Single-copy DNA-DNA hybridization was used to establish phylogenetic relationships among 13 species of waterfowl (Anatidae) chosen from 10 tribes. On the basis of UPGMA clustering of [delta]T m to distances, we suggest that the tribes Anatini, Aythyini, Tadornini, Mergini, and Cairinini diverged more recently than the Anserini, Stictonettini, Oxyurini, Dendrocygnini, and Anseranatini. The Freckled Duck (Stictonetta naevosa, tribe Stictonettini) is only distantly related to the other Anatidae. Presumably the lineage diverged very early. The sheldgeese (tribe Tadornini) and the true geese (Anserini) are only remotely related. The Oxyurini, considered to be in the subfamily Anatinae, is remotely related to the other Anatidae. The Dendrocygnini form an isolated tribe with no close relationship to the swans and geese (subfamily Anserinae). We found that the screamers (Anhimidae) are distantly related to the Anatidae. A method to estimate missing cells in a data matrix of pairwise distances is presented. Received 1 May 1987, accepted 16 February 1988.

Institute of Molecular Biophysics, The Florida State University, Tallahassee, Florida 32306 USA RECENTLY, nucleic acid sequence analysis has been used to estimate relationships of many or- ganisms. The analysis can include direct se- quencing of individual cloned genes or gene products, restriction enzyme analysis of ho- mologous DNA fragments, and general analysis of unique DNA by single-copy DNA-DNA hy- bridization. In higher eukaryotes, perhaps only 5-10% of the DNA (including gene-coding re- gions and a small fraction of the single-copy DNA) is under selection (Britten 1986). The ma- jor proportion of DNA, unique as well as re- petitive, consists of nucleotide sequences that are never expressed as proteins, have no known function, and are therefore presumably free to evolve (Britten 1986). Although both single-copy and repeated DNA comparisons have been used in studies of phy- logenetic relationships in birds (Shields and Strauss 1975; Eden et al. 1978; Sibley and Ahl- quist 1982, 1983), the total DNA of higher eu- karyotes is composed of many subsets of DNA that show varying degrees of reiteration (Brit- ten and Kohne 1968). These include the ex- tremes of specific sequences, present as only one copy per genome through sequences re- peated millions of times. The repeated families show sequence as well as copy-number evolu- tion. Sequences that occur only a few times in one organism may occur many times even in a closely related species (Zwiebel et al. 1982). De- pending on reassociation conditions, the re- petitive DNA fraction in bird genomes makes up about 10-40% of the total DNA (Eden et al. 1978) but contains only a small percentage of the total sequence diversity. The remaining 60- 90% are unique sequences and represent about 98% of the total sequence complexity (Sibley and Ahlquist 1983). The most convenient procedure for sequence comparison of total DNA of different species involves measurement of the thermal stability of reannealed heteroduplex single-stranded DNA. Because repetitive DNA sequences rean- neal first, such sequences might contribute in a disproportionate way to the thermal stability of resulting duplexes (Sibley and Ahlquist 1983). For this reason the repetitive DNA fraction is usually removed. Nevertheless, repetitive se- quences, because of rapid evolution, can also provide valuable data on the relationship of recently evolved species (Gillespie 1977, Ar- nason and Widegren 1986). Although the fossil record of birds is incom- plete, that of waterfowl appears better than in many other families (Olson and Feduccia 1980). Paleontological investigations show that the major waterfowl radiation probably took place within the last 60 million years (Delacour and Mayr 1945, Olson and Feduccia 1980). These events fall within the sensitivity range of DNA- DNA hybridization proposed by Sibley and Ahlquist (1983). We propose a phylogeny of the Anatidae based on single-copy DNA relationships among samples from 13 of the 148 species of waterfowl belonging to 10 traditionally defined tribes. Single-copy DNA sequence relationships among these waterfowl and three other avian species (Northern Screamer, Chauna chavaria; Spur- winged Plover, Vanellus spinosus; and domestic chicken, Gallus gallus) were investigated. MATERIALS AND METHODS Material came from Mergus Waterfowl Farms (Tal- lahassee, Florida). Species were the White-faced Whistling-Duck (WW) (Dendrocygna viduata), Plumed Whistling-Duck (D. eytoni), Bar-headed Goose (BG) (Anser indicus), Red-breasted Goose (Branta ruficollis), Ruddy Duck (RD) (Oxyura jamaicensis), Orinoco Goose (OG) (Neochen jubata), Hooded Merganser (HM) (Lo- phodytes cucullatus), Wood Duck (WD) (Aix sponsa), White-eyed Pochard (WP) (Aythya nyroca), Mallard (ML) (Anas platyrhynchos), White-cheeked Pintall (WC) (Anas bahamensis), and the nonanatid Spur-winged Plover and domestic chicken. Northern Screamer (SC) material came from Mary Dam (Haines City, Florida), the Magpie Goose (MG) (Anseranas semipalmata) and Trumpeter Swan (TS) (Cygnus buccinator) from Mi- chael Lubbock (Sylva, North Carolina), and the Freck- led Duck (FD) (Stictonetta naevosa) from the Wildfowl Trust (Slimbridge, United Kingdom). Gamma-labeled ATP 32 was obtained from ICN (Irvine, California). Isolation of DNA.--DNA was isolated from blood cells by a procedure modified from Blin and Stafford (1976). Blood was obtained from a wing vein and collected in a heparinized syringe. After one-fold di- lution with 0.15 M NaC1 containing 0.05 M EDTA at pH 7.6 (buffer A), cells were collected by centrifu- gation, resuspended in 25 volumes of buffer A, and lysed by the addition of 0.3% sodium dodecyl sulfate. An equal volume (with respect to the total lysate) of redistilled phenol, equilibrated with 0.1 M Tris-HCL buffer pH 9.0, was added, and the mixture was shaken in a gyrotory waterbath overnight at room tempera- ture (25øC). After the addition of one-half volume of chloroform, the phases were separated by centrifu- gation, the aqueous phase collected, and the crude DNA precipitated by the addition of an equal volume of absolute ethanol. The DNA was washed with 70% ethanol and dissolved in TE (0.01 M Tris-HCL buffer pH 7.6, 0.001 M sodium EDTA) (5 ml/ml original blood). After complete solution was obtained, 0.5 ml of 1.0 M Tris-HCL buffer pH 8.0 was added followed by 100 g of pancreatic ribonuclease. The mixture was incubated for 30 min at 37øC, 5 mg of pronase was added, followed immediately by 0.2 ml of 10% sodium dodecyl sulfate, and the solution was incubated for 2 h at 60øC. The mixture was again deproteinized by shaking overnight with an equal volume of phenol saturated with 0.1 M Tris-HCL buffer pH 9.0, the phases were separated by centrifugation, and the DNA was precipitated with two volumes of cold ethanol after the addition of ammonium acetate to a final concentration of 2.5 M. After briefly washing with 70% ethanol, the DNA was dissolved in TE buffer and the concentration adjusted to 1.0 mg/ml. Isolation of single-copy DNA.--A DNA fraction en- riched for single-copy material was isolated as de- scribed by Sibley and Ahlquist (1983). DNA was sheared by sonication into fragments with an average length of 500 nucleotides. Fragment size was deter- mined by electrophoretic comparison with calibrated markers. The DNA was denatured by immersion in boiling water for 10 min and reannealed at 50øC in 0.48 M sodium phosphate buffer (pFI 7.0) to a Cot of 1,000. A single stranded fraction was isolated by hy- droxyapatite chromatography at 50øC. The fraction was dialyzed against TE buffer, concentrated with iso- butanol (Maniatis et al. 1982), and precipitated with ethanol. After washing with 70% ethanol the DNA was dissolved in TE buffer and the concentration ad- justed to 1.0 mg/ml. Radio-labeling of single-copy DNA.--Single-copy DNA was labeled with 32P at the 5' terminus by incubation with polynucleotide kinase and gamma-labeled ATP 32. A 5' terminus labeling kit was obtained from Bethesda Research Laboratories and used according to the in- structions provided by the manufacturer. The specific activity of the radio-labeled DNA prepared by this procedure averaged 5 x 106 counts/g of DNA. The labeled DNA was freed of unincorporated labeled ATP by the spin-column technique (Maniatis et al. 1982). The maximum amount of unincorporated la- beled ATP contamination was less than 3%. Terminal labeling was substituted for iodination (Sibley and Ahlquist 1983) for safety reasons. Hybridization of single-copy tracer DNA with driver DNA.--Hybrids were formed from a mixture com- posed of 1 part (=200 ng) radio-labeled single-copy DNA and 1,000 parts (=200 g) sheared, whole DNA (Sibley and Ahlquist 1983). Hybrids were denatured by immersion in boiling water for 10 min, precipi- tated with ethanol, and redissolved in 0.5 M sodium phosphate (pH 7.0). Reannealing was carried out at 60øC until a Cot of 8,000 was reached. The formation of double-stranded hybrid DNA was analyzed by thermal elution from hydroxyapatite. Thermal elu- tion columns were submerged within an insulated column box connected to a Haacke temperature-con- trolled (ñ0.1øC) circulating waterbath. After appli- cation of the samples to the column (200 g of DNA/ ml of hydroxyapatite) at 55øC in 0.03 M sodium phos- phate, the column was washed with 40 ml of 0.12 M sodium phosphate (pH 7.0) to remove the unbound tracer. The last fraction of the wash (5 ml) was counted in a liquid-scintillation counter, and radioactivity was determined to be no higher than the normal back- ground, indicating that the unbound tracer was elut- ed from the column. At each of the eight temperatures (60-95øC in 5øC increments) the single-stranded DNA was eluted with 10 ml of 0.12 M sodium phosphate 8C r LU r Z  80 z c 2c IOO  40 z 20 2 4 õ \ 70 80 90 T,C (pH 7.0) and counted in a liquid-scintillation counter. The average number of samples chromatographed si- multaneously was 6, with a maximum number of 10. We used Tm to contrast homoduplex thermal disso- ciation distribution with those of heteroduplexes (Sibley and Ahlquist 1983, Sheldon 1987). Figure 1 gives examples of thermal elution curves from which ATto values were extrapolated. Each 1 ø difference for AT values indicates a 1% divergence in nucleotide sequences (Bonner et al. 1973). All other statistics (percentage hybridization [%H], normalized percent- age hybridization [NPH], and T5oH) were calculated as described by Sheldon (1987). To calibrate AT and ToH values with time, one must assume that DNA evolves at a constant rate (Sib- ley and Ahlquist 1983). In view of recent evidence suggesting a nonconstant rate (Britten 1986, Sheldon 1987), we use ATto values only to estimate the branch- ing order. Construction of the data matrix and phylogenetic tree.-- A 13 x 13 matrix was constructed (Table 1) from the average ATto values for each pairwise comparison. The 23 missing AT values required for constructing a complete matrix were estimated using the following rationale. First, a distance function must satisfy the triangle inequality: d, < d,k + di. Second, some distance functions satisfy a stronger rule called the "ultrametric inequality" (Hartigan 1975): d. < max(d,k, djk). If a distance function satisfies the ultrametric in- equality, then a matrix of such pairwise distances has an exact representation as a tree. Conversely, any tree has an exact representation in terms of a matrix of distances that satisfy the ultra- metric inequality (Hartigan 1975). We estimated miss- ing values from the ultrametric inequality. For ex- ample, if d23 was missing, then d2, < max(d2k, dsk) for i I 60 70 80 90 T, C Fig. l. Top: Thermal dissociation curves in which the tracer DNA of the Hooded Merganser (Lophodytes cucullatus) was hybridized with the driver DNAs of Hooded Merganser (1), Orinoco Goose (Neochen ju- bata; 2), Mallard (Anas platyrhynchos; 3), Wood Duck (Aix sponsa; 4), Ruddy Duck (Oxyura jamaicensis; 5), and White-faced Whistling-Duck (Dendrocygna vidua- ta; 6). Bottom: Thermal dissociation curves in which the tracer DNA of the White-faced Whistling-Duck was hybridized with the driver DNAs of White-faced Whistling-Duck (1), Mallard (2), Hooded Merganser (3), Orinoco Goose (4), White-eyed Pochard (Aythya nyroca; 5), Wood Duck (6), Ruddy Duck (7), and Spur- winged Plover (Vanellus spinosus; 8). July 1988] Waterfowl Phylogeny 455 TABLE 2. Reciprocal ATto values. Mean a n b Range Anas platyrhynchos 0.7 12 1.1 Anas bahamensis 0.1 4 0.3 Aythya nyroca 0.5 7 0.8 Aix sponsa 0.3 9 0.7 Neochen jubata 0.4 8 1.7 Lophodytes cucullatus 0.3 8 0.8 Anser indicus 1.3 7 2.7 Cygnus buccinator 1.0 6 1.6 Stictonetta naevosa 1.0 5 0.6 Oxyura jamaicensis 0.5 9 0.9 Dendrocygna viduata 0.7 11 1.7 Anseranas semipalmata 1.0 6 2.4 Chauna chavaria 1.1 6 2.8  Average distance between average of a labeled Anas platyrhynchos distance from another species and average of the reciprocal test. b Number of species for which reciprocal tests were made with Anas platyrhynchos. Anas platvrhvnch (ML) [-- Arias bahomen$i$ (WC) I Ap? .yoc= (wP) Aix sponsa (WD) -- Neoche_n ]ubato (OG) I Lophodytes cucullatus (HM) Anse.__._[r indlcus (BG) Cygnus buccinotor (T$) $tictonetta hoerosa (FD) Cyura jamolcensis (RD) Dendrocyna viduata (WW) Anseran0$ $emiplmata (MG) Cheuna Choyaria ($C) 12.0 I0.0 8.0 6.0 4.0 2.0 0.0 % DIVERGENCE Fig. 2. Dendrogram generated by UPGMA based on DNA-DNA hybridization (ATto) of 13 species of waterfowl representing 10 tribes in 3 subfamilies. See Table 1 for the distance matrix and abbreviations. all k and hence is bounded above by the minimum of the maxima (Meeter pets. comm.). This estimate, the largest distance consistent with the ultrametric inequality, was used for the missing distances. In the estimate of the ATto for WC and WP (Table 1), 1.7 was the smallest maximum value of all distances between other taxa compared individually with WC and WP. To verify the accuracy of this method, a data matrix was constructed using only species for which com- plete data were available (Anas platyrhynchos, Aythya nyroca, A ix sponsa, Neochen jubata, Lophodytes cucullatus, Oxyura jamaicensis, and Dendrocygna viduata ). Five cells within this 7 x 7 complete matrix were chosen ran- domly, and estimated values were compared with ac- tual values. The mean difference between measured and estimated values was 0.26 (SD = 0.11). The dif- ference between measured and estimated values ranged from 0.1 to 0.4. The distance matrix was clustered by the unweight- ed pair-group method using arithmetic averages (UPGMA; Sneath and Sokal 1973). RESULTS Delta Tm values for a pair of taxa varied from 10% to 30% of the average value (Table 1). In almost all individual experiments, however, the order of the ATm was the same. Discrepancies occurred in some reciprocal values (Table 2). Many of the higher reciprocal ATto values were influenced by the one or two aberrant values relating to a particular species or an abnormally low homoduplex T value (see below and Dis- cussion). The NPH values ranged from 90% for species with a AT of less than 3.0, to 75-90% for those with a AT of less than 8.0 and more than 3.0, to about 70% for the hybrids of single- copy DNA between the Mallard and the Magpie Goose (Anseranas) or the screamer (Chauna). Hy- brids between single-copy DNA of the Mallard and the total DNA of the chicken or the Spur- winged Plover (Vaneflus) had NPH values of 40-50%. For unknown reasons NPH values among closely related species often exhibit a high degree of variance (Sheldon 1987). We found a high degree of NPH variance (10-20%) for both distantly related and closely related species. For this reason we used the T statistic in lieu of ToH. The T statistic does not incor- porate NPH values when calculated; therefore, any variance in NPH will not obscure the branching pattern. Species in the subfamily Anatinae (Mallard [Anatini], White-eyed Pochard [Aythyini], Hooded Merganser [Mergini], Wood Duck [Cairini], and Orinoco Goose [Tadornini]) ap- parently diverged more recently than the other taxa (Fig. 2). These birds exhibited AT values smaller than 3.2. In addition, the White-eyed Pochard was more closely related to the Mallard (ATto = 1.7) than to the Wood Duck, Hooded Merganser, or Orinoco Goose. The latter two species were consistently more closely related (AT = 2.0) to each other than to other species. The White-cheeked Pintail (Anatini) was also more closely related to the Mallard (AT = 1.1) than to the other species tested. The remaining species showed large AT val- ues with respect to each other and to the five Anatinae species. The Magpie Goose (Anser- anatinae) was the most distant species, with an average ATm value of 9.5 with respect to all other waterfowl. With respect to the Mallard, the White-faced Whistling-Duck (Dendrocygnini) showed a very early divergence (ZTm = 7.8), followed by the Ruddy Duck (Oxyurini; ZTm = 7.5), Freckled Duck (Stictonettini; zTm = 6.5), and Bar-headed Goose (Anserini; ZTm = 6.0). Discrepancies among reciprocal values were higher when the number of hybridizations for each comparison was limited. Yet, when the Bar-headed Goose and White-faced Whistling- Duck were compared (n = 6), a large discrep- ancy in reciprocal values remained. When the Bar-headed Goose tracer DNA was hybridized to the White-faced Whistling-Duck driver DNA, the average ZTm was 5.9 (n = 2, SD = 0.14). The reciprocal experiment was repeated twice and had an average ZTm of 7.7 (SD = 0.49). The experiment was repeated with a new tracer DNA for each species. The ZTm values were within 0.3 ø of the previous values. The overall ZTm av- erages for the tracer DNAs of the Bar-headed Goose and White-faced Whistling-Duck were 5.9 (n = 3, SD = 0.10) and 7.7 (n = 3, SD = 0.38). A similar condition existed in the Red-breasted Goose and the Plumed Whistling-Duck. The Trumpeter Swan was more closely related to the Bar-headed Goose and the Red-breasted Goose (ZTm = 4.0) than any of the waterfowl tested (ATmS > 5.8). The AT m value (4.0) indi- cates that this swan and goose lineage diverged relatively long ago. Species from the family Anatidae were com- pared with the screamers (Anhimidae), Spur- winged Plover (Charadriidae), and domestic chicken (Phasianidae) (data not shown). These data (including reciprocal values) indicated that ducks, geese, and Anserarias semipalmata are more closely related to screamers (average ZXTm = 11.3) than to Spur-winged Plovers (ZXTm = 14.4) or chickens (average ZXTm = 14.1). DISCUSSION We found that a single-copy DNA similarity based on relationships among different taxa of waterfowl was in good agreement with results based on other criteria. The tribes in the subfamily Anatinae apparently radiated more recently than the Anserinae, Oxyurini, Den- drocygnini, Stictonettini, and Anseranatinae, which diverged much earlier. Our analysis gen- erally agreed with the schemes advanced by Delacour and Mayr (1945), Johnsgard (1961), Brush (1976), and others (Frith 1955, 1964; Kear 1967; Sibley and Ahlquist 1972; Jacob and Gla- ser 1975; Bottjer 1983; Kessler and Avise 1984; Livezey 1986) who used anatomical, behavioral, and other chemical or immunological data. The branching pattern (Fig. 2) among certain species (e.g. Lophodytes cucullatus, Neochen jubata, and Aix sponsa) may not represent the true phylo- genetic pattern because of the high degree of variance and the closeness of ZTm values. Yet, for the species tested two main periods of ra- diation have occurred. We believe that the Mag- pie Goose (Anseranas) lineage diverged very early (von Boetticher 1940) and the whistling- ducks (Dendrocygnini) and the stifftails (Oxy- urini) diverged somewhat more recently from the main lineage. Specific results with respect to the early divergence of the Ruddy Duck (Oxy- ura) are interesting because in virtually all schemes for the classification of waterfowl this tribe has been included in the subfamily Ana- tinae. Delacour and Mayr (1945) considered the Oxyurini a predabbler; Johnsgard (1978) con- sidered the Oxyurini also to be in the Anatinae and to have diverged after the shelducks. Feath- er-protein data (Brush 1976) and morphological characteristics (Livezey 1986) have supported a possible link between the Oxyurini and the seaducks (Mergini). The large ZTm value be- tween members of these taxa is inconsistent with this interpretation. Feather protein and mor- phological similarities may be the result of con- vergent evolution due to diving in both tribes. The Freckled Duck gave large ZTm values (>6.0) compared with all other waterfowl. This species was thought to be an aberrant member of the tribe Anatini (Delacour and Mayr 1945). Our data support the hypothesis advanced by Frith (1964) and Brush (1976) that Stictonetta is only distantly related to the other Anatidae and that its lineage must have diverged early from the main lineage leading to other waterfowl. We found that different values for the ZTmS between a pair of species were generally stable, although never as stable as reported by other investigators (Sibley and Ahlquist 1983, Shel- don 1987). A number of factors could produce variance in ZTm values. One is the effect of ho- moduplex T variance on calibrations of hetero- duplex ZTm values. The majority of homoduplex Tm values fell close to the ideal 85øC proposed by Sheldon (1987), but a few were 1-2 ø below; none were higher. A correction factor was not employed (as done by Sheldon 1987) because the causes of variance (e.g. tracer DNA impurity or improper labeling) in the homoduplex Tm might equally affect the heteroduplex Tin. De- cisions about which values, if any, should be corrected were difficult because the number of hybridizations was low. The small number of experiments for each comparison and mechan- ical error may contribute to the increased vari- ance. Producing exact elution rates, column temperatures, and fraction volumes is difficult when performing DNA-DNA hybridization ex- periments. This was evident from the observed higher degree of interexperimental vs. intraex- perimental variance. Variance in reciprocal values was usually no greater than the degree of variance exhibited for interexperimental ATto values. The most se- rious discrepancies were between the geese and the whistling-ducks. We have no explanation for this anomalous result. Nonreciprocity in ATto values has been considered by other investi- gators as an indication of "experimental qual- ity" (Sheldon 1987). The consistent difference in reciprocal values (1.8) for these two species, however, indicates that some factor other than experimental error was involved. The rather large variance precludes the resolution of in- equalities of DNA evolution rates. The DNA of several species of waterfowl, in- cluding the Bar-headed Goose, contains many "sequence families" that consist of considerable amounts of repetitive DNA (McHugh, Madsen, and de Kloet in prep.). The reiteration frequen- cies of the sequences in these families can differ considerably even among closely related species. Certain sequences that are barely detectable in one species are found in multiple copies in oth- ers (sometimes accounting for 5-10% of the total DNA). This phenomenon of sequence differ- ences among closely related species also occurs in Drosophila (Zweibel et al. 1982). We have not determined the actual degree of sequence di- vergence among the members of such families of reiterated DNA in any species, but in some cases such sequences may have diverged farther than in others. This would create a disparity in the actual sequence complexity for species' tracer DNA. The conditions (0.5 M sodium phosphate at 60øC) for hybridization in general usage for taxonomic studies are considered to be of low stringency. Especially for heterologous hybrids, reciprocal tracer and driver combinations may give different results. The reiterated sequences in the driver may be partly responsible for the discrepancies. In experiments where single-copy instead of total DNA is used as the driver, the difference between the reciprocal values is re- duced by approximately 50% (Madsen and de Kloet in prep.). Despite the inconsistencies, the Anserini lin- eage apparently diverged relatively early in wa- terfowl evolution, although probably later than the Dendrocygnini. Although these tribes are classified in the same subfamily (Anserinae), they are not as closely related to one another as geese are to swans. We believe that the Orino- co Goose (Tadornini) and the Bar-headed Goose are only remotely related, and that their mor- phological and behavioral similarity is the re- sult of convergent evolution (Delacour and Mayr 1945, Johnsgard 1961). ACKNOWLEDGMENTS We thank Frances C. James for technical and edi- torial assistance and Duane A. Meeter for help with statistical analysis. We are indebted also to Lawrence G. Abele for technical help. Alan H. Brush, Anthony Bledsoe, and Jon Ahlquist provided useful criticisms of the manuscript. We thank Mike Lubbock (Sylva, North Carolina), Mary Dam (Haines City, Florida), and the Wildfowl Trust (Slimbridge, Gloucestershire, U.K.) for cooperation in obtaining blood samples. LITERATURE CITED ARNASON, U., & B. WIDEGREN. 1986. Pinniped phy- logeny enlightened by molecular hybridization using highly reiterated DNA. Mol. Biol. Evol. 3: 356-365. BLIN, N., & D. W. STAFFORD. 1976. A general method for the isolation of high molecular weight DNA from eucaryotes. Nucleic Acid Res. 3: 2303-2308. BONNER, T. I., P. J. BRENNER, B. R. NEUFIELD, & R. J. BRITTEN. 1973. Reduction in the rate of DNA reassociation by sequence divergence. J. Mol. Biol. 81: 123-135. BOTTJER, P.D. 1983. Systematic relationships among the Anatidae: an immunological study, with a history of anatid classification, and a system of classification. Ph.D. dissertation, New Haven, Connecticut, Yale Univ. BRITTEN, R. J. 1986. Rates of DNA sequence evolu- tion differ between taxonomic groups. Science 231: 1393-1398. --, & D. E. KOHNE. 1968. Repeated sequences in DNA. Science 161: 528-540. BRUSH, A.H. 1976. Waterfowl feather proteins: anal- ysis of use in taxonomic studies. J. Zool. London 179: 467-498. DELACOUR, J., & E. MAYR. 1945. The family Anatidae. Wilson Bull. 57: 109-110. EDEN, F. C., J. i . HENDRICK, & S.S. GOTLIEB. 1978. Hornology of single copy and repeated sequences in chicken, duck, Japanese Quail and Ostrich DNA. Biochemistry 17: 5113-5121. FRITH, H.J. 1955. The downy ducklings of the Pink- eared and White-eyed ducks. Emu 55: 310-312. --. 1964. Taxonomic status of Stictonetta naevosa (Gould). Nature 202: 1352-1353. GILLESPIE, D. 1977. Newly evolved repeated DNA sequences in primates. Science 196: 889-891. HARTIGAN, J.A. 1975. Clustering algorithims. New York, John Wiley & Sons. JACOB, J., & A. GLASER. 1975. Chemotaxonomy of Anseriformes. Blochem. Syst. Ecol. 2: 215-220. JOHNSGARD, 1 . A. 1961. The taxonomy of the Anati- dae--a behavioural analysis. Ibis 103a: 71-85. --. 1978. Ducks, geese and swans of the world. Lincoln, Univ. Nebraska Press. KFAR, J. 1967. Notes on the eggs and downy young of Thalassornis leuconotus. Ostrich 38: 227-229. KESSLER, L. G., & J. C. AVISE. 1984. Systematic rela- tionships among waterfowl (Anatidae) inferred from restriction endonuclease analysis of mito- chondrial DNA. Syst. Zool. 33: 370-380. LIVEZEY, B.C. 1986. A phylogenetic analysis of re- cent anseriform genera using morphological characteristics. Auk 103: 737-754. MANIATIS, T., E. F. FRITSCH, & J. SAMBROOK. 1982. Molecular cloning, a laboratory manual. New York, Cold Spring Harbor Lab. Press. OLSON, $. L., & A. FEDUCCIA. 1980. Presbyornis and the origin of the Anseriformes (Aves: Charadrio- morphae). Smithsonian Contrib. Zool. No. 323: 1-24. SHELDON, F.H. 1987. Rates of single-copy DNA evo- lution in herons. Mol. Biol. Evol. 4: 56-69. SHIELDS, G. $., a: N. A. STRAUSS. 1975. DNA-DNA hybridization studies of birds. Evolution 29: 159- 166. $1BLEY, G. C., & J. E. AHLQUIST. 1972. A comparative study of the egg white proteins of nonpasserine birds. Peabody Mus. Nat. Hist. Bull. 39: 1-276. , & -- 1982. The relationships of the Hawaiian honeycreepers (Drepaninini) as indi- cated by DNA-DNA hybridization. Auk 99: 130- 140. , & . 1983. Phylogeny and classification of birds based on data of DNA-DNA hybridiza- tion. Pp. 245-292 in Current ornithology, vol. ! (R. F. Johnston, Ed.). New York, Plenum Press. $NEATH, P. H. A., a: R. R. $OKAL. 1973. Numerical taxonomy. San Francisco, W. H. Freeman. YON BOETTICHER, H. !940. Bemerkungen zur Syste- matik der Anatiden. Verh. Ornithol. Ges. Bayern 22: 160-!65. ZWIEBEL, L. J., V. H. COHN, D. R. WRIGHT, a: G. P. MOORE. 1982. Evolution of single-copy DNA and the ADH gene in seven drosophilids. J. Mol. Evol. 19: 62-71.