Brew- er's sparrows (Fig. 2) with existing shrublands (Fig. 1). At the landscape scale, burned areas were associated with low similarities to the mean habitat vectors at sites where we observed Sage and Brewer's sparrows (Fig. 1). Maps generated from the X 2 probability dis- tribution of the generalized squared distances for Horned Larks and Western Meadowlarks were more conservative in predicting the spatial ex- tent of habitats than we expected because of these species' ubiquity in the samples (Fig. 3). However, areas of predicted habitats for Horned Larks and Western Meadowlarks included both greater spatial extent and more grassland regions than maps generated for Sage and Brewer's spar- rows. Cumulative frequency distributions represent- ed the proportion of the study area relative to the mean habitat vector for each species (Fig. 4). As expected, a smaller proportion of the hab- itat in the study area was similar to the mean habitat vector associated with Sage Sparrow, the least-observed and most habitat-specific species, when compared to Brewer's Sparrow, Horned Lark, or Western Meadowlark. We correctly predicted presence or absence at 79% of the verification sites for Sage Sparrows and Western Meadowlarks, at 82% of the sites for Brewer's Sparrows, but at only 36% of the sites for Horned Larks (Table 3). In most clas- sification errors, a species was present at a site where the selection model predicted absence. Horned Larks, present at 22 of 39 sites where TABLE 2. SUMMARY STATISTICS FOR HABITAT VARIABLES AT SURVEY SITES IN SOUTHWESTERN IDAHO FOR SAGE AND BREWER'S SPARROWS, HORNED LARKS, AND WESTERN MEADOWLARKS Sage Sparrow Brewer's Sparrow Horned Lark Western Meadowlark (N = 36) (N = 83) (N = 102) (N = 96) Habitat variable f{ SE f SE  SE  SE Agriculture (no. cells) 21.3 13.0 34.0 11.8 28.2 9.6 36.6 11.0 Habitat richness 6.1 0.1 6.1 0.1 6.1 0.1 6.0 0.1 Hydrography (no. cells) 37.4 4.8 45.4 3.7 48.1 3.7 51.1 3.6 Shrubs (no. cells) 854.0 46.5 647.6 38.2 466.2 34.7 502.5 33.4 Mean shrub patch (km 2) 47.0 9.1 30.2 5.2 22.6 4.6 14.5 2.5 Shrub patch variance 19.1 2.9 14.5 1.7 11.8 1.8 11.6 1.6 Habitat patchiness 405.6 33.2 439.6 22.4 462.5 22.6 471.3 21.2 Note: Sample sizes (N) are the number of sites where presence was recorded at 119 sites surveyed annually from 1992 through 1995. 108 STUDIES IN AVIAN BIOLOGY NO 19 A. Sage Sparrow Ix, I A. HornmJ Lark FIGURE 2. Spatial distribution of habitats for (A) Sage Sparrow and (B) Brewer's Sparrow in a 200.000- ha shrubsteppe region of southwestern Idaho. Habitat rank is the X 2 probability that habitats in individual map cells were similar to the multivariate mean habitat vector associated with the species presence at survey sites (ranks closest to 100 have the highest probability of similarity). FIGURE 3. Spatial dstribution of habitats for (A) Horned Lark and (B) Western Meadowlark in a 200.000-ha shrubsteppe region of southwestern Idaho. Habitat rank is the X 2 probability that habitats in in- dividual map cells were similar to the multivariate mean habitat vector associated with the species pres- ence at survey sites (ranks closest to 100 have the highest probability of similarity). absence was predicted, represented the extreme in these errors (Table 3). DISCUSSION Spatial distribution of habitats for the two shrubland-obligate species, Sage and Brewer's sparrows, was clearly related to existing shrub- lands. Because large-scale fires have converted shrublands to exotic annual grasslands with in- creased fire frequency, we expect that habitats for shrubland-obligate species will continue to DISTRIBUTION OF BIRD HABITATS--Knick and Rotenberry 109 lOO 9o  80 c 70 0' 60 ß 50 = 40 ; 30  20 10 '" -- Western Meadowlark --. Srewer'sSpsrrow /:.': , 'i'" , 25 50 75 100 125 150 Generalized squared distance Mean Habitat Vector FIGURE 4. Cumulative frequency distribution of generalized squared distances for Sage and Brewer's sparrows, Horned Larks, and Western Meadowlarks in a 200,000-ha shrubsteppe region of southwestern Ida- ho. Relative shift in distribution toward the right in- dicates greater proportions of areas that are less similar to habitats (mean habitat vector) where the species was observed. decline. We expect the current trajectory of hab- itat changes to have a particularly adverse effect on Sage Sparrow habitats. Since 1979 fires have destroyed more than 30% of the existing shrublands in our study area (U.S. Department of the Interior 1996). We do not know the fire regime of presettlement peri- ods, but large-scale fires, although present, like- ly were much less frequent than they now are because of the difference in grassland understo- ry (Whisenant 1990). Because of cheatgrass in- vasion into this system in the late 1800s and early 1900s, continuous fuels are now omni- present in the understory and facilitate fire- spread. In addition, cheatgrass cures earlier than native grasses, thus increasing the length of the fire season (Klemmedson and Smith 1964). The larger and more frequent fires in the present dis- turbance regime have either eliminated or wide- ly dispersed the existing seed sources of shrub species (Knick and Rotenberry 1997). The po- tential for recovery of shrubs, such as sagebrush, is far outpaced by the rate of loss. Thus, the system has lost much of the once-dominant shrubland and now exists in a new grassland state that potentially represents a habitat sink from which shrub recovery by natural means of gradual recolonization by seedling establishment is unlikely or extremely long-term. Our study demonstrated the potential of land- TABLE 3. ERROR MATRICES FOR PRESENCE OR AB- SENCE PREDICTED BY RESOURCE-SELECTION MODELS DE- VELOPED FOR SAGE AND BREWER'S SPARROWS, HORNED LARKS, AND WESTERN MEADOWLARKS AT 39 SITES IN A SHRUBSTEPPE REGION OF SOUTHWESTERN IDAHO Known Species Predicted Absent Present Total Sage Sparrow Absent 27 5 32 Present 3 4 7 Total 30 9 39 Brewer's Sparrow Absent 24 6 30 Present 1 8 9 Total 25 14 39 Horned Lark Absent 12 22 34 Present 3 2 5 Total 15 24 39 Western Meadowlark Absent 28 7 35 Present 1 3 4 Total 29 10 39 scape-scale attributes to determine habitats for shrubland-obligate species; we correctly predict- ed presence/absence at approximately 80% of the verification points. Both species persisted in burned areas when measured at a local scale (< 10 ha; Petersen and Best 1987, 1999), but local plots still were embedded in larger-scale land- scapes of shrubland. Loss of shrublands at our larger scale of investigation (1-km radius around each point) was reflected in complete absence of habitat for shrubland-obligate species, and those species were not present. Our technique clearly represented the spatial distribution of habitat for shrubland-obligate specialists, such as Sage and Brewer's sparrows, and for Western Meadowlark, a grassland spe- cies. For these specialists, the mean and covari- ance matrix represented the distribution of the habitats used by the species. For more generalist species, however, such as Horned Lark, the mean and covariance matrix more likely repre- sented the distribution of habitats in the study area rather than a species-habitat association. Thus, the generalized distance from the mean habitat vector may not represent the wide range (or variance) of habitats used by generalist spe- cies. As the species-habitat association becomes more general, the mapped distribution changes from the mean habitat vector of the species to represent the mean configuration of habitats in the study area. Based on verification surveys, our habitat maps were consistently conservative in predict- ing species presence, despite using a relatively liberal definition of habitats used (species pres- ence at a site in any of 4 yr) to define the mul- tivariate mean habitat. By using a narrower def- 110 STUDIES IN AVIAN BIOLOGY NO. 19 inition, such as species presence at a site in all years, map predictions likely would underesti- mate further the actual distribution of habitats. Alternatively, we could change the X 2 probabil- ity that defines presence or absence, or simply rescale the generalized squared distance into user-defined quantiles (e.g., Knick and Dyer 1997). When rescaled into quantiles, the cate- gories then represent a percent of the study area in each class (e.g., the top 10% of the study area) rather than a probability of similarity to the multivariate habitat mean. ACKNOWLEDGMENTS Our study was funded by the Challenge Cost Share program of the U.S. Bureau of Land Management in cooperation with the University of California at Riv- erside and Boise State University and by the U.S. Fish and Wildlife Service and U.S. National Biological Ser- vice (now Biological Resources Division, U.S. Geo- logical Survey). The vegetation map and GIS cover- ages were developed under the Global Climate Change programs of the U.S. Bureau of Land Management and U.S. National Biological Service. We appreciate ad- ministrative support from S. G. Coloff, M. R. Fuller, M. N. Kochert, and S. Sather-Blair. P. Bates, D. Dyer, W. Dyer, D. Johnson, G. Kaltenecker, T Katzner, R. Lara, J. Younk, and T. Zarriello assisted with various aspects of the study. We appreciate the manuscript re- view by P. D. Vickery and two anonymous reviewers. LITERATURE CITED ALLDREDGE, J. R., AND J. T RATI. 1986. Comparison of some statistical techniques for analysis of re- source selection. Journal of Wildlife Management 50:157-165. ALLDREDGE, J. R., AND J. T RATTI. 1992. Further com- parison of some statistical techniques for analysis of resource selection. Journal of Wildlife Management 56:1-9. AUGUST, P., J. MICHAUD, C. LABASH, AND C. SMITH. 1994. GPS for environmental applications: accuracy and precision of locational data. Photogrammetric Engineering and Remote Sensing 60:41-45. BURKEY, T V. 1989. Extinction in nature reserves: the effect of fragmentation and the importance of mi- gration between reserve fragments. Oikos 55:75-81. CLARK, J. D., J. E. DONN, AND K. G. SMITH. 1993. A multivariate model of female black bear habitat use for a geographic information system. Journal of Wildlife Management 57:519-526. DANIELSON, B. J. 1992. Habitat selection, interspecific interactions and landscape composition. Evolution- ary Ecology 6:399-411. FRETWELL, S. D. 1972. Populations in a seasonal en- vironment. Princeton University Press, Princeton, NJ. HANSSON, L., AND 1 ANGELSTAM. 1991. Landscape ecology as a theoretical basis for nature conserva- tion. Landscape Ecology 5:191-201. HOBBS, N. T., AND r. A. HANLEY. 1990. Habitat eval- uation: do use/availability data reflect carrying ca- pacity? Journal of Wildlife Management 54:515- 522. JOHNSON, D. H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65-71. KLEMMEDSON, J. O., AND J. G. SMITH. 1964. Cheatgrass (Brornus tectorurn L.). Botanical Review 30:226- 262. KNICK, S. r., AND D.C. DYER. 1997. Distribution of black-tailed jackrabbit habitat determined by GIS in southwestern Idaho. Journal of Wildlife Manage- ment 61:75-85. KNICK, S. r., AND J. T ROTENBERRY. 1995. Landscape characteristics of fragmented shrubsteppe habitats and breeding passerine birds. Conservation Biology .9:1059-1071. IG'qICK, S. T, AND J. T ROTENBERRY. 1997. Landscape characteristics of disturbed shrubsteppe habitats in southwestern Idaho (U.S.A.). Landscape Ecology 12:287-297. KNICK, S. r., J. T. ROTENBERRY, AND r. J. ZARRIELLO. 1997. Supervised classification of Landsat thematic mapper imagery in a semi-arid rangeland by non- parametric discriminant analysis. Photogrammetric Engineering and Remote Sensing 63:79-86. KOCHERT, M. N., AND M. PELLANT. 1986. Multiple use in the Snake River Birds of Prey Area. Rangelands 8:217-220. Ct, H., AND J. E REYNOLDS. 1994. A simulation exper- iment to quantify spatial heterogeneity in categorical maps. Ecology 75:2446-2455. MANLY, B., L. McDONALD, AND D. THOMAS. 1993. Re- source selection by animals. Chapman and Hall, London, U.K. OPDAM, P. 1991. Metapopulation theory and habitat fragmentation: a review of holarctic breeding bird studies. Landscape Ecology 5:93-106. PETERSEN, K. C., AND C. B. BEST. 1987. Effects of pre- scribed burning on nongame birds in a sagebrush community. Wildlife Society Bulletin 15:317-325. PETERSEN, K. C., AND C. B. BEST. 1999. Design and duration of perturbation experiments: implications for data interpretation. Studies in Avian Biology 19: 230-236. RALPH, C. J., J. R. SAUER, AND S. DROEGE (EDITORS). 1995. Monitoring bird populations by point counts. USDA Forest Service Gem Tech. Rep. PSW-GTR- 149. USDA Forest Service, Albany, CA. ROTENBERRY, J. T. 1986. Habitat relationships of shrubsteppe birds: even "good" models cannot pre- dict the future. Pp. 217-221 in J. Verner, M. L. Mor- rison, and C. J. Ralph (editors). Wildlife 2000: mod- eling habitat relationships of terrestrial vertebrates. University of Wisconsin Press, Madison, WI. ROTENBERRY, J. T., AND S. T KNICK. 1999. Multiscale habitat associations of the Sage Sparrow: implica- tions for conservation biology. Studies in Avian Bi- ology 19:95-103. TEMPLE, S. A., AND J. R. CARY. 1988. Modeling dy- namics of habitat-interior bird populations in frag- mented landscapes. Conservation Biology 2:340- 347. URBAN, D. L., AND H. H. SHUGART, JR. 1986. Avian demography in mosaic landscapes: modeling para- digm and preliminary results. Pp. 273-280 in J. Ver- ner, M. L. Morrison, and C. J. Ralph (editors). Wild- life 2000: modeling habitat relationships of terres- DISTRIBUTION OF BIRD HABITATS--Knick and Rotenberry 111 thai vertebrates. University of Wisconsin Press, Madison, WI. U.S. DEPARTMENT OF THE INTERIOR. 1979. Snake River Birds of Prey special research report to the Secretary of the Interion U.S. Bureau of Land Management, Boise, ID. U.S. DEPARTMENT OF THE INTERIOR. 1995. Snake River Birds of Prey National Conservation Area manage- ment plan. Bureau of Land Management, Lower Snake River District Office, Boise, ID. U.S. DEPARTMENT OF THE INTERIOR. 1996. Effects of military training and fire in the Snake River Birds of Prey National Conservation Area. BLM/ IDARNG research project final report. U.S. Geolog- ical Survey, Snake River Field Station, Boise, ID. V^N HORNE, B. 1983. Density as a misleading indi- cator of habitat quality. Journal of Wildlife Manage- ment 47:893-901. VERNER, J., M. L. MORRISON, AND C. J. RALPH (EDI- TORS). 1986. Wildlife 2000: modeling habitat rela- tionships of terrestrial vertebrates. University of Wisconsin Press, Madison, WI. VICKERY, P. D., M. L. HUNTER, JR., AND J. V. WELLS. 1992. IS density an indicator of breeding success? Auk 109:706-710. WHISENANT, S. G. 1990. Changing fire frequencies on Idaho's Snake River plains: ecological and manage- ment implications. Pp. 4-10 in E. D. McArthur, E. M. Romney, S. D. Smith, and P. T Tueller (editors). Proceedings: symposium on cheatgrass invasion, shrub die-off, and other aspects of shrub biology and management. USDA Forest Service Intermountain Research Station, Ogden, UT WIENS, J. A. 1989. Spatial scaling in ecology. Func- tional Ecology 3:385-397. Studies in Avian Biology No. 19:112-121, 1999. HABITAT RELATIONS AND BREEDING BIOLOGY OF GRASSLAND BIRDS IN NEW YORK CHRISTOPHER J. NORMENT, CHARLES D. ARDIZZONE, AND KATHLEEN HARTMAN Abstract. In 1994 we began a study of the habitat relations and breeding biology of grassland birds in western New York. Most fields contained fewer than four grassland species, with Bobolink (Doli- chonyx oryzivorous) and Savannah Sparrow (Passerculus sandwichensis) being the two most common species. Species of management concern in the Northeast, such as Henslow's Sparrow (Ammodramus henslowii) and Upland Sandpiper (Bartramia longicauda), were absent from the study area. Bird- habitat models generated through Principal Components Analysis and stepwise multiple regression indicated that field area, or variables correlated with area, explained most of the variation in overall grassland bird species richness (partial r 2 = 0.43) and abundance (partial r 2 - 0.60) and in the abun- dance of Bobolinks and Savannah Sparrows. Grassland birds were generally absent from fields smaller than 5 hectares. Areas with few shrubs and low horizontal heterogeneity supported more grassland bird species than did fields with more shrubs and high horizontal heterogeneity, and fields with shorter, less dense vegetation had more individuals than did fields with taller, dense vegetation. Few grassland birds occurred in fields planted in switchgrass (Panicum virgatum) monocultures. More than 90 percent of all known nesting pairs fledged young by the end of the first week in July. Nest success was generally high; the proportion of nests fledging one or more young was 0.76 for Savannah Sparrows, 0.54 for Bobolinks, and 0.67 for Eastern Meadowlarks (Sturnella magna). Grassland bird populations in this study may benefit from management practices that increase field area, control shrub invasion, and encourage the growth of grasses other than switchgrass. The current low levels of grazing at Iroquois National Wildlife Refuge, with cattle allowed in pastures only after 15 July, do not appear to be harmful to grassland bird populations. LAS RELACIONES ENTRE LOS H,BITATS Y LA BIOLOGA REPRODUCTIVA DE AVES DE PASTIZAL EN NUEVA YORK Sinopsis. En 1994 iniciamos un estudio de las relaciones entre los hibitats y la biologœa reproductiva de aves de pastizal en el oeste de Nueva York. La mayorfa de los campos tenfan menos de cuatro especies de pastizal, con el Tordo Arrocero (Dolichonyx oo'zivorous) y el Gorri6n Sabanero (Passer- culus sandwichensis) como las dos especies mils comunes. Las especies de importancia para manejo en el noreste, como el Gorri6n de Henslow (Ammodramus henslowii) y el Zarapito Ganga (Bartramia longicauda) estaban ausentes del irea de estudio. Los modelos de hibitat para aves producidos por el Anfilisis de Componentes Principales y las regresiones mfiltiples de escala indicaron que el irea del campo (o las variables correlacionadas con el irea) daban cuenta de la mayor parte de la variaci6n de la riqueza total de especies de aves de pastizal (parcial r 2 = 0,43), de la abundancia total de elias (parcial r 2 - 0,60) y de la abundancia de los Tordos Arroceros y los Gorriones de Henslow. Las aves de pastizal generalmente estaban ausentes en los campos de menos de 5 hectfireas. Las ireas con pocos arbustos y una escasa heterogeneidad horizontal mantenfan mis especies de aves de pastizal que los campos con mis arbustos y una abundante heterogeneidad horizontal; los campos con vege- taci6n mis baja y menos densa tenfan mis individuos que los campos con vegetaci6n mis alta y densa. Habfa pocas aves de pastizal en campos sembrados con monoculturas de Panicurn virgatum. Mis de un 90 por ciento de todas las parejas conocidas con nidos produjeron pollos para el fin de la primera semana de julio. E1 6xito de los nidos fue generalmente alto; la proporci6n de los nidos que produjeron un pollo o mis fue 0,76 para los Gorriones Sabaneros, 0,54 para los Tordos Arroceros y 0,67 para los Praderos Orientales (Sturnella magna). Las poblaciones de aves de pastizal pueden beneficiarse con las pricticas de manejo que aumenten el irea de los campos, controlen la invasi6n de arbustos y estimulen el crecimiento de hierbas que no sean Panicum virgatum. Los bajos niveles actuales de apacentamiento en el Refugio Nacional de Fauna Iroquois, con ganado permitido en las praderas solamente despu6s del 15 de julio, no parecen ser dafiinos para las poblaciones de aves de pastizal. Key Words: Bobolink; breeding biology; Dolichonyx oryzivorous; Eastern Meadowlark; grassland birds; habitat selection; New York; Passerculus sandwichensis; Savannah Sparrow; Sturnella magna. Populations of many grassland bird species in the United States have declined significantly since the mid-1960s (Robbins et al. 1986, Knopf 1994). Although declines of North American breeding birds may vary across geographic re- gions (James et al. 1992; Peterjohn and Sauer 1994, 1999; Herkert 1995a), the trend evident for grassland birds is consistent across North America (Robbins et al. 1986, Bollinger and Ga- vin 1992, Smith and Smith 1992, Askins 1993, Peterjohn and Sauer 1999). Reasons for declines of grassland birds in the northeastern United 112 GRASSLAND BIRDS IN NEW YORK Norment et al. TABLE 1. TYPES AND SIZES OF FIELDS CENSUSED FOR GRASSLAND BIRDS IN WESTERN NEW YORK, 1995 113 Sample sizes Iroquois Montezuma Braddock Bay Habitat type NWR a NWR b WMA Total Size (ha) Cool-season grassland c 8 I 9 5.1-20.1 Warm-season grassland 6 1 7 1.3-44 Pasture 3 3 19.0 98.4 Fallow farm field 4 1 5 5.0-14.0 Forb-dominated field 5 2 7 3.0-32.9 Old field with shrubs 8 4 12 2.0-14.6 Total 34 8 1 43 1.3-98.4 Includes Oak Orchard and Tonawanda WMAs. Includes NYSDEC lands in the Northern Montezuma Wetlands Complex. Habitat descriptions given in Appendix. States include farmland abandonment, decline of hayfield area, and earlier and more frequent hay- cropping rotations (Andrle and Carroll 1988, Bollinger and Gavin 1992). Many species of grassland birds are area sensitive and are partic- ularly vulnerable to loss of grassland habitat (Smith and Smith 1990, Vickery et al. 1994). In the Northeast, grassland habitat has declined by about 60% since the 1930s (Vickery et al. 1994). In 1994 we began a study of grassland birds on lands in western New York administered by the U.S. Fish and Wildlife Service (USFWS) and New York State Department of Environ- mental Conservation (NYSDEC). Our objectives were to determine grassland bird species rich- ness and abundance, breeding biology, and hab- itat relations on these lands. Studies on breeding biology focused on determining nest success and chronology, whereas bird-habitat relations were examined at both the local (vegetation) and landscape levels. Results of this study will be used to evaluate the status of grassland bird pop- ulations on public lands in the Great Lakes Plain of western New York and to suggest manage- ment alternatives to increase grassland bird pop- ulations in the region. STUDY AREA We began our study in May 1994 at Iroquois Na- tional Wildlife Refuge (NWR) and the contiguous NYSDEC Tonawanda and Oak Orchard Wildlife Man- agement Areas (WMAs), located about 65 km west of Rochester, New York, in the Great Lakes Plain eco- zone of New York (Andrle and Carroll 1988). The area comprises approximately 8,000 ha of wetlands and up- lands that historically has been managed to provide habitat for breeding and migratory waterfowl (Iroquois NWR 1993). More than 1,000 ha of potential upland habitat for grassland birds also exist in the area. This potential habitat includes fields managed as cool-sea- son grasslands, warm-season grasslands dominated by switchgrass (Panicum virgatutn), old fields with a grass/forb/shrub mix, fallow farm fields, forb-domi- nated fields, and pastures (Iroquois NWR 1990; Table 1; see Appendix for a description of habitat types). The various fields range in size from 0.5 to 98 ha, inter- spersed in a landscape matrix of wetlands, croplands, and hardwood forests. In 1995 we expanded the study to include two ad- ditional sites: a 44-ha warm-season grassland at Beat- tie Point in the NYSDEC Braddock Bay WMA, 11 km west of Rochester, New York, and approximately 55 ha of upland habitat in the Northern Montezuma Wet- lands Complex, about 50 km west of Syracuse, New York, and administered by the USFWS and NYSDEC. The grassland at Braddock Bay is on the southern shore of Lake Ontario and is bordered on three sides by extensive wetlands. The upland habitat in the Northern Montezuma Wetlands Complex is surround- ed by a mixture of wetlands, agricultural fields, and deciduous forest. METHODS We determined grassland bird species abundance and richness using fixed 50-m-radius point counts. We established 59 census points in 34 fields in 1994 and 82 points (the same 59 points plus an additional 23) in 43 fields in 1995. These fields represented the range of shmb/grassland habitats found in the study area (Ta- ble 1). To control for area-related differences in sam- pling intensity, we placed no more than one census point in fields smaller than 10 ha and maintained a density of approximately one census point per 7 ha in larger fields. In fields with more than one census point, We separated point centers by at least 200 m to mini- mize recounts. Each point was censused five times a year for 10 min per census. We conducted censuses between 0600 and 1000 eastern standard time from 15 May to 1 July. For each point, we recorded the number of species, individuals per species, and total number of individuals seen and/or heard during the 10 min. For fields with more than one census point, we aver- aged bird abundance across points and censuses to ob- tain the mean number of individuals per census per point for the field. We also searched the study area for species of management concern at either the state or federal level (e.g., Henslow's Sparrow [Ammodramus henslowii] and Upland Sandpiper [Bartramia longi- cauda]) by walking transects and broadcasting songs in likely habitat. We monitored nests of grassland and old-field spe- cies at Iroquois NWR in 1994, 1995, and 1996 to de- termine nest success and chronology for grassland 114 STUDIES IN AVIAN BIOLOGY NO. 19 birds; we restricted intensive nest searches and moni- toring to this site because of time constraints. We io- cated nests either by dragging ropes or by following birds to their nests. All nests located were marked with a small piece of flagging 5 m north of the nest and were checked at approximately 3-d intervals until fledging. We recorded the number of eggs and/or nes- tlings and checked for the presence of brood parasitism by Brown-headed Cowbirds (Molothrus ater). For spe- cies with a sample size larger than 10 per year, we used Mayfield's (1975) method to calculate nest suc- cess based on exposure. We evaluated data from the 1995 field season on grassland bird-habitat relations at both the local, or in- field, and landscape levels using methods similar to those of Wiens and Rotenberry (1981) and Pearson (1993). Between 18 and 25 May 1995, we measured vegetation at 10-m intervals along 50-m transects ex- tending out from each census point in the four cardinal directions (N = 20 samples/point). At each sampling point we passed a 3-mm-diam, 1-m-long rod vertically through the vegetation perpendicular to the ground and counted the number of contacts made by four classes of vegetation (grass, forb, shrub, and dead). These measurements were used to derive 12 in-field vari- ables: (1) mean vegetation height; (2) maximum veg- etation height; (3) coefficient of variation of vegetation height, which is a measure of horizontal heterogeneity; (4) proportion of ground cover; (5) number of vege- tation contacts --< 25 cm; (6) number of vegetation con- tacts > 25 cm; (7) total vegetation contacts; (8) total forb contacts; (9) total grass contacts; (10) total shrub contacts; (11) total dead contacts; and (12) total num- ber of shrub stems intersected by the transects. In 1995 we quantified 10 landscape-level variables using a combination of Geographic Information Sys- tems (GIS) technology and interpretation of U.S. De- partment of Agriculture Agricultural Stabilization and Conservation Service (ASCS) 1:20,000 aerial photo- graphs. For each field we calculated three variables: field area (which was log transformed before use in subsequent analyses), field perimeter, and distance from the center of the field to the nearest field-forest edge. We quantified seven additional landscape-matrix variables in a 500-m radius from the edge of each field. These variables were measured from ASCS aerial pho- tographs with a simple dot grid transparent overlay and were based on the proportion of area occupied by sev- en different habitat types: (1) old field with shrubs, (2) forb-dominated field; (3) cool-season grassland and pasture, (4) wetland, (5) cropland, (6) deciduous forest, and (7) warm-season grassland. Habitat types were de- termined during ground surveys; patches were then classified based on the predominant habitat type (> 50%) in the patch. The vegetation and landscape measurements from 1995 produced sets of 12 and 10 variables, respective- ly. We used Principal Components Analysis (PCA) to simplify the structure in each variable set by reducing the original number of variables to a smaller set of new, uncorretated variables or axes (factors). All veg- etation and landscape variables except proportion of ground cover were used in the PCAs; ground cover was excluded because it showed almost no variation among fields. PCAs were performed on correlation matrices; the initial solution was then rotated to pro- vide a clearer interpretation of the loadings, and those factors with eigenvalues greater than 1.0 were used in subsequent analyses of bird-habitat relationships (Wiens and Rotenberry 1981, Pearson 1993). We then constructed statistical models to describe the variation in bird communities using factor scores and abun- dance/species-richness data for each field; abundance/ species-richness data were based on means for all 1995 censuses in each field. We focused primarily on grass- land birds, which included species in the North Amer- ican grassland avifauna of Mengel 1970 (see also Knopf 1994), with the addition of Bobolink (Doli- chonyx oryzivorous). Response variables included number of grassland species (hereafter referred to as "species") observed in the field during the season; mean number of grassland birds per census per point for each field; and for each common grassland species in the study area (Savannah Sparrow [Passerculus sandwichensis], Bobolink, and Eastern Meadowlark [Sturnella magna]), mean number of individuals per census per point for the field. Stepwise multiple re- gression was then used to select and evaluate the pow- er of specific vegetation and landscape factors in ex- plaining variation among fields in 1995 response var- iables. In addition, correlation coefficients were cal- culated for the relationship between the abundance of individual bird species and scores for each field on the most important factors (Wiens and Rotenberry 1981); for comparative purposes, nongrassland species were included in this analysis. Although we restricted most bird-habitat analyses to the larger 1995 data set, we did test the 1994 data on species richness and abundance for their response to field area using simple linear regression. RESULTS SPECIES RICHNESS AND ABUNDANCE We observed five grassland bird species in the study area in 1994 and 1995: Northern Hamer (Circus cyaneus), Upland Sandpiper, Savannah Sparrow, Eastern Meadowlark, and Bobolink. Only the last three species were observed reg- ularly (> 0.5 individuals/census/point) in at least one field. Savannah Sparrows and Bobolinks were widely distributed throughout the study area, but Eastern Meadowlarks were observed regularly in only 4 of 34 fields censused in 1994 and in 4 of 43 fields censused in 1995. Other grassland species of management concern in the region, including Henslow's Sparrow, Grasshopper Sparrow (Ammodramus savannarum), and Ves- per Sparrow (Pooecetes gramineus), were not observed in the study area, although Henslow's and Grasshopper sparrows have occurred spo- radically at Iroquois NWR in the past (E. Der- leth, pers. comm.). The total number of species observed in a field and the average number of individuals per census per point increased with field area in both 1994 and 1995 (Table 2). We saw few species GRASSLAND BIRDS IN NEW YORK--Norment et al. 115 TABLE 2. LINEAR CORRELATIONS (R 2) BETWEEN LOG OF FIELD AREA AND VARIOUS INDICES OF GRASSLAND BIRD ABUNDANCE IN WESTERN NEW YORK, 1994 AND 1995 1994 1995 (N = 34) P (N = 43) P Species richness a 0.591 < 0.001 0.508 < 0.001 Number of individuals/census/point a Total grassland birds 0.604 < 0.001 0.365 < 0.001 Savannah Sparrow 0.551 < 0.001 0.354 < 0.001 Bobolink 0.395 < 0.001 0.261 < 0.001 Eastern Meadowlark 0.144 0.051 0.077 0.065 a Grassland species only (Northern Hamer, Upland Sandpiper, Savannah Sparrow, Bobolink, and Eastern Meadowlark). or individuals in fields smaller than 5 ha (Fig. 1). The mean number of Savannah Sparrows and Bobolinks per census per point increased with area in 1994 and 1995 (Table 2, Fig. 2), with few individuals occurring in fields smaller than 5 ha. Abundance of these species did not in- crease, however, in larger old-field or warm-sea- son grassland habitats (Fig. 2). The relationship between area and abundance was weak for East- ern Meadowlarks (Table 2), although this result may have been affected by the small number of fields where this species was recorded; it was not observed in fields smaller than 13 ha in ei- ther 1994 or 1995 (Fig. 2).  A ' 9- 8 = 7 4 . 3 0 Area Cool ß season grassland Warm i season grassYand I Fallow field I Old-field ß Pasture ' Forb- I ß dominaicl 100 field B 'O 3 - M 00 ß  BP  2- -e e ß E z 0 i ..... - 0 10 100 Area (ha) FIGURE 1. Mean number of grassland individuals per census per point (A) and number of grassland spe- cies (B) plotted against area (log transformed) in west- ern New York fields, 1995. M = abandoned hayfield at Montezuma NWR; BP = Beattie Point warm-season grassland at Braddock Bay WMA (see "Results"). Grassland bird abundance and species rich- ness were consistently lower in warm-season grasslands, including in the 44-ha field at Brad- dock Bay WMA, than in cool-season grasslands and pastures in the study area (Figs. 1 and 2). Common species in warm-season grasslands in- Cool 1 season grassland Savannah Sparrow I Warm M grassland Fallow field ß * ß Old-field --0 00 ß Forb- 0.5 , Eastern Meadowlark 0.4  0,3 0.2 . o,1 o,o o 10 100 Area (ha) FIGURE 2. Mean number of individuals per census per point plotted against field area (log transformed) for Savannah Sparrows, Bobolinks, and Eastern Mead- owlarks in western New York grasslands, 1995. M = abandoned hayfield at Montezuma NWR; BP = Beat- tie Point warm-season grassland at Braddock Bay WMA (see "Results"). 116 STUDIES IN AVIAN BIOLOGY NO. 19 TABLE 3. LANDSCAPE (L) FACTORS AND FACTOR LOADINGS GENERATED BY PRINCIPAL COMPONENTS ANALYSIS FOR GRASSLAND BIRDS IN WESTERN NEW YORK IN 1995 Landscape factors LI L2 L3 L4 L5 Eigenvalue 2.831 2.676 1.291 1.276 1.221 Proportion of total variance explained 0.283 0.168 0.129 0.128 0.122 Cumulative proportion of variance explained 0.283 0.451 0.580 0.708 0.830 Variable Field area 0.551 -0.085 0.049 -0.180 -0.013 Field perimeter 0.532 -0.071 0.108 -0.279 -0.079 Distance to nearest field/forest edge 0.512 0.054 0.129 0.098 -0.106 Proportion warm-season grassland -0.214 0.047 0.169 0.560 0.196 Proportion deciduous forest 0.192 -0.625 0.024 -0.035 -0.011 Proportion cool-season grassland -0.150 -0.437 0.074 0.076 0.564 Proportion cropland -0.148 0.014 -0.702 -0.189 -0.278 Proportion forb-dominated field -0.114 0.203 -0.267 0.653 -0.098 Proportion old field 0.082 -0.019 0.576 0.052 0.614 Proportion wetland 0.058 0.599 0.199 -0.308 0.402 Note: Only factors with eigenvalues > 1.0 are shown. cluded Swamp Sparrow (Melospiza georgiana), Song Sparrow (M. melodia), and Field Sparrow (Spizella pusilia). Two species of management concern in the Northeast (Schneider and Pence 1992) used switchgrass fields during the study: Northern Harriers nested in switchgrass fields at Tonawanda and Braddock Bay WMAs, and Sedge Wrens (Cistothorus platensis) held terri- tories in switchgrass fields at Iroquois NWR and at Braddock Bay WMA. The one field in the study area with a dense growth of alfalfa (Medi- cago sativa), a 10.1-ha former hayfield at Mon- tezuma NWR, supported a much greater abun- dance of grassland birds than predicted on the basis of area alone (Fig. 1). MULTIVARIATE ANALYSIS OF BIRD-HABITAT RELATIONSHIPS PCA produced five landscape and three veg- etation factors with eigenvalues greater than 1.0; these accounted for 83.0 and 77.6% of the total variation, respectively (Tables 3 and 4). These factors were interpreted by examining loadings on the original variables. Among the landscape factors, L1 clearly represented area, with three variables related to field area (field area, field perimeter, and distance from the center of a field to the nearest field/forest edge) having high pos- itive loadings on the axis (Table 3). Fields with high scores on factor L2 were surrounded by TABLE 4. VEGETATION (V) FACTORS AND FACTOR LOADINGS GENERATED BY PRINCIPAL COMPONENTS ANALYSIS FOR GRASSLAND BIRDS IN WESTERN NEW YORK IN 1995 Vegetation factors VI V2 V3 Eigenvalue Proportion of total variance explained Cumulative proportion of variance explained Variable Total vegetation contacts Vegetation contacts > 25 cm Mean vegetation height Total dead contacts Maximum vegetation height Vegetation contacts --< 25 cm Total shrub contacts Total shrub stems Coefficient of variation of vegetation height Total grass contacts Total forb contacts 4.299 3.136 1.214 0.391 0.285 0.110 0.391 0.676 0.776 -0.430 -0.229 0.027 -0.415 -0.074 0.151 -0.413 0.145 0.129 -0.402 -0.184 0.207 -0.323 0.351 0.180 -0.320 -0.293 -0.084 -0.192 0.373 -0.533 -0.184 0.374 -0.523 -0.126 0.391 0.209 -0.106 -0.365 -0.404 0.080 0.336 0.337 Note: Only factors with eigenvalues > 1.0 are shown. GRASSLAND BIRDS IN NEW YORK Norment et al. 117 TABLE 5. STEPWISE MULTIPLE REGRESSION MODELS OF GRASSLAND BIRD-HABITAT RELATIONSHIPS IN WESTERN NEW YORK Habitat variables entered into Bird variable model a Partial r 2 r 2 Species richness L l 0.43 0.51 V2 0.08 Abundance L1 0.60 0.66 VI 0.06 Savannah Sparrow LI 0.57 0.62 VI 0.05 Bobolink L1 0.43 0.43 Eastern Meadowlark No variables entered into model at P < 0.05 Note: All variables given have P < 0.05; r 2 is the proportion of the total variation in the particular bird variable explained by the model. a L1 - area, V2 - vegetation heterogeneity/shrub density, VI = vege- tation height/density. large amounts of wetland habitat and small amounts of deciduous forest habitat. Factor L3 rep