Factors that might limit distribution and ability to thermoregulate were examined in a male Broad-tailed Hummingbird (Selasphorus platycercus) wintering in Arizona, unusual for this species. As in the breeding season, territorial activity and feeding did not begin until incident daylight intensity exceeded 11 lux. The daylength available for attaining positive energy balance was one-third shorter than in the summer. Meal size increased through the day. By evening, cumulative intake of sucrose was sufficient to fuel maintenance of a high, non-torpid, temperature for 14 h of nocturnal fast, even without the small flying insects and nectar from introduced Salvia, taken to supplement the feeder solution. In contrast to the subordinate status of broad-tails wintering in Mexico, this bird excluded larger Anna's and smaller Costa's Hummingbirds from his territory. His January-February molt timing was similar to that of a population wintering in Jalisco, fitting the annual cycle of northern breeders. The Tucson male returned the following autumn, proving his ability for extralimital survival and orientation.

Department of Ecology and Evolutionary Biology University of Arizona Tucson, Arizona 85721 USA UN SELASPHORUS PLATYCERCUS INVERNANDO FUERA DE LIMITES: DESORIENTADO O ANUNCIADOR DE CAMBIO? Sinopsis.--Se examinaron factores que puedan limitar la distribuci(n y la habilidad para termorregularse en un Selasphorus platycercus macho invernando en Arizona, poco usual en esta especie. Como en la &poca de apareamiento, la actividad territorialista y la alimentacibn no empez6 sino basra que la intensidad de luz diurna incidental sobrepasaba las 11 unidades lux. El largo de dia accesible para obtener un balance de energia positivo era un tercio m/rs corto queen el verano. E1 tamafio de la comida aument6 a trav&s del dia. Por la tarde el ingreso acumulado de sucrosa era suficiente para mantenet una temperatufa alta, no en torpor, por 14 horas de ayuno nocturno, inclusive sin los pequefios insectos voladores y el nectar de la Salvia introducida suplementando la soluci6n de comederos. En contraste a la condici6n subordinada de los Selasphorus platycercus invernando en M&xico, este ave excluy6 colibr/es como Calypte anna (mayor) y C. costae (menor) de su territorio. Su tiempo de muda entre enero y febrero fue similar al de una poblaci6n invernando en Jalisco, ajusfftn- dose al ciclo anual de los anidantes del norte. E1 macho de Tucson retorn6 el otofio posterior, demostrando su extralimitadas capacidades de supervivencia y orientaci6n. Individual birds, encountered beyond seasonal or geographical limits of species' distribution ("extra-limitals"), excite field ornithologists. Gen- erally missing from the excitement, however, has been discussion of the relationship of possibly risky extralimital occurrence to the evolution of distribution range. Why are they here? (How) do they survive? A bird might be "extralimital" or "accidental" because it was (1) disoriented, (2) unable to store enough fat for normal completion of migration, (3) displaced from migration route by weather, (4) flight-impaired or (5) an escaped captive. Alternatively, the bird could have been (6) a potential pioneer, however unwittingly. Small size and associated energetic constraints make hummingbirds seem unlikely prospects for survival beyond their normal environments. Nevertheless, extralimital observations of hummingbirds are now fre- quent, perhaps related to availability of artificial feeders and cultivated flowers. Anna's Hummingbird (Calypte anna) was once essentially limited to California (Bent 1940), but appeared in Seattle, Washington in 1964. Three were reported in winter 1970-1971, and 15 on the 2 Jan. 1993 Christmas Census (Mattocks 1993, Zimmerman 1973). If previously con- strained by energy balance problems, distributions may change with an- thropogenic climate change. Hence some extralimitals could function as indicators of climate change. Womack (1993) compiled a preliminary list of "extralimital hummingbirds" reported between 1988 and 1993. As she indicated, some reports may be erroneous or duplicates. Others may have been merely tardy transients that did not stay all winter. Extra-limitalism is common, even with delection of some of the 900 records. Valuable insight may await us in the study of extra-limitals. Standard references describe winter ranges but provide little information on how birds function for three quarters of the year. The Broad-tailed Humming- bird (Selasphorus platycercus) normally winters from northern Mexico highlands, southward into the mountains of central and southern Mexico. It is only rarely seen as far north as Big Bend National Park, Texas (Amer- ican Ornithologists' Union 1983, Johnsgard 1983). Davis and Russell (1990) cited three broad-tail records from southern Arizona in December, but no January records to indicate that these birds survived the winter. By wintering 20 ų south of typical Colorado breeding sites, the broad- tail can not only find flowers in mid-winter, but gain 1.4 h of additional daylight for feeding (Calder and Calder 1992). Survival of such a very small bird north of its normal winter range provided a unique opportu- nity to evaluate distributional constraints imposed by temperature, energy supply, daylength and interspecific biology. I located the winter territory of an adult male Broad-tailed Humming- bird on 24 Nov. 1990, in a small fragment of native desert-scrub vegetation in a suburb on Tucson's north side. Normally, broad-tails are seen in this habitat only during spring migration; in the Santa Catalina Mountains north of Tucson, broad-tails are not observed until late February at the earliest (Calder 1974). I monitored his body mass, behavior and relevant physical variables to analyze his winter situation. METHODS Site.--This male (band T22474, 2 December) had claimed a territory in a residential front yard, at 780 m elevation on the southern bajada of alluvium from the Santa Catalina Mountains. The native Sonoran desert- scrub vegetation (Lower Sonoran [Lowe 1967]), now restricted to about one-fifth of its original area by houses, driveways and landscaping, includ- ed foothill paloverde (Cercidium microphyllum), mesquite (Prosopisjuliflo- ra), white-thorn (Acacia constricta), graythorn (Ziziphus obtusifolia), stag- horn cholla (Opuntia versicolor), prickly pear (Opuntia phaeacantha), sa- guaro ( Carnegia giganteus) and bur-sage (Franseria deltoidea). Introduced as a landscaping plant, Salvia greggii bloomed profusely and attracted the broad-tail, Anna's (resident; breeding in January-February) and Costa's (Calypte costae) Hummingbirds until the flowers were nipped by record cold weather (-4 C) on 23 December. On 1 December, nectar samples from the Salvia had an average optical density equivalent to a 38 per cent (by weight, = 1.3 mol/1) sucrose solution. The broad-tail ob- tained an unmeasured portion of his diet from small flying insects, but his energy balance was heavily subsidized by a feeder (Perky-Pet "Little Beginner" #214) containing solutions varying from 28 to 36% sucrose (0.97 to 1.25 mol/1). This was, without doubt, crucial to his energy bal- ance after the demise of the Salvia. Nectar and feeder solution concen- trations were measured with a pocket refractometer (Atago N2, range 28- 62%, calibrated to a reference saline solution). Observations.--I recorded arrivals ( n = 41 d), daytime activity and feed- ing (n = 8 d, 1 d continuous dawn-dusk), and last activity (n = 25 d) at intervals of 2 to 6 d, 1-20 Dec. 1990 and 11 Jan.-24 Feb. 1991 (3 wk absence to study broad-tails wintering in Jalisco). I timed events with a chronometer-stopwatch set to National Bureau of Standards Radio WWV. The broad-tail's arrival at dawn from the northeast (roost site unknown) was signalled by his inbound wing trill (until molting) and by rapid vocal chipping as he approached and perched in a paloverde or acacia. I mea- sured light intensity with a Gossen "Luna-Pro" light meter with diffusing hemisphere, the logarithmic scale of which was readable to the nearest 0.9 lux in the 11-22 lux ambient range of first activity. I monitored air temperatures to __+0.5 C with a mercury max-rain ther- mometer suspended 1.5 m above ground, approximating his dawn perch- ing height. These local minima correlated well (r = 0.89; P ( 0.001; n -- 38), averaging 0.1 C warmer (range --+4 C), than the National Weather Service low recorded 30 km south and 50 m lower in the Tucson thermal inversion. During my 3-wk absence in Mexico for field work, the mini- mum at the site was -4 C; the official low on 23 December was -6 C (Fig. 1). Weighing.--I determined body mass via a perch affixed to the pan of an electronic balance (Carpenter et al. 1983). The Mettier PE-300-SE re- mote-reading balance was readable to 0.01 g, tared often to correct for zero drift, and within _0.01 g (checked against reference masses at day- break, mid-day, and when temperature or other changes warranted reas- surance). A 5-s averaging function on the balance dampened fluctuations during windy periods. Once the bird was accustomed to my presence near the feeder, I substituted a more convenient battery-powered balance (Ohaus CT200, reliable to 0.01 g and checked as above) for dawn and final weighings. RESULTS AND DISCUSSION Daylength utilization and territorial behavioz.Feeding onset at dawn appeared to be a function of incident light, rather than ambient temper- ature (note greater variability in low temperatures [Fig. 1 ] than for light intensities and arrival times on territory for the first, modest meal [Fig. 2a]). This was similar in pattern to breeding males' first activity during 15 DEC 1990 FEB 1991 I ß ß 10- -5 -lO -4o -20 o 20 40 60 80 1 o0 DAYS FROM WINTER SOLSTICE FIGURE 1. Dawn air temperatures on the winter territory of a male Broad-tailed Humming- bird (triangles) may be compared to the National Weather Service official low temper- atures for Tucson during the winter of 1990-1991. The low on 23 December (2 d after winter solstice) was a new record for that date. the summer in Colorado, where males returned from nocturnal roosts about 18 min before sunrise, at 11 lux average incident light intensity. After the summer solstice, first activity coincided roughly with 11 lux in- tensity as days shortened (Calder 1975). In winter, the Tucson bird re- turned to his territory at a mean light intensity of 18 lux (+7.2 SD), averaging only 1.8 min after 11 lux was recorded (Fig. 2A). Thus his winter activity could have been predicted from ambient light intensity. I expected him to need to exploit the entire winter daylength in Tuc- son, since it was 4 h shorter than the summer activity-daylength at 39ųN, but he abandoned his territory 1 h or more before sunset on several occasions (Fig. 2B). Perhaps he was feeding elsewhere, but the lower ac- tivity level when not breeding and access to unlimited feeding had already allowed adequate refuelling in 10 h to support an 14 h nocturnal fast, permitting early diurnal abandonment of territory (calculations below). Migrant hummingbirds, including the broad-tail, subordinate to resi- dent species (many of which are larger) while wintering in the Volcan de Colima region of south-western Mexico (DesGranges 1979, DesGranges and Grant 1980). The Tucson broad-tail, however, did not defer to Anna's Hummingbirds, whose average body mass was 24% larger (4.44 g). Anna's or Costa's were promptly chased away by the broad-tail if they approached the Salvia or feeder. They were able to feed there only before the broad- tail arrived at first light, during his infrequent mid-day absences, and after he departed for the night. lOO 80 I-- 1/23 60 ,17 40 I- 20 0 -20 -40 140 i i i i i i -20 0 20 40 60 80 100 DAYS FROM SOLSTICE 120 120 100 80 60 40- 2O 0 -40 -0 0 2'0 4'0 6'0 8'0 160 120 DAYS FROM SOLSTICE FIGURE 2A. The time of arrival by a male Broad-tailed Hummingbird on his winter territory is indicated by squares. Each * signifies the time at which light intensity reached 11 lux. 4.4 4.2- 4- 3.8- 3.6- 3.4- 3,2- [] [] ß FED ARR 35.4% SUCROSE ad lib 3 4 1'0 1'2 1'4 1'6 1'8 2'0 22 20 JANUARY 1991 (HOURS) FIGURE 3. The body mass of the wintering male was recorded upon arrival (empty squares, ARR) and at conclusion of each meal (filled squares, FED). The slope of the regression for ARR is assumed to represent gradual fat accumulation from previous meals. By 1700 hours, this is calculated to be sufficient to meet the energy cost of nocturnal homeoth- ermy. Thus hypothermic torpor does not appear to be necessary (see text). Slope dif- ferences between regression lines suggest a tendency for meal sizes to increase progres- sively through the day. Food intake and body mass patterns.--Mass gains were evaluated in the context of (a) published data for the same species breeding in Colorado and wintering in Mexico, and (b) total energy intake compared with dou- bly-labelled water determinations for the Anna's Hummingbird. I assume that gains represented lipid storage, for reasons given in Calder (1994). During the breeding season, territorial male broad-tails gain only 2-3 mg/h, adding less than 2% to body mass all day. This weight-control facilitates rapid acceleration and territorial performances. Not until dusk do they indulge in a burst of intense feeding (Calder et al. 1990). Nesting females gained steadily through the day at a rate of 15 ń 2.3 mg/h (Cal- der 1994, Calder and Calder 1992). In a wintering population on Volcan The upper line shows time of local sunrise. This relationship between light intensity and onset of territorial activity is indistinguishable from summer behavior, except that the daytime activity period is considerably shorter. 2B. The male Broad-tailed Hummingbird usually abandoned his winter territory and departed (triangles) well in advance of local sunset (upper curve). Nevado de Colima, Jalisco, Mexico, capture body masses of both sexes tended to increase progressively on an hourly time scale when conditions for energy balance were good (Calder 1994). The Tucson-wintering ma- le's progressive storage followed a similar pattern, as if anticipating the cost of nocturnal temperature regulation (Fig. 3). Body mass upon arrival at the feeder perch increased linearly through the day (n = 38, r = 0.92, P < 0.001), the regression on hour for 20January being: mass = 3.04 + 0.045 x (hour). This regression did not differ significantly when several days' data from intermittent monitoring were included (Calder 1994; n = 115, r = 0.876, P < 0.001): mass -- 3.12 + 0.038 X (hour). These mass-gain rates of 4.5 and 3.8 mg/h exceeded what was observed in the Jalisco population, not surprising, given the unlimited artificial sucrose supply. This rate was at the high end of the range for maximum lipid deposition by migrant rufous hummingbirds (Selasphorus rufus), of 0.3 to 0.5 g/d (hourly 2.5-4.2 mg/h; Carpenter et al. 1983, Carpenter and Hixon 1988). I determined meal sizes and total daily intake from the bird's body mass changes over an entire day (20 January). At dawn, he perched at the feeder often but briefly. Meal sizes were scant to unmeasurable, either because his digestive system was not ready for a larger intake of cold sucrose solution (dawn air 5.5 C), and/or because a higher priority went to re-establishment of his territorial claim. Meal size increased with time from 5 min after his first feeding, to average 0.15 g + 0.10 SD (n -- 37) for the day, within the range of meal sizes of a territorial male Anna's Hummingbird using a feeder (Carpenter et al. 1991). Total intake for the day, 5.68 g of 35.4% sucrose solution (2.01 g sucrose), contained 33.1 kJ of energy, plus any net gain from feeding on aerial insects, which I ob- served in mid-day, unquantified. This was very close to the 1.98 g of sugar and 31.8 kJ of energy for Anna's Hummingbird in September, calculated from doubly-labelled water turnover (Powers and Nagy 1988). The ther- moregulatory energy cost in their study would have been less in warmer weather (15.5 C minimum), reducing the potential for heat loss to 71% of what the Tucson broadtail faced ([40-15.5 C]/[40-5.5 C] = 0.71). Their 4.48 g Anna's, however, weighed 31% more than this broad-tail, so would have required 21% more energy (metabolic scaling (1.310.72 = 1.21; Calder 1974). Thus size and seasonal differences tend to cancel. On 20 January, the broad-tail gained body mass from 3.40 at daybreak to 3.80 g at dusk, representing 13.7 kJ of fat reserves, the most likely form of gain (see Calder 1994). In two final feedings, he increased in mass from 3.80 to 4.11 g (1705 h), then from 3.97 to 4.28 g (1717 h). As unconverted sucrose and water, 0.48 g of 35.4% sucrose would have added 2.7 kJ for a total of 16.4 kJ, or 1.19 kJ/h for the ensuing 13.8-h fast, compared to a 1.06 kJ/h requirement calculated for his size and a body- air temperature difference of 40-5.5 C (Calder 1984: eqn. 8-15). Life history and annual cyc/e. mMolt timing in the Tucson winterer was typical of a northern breeder wintering in Jalisco. Recaptured for exam- ination on 2 February, his gorget was still dull-bronzy, as is seasonally characteristic of birds wintering in Jalisco (unpubl. data). Timing of flight feather replacement also coincided with that in Jalisco (Calder and Calder 1992). His 7th primaries were half-emerged, and the central rectrices (1, 2, 3) were emerging. By 6 February, his wings no longer whistled, so he had probably dropped the old sound-generating tenth primaries. Limitations.--The range in minimum temperatures in this study was similar to that encountered by Broad-tailed Hummingbirds in montane summers of Colorado (Calder 1973, 1994). Therefore, in terms of tem- perature per se, the broad-tail could survive well to the north of its normal wintering range. Food supply thus seems more likely than temperature to be the factor that normally limits northern distribution of humming- birds in winter. Watered landscaping and hummingbird feeders ease this limitation. Had the shorter days been a severe problem for adequate food intake, this bird would not have vacated his territory when an hour or more of daylight remained for feeding. His high rate of mass increase with artificial subsidy of sucrose solution provided enough energy to ob- viate any need for nocturnal hypothermia. This male was last seen on 10 Apr. 1991 and was assumed to have migrated. He reappeared irregularly (band number confirmed 5 Octo- ber) between 29 September and 15 December, but one or more in a series of seven Rufous Hummingbirds (29 September-3 April) may have pre- empted this winter territory. Interestingly, neither the seven Rufous nor the Calliope Hummingbirds returned for a second winter. Had they been disoriented or blown off course, or did they just shift wintering sites for some sensible reason? Disorientation hardly seems an appropriate explanation for this extra- limital wintering, given his demonstrated ability to return the next season. Furthermore, timing, essential for celestial navigation, was good on sea- sonal (e.g., molting) and daily (e.g., morning arrivals) scales. It would have been quite a coincidence for him to have been blown off-course to the same place in two consecutive years. Success in chasing away intruders suggests that flight and maneuverability were unimpaired. Escape from captivity was improbable. Daily mass gain rates equal to those of migratory refuelling (matching or exceeding other records, from feeders, flowers, and insects on a daily basis, make it seem rather unlikely that normal migration was canceled by inability to store energy. Thus his extralimital winter seems to have been a pioneering event, which though not neces- sarily of thoughtful purpose, may have been what expanded humming- bird distribution from tropical origins to 60ųN and 54ųS (Calder 1976, Greenewalt 1960). Broad-tails breed within 20 min flying time in the Santa Catalina Mountains north of the city, so elevational migration could have replaced the normal latitudinal migration. ACKNOWLEDGMENTS The hospitality of Tony and Pat Philipps to hummingbird and watcher made this study possible. I am grateful for their feeder-filling and tolerance of a sheriffs report of a dawn prowler and the prowler's requests to use their bathroom. This project was supported in part by Biomedical Research Support Grant S07RR07002 from the Division of Research Re- sources, National Institutes of Health and by grant 4283-90 from the National Geographic Society. Perky-Pet feeders were generously provided by Bill Barber. Bands were cut with a slicer designed, built, and generously loaned by Lew Graves. I thank Francisco Ornelas for correcting my sinopsis. Federal banding permit 8081; Arizona permit CLDR0000041. LITERATURE CITED AMERICAN ORNITHOLOGISTS' UNION. 1983. Check-list of North American Birds. 6th ed. Amer- ican Ornithologists' Union. Washington, D.C. 877 pp. BENT, A. C. 1940. Life histories of North American cuckoos, goatsuckers, hummingbirds and their allies. Bulletin 176, Smithsonian Institution United States National Museum, Wash- ington, D.C. CaDER, W. A. 1973. Microhabitat selection during nesting of hummingbirds in the Rocky Mountains. Ecology 54:127-134. 1974. Consequences of body size for avian energetics. Pp. 86-151, in R. A. Paynter, Jr. ed. Avian energetics. Publ. Nuttall Ornithological Club No. 15. Cambridge, Massachu- settsß 1975. Daylength and the hummingbird's use of time. Auk 92:81-97. ß 1984. Size, function, and life history. Harvard Univ. Press, Cambridge, Massachusetts. 1994. When do hummingbirds use torpor? Physiol. Zool. 67:1051-1076. , ND L. L. CaDER. 1992. Broad-tailed Hummingbird. In The birds of North America, No. 16. A. Poole, P. Stettenheim, and E Gill, eds. Philadelphia: The Academy of Natural Sciences; Washington, D.C.: The American Ornithologists' Union. , L. L. CaDE, tND T. D. lhaZE. 1990. The hummingbird's restraint: a natural model for weight control. Experientia 46:999-1002. CARPENTER, F. L., D.C. PATON, AND M. A. HIXON. 1983. Weight gain and adjustment of feeding territory size in migrant humming-birds. Proc. Natl. Acad. Sci. USA 80:7259-7263. , AND M. A HIXON. 1988. A new function for torpor: fat conservation in a wild migrant hummingbird. Condor 90:373-378. , M. A. HIXON, A. HUNT, AND R. W. RUSSELL. 1991. Why hummingbirds have such large crops. Evol. Ecol. 5:405-414. D^ws, W. A., AND S. M. RUSSELL. 1990. Birds in southeastern Arizona. Tucson Audubon So- ciety, Tucson. 154 pp. DESGRANGES J.-L. 1979ß Organization of a tropical nectar feeding bird guild in a variable environment. Living Bird 7:199-236. , AND P. R. GRANT. 1980. Migrant hummingbirds' accommodation into tropical com- munities. Pp. 395-409, in A. Keast and E. S. Morton, eds. Migrant birds in the neotropics: ecology, behavior, distribution, and conservation. Smithsonian Press, Washington, D.C. GREENEWALT, C. H. 1960. Hummingbirds. Doubleday, New York, New York, 162 pp. JOHNSGaD, P.A. 1983. The hummingbirds of North America. Smithsonian Press, Washington, D.C. 303 pp. Lows, C. H. 1967. The vertebrates of Arizona. Univ. Arizona Press, Tucson, Arizona, 270 pp. MATrOCKS, P. 1993ß Seattle, Washington Am. Birds 47:917-918ß POWERS, D. R., AND K. A. NAGY. 1988. Field metabolic rate and food consumption by free- living Anna's Hummingbirds (Calypte anna)ß Physiol. Zool. 61:500-506. WOMC E. 1993. Extralimital hummingbirds. Hummingbird Hotline No. 29 (Nov.). 1022 So. Sycamore Dr., Grove, Oklahoma 74344. ZIMMer, D. A. 1973. Range expansion of Anna's Hummingbird. Am. Birds 27:827-835. Received 28 Nov. 1994; accepted 6 Feb. 1995.