GEOGRAPHIC VARIATION AND SEXUAL DIMORPHISM IN THE TREMBLERS (CINCLOCERTHIA) AND WHITE-BREASTED THRASHER (RAMPHOCINCLUS)

ROBERT W. STORER Museumof Zoologyand Departmentof Biology,University of Michigan,Ann Arbor,Michigan 48109 USA

ABSTRACT.--Icompared interisland variation in color and in length of wing, tarsus,and bill of the LesserAntillean thrashers,Cinclocerthia and Ramphocinclus(Mimidae). Statistical analysesshow that someearly-taken specimens were missexed.I recognizetwo speciesof trembler,Cinclocerthia ruficauda (Brown Trembler) and C. gutturalis(Gray Trembler), and one of Ramphocinclus(White-breasted Thrasher). Brown Tremblersfrom Sabato Monserrat ("C. r. pavida")are not consideredseparable from those of Guadeloupe(C. r. tremula). I summarizeforaging methods and give possibleexplanations for the smaller amount of sexualdimorphism in bill length in Cinclocerthiaon islands where Ramphocinclusoccurs. Received6 June1988, accepted1 December1988.

THE avifauna of the West Indies is rich and land of Antigua (Steadmanet al. 1984, Pregill well-known. Yet it has been underused as a et al. 1988). sourceof information for zoogeographicand The two forms of the White-breasted Thrash- evolutionarystudies. The LesserAntilles, which er have been consideredseparate species (Ridg- extend in an arc south from the Virgin Islands way 1907), but they are currently considered toward Trinidad, are home to several endemic conspecific(Hellmayr 1934, Davis and Miller genera,including the tremblers(Cinclocerthia) 1960). They inhabit the adjacentislands of Mar- and the White-breasted Thrasher (Ramphocin- tinique (Ramphocinclusbrachyurus brachyurus) and clusbrachyurus). St. Lucia(R. b.sanctaeluciae). Thus Ramphocinclus Tremblersare long-billed, reddish brown or is everywhere sympatricwith Cinclocerthiagut- brownish gray thrashersfound from the islands turaliswhereas C. ruficaudaoccurs alone both to of Saba and St. Eustatius south to St. Vincent. the north and to the south of the islands on At present, they are found primarily in wet, which Ramphocinclusand Cinclocerthiaare sym- upland forestsand less frequently in second pattic. growth (Bond 1936, American Ornithologists' Union 1983).In earlierclassifications (e.g. Ridg- way 1907),the grayishforms on Martinique and MATERIALS AND METHODS St. Luciawere both consideredseparate species. I noticed both long- and short-billed birds in most Current arrangements(Hellmayr 1934, Davis of the populationsof tremblers in the collectionsof and Miller 1960) classifyall the tremblersas a the AmericanMuseum of Natural History.This dif- single specieswith six subspecies.From north ference was at least in part correlatedwith sex; the to south,these subspecies are Cinclocerthiaruff- long-billed birds were usuallysexed as females,and caudapavida (Saba, St. Eustatius,St. Kitts, Nevis, the short-billed birds as males. Data from this and and Monserrat), C. r. tremula(Guadeloupe), C. other museumsshowed that there were apparentex- r. ruficauda(Dominica), C. r. gutturalis(Marti- ceptions,especially among the specimensfrom Domi- nique), C. r. macrorhyncha(St. Lucia), and C. r. nica whence seven short-billed birds were sexed as tenebrosa(St. Vincent). For reasons given below, femalesand four long-billed ones as males (nearly the grayishtremblers of Martinique and St. Lu- one-fourthof the early-taken,sexed specimens from that island). Other preliminary data suggestedthat cia are here considereda separatespecies, Cin- sexualdimorphism was reducedon Martinique and clocerthiagutturalis (the Gray Trembler), and the especiallySt. Lucia, where Ramphocinclusoccurs with reddish brown birds of the other islands are Cinclocerthia, so I examined and measured most avail- consideredto be C. ruficauda,(the BrownTrem- able specimensof thesegenera. bler). A fossil trembler is also known from de- Of the 391 specimensexamined, 359 adults(280 of posits between 2,500 and 4,500 ¾.B.P.on the is- Cinclocerthiaand 79 of Ramphocinclus)were usedin the 249 The Auk 106: 249-258. April 1989 250 ROBERTW.STORER [Auk,Vol. 106 analyses.This is not a largesample by modernstan- I believe the assumptionthat birds were missexed dards,but it isso much larger than the seriesavailable is valid.First, there is a decidedgap between the data to earlier reviewers(Ridgway [1907] reported on 61 for malesand femalesin Schwartz'ssample, and sec- specimens,and Hellmayr [1934],on 50) that, for the ond, 10 of the 14 presumablymissexed specimens firsttime, variation within severalof the populations weretaken by two collectors,A. Hyatt Verrill (7) and can be treatedstatistically. SelwynBranch (3). (Two birds taken in 1905,presum- Measurementstaken included wing length (chord), ablyby Verrill,came to the Universityof Michigan tail length, tarsuslength, length of bill from the an- Museumof Zoologyby way of thecollection of Henry terior borderof the nostril,and culmenfrom the junc- K. Coale,who removedthe original labelsand sub- tion of the bill with the skull. Of these,wing length stitutedhis own. It is possiblethat Coalemay have and length of bill from nostrilproved the mostuseful; switchedthe labels,assuming that the longer-billed tail lengthwas so greatly affected by wearthat many bird wasthe male.)On the other hand,Bond's (1952) measurementsof it wereuseless; and bill lengthfrom questioningor discreditingmany of Verrill's bird rec- skull proveddifficult to duplicate.The ratioof wing ordsfrom Dominicasuggests that he wasnot a careful length to length of bill from nostril was calculated worker. for each bird. Thepreliminary results were tested further by ob- BeforeI attemptedan analysis of thevariation among tainingdiscriminant functions for wing length, tarsus the populations,it was necessaryto establishthe ex- length,and bill fromnostril for the Schwartzsample tent of sexualdimorphism in Cinclocerthia.Although andusing these to testthe remaining birds from Dom- Ridgway (1907) reported sexualdimorphism in bill inica.In all cases,the resultsfrom the preliminary length, I found long-billed birds sexedas malesand analysiswere confirmedby discriminant-function short-billedbirds as females. This raisedthe question analysis.A similar analysiswas performedon the of whether these birds were missexed or whether birds from Guadeloupe.The resultsindicated that a individual variation in this genusis unusuallylarge. single female (MCZ 66485) was missexed.However, Missexedbirds are probablymore numerous in col- the bill lengthwas within the rangeof femalesand lectionsthan is generally realized. They can be de- outsidethe rangeof malesfor thatpopulation, where- tectedmost easily in specieswith little or no overlap asthe measurementsfor wing and tarsuswere within in one or more measurements(e.g. grebes;Storer and the rangeof bothsexes. Because dimorphism is great- Getty 1985, Storer 1987). Becausethe males of most est in bill length and becausethere were no other bird specieshave longer bills than the females,col- individualsshowing overlap, this bird was consid- lectorsprobably missexmore individuals of species ered correctlysexed as a female. like the tremblers in which the reverse is true. I alsotested the sexingof tremblersfrom St. Lucia I usedthe largestsample, 99 adult specimensfrom by using discriminant functions. A single bird Dominica,for preliminarytests. Thirty-six birds from (BM[NH] 94.1.2.242,collected in February1893) sexed that island collectedby Albert Schwartzin 1961and asa femalewas classifiedas a male by discriminant- 1962 yielded the following data: bill from nostril-- function analysisand was, therefore, consideredmis- 24 males, 20.1-22.3 mm; 10 females, 23.7-27.5 mm; sexed. The sexes of other birds I assumed to have been wing length/bill from nostril--23 males,4.16-4.86; 9 missexed or that I sexed on the basis of measurements females, 3.25-4.01. were verified by this test. Assuming that these birds were correctly sexed,I A preliminary discriminant-functiontest for the comparedthese data with the measurementsof all the Dominica sampleon birds from St. Vincent showed earlier-takenbirds. The bill lengthsof four olderspec- that two Schwartzbirds from the latterisland might imens (22.4, 22.4, 22.6, 23.3 ram) fell above those for have been missexed. Discriminant functions calcu- males and below those for females in the Schwartz lated from the St. Vincent birds not collectedby series. Of these, all but the third were well within Schwartzwere usedto checkthe Schwartzsample, the range of wing/bill ratios (4.32, 4.35, 4.12, 4.23, and the likelihoodof missexingwas confirmed. The respectively) for males in the Schwartz series. All bill measurementsof these two birds, sexedas males, exceptthe first bird (which was unsexed)were sexed fell within the range of females.These birds were as malesby the collectors.The third bird is nearerto taken in February,when gonadsare small and the Schwartz'smales than to hisfemales in bothbill length chanceof missexingmost likely, so they were con- and wing/bill ratio. Therefore, I consideredall four sideredmissexed. The sexes assigned to birdsunsexed to be males. Four birds sexed as males were within by the collectorsand to birds believed to have been the range of femalesin both charactersand nine "fe- missexedare listed in Appendix 1. males" fell within the range of males.A tenth bird After sexingwas checkedand corrected,I calculat- sexed as a female had a bill for which bill-from-nostril ed means and standard deviations for each sex of the measurement could not be taken but had a total bill populationof trembler on each island. Descriptive length that waswithin the rangeof malesand outside statistics for three combination of islands were also the range of females.Therefore, I consideredall 10 calculated:those traditionally included in the race birds to have been missexed. pavida(Saba through Monserrat), the islandsof St. April1989] ThrasherVariation 251

Eustatius,St. Kitts, and Nevis (which were probably Banks,J. W. Fitzpatrick,T. R. Howell, N. K. Johnson, connected at times of lowered sea levels in the Pleis- B. L. Monroe Jr., H. Ouellet, J. V. Remsen, and the tocene),and the islandsincluded in pavidaplus Gua- author. deloupe.Pairwise comparisons (Shefie tests) were used to determinethe significanceof interislanddiffer- RESULTS ences.Dimorphism indices for eachcharacter of each islandpopulation were calculatedas the difference There are no clear trends in measurements of betweenthe meansfor the sexesdivided by the mean CincIocerthia,from north to south (Table 1), nor for the means of the sexesand multiplied by 100 is there a color cline. Sexualdimorphism ranges ($torer 1966). from ca. 2-5% for wing and tail length, is less Becauseof the large overlap in measurementsbe- tween the sexes in the White-breasted Thrasher, I did (0.4-3.3%)in tarsuslength, and is large (12.8- not attemptto checkfor missexedbirds or to assign 19.6%)in length of bill from nostril.The last is sex to those unsexed by the collectors.Means and greatestin the three northern races,least in standard deviations for the sexed birds in both pop- macrorhyncha,and intermediate in the othertwo. ulations were calculated, as was the mean for the The indexfor wing/bill-from-nostrilratio is also meansof the sexesof both this speciesand the Gray least in macrorhynchabecause of the small di- Trembler on the two islandswhere they are sympat- morphismindex in bill length. tic. The last were usedto calculatea speciesindex in In RamphocincIus,the Martiniquerace (brachy- the sameway asthe dimorphismindex wascalculat- urus) is consistentlysmaller than that on St. ed. Lucia (sanctaeIuciae),and, except in culmen In Cinclocerthia,there was no statisticallysignificant relationshipbetween the distancebetween islands length,considerably less dimorphic (Table 1). and the amount of differencebetween populations Tremblers from Saba,St. Christopher, Nevis, on adjacentislands (comparing mean measurements and Monserrat("pavida") did not differ consis- for winglength, tarsus length, and length of bill from tentlyin color.Birds from Guadeloupe (tremuIa) nostril for each sex independentlywith the interis- reportedlyare darkeroverall and have grayer, land distancesusing a correlationmatrix). The anal- lesstawny, chests than "pavida"(Ridgway 1907). ysiswas simplified because the islandslie alongwhat Although Schwartz'sseries from Guadeloupe is essentiallya single axis. is, on average,slightly darker than his seriesof Similarly,I useda correlationmatrix to testfor a "pavida,"there is greatoverlap in thischaracter; statisticallysignificant relationship between island size and each of the three linear measurements.Again, no and the color of the underparts,including the such relationship was found. chest,is quite variable,with nearly complete A discriminant-function analysis was used to pre- overlapbetween the Guadeloupe birds and those dict the percentageof specimensthat couldbe iden- from the islands to the north. tified on the basisof the lengthsof wing, tarsus,and According to Ridgway (1907), birds from bill from nostril. This was performedfor pairs of is- Guadeloupeare dark like thosefrom St. Vincent landsby sex.Discriminant functions were calculated (tenebrosa)and thus darker than birds from for the three measurementsfor each island pair. In- Dominica(ruficauda). While the birdsfrom Gua- dividualspecimens were scored by multiplyingtheir deloupeaverage darker above than thosefrom measurementsby these functions and summing the Dominica, there is considerable overlap; and products.These scores, based on the linear functions of the three variables, were used to calculatethe pre- there is a nearly completeoverlap in the color dictedrange of scoresfor eachisland. Individual scores of the underparts. werecompared with the predictedrange of scoresfor Birds from St. Vincent are the darkest of the each island, and the number of specimenscorrectly rufouspopulations. They are sufficiently deeper or incorrectlyclassified was expressed as a percentage. rufouson the backand darker grayon the crown I tested Monserrat vs. Guadeloupe,Guadeloupe vs. and nape than the birds from Dominica to war- Dominica,Guadeloupe vs. St. Vincent, Dominicavs. rant subspecificrecognition. Below, the St.Vin- St. Vincent, and Martinique vs. St. Lucia. Becauseof cent birds averagedarker and tend to be more their very differentcoloration, the grayforms on Mar- heavily markedwith broad,diffuse streaks, but tinique and St. Luciawere not testedagainst any of the rufous forms. there is considerableoverlap between the two The seriesof specimensfrom the SchwartzCollec- populationsin thesetwo characters. tion (now at the Museum of Natural History, Loui- The Gray Tremblersfrom Martinique (gut- siana StateUniversity), plus a pair of the Martinique turalis)and St. Lucia (macrorhyncha)differ mark- Gray Trembler from the Field Museum of Natural edly in overallcolor from the BrownTremblers History,were compared for colordifferences by R. C. and differ consistentlyfrom each other in the 252 ROBERTW. STORER [Auk, Vol. 106

TABLE1. Sample sizes, ranges, means, and standard deviations of measurements(mm) and dimorphism indicesof the subspeciesof tremblers and White-breastedThrashers.

Dimor- Males Females phism n Min-max œ SD n Min-max œ SD index • Wing length C. r. pavida 25 95.2-106.8 101.26 2.89 10 95.4-102.8 98.86 2.53 2.40 C. r. tremula 15 96.0-105.1 101.49 2.68 20 92.7-102.9 97.93 2.55 3.57 C. r. ruficauda 60 87.4-101.8 95.83 2.92 28 87.6-97.6 92.64 2.75 3.39 C. g. gutturalis 10 99.2-115.5 107.76 5.59 2 97.4-110.7 104.05 9.40 3.50 C. g. macrorhyncha 30 99.4-112.6 106.20 3.42 17 96.2-109.3 102.63 3.19 3.42 C. r. tenebrosa 23 93.1-101.3 97.19 2.22 18 87.6-97.6 92.68 2.84 4.75 R. b. brachyurus 16 91.3-102.3 96.11 3.77 9 89.4-105.0 95.67 4.33 0.46 R. b. sanctaeluciae 18 98.0-111.7 106.43 4.01 10 99.5-108.5 103.38 2.80 2.91 Tail length C. r. pavida 21 81.2-93.8 87.79 3.57 7 82.0-88.2 85.01 2.49 3.21 C. r. tremula 13 81.8-90.6 87.36 2.67 14 79.5-89.4 85.58 2.66 2.06 C. r. ruficauda 44 71.0-84.1 79.58 2.51 17 70.3-82.6 77.22 3.05 3.01 C. g. gutturalis 8 81.1-95.6 87.79 4.80 1 83.5 C. g. macrorhyncha 20 80.0-94.5 87.71 3.74 10 78.3-92.5 85.18 4.06 3.93 C. r. tenebrosa 15 77.4-86.5 82.15 2.46 12 72.3-83.3 78.13 3.21 5.02 R. b. brachyurus 13 69.2-83.6 76.11 4.86 9 69.0-82.7 75.60 5.05 0.67 R. b. sanctaeluciae 14 77.5-92.5 85.11 3.71 8 76.4-86.9 81.44 3.71 4.41 Tarsus length C. r. pavida 31 29.5-32.9 31.29 0.90 12 29.4-32.9 30.83 1.09 1.48 C. r. tremula 20 30.1-34.9 31.61 1.15 21 29.9-33.3 31.49 0.84 0.38 C. r. ruficauda 62 27.3-31.2 29.25 0.77 27 26.4-30.9 29.06 0.85 0.65 C. g. gutturalis 9 31.5-34.3 32.76 0.94 1 29.3 C. g. macrorhyncha 32 29.3-33.5 31.28 1.05 17 29.9-33.1 31.32 0.86 -0.13 C. r. tenebrosa 23 29.3-32.2 30.84 0.65 17 28.9-31.0 29.84 0.66 3.30 R. b. brachyurus 16 29.1-32.3 30.81 1.01 10 29.6-31.7 30.52 0.67 0.88 R. b. sanctaeluciae 21 32.6-36.0 34.26 0.92 11 31.1-35.4 33.59 1.68 1.97 Culmen length C. r. pavida 18 32.6-38.4 35.74 1.48 7 39.4-42.8 41.23 1.12 -14.27 C. r. tremula 6 33.7-36.6 33.47 1.04 11 38.6-44.6 41.68 1.86 -16.10 C. r. ruficauda 33 30.5-34.7 32.54 1.09 13 35.7-40.6 38.63 1.45 -17.11 C. g. gutturalis 3 34.1-37.5 35.47 1.80 1 39.5 C. g. macrorhyncha 13 38.7-43.7 40.79 1.46 8 40.8-47.6 45.18 2.16 -10.21 C. r. tenebrosa 9 33.6-35.0 34.24 0.42 9 37.6-41.0 38.88 1.09 - 12.69 R. b. brachyurus 9 27.3-30.7 29.22 0.99 5 26.0-28.6 27.58 1.06 5.77 R. b. sanctaeluciae 5 30.0-33.3 31.46 1.52 2 29.5-30.0 29.75 0.35 5.59 Bill from nostril C. r. pavida 32 21.2-25.6 23.35 1.09 11 26.5-30.0 28.18 1.01 - 18.75 C. r. tremula 16 22.0-24.9 23.18 0.82 19 25.8-31.6 28.22 1.71 - 19.61 C. r. ruficauda 60 19.7-23.3 21.28 0.76 28 23.7-27.6 25.89 1.05 - 19.55 C. g. gutturalis 10 21.0-24.2 22.66 0.87 2 26.7-26.8 26.75 0.07 -16.56 C. g. macrorhyncha 29 25.1-30.2 27.62 1.27 16 29.7-33.6 31.38 1.31 - 12.75 C. r. tenebrosa 20 21.2-23.7 22.40 0.62 17 24.8-29.0 26.35 1.16 -16.21 R. b. brachyurus 17 16.0-19.4 17.99 0.87 10 16.2-20.6 17.83 1.22 0.89 R. b. sanctaeluciae 20 18.2-22.0 20.08 1.00 11 17.9-21.2 19.51 1.04 2.80 Wing/bill C. r. pavida 25 3.9-4.9 4.37 0.27 9 3.2-3.7 3.50 0.14 22.11 C. r. tremula 12 4.0-4.7 4.36 0.22 19 3.1-3.9 3.50 0.22 21.88 C. r. ruficauda 58 3.5-5.0 4.50 0.24 27 3.3-4.0 3.59 0.19 22.50 C. g. gutturalis 10 4.2-5.4 4.73 0.31 2 3.6-4.2 3.90 0.36 19.24 C. g. macrorhyncha 27 3.6-4.3 3.85 0.17 15 3.0-3.5 3.28 0.16 15.99 C. r. tenebrosa 20 4.0-4.8 4.34 0.19 17 3.1-3.8 3.51 0.18 21.15

Obtained by dividing the differencebetween the meansfor the sexesby the mean for the meansof the sexesand multiplying by 100. April 1989] ThrasherVariation 253

T^I•I,E2. Percentageseparability of tremblersfrom pairs of islands,based on discriminant-functionanalysis of lengths of wing, tarsus,and bill from nostril.

Percentagecorrectly classified Island pair Males Females Color difference Monserrat vs. Guadeloupe 62.5 a vs. 58.3 100 b vs. 83.3 + Guadeloupe vs. Dominica 100 vs. 98.2 88.9 vs. 100 + Guadeloupe vs. St. Vincent 100 vs. 100 83.3 vs. 94.1 - Dominica vs. St. Vincent 87.7 vs. 95.0 80.0 vs. 64.7 + Martinique vs. St. Lucia 100 vs. 100 100 b vs. 100 + ßThe pairs of figuresare in the samesequence as thoseof the islands. bSample size = 1; all others >- 8.

colorof the underpartsas described by Ridgway and flattened cranium are related to probing (1907). The color differences between the two narrow spaces,and the narrow antorbital re- populations of the White-breasted Thrasher gion and the eyes are "oriented for closebin- (Ramphocinclus)are also consistent and as de- ocularvision." The shortlegs are "probablyan scribedby Ridgway (1907). advantage for perching on vertical surfaces." While the Gray Tremblers differ most mark- His comment that the reduced sternum and edly in color from all the Brown Tremblers, wings are "possiblycorrelated with the reduced they also differ more in color between them- need for extendedflight" may be true, but the selvesthan do any two populationsof the Brown straightedge of the shallow,tapered keel of the Trembler. Discriminant-function analysis of sternum resembles that of woodpeckers and measurements(Table 2) indicatesthat they also woodcreepers(Dendrocolaptidae) and may be differ more from each other than do any pair an adaptationfor bringing the centerof gravity of populationsof the Brown Trembler in these of the bird nearerto the substrateto which they characters. cling. The specificadvantage of this featurecan- notbe determinedwithout a moredetailed study DISCUSSION of the bird's methods of locomotion, because Zusi's (1969) figure of a trembler clinging to a A potential selective advantage to sexual di- tree trunk showsthe bird perched horizontally morphism and an explanation for sexual dif- on the trunk rather than vertically like a wood- ferencesin bill length may be related to for- pecker. aging. In his work on the Brown Trembler on Zusi was unable to determine the sex of the Dominica, Zusi (1969) found that these birds birds that he watchedforaging, so there is no forage in a variety of ways. They take small information on foraging differencesrelated to fruits in trees and bushes, toss leaves to uncover the sexualdimorphism in bill length. Specu- small animals on the ground, probe in rotting lation on this subjectmay provide hypotheses logs, take small prey from crevices on tree to be tested in future fieldwork. trunks, or probe among tangled vines, clumps The dimorphism in bill length is presumably of dead leaves,or epiphytes. related to differencesin its use in probing, a Zusi believed the Brown Trembler's three method of foraging not shared with the trem- most important potential competitors for food bler's potential competitors.The longer bill of on Dominica to be the Scaly-breastedThrasher the femalesincreases the range of depths avail- (Margaropsfuscus), the Pearly-eyedThrasher (M. able to them and thus the potential for an in- fuscatus),and the Forest Thrush (Cichlherminia creasedfeeding rate. This may be important for lherminieri).The bills of these species differ females when extra energy is needed for the markedly from those of the tremblers. The bill development of eggs and during incubation, of the Scaly-breastedThrasher is much shorter, when foraging time may be limited. It might and thoseof the other two (especiallythe Forest also be important if the female has the larger Thrush)considerably heavier. Zusi (1969)found share in feeding the young, but the relative other structuraladaptations related to the trem- roles of the sexesin the care of the young are bler's arboreal foraging methods.The long bill not known. The latter hypothesis does not, 254 ROBERTW. $TORER [Auk,Vol. 106

however,explain why the males'bills areshort- on the windward coast,showing a distinctpref- er. This may be related to a possibledifference erence for a low woodland with thin, crowded in the utilization of foraging places. Perhaps tree trunks, no ground cover and abundant leaf the males are better adapted for working in litter," and adds that it "feeds chiefly on the shallowcrevices or foragingon the groundand ground and mainly by turning over leaf litter," the femalesfor probing deeply in epiphytes.A but that "one bird was seen to pick berriesoff foraging study with color-marked birds on a terminal twig." He continuedthat it "was for- Dominica, where the birds are still numerous, merly much more widespread, at least on St. should answer this question. Lucia,where Semper(1872) found it common." On Martinique and St. Lucia, tremblers are Semper(1872) only saysthat he met them "busi- much grayer than elsewhere,are lesssexually ly searchingamongst the bushesnear the ground dimorphic, and are sympatric with the White- and in low trees." According to his field notes breastedThrasher, which is only known from for Martinique,Fred A. Ober (in Lawrence1878) thesetwo islands.Competitive interactionsbe- collected White-breasted Thrashers at Trois Is- tween the two speciesare possiblebecause the lets and observed one in the Jardin des Plantes bill of the White-breasted Thrasher is more sim- at St. Pierre. He commentedthat it "loves deep ilar in shape to that of the trembler than any woods and the borders of streams." Near the of the other sympatricthrashers or thrushes.At mouth of the Rivi•re Chaloupe, St. Lucia, Zusi present,the numbers of both speciesare much (in litt0 found White-breasted Thrashers both reduced on both islands. In spite of the limited in spindly, short, deciduoustrees 3-6 m tall and amount of overlap today, the earlier literature near or on a canyon floor where the trees were indicatesthat both speciesoccupied a wider 18-21 m tall and greener. range of habitatsbefore the islandswere greatly Fromthe aboveevidence, it appearsthat Gray modified by European settlers,and that they Tremblersonce occurred in the dry scrubforest were probablysympatric over large areas. where White-breastedThrashers occur today On Dominica, where the mongoose (Her- and that the latter once occurredin deeper for- pestes)is absent and where Brown Tremblers est than they do now. Within the forests,the are not uncommontoday, they occupya fairly Gray Tremblersevidently foragedhigher in the wide range of habitats.Zusi (1969) found them trees and the White-breastedThrashers stayed in montane forests,secondary forests, and plan- on or near the ground. tations, as well as rain forests, the most usual The bill of the White-breasted Thrasher is habitat on other islands, but he did not find much shorter and the cranium much higher them in dry scrubwoodland during the leafless than those of the tremblers (Zusi 1969). Pre- season.According to Diamond (1973),tremblers sumablythis limits the ability of thesebirds to evidently occupy "a slightly wider range of probe in epiphytes or in crevices.The keel of habitats on St. Lucia than on Dominica," and the sternum in the White-breasted Thrasher is Danforth (1935) found Gray Tremblers "locally deeperand rounded in outline, unlike the shal- commonchiefly in the humid virgin forest, al- low, straight-edgedkeel of the tremblers(pers. though found to someextent in secondgrowth obs.).Presumably, the White-breastedThrasher and in a low, rather dry, brushytype of forest." is lesswell-adapted for clinging to tree trunks Bond (1928) found Gray Tremblers"a consid- than the trembler. On the other hand, it is rel- erable distance from virgin forest . . . also in atively long-legged (wing/tarsus ratio 3.1 vs. the low forest which peters out into arid scrub 3.4 for the tremblers),indicating that it is a more in northern St. Lucia." Zusi (in litt.) found both terrestrialform, which is corroboratedby Bond's speciesalong the Rivi•re Sourci•re, St. Lucia, statement(1957) that they "are largely terres- where White-breasted Thrashers were in bush- trial." Diamond's(1973) statementthat they fed es or feeding on the ground whereas Gray by turning over leaf litter, as Zusi (1969) re- Tremblers were in taller, leafy trees or quite portedBrown Tremblers did on Dominica,sug- low in slender, scrubby trees. At this site he geststhat the White-breastedThrasher may have only observedtremblers foraging in dried leaves taken over this feeding niche on Martinique caught in a tree fork. and St. Lucia, "forcing" the tremblerson those There is less information on the White- islandsto spendmore time foraging in the trees. breastedThrasher. Diamond (1973)says that "on This is supportedby Zusi'sobservations (in litt0 both islandsit now occursonly on scrubforests of White-breastedThrashers' clearing small areas April 1989] ThrasherVariation 255

of leavesby graspingand tossing,or by broom- It ispossible that, like the GrayTrembler, the sweepingwith vigorousside-to-side sweeps of White-breasted Thrasher arose on the islands the bill, or by tossingto the front. of the limestone arc and were successful in col- The differencein color between the Gray and onizing only Martinique and St. Lucia,possibly Brown tremblers is not easy to explain. Some becausethe low scrub vegetation on these is- birds, like the Fox Sparrow (Passerellailiaca), landsfit the ecologicalconditions of the islands show color differencesthat appear to be asso- on which they arose and to which they were ciated with habitat differences (Swarth 1920). adapted. The gray color of the tremblers on Martinique The original source of the White-breasted and St. Lucia may have resultedfrom a habitat Thrasher is unclear. Although its juveniles are difference, but if this is so it is difficult to ex- dark-breasted like those of tremblers, Zusi (1969) plain on the basisof their presentdistribution, believed it nearest to the Mexican and Central because the tremblers on these islands are now Americangenus Melanotis, whereas Cinclocerthia largely restrictedto wet forests,the habitat in may be distantly related to Margarops(includ- which they aremost numerous on other islands. ing Allenia), the three speciesforming an en- However, it may not always have been thus. demic West Indian group, whoserelationships If dry scrub woodland was originally more with continental mimids remain to be deter- extensiveon Martinique and St. Lucia than on mined. A genetic assayof these genera, their the other islands on which tremblers occur and island populations, and their mainland rela- if tremblerswere numerousin this habitat,they tives should prove most valuable in solving might have been more easily seen by aerial these problems. predatorssuch as barn-owls (Tyto alba) there than in the rain forest. Thus, the advantageof TAXONOMIC CONCLUSIONS a grayish plumage in scrub woodland might have outweighed its possibledisadvantage in The tremblerforms gutturalis (Martinique) and rain forest.The darker,sooty color of the White- macrorhyncha(St. Lucia) differ markedly from breasted Thrashers may be related to their trembler populationson the other islandsand spendingmuch time in shadowson or near the to a lesser extent from each other. Both have ground. been considered separate species (Ridgway White-breasted Thrashers are now birds of 1907). Both are brownish gray rather than red- the scrub woodlands, and sexual dimorphism dish brown in general color, but they differ in the Gray Trembler is reduced on St. Lucia. from each other in details of the color of the These facts suggest the possibility that the underparts.The birds from Martinique average White-breastedThrasher, which hasa similarly slightly larger in wing and tarsuslengths than shapedbut shorterbill, may have been more in those from St. Lucia, but the latter have much competitionwith males than with the longer- longer bills and are lesssexually dimorphic in billed females of the Trembler. Thus sexual di- this character. The differences between these morphismin the latter may have been reduced two populations and the rufescent ones are through a greater increasein the length of the comparableto or greater than those between males' than the females' bills. other pairs of thrasher species(e.g. the Brown An alternate hypothesis is that the Gray Thrasher,Toxostoma rufum, and the Long-billed Trembler may have arisen on one of the low- Thrasher, T. longirostre).In addition, Zusi (in lying limestone islandseast of the volcanicarc litt.) found differences in vocalization and in on which Brown Tremblers now occur and later trembling between the Brown Trembler on may have colonized Martinique and St. Lucia. Dominica and the Gray Trembler on St. Lucia. Once there, it may have replaced an existing On Dominica the tremblers were silent while populationof the Brown Trembler. A possible foraging,whereas on St. Luciathey gave"a loud source of a gray ancestral stock is Barbados, call of repeated notes when flying to another where little original habitat is left and tremblers tree (the 'song'described by Bond[1936] as like are not known to exist.The presenceof a fossil a Carolina Wren's [Thryothorusludovicianus]). trembleron Antigua is evidencethat tremblers Trembling differsin that, on St. Lucia,it is often formerly occurredon low-lying islands of the confined to the tail." Evidence from the tapes limestone arc. The finding of such a fossil on of Roch• (1971)and Hardy et al. (1987)further Barbadoswould provide supportfor this idea. indicates vocal differences between the two 256 ROBERTW. $TORER [Auk,Vol. 106 tremblers. I therefore recommendthat they be ida"), I think it possiblethat this specimenac- separated from the other tremblers as Cinclo- tually camefrom Guadeloupe.In any case,there certhiagutturalis (Gray Trembler) and that the seemsto be insufficientevidence to assignany English name for the remaining populationsbe type locality to "sola." Brown Trembler. Ridgway (1907) considered the two forms of The birdsfrom the otherislands (C. ruficauda) the White-breastedThrasher (Ramphocinclus) to are more similar to each other in color and are be distinct species.They differ considerablyin separatedprimarily on the basis of measure- size and color. The St. Lucia birds (sanctaelu- ments (Table 2). Populationsfrom Sabato Gua- ciae)are the larger in all measurements(Table deloupe do not differ sufficiently in measure- 1), and the sidesand flanks of the Martinique mentsfrom islandto islandto warrant separation birds (brachyurus)are much paler than the up- into subspecies.On the basisof color,birds from perparts, not the same shade as in the St. Lucia Guadeloupe(tremula) were formerly separated birds.More recentauthors (Hellmayr 1934,Da- from those from the islandsto the north ("pav- vis and Miller 1960) have considered the two ida") by their supposedlydarker color and their conspecific. grayer,less tawny or ochraceouschests, but these While it is virtually impossibleto determine charactersdo not hold up in series.In spite of if the specieslevel has been reached between their being separatedgeographically by the two bird populationson adjacentislands, I believe large, grayishforms on Martinique and St. Lu- that the relationshipsamong the forms of these cia, birds of the other two reddish brown races, speciesare best representedby the following ruficauda(on Dominica) and tenebrosa(on St. arrangement: Vincent) are more similar to each other in size Cinclocerthiaruficauda tremula (including pav- than to any other form. They averagesmaller ida) than tremulaincluding "pavida,"but they differ Cinclocerthiaruficauda ruficauda from eachother in color, tenebrosabeing darker in overall color. Cinclocerthiaruficauda tenebrosa Cinclocerthiagutturalis gutturalis Bangs(1929) describeda race of trembler as Cinclocerthiagutturalis macrorhyncha C. r. sola,presumably from a small island off Ramphocinclusbrachyurus brachyurus Guadeloupe.He claimed that it differed from Ramphocinclusbrachyurus sanctaeluciae the Guadeloupeform in being paler like "pav- ida" but had a much longer bill. I have exam- ined the type and only specimenreferred to the ACKNOWLEDGMENTS new subspeciesin the Museum of Comparative For the loan and use of material, I am much en- Zoology.Its coloris similarto that of a specimen debted to the curators of the collections which I was in the same collection from St. Kitts. My mea- permitted to examine.These collections and the num- surementsof the type are: wing 97.4 mm, tail ber of specimensexamined are listed in Appendix 2. 84.4 mm, culmen 41.9 mm, and bill from nostril I thank S. M. Goodmanfor the computer analyses 28.3 mm. The wing/bill-from-nostril ratio is 3.4. and for preparing the tables. R. L. Banks,J. W. Fitz- All are well within the rangeof femalesof pav- patrick, T. R. Howell, N. K. Johnson,B. L. Monroe Jr., H. Ouellet, and J. V. Remsen checked color dif- ida and tremula(cf. Table 1). In his description ferences among the populations and made helpful of "sola," Bangs (1929) discussedthe prove- comments on the manuscript. R. L. Zusi read the nance of the unique type, the only paratype of manuscript and kindly provided unpublished data. C. r. tremulaof Guadeloupe.He found the para- Helpful commentswere alsoprovided by J. C. Barlow type sodifferent from a seriesof specimensfrom and K. C. Parkes. Guadeloupe in coloration and bill length that he named it and said that it "probably [came] LITERATURE CITED from somesmall island near Guadeloupe,pos- AMERICAN ORNITHOLOGISTS' UNION. 1983. Check-list sibly Desirade."Bond (1956), without present- of North American birds, 6th ed. Washington, ing reasons,"suggested" Monserrat as the type D.C, A.O.U. locality.In view of the sexualdimorphism and BANGS,O. 1929. A Trembler new to science. Proc. considerable individual variation in coloration, New England Zool. Club 11: 39-41. which neither Bangsnor Bondrecognized, and BOND, J. 1928. On the birds of Dominica, St. Lucia, the agreementin measurementsof the type of St. Vincent, and Barbados, B. W. I. Proc. Acad. "sola"with femalesof tremula(including "pav- Nat. Sci. Philadelphia 80: 523-545. April 1989] ThrasherVariation 257

1936. Birds of the West Indies. Acad. Nat. RIDeWAY, R. 1907. The birds of North and Middle Sci. Philadelphia. America.U.S. Natl. Mus. Bull. 50, pt. 4. 1952. Secondsupplement to the check-list ROCHg,J.C. 1971. Oiseauxdes Antilles: 1. Lespetits of birds of the West Indies (1950). Acad. Nat. Sci. Antillesßde Grenadea la Guadeloupe.Phonodisc Philadelphia. (33#) ER 706. --. 1956. Check-list of birds of the West Indies. SEMPER,J.E., with notes by P. L. SCLATER.1872. Ob- Acad. Nat. Sci. Philadelphia. servations on the birds of St. Lucia. Proc. ZooL --. 1957. Notes on the White-breasted Thrasher. Soc. London 1872: 647-653. Auk 74: 259-260. STEADMAN,D. W., G. K. ?REGILL,•r S. L. OLSON. 1984. DANFORTH,S.T. 1935. The birds of Saint Lucia. Univ. Fossilvertebrates from Antigua, LesserAntilles: Puerto Rico Monogr., Ser. B, no. 3. evidence for late Holocene human-caused ex- DAWS,J., & A. H. MILLER.1960. Pp. 440-458in Check- tinctions in the West Indies. Proc. Natl. Acad. Sci. list of birds of the worldß vol. 9 (E. Mayr and J. USA 81, pp. 4448-4451. C. Greenway Jr., Eds.). Cambridge, Massachu- STORER,R. W. 1966. Sexualdimorphism and food setts,Mus. Comp. Zool. habits in three North American accipiters.Auk DIAMOND,A.W. 1973. Habitats and feeding stations 83: 423-436. of St. Lucia forest birds. Ibis 115: 313-329. ß 1987. Morphology and relationshipsof the HARDY, J. W., J. C. BARLOW,& B. B. COFFEYJR. 1987. Hoary-headedGrebe and the New Zealand Dab- Voices of all the mockingbirdsßthrashers, and chick. Emu 87: 150-157. their allies,family Mimidae. (Cassette)ARA 12. ß & T. GEttYß 1985. Geographicvariation in HELLMAYR,C. E. 1934. Catalogue of birds of the the LeastGrebe (Tachybaptus dominicus). Ornithol. Americas. Field Mus. Nat. Hist. Publ., Zool. Ser. Monogr. 36: 31-39. 13, pt. 7. SWARTH,H. S. 1920. Revision of the avian genus LAWRENCEßG.N. 1878. Catalogueof the birds col- Passerellaßwith specialreference to the distribu- lectedin Martinique by Mr. Fred A. Ober for the tion and migrationof the racesin California.Univ. Smithsonian Institution. Proc. U.S. Natl. Mus. 1: Calif. Publ. Zool. 21: 75-224. 349-360. ZusI, R. L. 1969. Ecology and adaptationsof the ?RIGILL, G. K., D. W. STEADMAN, S. L. OLSON, •r F. V. Trembleron the Islandof Dominica.Living Bird, GRADY. 1988. Late Holocene vertebrates from 8: 137-164. Burma Quarry, AntiguaßLesser Antilles. Smith- sonJan Contrib. Zool. 463. 258 ROBERTW. STORF• [Auk,Vol. 106

APPENDIX1. List of unsexed specimensand those APPENDIX2. List of specimensexamined and acro- assumed to have been missexed and the sex as- nyms used for collections. signed to each for the purposeof analysisin this paper. Specimens examined Cinclocerthiaruficauda pavida. Saba (14): SW 7, USNM Unsexed birds assumed to be males.--Cinclocerthia 4, ANSP3. St.Eustatius (1): USNM 1. St.Christopher ruficaudatremula: AMNH 504515;FMNH 29046;MCZ (10):SW 1, FMNH 6, USNM 1 (type),MCZ 1, BM(NH) 66489.C. r. ruficauda:MCZ 115101.C. r. tenebrosa:ANSP 1. Nevis (10): SW 8, BM(NH) 2. Monserrat (10): SW 160590.C. gutturalisgutturalis: FMNH 29053,29055. C. 5, AMNH 1, USNM 4. C. r. "sola.""Guadeloupe" 1 g. macrorhyncha:FMNH 29073, 29076. (type):MCZ. C. r. tremula.Guadeloupe (50): SW 13, Unsexed birds assumed to be females.--C. r. tremula: AMNH 7, FMNH 6, USNM 8, MCZ 10 (including FMNH 29044, 29047;ANSP 108764.C. r. ru]•cauda: type),BM(NH) 1, ANSP 1, PA 4. C. r. ruficauda.Domi- BM(NH) 91.7.25.6. C. r. tenebrosa:FMNH 29060. C. nica (99): SW 36, AMNH 12, FMNH 4, USNM 6, MCZ gutturalismacrorhyncha: AMNH 504516;FMNH 29072. 6, BM(NH) 8, ANSP 3, YL 16, UMMZ 3, ROM 4, CM Birds sexed as females, assumed to be males.--C. r. 1. C. r. gutturalis.Martinique (17): AMNH 3, FMNH pavida:BM(NH) 40.5.13.3,uncatalogued; USNM 80954. 5, USNM 1, MCZ 1 (type), BM(NH) 2, ANSP 2, PA 3. C. r. tremula:MCZ 76365; ANSP 628. C. r. ruficauda: C. r. macrorhyncha.St. Lucia (55): SW 16, AMNH 8, AMNH 406603, 504507; BM(NH) 89.6.10.8; FMNH FMNH 7, USNM 5, MCZ 7, BM(NH) 5, ANSP 3, PA 124862; MCZ 113588, 113589; ROM 339188; UMMZ 1 (type), UMMZ 2, CM 1. C. r. tenebrosa.St. Vincent 135042; YL 25559, 25563. C. r. tenebrosa:BM(NH) (41):SW 9, AMNH 5 (includingtype), FMNH 5, USNM 98.2.8.26.C. gutturalisgutturalis: AMNH 174754,748590; 7, MCZ 4, BM(NH) 7, ANSP 4. MCZ 76311; ANSP 9068; PA 1884218. C. g. macror- Ramphocinclusbrachyurus brachyurus. Martinique (39): hyncha:AMNH 174753, 39222; BM(NH) 86.8.2.192, SW 10, AMNH 5, FMNH 5, USNM 3, MCZ 4, BM(NH) 94.1.2.242; CM 111640; MCZ 27381, 27382, 229536; 6, ANSP 4, PA 2. R. b. sanctaeluciae.St. Lucia (44): SW USNM 80902. 3, ANSP 6, FMNH 6 (including type), USNM 5, Birds sexed as males, assumed to be females.--C. r. BM(NH) 5, ANSP 2, YL 6, PA 1, UMMZ 4, CM 1. pavida:AMNH 174756;BM(NH) 91.1.25.2; ANSP 86438. Acronyms used for collections C.r. ruficauda:FMNH 124861;ANSP 81191; ROM 22734; AMNH = American Museum of Natural History, UMMZ 134382. C. r. tenebrosa:BM(NH) 98.2.8.25; SW New York City; BM(NH) = British Museum (Natural 3307,3310. C. gutturalisgutturalis:BM(NH) 86.28.2.190; History),Tring; CM = CarnegieMuseum, Pittsburgh; ANSP 9067. C. g. macrorhyncha:BM(NH) 86.8.2.193; FMNH = Field Museumof NaturalHistory, Chicago; MCZ 29537, 29538. MCZ = Museum of Comparative Zoology, Cam- bridge, Massachusetts;PA = Paris Museum;ANSP = Academyof NaturalSciences of Philadelphia;ROM = Royal Ontario Museum, Toronto; SW = Collection of Albert Schwartz(now at the Museumof Zoology, LouisianaState University, Baton Rouge);UMMZ = University of Michigan Museum of Zoology,Ann Arbor; USNM = National Museum of Natural His- tory, Washington,D.C; YL = Yale PeabodyMuseum, New Haven.