Herpetologists' League

The Diets of Hispaniolan Colubrid II. Prey , Prey Size, and Phylogeny Author(s): Robert W. Henderson, Teresa A. Noeske-Hallin, Brian I. Crother, Albert Schwartz Source: Herpetologica, Vol. 44, No. 1 (Mar., 1988), pp. 55-70 Published by: Herpetologists' League Stable URL: http://www.jstor.org/stable/3892198 Accessed: 16/09/2008 16:57

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=herpetologists.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Herpetologists' League is collaborating with JSTOR to digitize, preserve and extend access to Herpetologica.

http://www.jstor.org March 1988] HERPETOLOGICA 55

STICKEL, L. F., W. H. STICKEL, AND F. C. SCHMID. WRIGHT, A. H., AND A. A. WRIGHT. 1957. Hand- 1980. Ecology of a Maryland population of black book of Snakes of the United States and Canada. rat snakes (Elapheo. obsoleta). Am. Midl. Nat. 103: Cornell University Press, Ithaca, New York. 1-14. ZAPPALORTI, R. T., AND J. BURGER. 1986. The im- USDA. 1980. A Soil Survey of Ocean County, New portance of man-made habitats to pine snakes. En- Jersey. U.S. Department of Agriculture, Somerville, viron. Cons. 12:358-361. New Jersey. Accepted: 13 April 1987 WEATHERHEAD, P. J., AND M. B. CHARLAND. 1985. Habitat selection in an Ontario population of the Associate Editor: Raymond Semlitsch , Elaphe obsoleta. J. Herpetol. 19:12-19.

Herpetologica, 44(1), 1988, 55-70 ? 1988 by The Herpetologists'League, Inc.

THE DIETS OF HISPANIOLAN COLUBRID SNAKES II. PREY SPECIES, PREY SIZE, AND PHYLOGENY

ROBERT W. HENDERSON', TERESA A. NOESKE-HALLIN', BRIAN I. CROTHER2, AND ALBERT SCHWARTZ3 'Section of VertebrateZoology, Milwaukee Public Museum, Milwaukee, WI 53233, USA 2Departmentof Biology, University of Miami, P.O. Box 249118, Coral Gables, FL 33124, USA 3BiologyDepartment, Miami-Dade Community College-North Campus, Miami, FL 33167, USA

ABSTRACT: Eight species of xenodontine colubrid snakes (n = 1874 specimens) from Hispaniola were examined for prey remains and yielded 557 prey items of which 63.1% were lizards of the iguanid Anolis. With the exception of Darlingtonia haetiana, an Eleutherodactylus frog specialist, all of the species in our sample preyed heavily upon anoles. In general, Hispaniolan colubrids were opportunistic predators, but the widespread exploitation of a single prey genus (Anolis) may be unique to the West Indies. Frequently exploited prey species were geographically widespread and generally found at high densities (Osteopilus dominicensis, Anolis coelestinus, A. cybotes, Ameiva chrysolaema). Active foraging snakes (Alsophis, Antillophis) were more eury- phagic, while sit-and-wait strategists (Hypsirhynchus, oxyrhynchus) were trophically specialized. All of the species in our sample tended to eat relatively small prey items, even though larger individuals of a given prey species were available. The historical (phylogenetic) component of trophic ecology of these snakes is discussed. Key words: Anolis; Eleutherodactylus; Phylogeny; Hispaniola; Snake diets

DESPITE its richness and geographical addition, anoles are uniformly exploited as uniqueness, the Antillean snake fauna has, food by colubrid snakeson all majorislands until recently, been ignored ecologically. and island groups in the Antilles (50-60% Investigation of the diets of West Indian of all prey items; Henderson and Crother, boid, tropidophiid, colubrid, and viperid 1988). Hispaniola has the richest snake snakes has revealed a spectacular (al- fauna and one of the largest anole faunas though not altogether surprising) trophic (second only to Cuba) of any West Indian relationship between the snakes and the island, and it has been singled out for de- iguanid lizard genus Anolis. Nearly 57% tailed studies of its boid (Henderson et al., of all prey items (n = 707) and 75.8% of 1987b) and colubrid (e.g., Henderson, all lizards recovered from the digestive 1984a; Henderson et al., 1987a,c) snakes. tracts of West Indian colubrid snakes were Besides a rich anole fauna, Hispaniola har- anoles (Henderson and Crother, 1988). In bors the richest frog and lizard faunas in 56 HERPETOLOGICA [Vol. 44, No. I

the West Indies: i.e., a vertebrate fauna items for col- C)c very rich in potential prey 0 ubrid snakes. 0 + Hen- 0~~~~~~~~~~~~0 In this paper, a continuation of derson (1984a), we describe the prey exploited by the xenodontine col- "0 0 -~~~~> species C) o i~~~~~~~~~t ) c ubrid snakes of Hispaniola, the West In- dian island with the richest colubrid fauna. Specifically, we address not only questions CD~~~~~~C "0 CL. ~- originally posed by Henderson (1984a) but CO ~ "C~ C)"~ 0 -.~~~~~~~0.~~~~~0c also ask: (1) What prey species make the greatest contributions (frequency and vol- colubrids? C) 0 ~~~ -C C)CO ume) to the diets of Hispaniolan (2) What characteristics(if any) are shared * 0~~~~W0 by frequently eaten prey species? (3) Do Hispaniolan colubrids, despite having po- tential access to rich frog and lizard (other than Anolis) prey faunas, still predomi- 4- * "~~~~~~~~~~~~~~~ C)~~~~~~- nantly exploit anoles as their primary food source? (4) Is there an historical (phylo- C) ~~~~~~~~~~~~~~~f genetic) component to trophic ecology of Co ~ ~ ~ L) ~~~C) ~6i(= C these snakes?

0 MATERIALS AND METHODS '44 +I" ~~~~~~~~ A sample of 1874 colubrid snakes (rep- C). :) C resenting six genera and eight species) was examined for prey items. All were from -~~ J~~ 0 t~~o AC 00 Hispaniola except for Alsophis cantheri- a0C N gerus, which is native to Cuba and the "0 ~~~~~~005u7Z +I -, Cayman Islands. Hispaniola harbors two CO bJD c C)~v O+ C"~ endemic species of Alsophis, but because of their rarity in museum collections, we chose the closely related A. cantherigerus 0 C's0 1 to represent the genus. It is common in C4 1 0- collections and is intermediate in -Z CZ museum E~~~~~0., size between the two HispaniolanAlsophis (A. anomalus and A. melanichnus). Ta- C) CO C) 0 >O~ ~ C) 0 bles 1 and 2 provide brief summaries of OCO co 0 the natural history of Hispaniolan colu- brids and A. cantherigerus. of dissection, determination 0 0 ' Techniques of prey volumes, and recording morpho- C) Q logical measurements were explained in "0 " detail in Henderson (1984a). Many prey C) >>CZ 0 ( items were not identifiable to species, and we generally did not determine volumes for such items. There is probablysome bias in our prey identifications.Some common, frequently exploited species (e.g., Anolis cybotes) were easily identified, even from largely digested remains. Other less com- mon and infrequently consumed species -t

TABLE 2.-Comparison of different aspects of the biology of three species of Uromacer. All are arboreal and diurnal.

Variable U. catesbyi U. frenatus U. oxyrhynchus Source Head and body blunt snout; heavy body; wide attenuated snout; slender attenuated snout; slender Henderson et al., 1981 morphology head body; narrow head body; narrow head Henderson et al., 1987c Skull morphology least specialized highly specialized highly specialized Maglio, 1970 Internal topogra- anterior organ position intermediate organ position, posterior organ position Martinez et al., 1985 phy (= primitive) but closer to oxyrhynchus Foraging mode active and sit-and-wait sit-and-wait sit-and-wait Henderson 1982 Henderson et al., 1987c Binocular field of presumably the narrowest of presumably wide very wide Walls, 1942 | vision genus Striking distance strikes from closer distance; strikes from greatest distance intermediate between catesbyi Ulrich and Ford, 1985 0 and accuracy misses more frequently; and rarely misses and frenatus 0 chases prey C Diet tree frogs; trunk and ground terrestrial Ameiva; grass, primarily trunk and ground Henderson et al., 1987c Anolis trunk, and ground Anolis Anolis, some grass Anolis Mean prey size 4.0 ? 1.0 2.8 ? 0.3 2.2 ? 0.3 Henderson et al., 1987c (cm3) Movement ecology moves long distances on probably intermediate be- rarely travels on ground; uses Henderson et al., 1981, 1982 ground; uses heavy tween catesbyi and oxyrhyn- slender branches branches chus Distribution islandwide "south" island "north" island with invasion Horn, 1979; Schwartz, 1980 of "south"

C,' -. 58 HERPETOLOGICA [Vol. 44, No. 1 were difficult to identify from body frag- History (USNM). The snakes were col- ments. lected at many localities throughout His- Index of relative importance (IRI) was paniola over a span of about 80 yr. Some used to evaluate the relative importance comparisons were made with diets of His- of food items in snake diets (Pinkas et al., paniolan boids (Epicrates) of which 214 1971). IRI was calculated by summing the were examined (Henderson et al., 1987b). numerical and volumetric percentage val- Species of snakes were categorized by ues of a food item (= prey species) and foraging mode on the basis of quantitative multiplying by percentage of occurrence (Henderson et al., 1982) and qualitative (N + V)F = IRI observations in the field by the authors (Henderson and Schwartz) and by col- where N = numerical percentage, V = leagues. We have designated each snake volumetric percentage, F = frequency of species as either a sit-and-wait (ambush) occurrence percentage, and IRI = index or active forager. We are aware that there of relative importance. exists a continuum of modes between these Prey species diversity (H') for each snake two extremes, but our data preclude mak- species was calculated with the Shannon- ing fine distinctions. Weiner information theoretic measure: Abbreviations used for morphometric variables are: SVL = snout-vent length; s HL = head length; MBC = mid-body cir- H' = - pilogep, cumference; SAW = snout anterior width; i=l SBW snout base width; SL = snout length; HW = head width. where S is the number of prey categories Sample sizes for each species are (num- (= species) and pi is the proportion of the ber of snake specimens examined/number total number of prey consisting in the ith of prey items): Alsophis cantherigerus category. (169/40); Antillophis parvifrons (649/ Data were analyzed using one-way anal- 199); Darlingtonia haetiana (131/46); ysis of variance (ANOVA), Spearmanrank Hypsirhynchus ferox (201/49); Ialtris correlation(one-tailed), regression, and chi- dorsalis (37/9); (258/ square. Differences among group means 62); U. frenatus (257/108); U. oxyrhyn- were determined with Student-Newman- chus (172/44). Because of the small sample Keuls test (SNK). Principal component size for I. dorsalis, we excluded it from analysis (PCA) with promax rotation ex- some of our analyses. amined the relationshipamong snake mor- Detailed analyses of diets for individual phologicalvariables; estimated factorscores species or genera have appeared elsewhere of prey genera were calculated along each (Henderson, 1984b; Henderson and principal component axis. Data for prey Schwartz, 1986; Henderson et al., 1987a, volume were logarithmically transformed c). Morphometric data appear in Hender- prior to analysis to normalize their distri- son (1984a:Table 1). bution. All analyses were performed using the Statistical Analysis System package RESULTS (SAS, 1985), with a = 0.05. Specimens from the following collec- Prey Species tions were examined for prey remains: Examination of preserved snakes yield- American Museum of Natural History ed 557 prey items: 17.7% frogs, 79.6% liz- (AMNH), Albert Schwartz Field Series ards, 0.4% snakes, 0.2% birds, and 2.1% (ASFS), Florida State Museum at the Uni- mammals. Most prey items were identified versity of Florida (UF-FSM), Museum of to genus, but only 326 were identified to Comparative Zoology at Harvard Univer- species; we were able to calculate volume sity (MCZ), Milwaukee Public Museum for 260 of those identified to species. (MPM), the Richard Thomas (RT) collec- Twenty-three (44.2%) of the minimally tion, and the National Museum of Natural 52 prey species in our sample are repre- March 1988] HERPETOLOGICA 59

TABLE 3.-Some ecological characteristics of prey species frequently eaten by Hispaniolan colubrid snakes. Means are followed by ? 1 SE, n is in parentheses.

x size (vol.) Species consumed Distribution Adaptive zone When active Eleutherodactylus 0.9 ? 0.0 nearly islandwide ground-dweller to low nocturnal abbotti (7) vegetation Osteopilus 4.4 ? 1.5 islandwide primarily arboreal, occa- nocturnal dominicensis (27) sionally on ground Anolis coelestinus 3.1 ? 0.4 widespread on "south" is- scansorial; leaf surfaces, diurnal (22) land occasionally on ground Anolis cybotes 3.6 ? 0.3 islandwide scansorial; ground, tree diurnal (56) trunks Anolis distichus 1.3 ? 0.1 islandwide scansorial; tree trunks diurnal (20) Anolis olssoni 1.0 ? 0.1 widespread on "north" is- scansorial; grass, low diurnal (24) land; enters "south" is- shrubs land Anolis 0.9 ? 0.1 islandwide scansorial; grass, low diurnal semilineatus (9) shrubs Leiocephalus 7.0 ? 1.6 widespread on southwest ground-dweller diurnal melanochlorus (6) peninsula Leiocephalus 5.2 ? 1.2 widespread on "north" is- ground-dweller diurnal personatus (6) land Ameiva 8.3 ? 1.6 islandwide at low eleva- ground-dweller diurnal chrysolaema (31) tions sented by a single record, and 38 species A. olssoni) did not contribute proportion- (73.1%) are represented by fewer than five ately in volume. Other species contributed records. Because we are here concerned a disproportionate volume compared to primarily with trends and patterns in prey their frequency of exploitation (e.g., exploitation and consequences of foraging Ameiva chrysolaema and Mus musculus). strategies (of predators and prey), our Anolis cybotes was consumed frequently analyses will concentrate on those prey and contributed a high proportionof prey species that were most frequently eaten. volume. Table 3 summarizes some ecological Principal component analysis of snake characteristics of 10 prey species that were morphometric data revealed two axes ex- frequently exploited. Most have wide dis- plaining 95% of the variation in the data. tributions on Hispaniola, and except for Measurementsof snake body width (MBC, the frogs Eleutherodactylus abbotti and HW, SAW, SBW) had high loadings on Osteopilus dominicensis all of the species the first axis, and measures of elongation are diurnal. The Anolis lizards are all scan- (SVL, SL, HL) had high loadings on the sorial, but some are more arboreal than second axis (Table 4). The analysis further others; A. cybotes is the single most ubiq- revealed that the prey genera Eleuthero- uitous species in terms of geographical dis- dactylus and Sphaerodactylus (i.e., small tribution and in vertical distribution with- prey species) are taken by those snakesthat in a given habitat. In terms of mean size are the shortestand most slender (e.g., An- (volume), the diurnal arboreal prey species tillophis, Darlingtonia) (Fig. 2). The most tend to be smaller than the diurnal ter- frequently consumed prey genera fall into restrial prey species (P < 0.0001, F = 73.1, the center of Fig. 2, implying that they ANOVA). are most often eaten by those species that Figure 1 compares the frequency with are not at morphometric extremes. The which 10 species occur in our overall sam- largest prey items (rodents, birds) were ple compared to the volume of food con- taken by those species that were longest tributed by each. Some species that were and had the widest heads and bodies (e.g., frequently eaten (e.g., Anolis distichus and Alsophis, Ialtris). 60 HERPETOLOGICA [Vol. 44, No. 1

30 D Percent Frequency 25[ DlPercent Volume

20

215

10A

10 5 ...... < 27 00 08

Eleuth. Osteopilus Anolis Anolis Anolis Anolis Anolis Leioceph. Ameiva Mus abbotti domin. coelest. cybotes distichus olssoni semi/in. spp. chrysol. musculus

Fic. 1.-Percent (by frequency and volume) contribution of 10 species of prey commonly eaten by Hispaniolancolubrid snakes. Abbreviations are: Eleuth. = Eleutherodactylus;domin. = dominicensis;coelest. coelestinus; semlin. = semilineatus; Leioceph. = Leiocephalus; chrysol. = chrysolaema.

TrophicNiche Breadth ed species of Uromacer and Antillophis Shannon-Weinervalues (H') for prey parvifrons, each species of snake has its speciesdiversity for each snakespecies in highest IRI value for a different genus of orderfrom broadestto narrowesttrophic prey. Alsophis cantherigerus and A. par- niche breadth are Antillophis parvi- vifrons exploit the most prey genera while frons = 2.304, Alsophiscantherigerus = Darlingtonia haetiana and U. oxyrhyn- 2.233, Uromacerfrenatus = 1.970, Dar- chus record the highest IRI values for sin- lingtonia haetiana = 1.861, Uromacer gle prey genera (respectively, Eleuthero- oxyrhynchus= 1.767, Uromacercates- dactylus and Anolis). byi = 1.373,Hypsirhynchusferox = 1.189. IRI values for prey species suggest that Threeof the four snakeswith the highest single prey species are sometimes very im- H' valuesare active foragers, and the three portant in diets of snakes. Uromacer ca- snakeswith the lowestH' value are exclu- tesbyi feeds largely on Osteopilus domini- sively (U. oxyrhynchus,H. ferox) ambush censis, Hypsirhynchus ferox on Ameiva foragersor, at least, a part-timeambush chrysolaema and A. parvifrons, and U. forager(U. catesbyi).Spearman rank cor- oxyrhynchuson Anolis cybotes. Uromacer relationprocedure indicated no significant > correlations(P 0.05) between H' and TABLE 4.-Correlations of snake morphologicalvari- samplesize, H' for prey speciesdiversity, ables with the first and second principal component and H' for prey generadiversity for each axes. snake species (Henderson,1984a) or for H' and any of severalmorphological vari- Variable Factor 1 Factor 2 ables (SVL,MBC, HW, SA). MBC 0.999 -0.046 HW 0.875 0.163 SAW 0.755 0.019 Index of RelativeImportance SBW 0.649 0.318 IRI valuesfor selectedprey generaand SVL -0.057 0.990 species are tabulatedin Tables 5 and 6. SL 0.077 0.990 With the exceptionof the two long-snout- HL 0.377 0.689 March 1988] HERPETOLOGICA 61

Birds- 4.0

3.0

2.0

O *Mus 0 mi Medium

Ameiva- *Osteopilus

0.0 X *Leiocephalus Small *Anolis /iaphaerodactylus -1.0 Eleutherodactylus

-1.5 -0.5 +0.5 1.5 2.5 Factor 2

FIG. 2.-Plot of mean scores for seven genera of prey and birds on the first two principal component axes. Factor 1 represents snake head and body width, Factor 2 represents snout, head, and body length. frenatus recorded high IRI values for An- ards, mammals), 67.2% of all frogs eaten olis olssoni and A. chrysolaema, but it also were <2.0 cm3, and all of the mammals had relatively high values for a number of (n = 10) were > 12.0 cm3. Lizards occurred other lizard species. in every size class, but the vast majority (64.0%) were <3.0 cm3. Prey Size Figure 4 illustrates the contribution of Hispaniolancolubrids took a wide range four lizard genera to snake diets by prey of prey sizes (0.1-40.0 cm3), but most prey size class. All Sphaerodactylus were < 1.0 items consumed were <3.0 cm3. Among cm3 and so were many Anolis, but most the three most commonly exploited anoles of the anoles eaten were 1-5.9 cm3. Anoles (A. coelestinus, A. cybotes, and A. disti- were less common in larger classes of prey chus), 60.6% were <3.0 cm3 and 19.1% size, and as their number decreased, Lei- were >5.0 cm3. Conversely, in a sample ocephalus and Ameiva became more im- (n = 74) of the same three species collected portant; Ameiva was the only lizard genus by local residents from a single locality, represented in the prey size class >14.9 79.7% were >3.0 cm3 (Fig. 3). The stom- cm3. ach content sample of Ameiva chrysolae- Prey size is a function of snake head ma had 67.9% < 10.0 cm3, but the collected width (P < 0.0001, F = 71.7, regression). sample (n = 25) had 92% > 10.0 cm3 (Fig. Figure 5 illustrates the percent contribu- 3). tion of prey size classes to the diets of snakes Within major prey groups (frogs, liz- of various head widths. The snakes with

TABLE 5.-Index of Relative Importance values for eight genera of prey exploited by six species of colubrid snakes.

Uromacer Uromacer Uromacer Genus Alsop. Antil. Darling. Hypsir. catesbyi frenatus oxyrhyn. Eleutherodactylus 89.8 576.3 17,955.1 Osteopilus 177.4 4.9 6016.2 - Sphaerodactylus 21.7 72.7 - - 2.1 Anolis 1545.5 7564.7 55.1 1806.6 2601.9 7619.0 14,406.3 Leiocephalus 24.5 30.2 445.8 73.8 338.6 27.5 Ameiva - 31.9 - 4963.1 1109.6 255.3 Celestus - 74.5 Mus 2245.0 28.0 62 HERPETOLOGICA [Vol. 44, No. 1

TABLE 6.-Index of Relative Importance values for frequently eaten species of prey by five species of Hispaniolan colubrid snakes. Alsophis cantherigerus was excluded, because it does not share prey species with the Hispaniolan snakes.

Uromacer Uromacer Uromracer Species Antil. Darling. Hypsir. catesbyi frenatus oxyrhyn.

Eleutherodactylus abbotti 43.5 740.8 - - Osteopilus dominicensis 8.7 6016.2 Anolis armouri 15.7 91.2 - Anolis caudalis 3.1 14.7 12.6 Anolis chlorocyanus 3.5 18.4 21.6 Anolis coelestinus 538.4 190.2 216.1 Anolis cybotes 3433.9 1025.0 247.1 213.7 2541.7 Anolis distichus 2.4 402.6 25.1 1811.6 Anolis olssoni 1323.1 16.6 Anolis semilineatus 9.9 9.4 10.2 272.8 Leiocephalus melanochlorus - 73.8 120.6 - Leiocephalus personatus 2.7 354.6 6.2 Leiocephalus vinculum 43.0 Ameiva chrysolaema 26.6 6135.4 1189.4 Ameiva taeniura - 4.0 96.6 the narrowest heads took prey only from detail elsewhere (Henderson et al., 1987a, the smallest size classes, but snakes with c), but see "Discussion" below. broader heads took a wide range of prey sizes, althoughthe smallestprey items were Summary not exploited by the snakes with the broad- To gain a clearer picture of levels of est heads. trophic specialization, we ranked each Alsophis cantherigerus and Uromacer species of snake in our sample (except Ial- catesbyi took the widest range of prey sizes tris dorsalis) by each of the variables that (Fig. 6). The Anolis specialist U. oxyrhyn- we used to quantify trophic ecology (IRI chus took the highest percentage (71%)of for prey genera/species; H'; prey size) and prey from a single size class (1-2.9 cm3). added another simple ranking based on the Alsophis and Hypsirhynchus ferox, both number of taxonomic classes and orders of relatively heavy bodied, took more prey prey that each species ate. Snakes were from the larger size classes than did the scored two points for each prey class (Am- other species. Although there were signif- phibia, Reptilia, etc.) and one point for icant differences (P < 0.001, F = 8.8, each prey order or suborder (Anura, Squa- ANOVA) in mean prey size taken by each mata-Sauria, -Serpentes, Ro- snake species, there was considerableover- dentia, etc.) eaten. Thus, each species of lap. snake (except for Alsophis cantherigerus Two species, Antillophis parvifrons and which did not have an IRI for the species , showed geographic of prey calculated) had a total of five num- differences in size of prey items consumed, bers summed to get the final value; the less and, concomitantly, in SVL (P < 0.01, specialized, the higher the total score. For ANOVA). Figure 7 presents a frequency those Hispaniolan species that are mem- distribution of SVL's for the two species bers of Maglio's (1970) cantherigerus on one or two satellite islands and at var- species assemblage, their scores were: Uro- ious other localities. The largest specimens macer catesbyi, 20; Uromacer frenatus, of A. parvifrons came from Ile de la Go- 19; Hypsirhynchus ferox, 17; Uromacer nave and Ile-a-Vache, Haiti. Both of these oxyrhynchus, 13. The two species outside species had exploited larger prey (Leio- the assemblage had the following scores: cephalus, Ameiva, Mus) on the satellite Antillophis parvifrons, 24; Darlingtonia islands than they did elsewhere, and these haetiana, 19. By contrast, although having dietary differences have been described in one less value to be summed, Alsophis March 1988] HERPETOLOGICA 63

50 creasedif we had examinedlarger sam- DlCollected Sample ples, but we doubtif any trophically"im- 40 DStomach portant" species are absent from our Contents Ameivachrysolaema 30 sample.It is possiblethat we have under- estimated the importanceof some geo- 20 graphicallyrestricted prey species,but in gaining a broadview of trophicecology, 10 [Ti it is unlikelythat increasingsample sizes would modify our resultsor interpreta- X L_ tions. 40 . Anolisspp. All of the dominantprey species(Table 3), whetherfrogs or lizards,diurnal or noc- turnal,are geographicallywidespread on 20 the Hispaniolanmain island,and most of themoccur on one or moresatellite islands (Hendersonand Schwartz,1984), and fre- quently in high relative densities(R. W. <1 1-2.9 3-4.9 5-9.9 10-19.9 20-39.9 >39.9 (cm3) Henderson,A. Schwartz,personal obser- Prey Size vations).Most of the smallerprey species FIG. 3.-Percent of prey size classes found in stom- are scansorial,whereas the larger lizards ach content samples collected by us and in samples collected by natives. The collected sample of Ameiva (Leiocephalus, Ameiva) are ground- chrysolaema was taken at 3 km NW Oviedo, Ped- dwelling.The primarylizard species are ernales Prov., Dominican Republic; the Anolis sam- diurnal,and the frogs are all nocturnal. ple (coelestinus, cybotes, distichus) was taken at 15 Anolis lizards were eaten more fre- km NW Cabral, Barahona Prov., Dominican Repub- quently (63.1%)than any other type of lic. prey, althoughmost do not make a pro- portionatecontribution to prey volume cantherigerus had the highest total score consumed(Fig. 1). Other,less frequently (33) indicating trophic generalization. eaten speciescontributed proportionately morefood volume (e.g., 0. dominicensis, A. chrysolaema,M. musculus). DISCUSSION The IRIvalues (Tables 5, 6) indicatethat Prey Species all of the snake species exploit Anolis at Although at least 52 species of prey were least occasionally,and some species (A. exploited by the snakes in our sample, only parvifrons,U. frenatus, U. oxyrhynchus) about 20%of those species were eaten fre- rely heavilyupon anolesas a food source. quently. The total number of prey species Certainspecies of Anolis(e.g., coelestinus, in our sample would certainly have in- cybotes,distichus, olssoni) make large fre-

100 IISphaerodactylus 80 LilAnolis

c 0~~ 60 E~~~~Leiocephalus 2 60 " _ l- {Ameiva 40

20

0 <1 1-2.9 3-5.9 6-9.9 10-14.9 >14.9 LizardSize Class (cm3) FIG. 4.-Percent of exploitationby Hispaniolancolubrids on four genera of lizards in six size classes. 64 HERPETOLOGICA [Vol. 44, No. 1

70 L11-2.9 60 . 13-5.9^

50 ]21014.9

o0 >14.9 0~~~~~~~~ 30

20

10

4-5.9 6-6.9 779 8-8.9 9-9.9 10-10.9 >10.9 Snake Head Width (mm) FIG. 5.-Percent of exploitation of prey size classes by snakes with different head widths. quency and/or volume contributionsto the al., 1987c). Similarly, Osteopilus domini- diet of some snakes;A. cybotes contributed censis was a majorcontributor to prey vol- over 40% of the total volume of prey to ume, but only U. catesbyi benefitted. It the diet of Antillophis parvifronsand about was the anoles that uniformly fell prey 50% of the total volume of prey for Uro- most frequently to all Hispaniolan colu- macer oxyrhynchus. Conversely, the slen- brids in our sample (Darlingtonia hae- der grass anole Anolis olssoni contributed tiana the exception). over 30% of the prey items taken by U. Hispaniola has a rich anole fauna (Wil- frenatus, but only about 10% of the prey liams, 1983) with species occurring from volume. sea level to >2400 m. They occur in most Ameiva chrysolaema contributed more habitats from ground level to the crowns to the total prey volume of all snake species of tall trees, in sunlight, shade, and sun- combined, but it was only important in the shade mosaics. Many species are tolerant diet of Hypsirhynchusferox and, to a less- of disturbance and can be found on and er extent, Uromacer frenatus (but pri- in human habitations; not surprisingly, marily on Ile de la Gonave: Henderson et some of the most widespread species are

70 0C1.0 60 11-2.9

]3-5.9 50 B6-9.9

C 40[ | II |,10-14.9 a h>149

30

20

10

Al isophs Ant4loph;s Darling. Hypsirhyn. U. catesbyi U, frenatus U oxyrhynchus FIG. 6.-Percent exploitation of prey size classes by seven species of snakes. March 1988] HERPETOLOGICA 65

30 Uromacer frenatus

20

10

50 . ElIVarious localities 40 ? I lie de la Gonave U lle-a-Vache 30

2 0 Antillophis parvifrons

0 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 Snake Size Class (cm) FIG. 7.-Frequency distributionof size classes (SVL) for Uromacerfrenatus and Antillophis parvifrons at satellite (Ile de la Gonave, Ile-a-Vache) and other localities. thosethat are mosttolerant of habitatdis- ple of 778 prey items from West Indian turbance.Although species densities are boid, trophidophiid, colubrid, and viperid not known for any Hispaniolananoles, snakes (n = approximately 2700 speci- elsewhereWest Indianspecies are known mens), only 0.3% were invertebrates (and to reachvery high densities(Gorman and all from Lesser Antillean Liophis; Hen- Harwood, 1977; Ruibal and Philibosian, derson and Crother, 1988). That inverte- 1974;Schoener and Schoener,1980). Ano- brates may occasionally fall prey to His- lis cybotes is the single most frequently paniolan snakes would not be surprising, exploitedprey species by Hispaniolanboid but possibly the ubiquitous distribution of and colubridsnakes. It is a geographically anoles in a wide array of sizes precludes widespreadspecies, occurs in a varietyof the necessity of predation on invertebrate habitats,and has a wide vertical distri- size prey. It is also noteworthy that inver- bution(ground to crown)in manyhabitats. tebrate predation by snakes may be rela- It is occasionally found under rocks tively rare except in North America (Shine, (Schwartzand Henderson,1982), and ju- 1977), although it is the rule in snakes of venilesand femalesare frequentlyfound the families Typhlopidae and Leptotyph- on the ground.Anolis lizardsin general lopidae. are the most conspicuousvertebrates on Eleutherodactylus is the most speciose Hispaniola,ranging throughout the main vertebrategenus on Hispaniola;species oc- island,most satelliteislands, in most hab- cur from sea level to >2400 m and in a itats, and frequently at high densities. variety of habitats. The frogs are ground- Diurnalcolubrid snakes, regardless of for- dwelling and scansorial, occur in a wide aging mode, must encounter them fre- range of sizes, and can reach high densities quentlyand, obviously,eat them routine- (Stewart and Pough, 1982). With the ex- ly. ception of Darlingtonia and Antillophis, What Hispaniolancolubrids do eat has Hispaniolan snakes rarely eat them (Fig. now been well defined.What they do not 1). The snakes that do eat Eleutherodac- eat is worthy of discussion.We have not tylus are all active foragers: the diet of recordeda single invertebratein the diet Antillophis is comprised of about 12% of any Hispaniolancolubrid, and in a sam- Eleutherodactylusby frequency and about 66 HERPETOLOGICA [Vol. 44, No. 1

17% by volume (Henderson et al., 1987a); predominantly on anoles, but with increas- the diet of Alsophis is about 6% by fre- ing size comes increasing diet specializa- quency Eleutherodactylus; the diet of tion on the common, widespreadhylid frog Darlingtonia is virtually 100% by fre- Osteopilus dominicensis. Adult U. cates- quency and volume Eleutherodactylus; byi prey on large adult Osteopilus and the Ialtris dorsalis exploits Eleutherodactylus, ephemerally abundant, recently meta- but we have not examined sufficient stom- morphosed young (and usually in multi- achs to gain an accurate view of the im- ples) (Henderson et al., 1987c). Similarly, portance of these frogs in its diet. young H. ferox prey predominantly on an- Before the arrival of Europeans in the oles, but large adults feed almost exclu- late 15th century, the only Mus-sized sively on the teiid lizard Ameiva chryso- mammal on Hispaniola was the insectivore laema (Henderson, 1984b). Uromacer Nesophontes zamicrus (Miller, 1929). It frenatus and U. oxyrhynchus feed pre- was widespread on the island (J. A. Otten- dominantly on anoles, although on Ile de walder, personal communication) and la Gonave, U. frenatus prey heavily on A. likely fell prey to boid and colubrid snakes, chrysolaema. Perhaps most misleading is but most Hispaniolan mammals attained the H' value determined for Darlingtonia adult sizes too large for colubrids to swal- haetiana. This small species feeds almost low. Only large examples of the boid Ep- exclusively on Eleutherodactylus and its icrates striatus feed routinely on mam- egg clutches; it will feed occasionally on mals on Hispaniola (Henderson et al., anoles. It is the most trophically unique of 1987b). the Hispaniolan colubrids but, because it According to Henderson and Crother feeds on a wide variety of Eleutherodac- (1988), xenodontine colubrids "may have tylus species (Henderson and Schwartz, been the optimum 'type' of snake for ra- 1986), it scored a fairly high H' value. Al- diating in the West Indies . . .". They are though we advocate identifying prey items adapted morphologically, ecologically, and to the highest possible taxonomic level, we behaviorally to trophically exploit the kinds suggest that, just as prey identifications at of vertebrate prey that occur with the broad taxonomic levels may be misleading greatest diversity and at the highest pop- (Greene and Jaksic, 1983), so too can iden- ulation densities in the West Indies (i.e., tifications at narrow levels. lizards and frogs). Nussbaum (1984) sim- ilarly suggested that snakes that eat lizards Prey Size and birds are more common on oceanic Hispaniolan colubrids, although eating islands because lizards and birds are more a wide array of prey sizes, fed primarily common on oceanic islands. on small prey items. Although it was ex- pected that small snakes would consume Trophic Niche Breadth small prey items, large snakesroutinely ate The Shannon-Weiner values (H') for small prey items in addition to "large" each species of snake produced expected ones (Fig. 5). Even if our native-collected as well as unexpected results, and certainly sample is biased towards larger individu- results different from earlier calculations als, it still indicates that larger prey items done for the same prey items identified were available. only to genus (Henderson, 1984a). That Antillophis parvifrons and Uromacer the active foragers Antillophis parvifrons frenatus attain larger SVL's on Ile de la and Alsophis cantherigerus have the high- Gonave, Haiti than elsewhere in their est H' values was expected. At first glance, ranges (Fig. 7), and this is reflected in their Uromacer catesbyi and Hypsirhynchus diets (Henderson et al., 1987a,c); both ferox do not seem trophically specialized, species eat a greater proportionof Ameiva but examination of ontogenetic changes in chrysolaema on Gonave than elsewhere, their diets indicates that they both become and it is the only place that Antillophis quite restricted. Small U. catesbyi feed exploited rodents. We have the impression March 1988] HERPETOLOGICA 67 from fieldwork on Gonave that popula- of polarity (or evolution of the traits) are tions of Ameiva (and Leiocephalus) reach possible. The first (Fig. 8b) suggests that greater densities there than most other the generalized diet and active foraging areas in which we have collected. It is pos- mode are primitive, with the specialized sible that increased body size may be a diet and sit-and-wait foraging mode de- response to the presence of an abundant, rived. The alternative (Fig. 8c) suggests "large" prey species. Similar patterns of that the generalized diet and active for- insular trends in body size related to type aging mode are derived for Alsophis, (size) and abundance of prey have been whereas the specialized diet and sit-and- noted in viperid (Case, 1978) and elapid wait foraging mode are primitive. The (Schwaner, 1985) snakes. presence of these two equally parsimoni- ous hypotheses requires a decision based Trophic Ecology and Phylogeny on additional information. Ecological and behavioral characteris- Thomas (1976), in his work on the snake tics within monophyletic lineages can be genus Philodryas, suggested that Alsophis shown to be under phylogenetic con- was a close relative to the former, and was straints (thus have an historical explana- perhaps derived from that stock. By using tion) or be the result of recent, indepen- Philodryas as an outgroup, it was possible dent selective forces, through their fit on to choose between the competing hypoth- cladistic estimations of relationship (Cod- eses. Philodryas is an active foraging op- dington, 1985; Dobson, 1985; Greene, in portunist,as is Alsophis. Consequently, we press;Wanntorp, 1983). "Fit" refers to the feel this supports hypothesis 8b (Fig. 8), ratio of the number of steps required for because hypothesis 8c requires reversalsto the evolution of a character on the clado- explain the character state distributions. gram to the minimum number of possible The genus Uromacer shows a similar steps. pattern. According to Horn (1969) and Maglio (1970) completed the only phy- Maglio (1970), U. catesbyi is the sister tax- logenetic study encompassing all the West on to a U. frenatus-U. oxyrhynchus clade. Indian xenodontines. He concluded that Within this genus, foraging mode and diet four unique lineages are present, three of preferences reflect phylogeny (Alsophis which occur on Hispaniola.Alsophis, Hyp- was the outgroup). From the above anal- sirhynchus, and Uromacer belong to the ysis, active foraging habits are primitive "cantherigerus"assemblage, with only Al- and sit-and-wait foraging habits are de- sophis found elsewhere. Antillophis par- rived from Uromacer. Uromacer catesbyi, vifrons is part of the "andreae" assem- however, retains the active mode as well. blage, with its only other member on Cuba, Thus, two hypotheses of the evolution of and Darlingtonia belongs to the "funer- foraging habits within Uromacer are pos- eus" assemblage whose supposed relations sible. One suggests that U. catesbyi re- are with Arrhyton, which is restricted to tained the primitive state (active foraging) Puerto Rico, Cuba, and Jamaica. Trophic in combination with the derived state (sit- ecologies were only known for the His- and-wait foraging) and that the primitive paniola taxa, thus they were examined in state was lost in U. frenatus and U. oxy- terms of interspecific relationships within rhynchus (Fig. 8d). The alternative sug- Hispaniola. gests that the U. catesbyi condition is not Within the "cantherigerus" assem- the retention of the primitive trait, but is blage, Maglio (1970) concluded that Al- a reversalor convergency (Fig. 8e). It seems sophis is the sister taxon to a Hypsirhyn- unlikely that the presence of- the active chus-Uromacer clade (Fig. 8a). Alsophis foraging mode in U. catesbyi is due to an actively forages and is a generalist in diet, independent evolutionary event. The for- whereas Hypsirhynchusand Uromacerare mer (Fig. 8d) is a more tenable hypothesis. clearly more specialized in diet and forage The diet preference character for Uro- in a sit-and-wait fashion. Two hypotheses macer also reflects phylogeny. Specialized 68 HERPETOLOGICA [Vol. 44, No. 1

H

a.

\ ~~b. pc

act \~~~~~~~~~~~~~~~c

gen. act e.

FIG. 8.-Hypotheses of ecological character evolution as plotted on phylogenetic estimates of relationship for the Hispaniolan colubrids: (a-c) the cantherigerus assemblage; (d, e) Uromacer with foraging mode. Abbreviations:H = Hypsirhynchus, U = Uromacer, A = Alsophis, f = frenatus, o = oxyrhynchus, c = catesbyi, sw = sit-and-wait, act = active forager, spec = specialist, gen = generalist, swa = sit-and-wait and active. March 1988] HERPETOLOGICA 69 diet evolved from a generalized diet and havior and ecology. Evolution 39:1384-1388. became further differentiated within Uro- GORMAN, G. C., AND R. HARWOOD. 1977. Notes on macer. Uromacer catesbyi exploits Osteo- population density, vagility, and activity patterns of the Puerto Rican grass lizard, Anolis pulchellus pilus, whereas U. frentus and U. oxyrhyn- (Reptilia, Lacertilia, Iguanidae). J. Herpetol. 11: chus exploit lizards. 363-368. The trophic characteristics of the rela- GREENE, H. W. 1986. Diet and arboreality in the tives of the three remaining taxa, Dar- emerald monitor, Varanus prasinus, with com- and Antillophis, are not ments on the study of adaptation. Fieldiana: Zo- lingtonia, Ialtris, ology 31:1-12. known. This prevents analysis of the evo- GREENE, H. W., AND F. M. JAKSIC. 1983. Food- lution of their foraging habits or diet pref- niche relationships among sympatric predators: Ef- erences. Antillophis parvifrons and Ialtris fects of level of prey identification. Oikos 40:151- dorsalis are active foraging generalists, so 154. HENDERSON, R. W. 1982. Trophic relationships and they possess the primitive condition. Dar- foraging strategies of some New World tree snakes lingtonia is an active forager and is spe- (Leptophis, Oxybelis, Uromacer). Amphibia-Rep- cialized in diet, feeding almost exclusively tilia 3:71-80. on the frogs of the genus Eleutherodac- . 1984a. The diets of Hispaniolan colubrid tylus. This unique diet may reflect the snakes I. Introduction and prey genera. Oecologia 6:234-239. unique lineage, but without additional in- 1984b. The diet of the Hispaniolan snake formation other explanationsare plausible. Hypsirhynchus ferox (). Amphibia- Hypothesized patterns of polarity were Reptilia 5:367-371. derived without convergences. This sug- HENDERSON, R. W., M. H. BINDER, AND R. A. SAJ- DAK. 1981. Ecological relationships of the tree gests that the ecological and behavioral snakes Uromacer catesbyi and U. oxyrhynchus characteristics of the Hispaniolan xeno- (Colubridae) on Isla Saona, Republica Dominicana. dontines are best explained in historical Amphibia-Reptilia 2:153-163. terms, and not with independent adaptive HENDERSON, R. W., AND B. I. CROTHER. 1988. Bio- or selective explanations. geographic patterns of predation in West Indian colubrid snakes. In C. A. Woods (Ed.), Biogeog- Acknowledgments.-We are indebted to the per- raphy of the West Indies. E. J. Brill, Leiden. In sonnel in charge of several collections: Richard G. press. Zweifel (AMNH), Ronald I. Crombie and W. Ronald HENDERSON, R. W., B. I. CROTHER, T. A. Heyer (USNM), Peter Meylan, David Auth, and Wal- NOESKE-HALLIN, A. SCHWARTZ, AND C. F. ter Auffenberg (UF), Richard Thomas (RT), and es- DETHLOFF. 1987a. The diet of the Hispaniolan pecially Jos6 P. Rosado and Pere Alberch (MCZ). colubridsnake Antillophis parvifrons.J. Herpetol. Henderson's fieldwork on Hispaniola has been sup- 21:328-332. ported by Friends of the Milwaukee Public Museum, HENDERSON, R. W., J. F. DROUGHT, AND A. D. KATZ. the Fundacion Gulf & Western Dominicana, Albert 1982. Daily activity patterns of the Hispaniolan Schwartz, the American Philosophical Society, and tree snakes Uromacercatesbyi and Uromaceroxy- the Museo Nacional de Historia Natural in Santo Do- rhynchus (Colubridae). Ann. Mtg. Soc. Stud. Am- mingo. Craig R. Dethloff provided excellent assis- phib. Rept. and Herpetol. League, p. 78 (Abstract). tance in the dissection of snakes, and Christine Cora- HENDERSON, R. W., T. A. NOESKE-HALLIN, J. A. dini typed several drafts of the manuscript. We have OTTENWALDER, AND A. SCHWARTZ. 1987b. On benefitted greatly from the critical comments of Gary the diet of the boa Epicratesstriatus on Hispaniola, S. Casper, Henry S. Fitch, Max A. Nickerson, Jose A. with notes on E. fordi and E. gracilis. Amphibia- Ottenwalder, Richard A. Sajdak, Laurie J. Vitt, and Reptilia 8:251-258. Ernest E. Williams on earlier versions of the manu- HENDERSON, R. W., AND A. SCHWARTZ. 1984. A script, and we thank C. Guyer, D. M. Hillis, and J. guide to the identification of the amphibians and M. Savage for discussion and comments on the phy- of Hispaniola. Milwaukee Public Mus. Spec. logenetic aspects. Publ. Biol. Geol. 4:1-70. 1986. The diet of the Hispaniolan colubrid LITERATURE CITED snake, Darlingtonia haetiana. Copeia 1986:529- 531. CASE, T. J. 1978. A general explanation for insular HENDERSON, R. W., A. SCHWARTZ, AND T. A. body size trends in terrestrial vertebrates. Ecology NOESKE-HALLIN. 1987c. Food habits of three col- 59:1-18. ubrid tree snakes (genus Uromacer)on Hispaniola. CODDINGTON, J. H. 1985. [Review of] M. Ridley, Herpetologica 43:235-242. The explanation of organic diversity: The compar- HORN, H. S. 1969. Polymorphism and evolution of ative methods and adaptations for mating. Cladis- the Hispaniolan snake genus Uromacer (Colubri- tics 1:102-107. dae). Breviora (324):1-23. DOBSON, F. S. 1985. The use of phylogeny in be- MAGLIO, V. J. 1970. West Indian xenodontine col- 70 HERPETOLOGICA [Vol. 44, No. 1

ubrid snakes:Their probableorigin, phylogeny and SCHWARTZ, A., AND R. W. HENDERSON. 1982. Ano- zoogeography. Bull. Mus. Comp. Zool. 141:1-54. lis cybotes (Reptilia: Iguanidae): The eastern His- MARTINEZ, D. R., J. A. LUCIO, AND A. SCHWARTZ. paniolan populations. Milwaukee Public Mus. Con- 1985. Topografia interna de las culebras del genero trib. Biol. Geol. (49):1-9. Uromacer (Colubridae). Caribaea 1:48-58. SHINE, R. 1977. Habitats, diets, and sympatry in MILLER, G. S., JR. 1929. A second collection of snakes: A study from Australia. Can. J. Zool. 55: mammals from caves near St. Michel, Haiti. Smith- 1118-1128. sonian Misc. Coll. 82:1-30. STEWART, M. M., AND F. H. POUGH. 1982. Popu- NUSSBAUM, R. A. 1984. Snakes of the Seychelles. lation density of tropical forest frogs: Relation to Pp. 361-367. In D. R. Stoddart (Ed.), Biogeography retreat sites. Science 221:570-572. and Ecology of the Seychelles Islands. W. Junk, THOMAS, R. A. 1976. A Revision of the South Amer- The Hague. ican Colubrid Snakes Genus Philodryas Wagler, PINKAS, L., M. S. OLIPHANT, AND I. L. K. IVERSON. 1830. Ph.D. Dissertation, Texas A&M University. 1971. Food habits of albacore, blue fin tuna and ULRICH, N., AND N. B. FORD. 1985. The feeding bonito in California waters. Dept. Fish Game Bull. strike of three species of vine snake, Uromacer. California 152:1-105. Ann. Mtg. Soc. Stud. Amphib. Rept. and Herpetol. RUIBAL, R., AND R. PHILIBOSIAN. 1974. The pop- League, p. 81 (Abstract). ulation ecology of the lizard Anolis acutus. Ecology WALLS, G. L. 1942. The Vertebrate Eye and Its 55:525-537. Adaptive Radiation. Bull. Cranbrook Inst. Sci. 19. SAS INSTITUTE. 1985. SAS Users Guide: Statistics, WANNTORP, H. E. 1983. Historical constraints in 5th Ed. Cary, North Carolina. adaptation theory: Traits and non-traits. Oikos 41: SCHOENER, T. W., AND A. SCHOENER. 1980. Den- 157-159. sities, sex ratios, and population structure of four WILLIAMS, E. E. 1983. Ecomorphs, faunas, island species of Bahamian Anolis lizards. J. Anim. Ecol. size, and diverse end points in island radiations of 49: 19-53. Anolis. Pp. 326-370. In R. B. Huey, E. R. Pianka, SCHWANER, T. D. 1985. Population ecology of black and T. W. Schoener (Eds.), Lizard Ecology. Har- tiger snakes, Notechis ater niger, on offshore islands vard University Press, Cambridge, Massachusetts. of South Australia. Pp. 35-46. In G. Grigg, R. Shine, and H. Ehmann (Eds.), Biology of Australian Frogs Accepted: 23 March 1987 and Reptiles. Royal Zool. Soc., New South Wales. Associate Editor: Arthur Dunham SCHWARTZ, A. 1980. The herpetogeography of His- paniola, West Indies. Stud. Fauna Curacao Carib. Isl. 61(189):86-127.