Music & Video

A Morphometric Comparison of Island and Mainland Boas (Boa Constrictor) in Belize

A Morphometric Comparison of Island and Mainland Boas (Boa Constrictor) in Belize
of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
   A Morphometric Comparison of Island and Mainland Boas( Boa constrictor  ) in Belize S COTT  M. B OBACK  Relative to mainland  Boa constrictor  , boas from islands off the coast of Belize aredescribed as being smaller, having longer tails, more elongate snouts, andproportionately larger eyes. However, no systematic confirmation of these patternshas been made. A morphometric study was initiated to investigate the body size andhead shape variation between island and mainland boas in Belize. One hundred twenty-nine boas from five islands and the mainland were caught and measured. Univariateand multivariate analyses indicated that, in general, previous descriptions are accurate.Island boas are about half the length and one-fifth the mass of mainland boas. Incontrast to mainland boas, no sexual size dimorphism is evident in island boas. Thehead shape of island boas differs from that of mainland boas but this divergence is not consistent among populations. Some island boas have more attenuate snouts comparedto mainland boas whereas other island boas have larger eyes and narrower heads. Maleisland boas have longer tails compared to males from the mainland, but sucha difference is not found in females from the two localities. The morphology of islandboas is consistent with an arboreal habit and reduced prey size. Because these changeshave occurred over an extremely short time interval, this may be another example of the speed and magnitude of adaptation that is possible in squamates. P OPULATIONS of boas ( Boa constrictor  ) arecommon on offshore islands in Belize,Central America. Anecdotal information frompet-trade collectors visiting one of these islands(Crawl Cay) in 1970 and 1971 described islandboas as being dwarfed in size, with more elongatesnouts, and with longer tails compared with boason the mainland (Porras, 1999). Additionally,the eyes of island boas were described as largerthan those of mainland boas, but this varied fromindividual to individual. Two additional islandsoff Belize (Wee Wee Cay and Coco Plum Cay) were surveyed in 1990, and the five specimenscaptured from the two islands appeared to besimilar to those collected two decades earlierfrom Crawl Cay (B. Sears, unpubl. data). Thislatter report was significant because it suggestedpopulations of dwarfed boas occurred on multi-ple islands off the coast of Belize. In contrast tothe terrestrial habits of mainland boas (Mont-gomery and Rand, 1978), island boas were almost always found in trees and tree cavities and weresuggested to consume mostly birds (Lillywhiteand Henderson, 1993).Such observations are of scientific interest because this system could demonstrate themagnitude and speed of morphological adapta-tion that is possible (e.g., Losos et al., 1997; Aubret et al., 2004). A slender body form (lighterand shorter) and a relatively long tail are featurescharacteristic of arboreal snake species (Shine,1983; Lillywhite and Henderson, 1993), whereasextreme snout attenuation is a feature of highly  visual snake species that feed on fast-moving prey (Henderson and Binder, 1980). Thus, themorphology of island boas may be a responseto a more arboreal habit and altered prey resources. Similarly, anoline squamates haveevolved to fit various ecomorphological roles ondifferent islands in the Greater Antilles (Wil-liams, 1983), are predictable, and occur overstrikingly short time intervals (10–14 years; Lososet al., 1997). Additionally, the current theory onbody size evolution suggests that snake specieslonger than approximately one meter becomeshorter on islands, whereas species shorter thanone meter become longer on islands. Thesechanges may be due to the existence of anoptimal body size (Boback and Guyer, 2003).Since maximum size of mainland boas is about five meters (Stafford and Meyer, 2000), signifi-cantly smaller body size in island boas would alsosupport this theory.However, morphological information fromisland boas is scattered and has not beenquantified and evaluated in a rigorous manner.Further, the anecdotal descriptions are basedlargely on specimens collected from a singleisland (Crawl Cay) and may not be representativeof other island populations. The goal of thecurrent project was to describe the morphologi-cal variation in boas from five islands andmainland Belize. Specifically, I document body size, tail length, and head shape variationbetween island and mainland boas as a first steptowards understanding the ecological and evolu- Copeia,  2006(2), pp. 261–267 #  2006 by the American Society of Ichthyologists and Herpetologists  tionary significance of these characteristics inboas.M  ATERIALS AND  M ETHODS Boas were collected, measured, and releasedduring surveys conducted on the mainland andfive offshore cays of Belize during portions of three years: 2 August–15 August 2001, 13 May–1 July 2002, and 17 July–6 August 2003 (seeBoback, 2005 for study site details and map).These cays are located within the shallow lagoonbetween the Belize coast and the MesoamericanBarrier Reef. This lagoon extends 20–45 kmoffshore and varies in depth from 20 m in thenorth (Belize City) to 35–77 m in the south (Gulf of Honduras). Beneath this lagoon lies the Belizecoastal shelf, where Purdy (1974) dated peat samples to 6,000–10,000 yr B.P. Modern reef growth therefore post-dates the Holocene sea level rise (12,000–15,000 yr B.P.; Stoddart et al.,1982). Reef patches close to the sea surface likely accumulated sediments, ultimately leading to cay establishment. Because these cays were createdby exposure of reef patches and a rise of sea level,they are likely to be less than 8,000 years old andformed more or less simultaneously.Body length was recorded for each individual,divided into snout-vent length (SVL) to theclosest 0.5 cm and tail length to the closest mm. Body mass was measured with Pesola springscales and recorded to the nearest gram. Usingdial calipers, the following head measures wererecorded to the nearest 0.01 mm: head width  5  widest head width, head length 5 anterior tip of rostral scale to posterior tip of mandible, labial 5 posterior edge of posterior-most labial scale toanterior tip of rostral scale, interocular  5 narrowest dorsal distance between orbits, ocular 5 horizontal diameter of the ocular scale, nares–ocular  5  anterior edge of ocular to posterioredge of nares, rostral–ocular  5  anterior edge of ocular to anterior tip of rostral scale, andinternares  5  narrowest internarial distance. With the exception of 22 boas retained to obtainlitters for neonate measures, all boas werereleased at their point of capture.To test whether island boas were smaller in sizecompared to mainland boas, body mass and SVLmeasures were log 10 -transformed and the groups were compared using an analysis of covariance(ANCOVA). Log transformation was performedto make the relationship between mass and SVLlinear. For each ANCOVA, I tested for theequality of slopes and, if the results of this test  were non-significant, I tested for a difference inintercepts using a test of adjusted means. To test for sexual size dimorphism (SSD), mean SVLbetween adult males and females was comparedusing a 2-way ANOVA with sex and location(island versus mainland) as main effects andan interaction effect of sex*location. TheLSMEANS statement and the PDIFF option inSAS were used to construct a matrix of the least squares differences among all interactions of sexby location using a Tukey-Kramer adjustment formultiple comparisons. The smallest known re-productive boa was 61 cm in total length (Ronne,1996), and this length was used to delimit  juvenile and adult boas. Additionally, it wastested whether island boas (separate analysis foreach sex) had longer tails compared to mainlandboas (one-tailed, non-parametric,  t  -tests [Wil-coxon two-sample test] on the residuals fromregressions of tail length on SVL). This test wasused because these data violated the assumptionof normality.To compare head shape between island andmainland boas, the linear head measures wereanalyzed using both univariate and multivariatestatistics. Head measures were compared be-tween island and mainland boas by regressingeach measure on head length and comparing thegroup residuals using ANCOVA. This analysisallowed a univariate comparison of each of thehead measures independent of other measure-ments. To compare head shapes between loca-tions in multivariate space, discriminant analysis was performed and canonical functions werederived (canonical discriminant analysis, CDA)to describe overall head shape. Canonical dis-criminant analysis is related to principal compo-nents analysis (PCA), as both are dimension-reduction techniques. Canonical discriminant analysis differs from PCA in assuming groupsare assigned  a priori   through some externalmeans and aims to summarize the data intocanonical functions that maximize group separa-tion (Tabachnick and Fidell, 2001; SAS vers.8.2, 4 th ed. Vols. 1–2, Statistical Analysis SystemsInstitute, Inc. Cary, NC, 1990). Therefore,because these data were from known populations(islands and the mainland), CDA was the moreappropriate analysis for these data. I examined whether the five island populations and themainland population were significantly separatedin morphological space by performing a multivar-iate analysis of variance (MANOVA) using thefirst two canonical functions. This is an appro-priate analysis because canonical functions arelinear combinations of the srcinal head mea-sures that maximize group separation and, by definition, are uncorrelated with each other(Klecka, 1980). A likelihood ratio test wasperformed to determine significant canonicalfunctions to be used in the MANOVA and the262 COPEIA, 2006, NO. 2  interpretation of the canonical functions wasfacilitated by examining the canonical loadings(the larger the loading, the greater the contri-bution of the original head measure to that function; Klecka, 1980). Discriminant analysisalso allows one to evaluate how well the srcinal variables can classify individuals into their re-spective groups. Classification was a secondary goal of this analysis. Allometic growth can cause extreme bias inanalyses of shape differences (Bookstein et al.,1985). Based on previous analyses that showedthe majority of variation in these data was due to variation in size, I used size-adjusted headmeasures in the CDA, obtained by using residualscalculated from Ordinary Least Squares (OLS)regression of the log 10 -transformed values onhead length (Schulte-Hostedde et al., 2005). Logtransformation was employed here to assureindependence of variances. All statistical tests were performed using SAS(vers. 8.2, 4 th ed. Vols. 1–2, Statistical AnalysisSystems Institute, Inc. Cary, NC, 1990). For alltests,  a  was set at   P   5  0.05 with dispersionindicated with 6 1 SD unless otherwise indicated(Zar, 1984).R  ESULTS One hundred twenty-nine boas were capturedand measured: Crawl Cay   5  26, Lagoon Cay   5 12, False Cay   5  9, Peter Douglas Cay   5  4, West Snake Cay  5 40, Mainland 5 38. Because of low sample size, boas from Peter Douglas Cay wereexcluded from all analyses of among-population variation. Mainland boas were collected in areasadjacent to three main cities within Belize (BelizeCity, Belmopan, and Punta Gorda), but themajority was collected from a single region(Belmopan). Mainland boas were most oftendiscovered while crossing roads after rains and while moving about in city garbage dumpsduring nocturnal surveys.In comparing the relationship of mass andlength, I found a significant difference in slopesfor island and mainland boas (ANCOVA,  F  1,129 5 19.79,  P  , 0.0001). Mainland boas had a steeperslope (3.4) compared to island boas (3.0)indicating that island boas were lighter andshorter compared to mainland boas. Mainlandboas achieved a maximum SVL of 2.3 m anda maximum mass of 8.6 kg. In contrast, islandboas reached a maximum SVL of 1.5 m anda maximum body mass of 1.7 kg (Fig. 1). Further,all island boas were consistently shorter thanmainland boas (Lagoon Cay mean SVL 5 948 6 239 mm, Peter Douglas Cay mean SVL  5  884  6 158, Crawl Cay mean SVL  5  1220  6  63 mm,False Cay mean SVL  5  1108  6  157 mm, West Snake Cay mean SVL 5  1190  6 179 mm). While mainland boas exhibited SSD in body length (male mean SVL 5 1164 6 110 mm,  n  5 8; female mean SVL 1478 6 513 mm,  n  5 13;  P  5 0.008), no SSD in body length was evident withinisland boas (male mean SVL 5 1158 6 186 mm, n   5  41; female mean SVL  5  1098  6  195 mm,  n  5 28;  P  5 0.34; Fig. 1). In comparing mean SVLbetween mainland and island snakes, mainlandfemales were significantly longer than bothisland males ( P   5  0.0002) and island females( P   ,  0.0001). Mainland males were not signifi-cantly different in mean SVL from either islandmales or island females.There was no indication that female islandboas had longer tails compared to femalemainland boas (mean tail length residual forisland boas  5 2 1.67  6  7.4,  n   5  32; mean taillength residual for mainland boas 5 3.35 6 20.6, n   5  16; Wilcoxon two-sample test,  S   5  432,  P   5 0.19). However, island males did have signifi-cantly longer tails compared to mainland males(mean tail length residual for island boas 5 1.68 6  22.4,  n   5  45; mean tail length residual formainland boas  5 2 6.89  6  17.8,  n   5  11; Wilcoxon two-sample test,  S   5 235,  P   5 0.05).To compare head shape differences betweenisland and mainland boas, the relative differ-ences in each of the head measures wereevaluated by plotting each measure on headlength. Head measures had different amounts of  variation as indicated by the coefficients of  variation (CV). Mainland boas had larger CVsthan island boas for all measures (Table 1). Fig. 1. Frequency distributions of snout-vent lengths (left hand side) and masses (right handside) for Belize mainland and island  Boa constrictor  .Black bars 5 females (mainland  n  5 16, island  n  5 33), gray bars  5  males (mainland  n   5  11, island  n  5 45). BOBACK—ISLAND BOA MORPHOMETRICS 263   Analysis of covariance results indicated that island boas had disproportionately narrowerheads, longer labial lengths, longer nares–oculardistances, shorter internares distances, and largereyes compared to mainland boas. Compared tohead length, head width, labial length, nares–ocular distance, and internares distance all hadsignificantly shallower slopes in island boascompared to mainland boas, whereas oculardiameter had a significantly steeper slope inisland boas compared to mainland boas. Addi-tionally, ANCOVA results indicated a non-signif-icant difference in elevation (y-intercept) forinterocular distance between island and main-land boas. That is, in relation to head length,island boas had proportionally narrower intero-cular distances compared to mainland boas(Table 1).To further explore head shape variationbetween island and mainland boas, I used allhead measures in a CDA on size-adjusted headmeasures. Island and mainland populations of boas were significantly separated by these size-adjusted head measures ( n   5  125, Wilks’  l  5 0.39,  P  , 0.0001). A likelihood ratio test showedthat the first two canonical functions significantly contributed to separating the groups and cumu-latively explained greater than 91 %  of the variation in the data ( P   ,  0.0001 for bothfunctions). Successful classification of individualsto their population of origin varied amongpopulations (35–78 %  success) but only 15 %  of island boas were incorrectly classified to themainland population.The highest loadings for canonical function 1 were labial length (0.78) and nares–oculardistance (0.68; Table 2). Rostral–ocular andinternares distance also showed strong correla-tion with function 1. Hence this functiondescribes head and snout attenuation. Individu-als with low function 1 scores have short snoutsrelative to head length, whereas individuals withlarge function 1 scores have long snouts relativeto head length. The highest loadings for canon-ical function 2 were ocular diameter ( 2 0.82) andhead width (0.65; Table 2). Hence this functiondescribes eye size relative to head width. Individ-uals with low function 2 scores have large eyesand narrow heads, whereas individuals with largefunction 2 scores have small eyes and wide heads. A MANOVA using the first two canonical func-tions showed that populations of boas weresignificantly separated by canonical functions 1and 2 (Wilks’  l 5 0.43,  F  8,125 5 15.5,  P  , 0.0001). All island boa populations, except for Crawl Cay,differed from the mainland boa population inhaving larger eyes and narrower heads (all  P   values  #  0.05, Tukey’s Studentized Range test).Crawl Cay and False Cay boas differed frommainland boas in having more attenuated snouts(all  P   values  #  0.05, Tukey’s Studentized Rangetest; Fig. 2).D ISCUSSION The results of this study clarify anecdotalreports from collectors in the late 1970s anddemonstrate that a short, thin body and an T  ABLE  1. M EANS AND  C OEFFICIENTS OF  V   ARIATION  (CV)  FOR   R   AW   H EAD  M EASURES FOR   I SLAND AND  M  AINLAND  B OAS IN B ELIZE . ANCOVA results of relative head measures [log 10 (head measure)  3  log 10 (head length)] of island andmainland boas are indicated. For a detailed description of head measures see Materials and Methods section.Slope  5  results from equality of slopes test, Intercept   5  results from test of adjusted means. I  5  island, M  5 mainland. Bolded values are significant. Head measureMean I n   5  91 CV Mean M n  5  38 CV Slope Intercept I M df   F P   I M df   F P  Head width 23.83 25 26.08 50 0.95 1.07 1 12.12  0.0007  - - - - -Labial 31.82 25 31.43 46 0.95 1.01 1 4.65  0.033  - - - - -Interocular 14.45 23 15.09 41 0.86 0.89 1 0.73 0.39 0.21 0.25 1 3.34 0.07Ocular 5.18 20 4.86 23 0.74 0.51 1 34.31  , 0.0001  - - - - -Nares-ocular 11.59 25 11.79 47 0.97 1.02 1 4.59  0.034  - - - - -Rostral-ocular 15.31 25 15.58 46 0.95 1.00 1 2.33 0.13 0.34 0.43 1 1.94 0.16Internares 5.77 21 5.87 40 0.75 0.90 1 18.35  , 0.0001  - - - - -T  ABLE  2. C  ANONICAL  L OADINGS  (T OTAL  S TRUCTURE C OEFFICIENTS )  FROM  CDA   OF  S EVEN  H EAD  M EASURES .  VariableCanonical loadingsFunction 1 Function 2 Head width  2 0.2062 0.6504Labial 0.7776  2 0.2244Interocular  2 0.2937 0.2223Ocular  2 0.1569  2 0.8167Nares-ocular 0.6755 0.1955Rostral-ocular 0.5225 0.1369Internares 0.5798 0.1955 264 COPEIA, 2006, NO. 2  attenuated snout are general features of islandboas, but not all island populations share this last feature. Additionally, island boas have relatively larger eyes compared to mainland boas, but a large amount of variation exists for thischaracter. Lastly, these results confirm that islandboas have longer tails compared to mainlandboas, but only for males and not females.Dwarfism of island boas is consistent with thegeneral pattern for snakes (Case, 1978; Boback,2003) and is consistent with an optimal sizehypothesis (Boback and Guyer, 2003). Addition-ally, the altered body plan of island boas (thinnerand shorter) is consistent with a morphologicaltendency for arborealsnakes tobeslenderand/orlaterally compressed compared to terrestrialsnakes (Shine, 1983; Lillywhite and Henderson,1993). Few data are available for boa movementsand habitat preferences, but existing evidencesuggests that mainland boas are largely terrestrialand use mammal burrows for refuge and ambushsites (Montgomery and Rand, 1978). I observedthe majority (96 % ) of island boas in trees and treehollows (compared to 18 %  for mainland boas)and therefore these morphological differencesare consistent with adaptations for arboreality. A more attenuate snout in island boas couldindicate a trophic modification to different prey abundances and/or sizes on islands (Boback,2003). Mainland boas have a catholic diet composed of lizards, birds, and mammals(Greene, 1983; Greene et al., 2003; Boback,2004), whereas the most common prey item forisland boas is passerine birds (Boback, 2005).The significance of an elongate snout for a birdspecialist is not clear, in part because there arefew snake species available for comparison that specialize on avian prey. The few snake speciesthat are known to specialize on birds are alsoisland endemics (e.g.,  Bothrops insularis  , Duarte et al., 1995;  Macrovipera schweizeri  , Nilson et al.,1999;  Gloydius shedaoensis  , Sun et al., 2001), but morphological data for these populations ortheir hypothesized sister taxa are not available. Alternatively, the elongated snout of islandboas may be an adaptation to saurophagy.Henderson and Binder (1980) found that Vinesnakes ( Ahaetulla  ,  Oxybelis  ,  Thelotornis  ,  Uromacer  )that specialize on lizards have longer snoutscompared to those that specialize on frogs.Further, snout attenuation is a characteristic of highly visual arboreal snakes that feed in part onlizard prey (Henderson and Binder, 1980; Lilly- white and Henderson, 1993). Locals have ob-served island boas capture and consume lizardson Wee Wee Cay (P. Shave, pers. comm.),a Belizean island not surveyed in this study. Although no island boas were found with lizardsin their stomachs,  Norops sagrei   is extremely abundant on all islands and is likely to be animportant prey source, especially for juvenileboas. Two additional squamates ( Aristelliger geor- geensis   and  Phyllodactylus tuberculosus  ) were docu-mented on four of the five islands and couldalso be important components of the island boa diet. Fig. 2. Scatterplot of population centroids ( 6 2 SE) from canonical discriminant functions 1 and 2 forisland and mainland  Boa constrictor  . Representative images of boa heads from each locality are indicated withtheir corresponding population symbol. BOBACK—ISLAND BOA MORPHOMETRICS 265
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks