POPULATION STUDY

records of the Caribbean Fu- BOLTElica caribaea(1974)REPORTED in North THEAmerica, FIRST seven white-shielded around Fort Lauderdale, Florida, from January 28, to May 9, 1974. Several were photo- White-shielded graphed and on April 2 an immature male was collected(National Museum of Natural History 567252). Identifi- in North America: cation was basedupon the broad, high, and bulbous frontal shield that lacked a red, corneous callus; in all other re- a critical evaluation spectsthe birds resembledthe American Coot F. americana. Bolte sketched the shieldsof all sevenbirds. Only two were wholly white. The shieldsof the other five (including the specimen) were marked variouslywith irregular,reddish patches, and three were washed with Don Roberson and various shadesof yellow. These 1974 birds were acceptedas the first records Luis F. Baptista for North America (A.O.U. 1983). Additional white-shielded coots were found in Florida in each succeeding winter, althoughskepticism soon arose. In the winter of 1974-75, Stevenson (1975) reported "a spateof 'Caribbean Coot' records from Florida." This led

Figure I. TypeA, TypicalAmerican Coot. June 1978, Palo Alto, California/D. Roberson.

Volume42, Number5 1241 some of us to suspectthat most, if not records of white-shielded coots and re- In this paper we follow nomenclature all, of thesewere merely aberrant Am. portedunpublished white-shielded birds of the A.O.U. (1983), which considers Coots showingthe variationsof shield in Texas duringthe winter of 1982, with the Caribbean Coot F. caribaea of the color and shapeknown in that species." photographs.Alan J. Ryff (in littO re- Antilles and northwestern Venezuela a Yet Caribbean Coots continued to be porteda white-shieldedcoot with a tiny separate species from the American reported annually in Florida, all ap- red spoton the high point of the shield Coot F. americana, the breedingbirds pearing during winter. In March 1985, at Reifel Island, Vancouver, British Co- of North America, although some au- a white-shieldedbird paired with a typ- lumbia, on October 6, 1984. Still an- thors already considerthem conspecific ical American Coot and remained to other white-shielded was studied (e.g., Roselaar 1980). American Coots nest near Tierra Verde, eventuallyrais- with American Cootsin May-July 1985 occasionallynests as far south as Flor- ing severalbroods (L. Atherton, in litt.). in East Chicago, Indiana, where it ida, and in the Bahamas and Greater Away from Florida, white-shielded eventuallymated with a typical bird. It Antilles. Recently birds showing the coots identified as Caribbean Coots was not thought to have Caribbean or- callus typical of the American Coot were reported from Chattanooga,Ten- igins (Peterjohn 1985). DeSante and have been observednesting with Carib- nessee,from November 1981 to Feb- Pyle (1986) rejectedmore northerlyand bean Coots at St. Maarten, Netherlands ruary 1982(Hall 1982,photo published; westerly records,but accepted records Antilles, and apparent hybrids have acceptedby A.O.U. 1983) and near De- for the Caribbean Coot in Florida. The been produced (Norton 1982). Similar troit, Michigan, in April-May 1982. National Rare Bird Alert (NARBA) re- mixed nesting has been noted in the The latter bird paired with a typical ported "confirmed" reports of Carib- Virgin Islands (R. L. Norton, in litt.). American Coot, nested, and raised at bean Cootsin 1986 from Florida, Texas, Many have suggestedthe two "species" least three young (Payne and Master Illinois, and Arizona (fide J. Mofian). representpolymorphism in the Amer- 1983). Payne and Master suggested,as A review of white-shielded coots in ican Coot (Ripley 1977; Payne and had DeBenedictis (1982) earlier, that North America is therefore warranted. Master 1983; A.O.U. 1983). A similar white-shielded birds in North America In the late 1970s and early 1980s, we situation existsin the , where the simply representedsporadically occur- both independently observed white- two phenotypesare simply confidered ring morphs within northern popula- shielded, yellow-shielded, and inter- morphs of F. americana (Gill 1964) or tions of coots. Payne and Master con- mediate coots in California. We report separate species,F. americana and F. cluded that there was no evidence to here our findingsof variation in win- ardesiaca(Fjeldsa 1983). Likewisethe suggesta Caribbean origin for the tering populationsof coots in coastal Hawaiian population of F. americana Michigan bird. California and comment on their tax- alai has both morphs (Pratt 1978, Clark (1985) summarized published onomic significance. 1987),though in that population,white- shielded morphs are the commoner phenotype. Although our paper may have taxonomic implications, we wish primarily to proposean answerto two questions. Has the Caribbean Coot, whatever its taxonomic status, been documented to occur in North Amer- ica?Is there any evidenceto suggestthat white-shieldedbirds anywherein North America have Caribbean origins? In November 1983--January 1984, we surveyedcoots wintering on sdected small pondsin California. For the most part these were ponds where we had found atypical cootsin previousyears, though half were selected without knowing what was present. We rather arbitrarily divided the shieldcharacters observedinto five categories:

A = Normal American Coots with typical red callus (Figure 1). B = "Intermediates," with red callus but washedwith yellow on shield below callus(Figure 2), somehav- ing a very reduced callus (Fig- ure 3). C = Bulbous shields strongly washed yellow, sometimes with reddish splotches(Figure 4). D = Bulbous white shields, with ob- Figure 2. TypeB, "Intermediate': November1983, San Francisco,California/K. Hansen. vious reddish spots, streaks, or

1242 American Birds, Winter 1988 splotches,but not a callus (Fig- ure 5). E = Bulbous white shields with little or no reddish staining, limited (if present)to extreme upper edgeof shield (Figures 6, 7).

Although there is some subjectivity in assigningbirds to these categories, they provide a convenient means of quantifying the frequencies of the morphsin variouspopulations sampled. Our survey resultsare summarized in Table 1. These surveys indicate a fre- quencyof bulbous-shieldedbirds (types C, D & E) of about 1.4%, and a fre- quencyof atypicalbirds (typesB, C, D & E) of about 3%. Bolte's (1974) seven Caribbean Coots in Florida (out of ap- proximately 800 birds surveyed) fell into categoriesC, D & E, a frequency of 0.8%. Clark (1985) thought the fre- Figure 3. TypeB, "Intermediate".November 1986, Palo Alto, California/J. Morlan. quency of Caribbean-type birds in his surveysof Texas and Florida birds to be about 0.25%, but somewhat greater than 0.5% if "intermediates" (our type B) were included, The Ft. Lauderdale, Florida, Christmas Bird Count has re- portedsix CaribbeanCoots among 6910 coots over the past five years (1981- 1985), a frequencyof only 0.1%. These surveys suggest that Caribbean-type coots occur as frequently in California as in Florida. Apart from our findings, additional white-shielded coots have been reported in California away from the areasof our surveys:McGrath State Beach, Ventura County (completely white-shieldedbird, August 5, 1982; L. Atherton), Oakland (completely white- shielded bird, February 11, 1984; J. Morlan), Fresno (February 1984; K. Hansenflde J. Morlan), Palo Alto (two birds in November 1986; J. Morlan), and Santa Barbara (one white-shielded, one intermediate in November 1986; J. Dunn, R. Abbey,allper& comm.). Given our assumption that all Cali- fornia coastalbirds are American Coots, we questionedwhether any of the Flor- Figure 4. TypeC. December1983, Arcadia, California/D. Roberson. ida birds could claim true Caribbean origin. Florida birds with bulbous Table 1. Frequeucyof occurrenceof coot morphsfrom variousCalifornia localities. shields,like those in California, appear ShieM Type primarily in the wintering population Total from November to May (L. Atherton, Locale N A B C D E in littO and appear to show the same range of variation observedin Califor- Chain-of-Lakes, Golden Gate Park, San Francisco 134 119 9 4 1 1 nia birds. There is no empirical evi- Nearys Lagoon, Santa Cruz 86 85 0 0 0 1 denceto suggesta Caribbeanorigin for Los AngelesArboretum, Arcadia (one pond only) 19 17 1 I 0 0 Crespi Pond, Pacific Grove 121 121 0 0 0 0 any Florida bird, except for Bolte's Lake El Estero, Monterey 216 213 1 0 2 0 (1974) specimen. Therefore we com- Naval Golf Course, Monterey 116 116 0 0 0 0 pared the only North American puta- TOTAL 692 671 11 5 3 2 tive specimen of the Caribbean Coot

Volume42, Number 5 1243 Figure 5. TypeD (center).November 1983, San Francisco,California/K. Hansen.

Table 2. Measurement (in mm) of two taxa of Fulica: with 20 F. americana (ten males, ten femalesfrom California, Colorado, and Taxon Sex N Range )7 (+s.d.) sig.* Indiana) selectedrandomly from the WING collectionat the California Academyof Sciences,San Francisco, and 19 F. car- F. americana M 10 186-212 199.5 (8.96) F 10 185-209 194.4 (9.80) ibaea(ten males,nine femalesfrom the F. caribaea M 11 171-202 188.73 (9.15) P < .01 LesserAntilles and Haiti) from the col- F 9 164-197 176.89 (10.10) P < .005 lection of the National Museum of Natural History. CULMEN Males averagedlarger than females in all measuresin both taxa (Table 2). F. americana M l0 40.5-52.5 45.27 (3.82) F 10 36.6-54.6 43.58 (4.92) Male F. americana had longer wings F. caribaea M 11 46.6-57.8 52.43 (3.39) P < .01 than F. caribaea(199.50 versus188.73; F 9 42.8-51.3 47.24 (2.99) n.s. p < 0.01). Wings of female F. ameri- cana also were longer than wings of fe- BILL male F. caribaea(194.4 versus176.89; F. americana M l0 13.I - 16.8 14.89 (1.06) p < 0.005). The shorter wings of the F 10 13.7-16.5 14.95 (.95) Caribbean Coot may reflect the rela- F. caribaea M 11 13.5-16.3 14.86 (.74) n.s. tively sedentarynature of this taxon as F 9 12.9-15.3 13.91 (.81) n.s. compared to its mainland relatives. Male F. caribaea, however, averaged GONYS larger in culmen length (52.43 versus F. americana M l0 10.0-12.0 10.79 (.79) 42.27; p < 0.01) and tarsallength (62.46 F 10 8.0-11.2 9.93 (.98) versus59.11; p < 0.05) than F. ameri- F. caribaea M 11 9.9-12.3 11.04 (.72) n.s. cana. There is no difference,however, F 9 9.4-11.3 10.39 (.65) n.s. in bill length from nostril between the two taxa (14.89 versus14.86; p > 0.10). TARSUS The longerculmen in F. caribaeais thus F. americana M 10 55.7-63.5 59. l I (2.33) probably a reflection of the difference F 10 54.6-65. l 57.67 (3.95) in shieldheight. We found no other sig- F. caribaea M 11 56.1-65.6 62.46 (4.27) P < .05 nificant morphometric differencesbe- F 9 54.2-65.4 59.41 (3.98) n.s. tween the two taxa (Table 2). * 2-tailedt-test. F. americanacompared with F. caribaea. The wing chord of the Florida bird,

1244 AmericanBirds, Winter 1988 Figure 6. TypeE (center).November 1983, San Francisco,California/K. Hansen.

a male, was 196 mm. The publishedin- formation on wing chords of male American Coots range from 188-- 199.8 (• = 195.9; Godfrey 1966), 182- 199 (RipIcy 1977) and 174-202 (• = 190.3; Blake 1977). The published rangefor male Caribbean Coots is from 173-191 (• = 184.9;RipIcy 1977; Blake 1977). Clearly, the Florida bird's wing chord falls outsidethe publishedrange of F. caribaea, but we found much more variation in extremesof both spe- cies than previouslyreported. It does appear that most American Coot males have a wing chord from 190-199, whereas that of most Caribbean Coot males is between 184-189. On that basis the wing length of the Florida bird is much nearer the mean of the American Coot. We did, however, find three male Figure 7. TypeE. January 1984, Santa Cruz, Califomia/D. Roberson. F. caribaeawith wing chordsas long or longer than the Florida specimen:De- individual of the bulbous-shielded mary. In neither speciesdoes the outer cember5, 1928, PuertoRico (196 mm., morphs present in North American white web extend to the tip. In F. car- USNM 353805); July 26, 1937, Mont- populations. ibaea it is lacking completelyon four serrat(198 min., USNM 353826) and Blake (1977) suggesteda further of the 19 specimens;in the othersit ex- January 14, 1938, Montserrat (202 characterto distinguishthe two taxa is tends to within four to 45 millimeters min., USNM 353821). The tarsallength that F. caribaca has the outer web of of the tip, thus being very variable in of the Florida bird, 60 min., is also the outermost primary more broadly extent. Likewise, in F. americana this closer to the mean of F. americana edged with white and the secondaries character is variable. In our 20 ran- (59.11 mm., Table 2). Although we usuallymore conspicuouslytipped with domly selected specimens, the white cannot say with certainty that the Flor- white. We found neither of these char- outer web terminated from three to 46 ida specimenis not F. caribaea, the ev- acters to be useful. We measured the mm. from the tip. One of 75 specimens idenceat hand stronglysuggests it is an extent of white on the outermost pri- in the California Academy of Sciences

Volume 42, Number 5 1245 lacked white on the outer web (CAS X 17 mm., well within the range of the Hansen and JosephMofian for the use 73952 from Colorado) and five others Caribbean Coot Finally, Gullion re- of their photographs had the white reduced to a hair-thin ported handling four birds with a much white edge.Contrary to Blake, we find reduced callus and one bird entirely LITERATURE CITED the extent of white on outer web to be lacking a callus, having only a red spot AMERICAN ORNITHOLOGISTS' highlyvariable in bothtaxa. The Florida on an otherwisewhite frontal shield (as UNION. 1983. Check-list of North specimen lacks white entirely on its in our type D). American birds, 6th ed. Washington, outer primary, a character consistent The available evidence on white- D.C. with either form. Likewise, an addi- shielded birds in North America sug- BLAKE, E. R. 1977. Manual of neotropical tional character suggestedby Ridgway geststhat most,if not all, are males.The birds, Vol. 1. Univ. of ChicagoPress (1887), the much-wrinkledshield ofF. Florida specimen is a male and the BOLTE, W. J. 1974. CaribbeanCoot, Fuhca caribaea, in Florida. Am. Birds 28:734- caribaea, was found in several of the Michigan breedingbird wasa male (co- 735. California birdsstudied in the field (e.g., pulation observed; Payne and Master CLARK, C. T. 1985. CaribbeanCoot? Btrd- Figure 4). 1983). Many Florida birds have been ing 17:84-88. Clark (1985) suggestedthat the Ca- very aggressive(L. Atherton,in litt.), as DeBENEDICTIS, P. 1982. The changing ribbean Coot occurs in two forms: a have most birds in California (pers. seasons. Am. Birds 36:150. "Lesser Antilles form" with a much obs.). Indeed, Stevenson (1982) won- DeSANTE, D., and P. PYLE. 1986. Distri- butional checklist of North American broader shield and a "Haitian form" deredif Florida birds might not simply birds. Artemisia Press,Lee Vining, Cal- with a reduced shield. He further con- be "a few old males among this popu- ifornia. cluded that North America white- lation [that] have steppedup their out- FJELDSA, J. 1983. Geographicvariation in shielded birds were of the "Haitian put of testosterone." the Andean Coot Fulica ardesiaca. Bull form". We found no evidence to sup- Our experiencewith California birds Brit. Orn. Club 103:18-22. port this hypothesis. Rather, males indicatesthat wintering populationsof GILL, F. B. 1964. The shield color and re- lationships of certain Andean coots tended to have larger shields than fe- American Coot have a small percentage Condor 66:209-211. males, and the two specimenswe had of bulbous,white-shielded birds (Table GODFREY, W. E. 1966. The birds of Can- from Haiti were females. They had 1). Reviewof photographsof birdsfrom ada. National Museums of Canada, Bull small shields(8.3 X 16.0 mm and 8.1 Florida, Tennessee, Michigan, and No. 203. X 12.5 mm), consistentin sizewith two Texas show the same variability, in- GULLION, G. W. 1951. The frontal shield femalestaken in the LesserAntilles (8.5 cludinga preponderanceof yellowcolor of the American Coot. Wilson Bull 63 157-166. X 12.0 mm and 8.8 X 12.5 mm). Fur- and red splotcheswith but a few pure HALL, G. A. 1982. Appalachianregion Am thermore, the Florida male (11.5 X 18.3 white-shielded birds, as is shown by Birds 36:177. mm), saidby Clark to be of the "Haitian coots in California. Our examination of NORTON, R. L. West Indes region. Am form", is in fact quite consistent in the Florida specimensuggests that it is Birds 36:897. shield size with Lesser Antillian males simply a white-shielded morph of F. PAYNE, R. T., and L. L. MASTER. 1983 (which range from 10.9 X 17.3 mm to americana. Lyn Atherton (in litt.) re- Breeding of a mixed pair of white- 14.1 X 22.8 mm). ports that one white-shielded bird at shielded and red-shielded American Cootsin Michigan. WilsonBull. 95 467- Seasonaland agevariation in sizeand Cape Canaveralbore a greenneck collar 469. composition of the shield hasbeen doc- coded "K49"; her correspondence PETERJOHN, B. G. 1985. Middlewestern umented by Gullion (1951), who found unearthed the fact that the collar had Prairie region.Am. Birds 39:307. that the red callus attached to the shield been placed on it during the breeding PRATT, H. D. 1978. Do mainland coots of the American Coot is of a different season in Wisconsin! Karen West (in occur in Hawaii? 'Elepaio 38:73. compositionthan the underlyingshield, litt.,fide L. Atherton) reportedbanding PRATT, H. D. 1987. Occurrence of the which is simply continuous with the 15 coots without a callus out of 1041 North American Coot (Fulica americana americana)in the Hawaiian Islands,with maxilla. He further found that the size coots banded at Horicon National commentson the of the Ha- and shape of the shield was correlated Wildlife Refuge in 1981 and 1982; waiian Coot. 'Elepaio 47:25-28. with breeding activity, being larger and "K49" had been banded at three years RIDGWAY, R. 1887. A manual of North more swollenin both sexesduring nest- of age on July 23, 1982. Therefore, ir- American birds. J. B. Lippincott, Phfia- ing and when birds are pair-bondedand respectiveof the taxonomic statusofF. delphia. territorial, whether nesting locally or caribaea, we conclude that there is no RIPLEY, S. D. 1977. Rails of the world Da- vid Godine Boston. not. Some old coots retained a large evidence to show that coots of Carib- ROSELAAR, C. S. 1980. "American Coot" shield year-round. Microscopic exam- bean origin have occurredanywhere in in Handbook of the birds of Europe, the lnation of testes showed a direct cor- North America. Middle East and North Africa, Vol 2 relation between gonadal activity and We wish to thank Lyn Atherton, Jo- Stanley Cramp and K. E. L. Simmons, shield size. He was able to enlarge the seph Morlan, and J. Van Remsen, Jr., eds. Oxford Univ. Press. shieldexperimentally by administration for their thorough reviewsof an earlier STEVENSON, H. M. 1975. Florida region Am. Birds 29:680. oftestosterone.Implantation resultedin draft of this paper. We thank Laurence STEVENSON, H. M. 1982. Florida region rapid growth of the shieldin both sexes. C. Binford, William Bolte, Charles T. Am. Birds 36:288. Birds with enlargedshields tended to be Clark, Robert L. Norton, H. Douglas more aggressiveand maintain domi- Pratt, and Alan J. Ryff for correspon- --282 GroveAcre Ave., Pacific nance over smaller-shielded birds. The dence and thoughts on this problem. Grove CA 93950 (Roberson), size of the shield in both manipulated Mercedes S. Foster of the Smithsonian CaliforniaAcademy of Sciences, and control birds attained dimensions Institution graciouslyarranged the loan Golden Gate Park, San Francisco, as large as 12 X 15 mm. and even 14 of specimens. We also thank Keith CA 94118 (Baptlsta)

1246 American Birds, Winter 1988