Biological Journal of the Linnean Society, 2010, ••, ••–••

Home , Boodontinae, Eryx elegans, Grayia (snake), Natricinae, Rhinotyphlops lalandei

Biological Journal of the Linnean Society, 2010, ••, ••–••. With 3 ﬁgures

Snake diets and the deep history hypothesis

TIMOTHY J. COLSTON1*, GABRIEL C. COSTA2 and LAURIE J. VITT1

1Sam Noble Oklahoma Museum of Natural History and Zoology Department, University of Oklahoma, 2401 Chautauqua Avenue, Norman, OK 73072, USA 2Universidade Federal do Rio Grande do Norte, Centro de Biociências, Departamento de Botânica, Ecologia e Zoologia. Campus Universitário – Lagoa Nova 59072-970, Natal, RN, Brasil

Received 3 November 2009; accepted for publication 12 May 2010bij_1502 1..12

The structure of animal communities has long been of interest to ecologists. Two different hypotheses have been proposed to explain origins of ecological differences among species within present-day communities. The competition–predation hypothesis states that species interactions drive the evolution of divergence in resource use and niche characteristics. This hypothesis predicts that ecological traits of coexisting species are independent of phylogeny and result from relatively recent species interactions. The deep history hypothesis suggests that divergences deep in the evolutionary history of organisms resulted in niche preferences that are maintained, for the most part, in species represented in present-day assemblages. Consequently, ecological traits of coexisting species can be predicted based on phylogeny regardless of the community in which individual species presently reside. In the present study, we test the deep history hypothesis along one niche axis, diet, using snakes as our model clade of organisms. Almost 70% of the variation in snake diets is associated with seven major divergences in snake evolutionary history. We discuss these results in the light of relevant morphological, behavioural, and ecological correlates of dietary shifts in snakes. We also discuss the implications of our results with respect to the deep history hypothesis. © 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010.

ADDITIONAL KEYWORDS: canonical phylogenetic ordination – community ecology – niche theory – phylogenetic structure – reptile ecology.

INTRODUCTION researchers have identified ecological shifts within and among major clades of organisms (Melville, Schulte & The structure of animal communities has been of Larson, 2001; Glor et al., 2003; Vitt et al., 2003; Vitt & interest to ecologists for more than half a century and Pianka, 2005), demonstrating that some ecological is central to understanding why there are so many traits have deep historical origins. We now know that species (Andrewartha & Birch, 1954; Hutchinson, much of the structure of some present-day communi- 1959). Ongoing species interactions, primarily compe- ties, with the exception of island communities (Losos tition and predation, dominated explanations during et al., 1998), results from the ability of species with much of the 20th Century (Pianka, 1973; Cody, 1974; deep historical roots to coexist (Vitt, Zani & Espósito, Schoener, 1974; Morin, 1983). The notion that present- 1999; Vitt & Pianka, 2005). day community structure might reflect species’ inter- Two very different hypotheses have been proposed actions in the past was introduced by G. E. Hutchinson to explain the origins of ecological differences among (Hutchinson, 1959) but not fully appreciated until species within present-day communities. The first phylogenetics merged with ecology, allowing research- hypothesis centres on recent effects, as closely-related ers to identify major shifts in ecological traits of entire taxa diverge to partition available resources in clades (Cadle & Greene, 1993; Losos, 1994; Webb et al., response to shifts in resource availability, inter-specific 2002). Using phylogenies to analyze ecological traits, competition or predation (competition–predation hypothesis). According to the competition–predation hypothesis, species interactions drive the evolution of *Corresponding author. E-mail: [email protected] divergence in resource use and niche characteristics

(food, time, and microhabitat) among species in local S1). Representatives of all major clades and subclades assemblages. This hypothesis predicts that ecological were included. Our approach was to compile dietary traits of coexisting species are independent of phylog- data for snake species representing both ecological and eny and that major shifts in niche preference result phylogenetic diversity. For dietary analyses, we iden- from interactions among species within present-day tified 34 discreet prey categories, varying from fish assemblages. This hypothesis has been suggested to eggs to large vertebrates. These comprised: lizards, explain local community structure of Amazonian mammals, anurans, birds, fish, snakes, amphibian snakes (Henderson, Dixon & Soini, 1979). eggs, reptile eggs, bird eggs, crustaceans, gastropods, The second hypothesis (deep history hypothesis) annelids, caecilians, chilopods, salamanders, amphis- suggests that divergences deep in the evolutionary baenians, carrion, tortoises, crocodilians, fish eggs, history of organisms (rather than recent effects) invertebrate eggs, diplopods, coleopterans, neuropter- resulted in sets of ecological traits (or niche prefer- ans, arachnids, unidentifiable hexapods, isopterans, ences) that are maintained for the most part in dermapterans, lepidopterans, dipterans, hemipterans, species represented in present-day assemblages (Vitt orthopterans, hymenopterans, and other arthropods. & Pianka, 2005). The deep history hypothesis posits Many studies contained quantitative data on prey that ecological traits of coexisting species can be pre- items, often identified to species. However, some dicted based on phylogeny regardless of the commu- studies placed prey into broad categories (e.g. frogs, nity in which individual species presently reside. For lizards, birds, and mammals). Some studies provided example, just five divergences in the evolutionary data on the kinds of prey eaten but with no quantita- history of lizards (nonsnake squamates) account for tive measure of relative proportions of each prey 80% of the variation among clades in diets (Vitt & category. Because of this great variation in quality of Pianka, 2005, 2007). Other major ecological traits can data among published papers, we used the presence or be traced to origins deep in the evolutionary history of absence for the analyses. The advantage is that we can squamates (Vitt et al., 2003). include a large number of snake taxa in our analyses. In the present study, we test the deep history The disadvantage is that we lose some dietary resolu- hypothesis along one niche axis, diet, using snakes as tion and, as a result, our analysis provides a conser- our model clade of organisms. Snakes are gape-limited vative estimate of dietary divergence among snake predators, and are best known because of the diversity clades. of vertebrate prey that they consume (Shine, 1991; Greene, 1997). Many snakes, however, feed on invertebrate prey (Webb & Shine, 1993; Webb et al., 2000), PHYLOGENETIC RECONSTRUCTION and a large number of species are dietary specialists. We constructed a phylogenetic hypothesis for the rela- Snakes presumably originated in the Mesozoic (Jiang tionships of the 196 snake species based on several et al., 2007) and occur on all continents except Antarc- recent studies representing a balanced view of snake tica (Greene, 1997). They comprise the largest clade evolutionary history (Heise et al., 1995; Kraus & within squamate reptiles (Serpentes, with over 3100 Brown, 1998; Burbrink, Lawson & Slowinski, 2000; species) and occupy almost every habitat in the world, Vidal et al., 2000, 2007; Wilcox et al., 2002; Lawson including deserts, tropical and temperate forests as et al., 2005; Vidal & Hedges, 2005; Zaher et al., 2009). well as grasslands, freshwater (streams, rivers, lakes), We included studies that used nuclear genes, mito- and the oceans (Shine, 1991; Greene, 1997; Pough, chondrial DNA, and morphological characters to 2001; Vitt & Caldwell, 2008). Historical shifts in snake construct the phylogenetic hypothesis. Parsimony, diets probably resulted in adaptive radiations contrib- maximum likelihood, nonparametric bootstrapping, uting to the diversity of snake species observed in Neighbour-joining, and Bayesian analysis produced present-day snake assemblages. Dietary divergence is highly supported phylogenetic relationships in most often correlated with shifts in morphology (Schluter & instances. Our assumption in the subsequent analy- Grant, 1984), behaviour (Fryer & Iles, 1972), and ses is that this phylogeny accurately represents the ecology (Smith et al., 1978). best available reconstruction of the evolutionary history of snakes.

MATERIAL AND METHODS STATISTICAL ANALYSIS SNAKE DIET DATA We used canonical phylogenetic ordination, CPO Dietary data were collected from available literature (Giannini, 2003), a method derived from canonical for 196 species of snakes, including representatives correspondence analysis (Ter Braak, 1986), to test from all ecological biomes and all six continents that the hypothesis that an association exists between contain snakes (see Supporting Information, Appendix snake evolutionary history and snake diets. CPO is a

© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, ••, ••–•• SNAKE DIETARY SHIFTS 3 constrained ordination method that associates a fossorial whereas most modern snakes are terrestrial dependent matrix (diet in this case) with an indepen- or arboreal), they are often found in the same geo- dent matrix (phylogeny) maximizing the correla- graphic regions (Martins & Oliveira, 1998; França & tion between the two datasets. The significance of Araújo, 2007; França et al., 2008). The next six clades the association is then determined by comparisons contributing to significant dietary divergences are, in with null models generated by Monte Carlo simula- rank order, Leptotyphlopidae, Homalopsinae, Natrici- tions. The goal of the analysis is similar to a Mantel nae, Aparallactinae, Tachymenini, and Pareatinae test, in the sense that a relationship between two (Table 2). Five other clades, Viperidae, Colubrinae, matrices is established. However, CPO detects varia- Dipsadinae, Boidae, and Elapinae, are significant con- tion overlooked by Mantel test and assigns sig- tributors to dietary variation but each explains less nificance to shifts in dependent variables associated than 3% of the total dietary divergence. Advanced with each divergence point in the phylogeny, rather snakes as a group are distributed directly opposite to than producing a significance value across the entire blind snakes across a gradient of prey types varying tree (Giannini, 2003). For detailed information on from vertebrates to arthropods (Fig. 2). It is also the method and some applications please refer to shown that natricines and homalopsines are strongly (Giannini, 2003; Vitt & Pianka, 2005; Werneck, Colli tied to the frog-salamander-fish end of a vertebrate & Vitt, 2009). prey gradient that extends from terrestrial verte- We used presence–absence data in our diet matrix, brates to aquatic vertebrates and gastropods. Tachy- where ‘1’ indicated the presence and ‘0’ indicated the menini and pareatine snakes are weakly associated absence of the prey item in the diet. Because snake with the aquatic end of the vertebrate prey gradient, body size can affect size (and presumably type) of prey whereas aparallactine snakes are weakly associated ingested (Shine & Thomas, 2005), we used maximum with the terrestrial vertebrate portion of the verte- snout–vent length for each snake species as a cova- brate prey gradient. riate in our analysis to minimize the effect of body A plot of snake species scores on the two canonical size. The analysis was performed using CANOCO, axes describing dietary variation reveals just how version 4.53 (Ter Braak & Smilauer, 2002). Monte extreme blind snakes are compared to advanced Carlo permutation tests were performed in stepwise snakes (Fig. 3). Most species of blind snakes feed analysis on each clade using 9999 permutations. Sym- primarily on larvae and pupae of termites and ants, metric scaling and unimodal methods were used, and with species of leptotyphlopids feeding on some other rare prey categories were downweighted. Each clade arthropods as well. Relatively few advanced snakes was tested one at a time manually to obtain F- and have exploited the high prey diversity and abundance P-values. After each clade was tested, significant of arthropods. Within advanced snakes, natricines and clades were included in the model and the subsequent homalopsines have independently converged to feed on clade that most reduced the variance was tested and aquatic vertebrate prey, including frogs, salamanders, included if statistically significant (P < 0.05). and fishes, whereas most remaining advanced snakes feed primarily on terrestrial vertebrates, including birds, mammals, reptiles, and some amphibians. A few RESULTS have specialized on gastropods (Fig. 3). Snakes used in our analysis ate a wide diversity of prey that varied in size from small social insects (e.g. DISCUSSION termites and ants) to large mammals and reptiles. More than 20% of the 181 species of alethinophidian Although the origin of snakes within squamate rep- snakes (referred to as advanced, nonblind snakes) ate tiles remains uncertain, Serpentes is clearly a mono- lizards, mammals, frogs, birds, fish, and snakes, with phyletic clade that arose in the Mesozoic Era, more than half eating lizards (Table 1). Almost 70% of diverging into the two major clades, Scolecophidia the variation in diets among snake clades is associ- (blind snakes) and Alethinophidia (advanced snakes), ated with seven major divergences in snake evolu- probably during the Hauterivian of the early Creta- tionary history (Fig. 1, Table 2). No scolecophidians ceous, approximately 133 Mya (Caldwell & Lee, 1997; (blind snakes) ate vertebrates. All ate insects and Caprette et al., 2004; Burbrink & Pyron, 2008). More their eggs, with some eating spiders, centipedes, and than 500 species of blind snakes are known, compris- millipedes. ing three ecologically similar families (Greene, 1997). The most ancient divergence in snake evolutionary By contrast, advanced snakes have diversified into history, the blind snake/advanced snake divergence, more than 2300 species (approximately 80% of all accounted for almost 25% of the dietary divergence. extant snakes) in thirteen or more families, with Although blind snakes and advanced snakes typically species occupying almost all imaginable microhabi- separate in spatial niche space (all blind snakes are tats in temperate and tropical regions of the world

Table 1. Frequency of snake species within each major clade feeding on prey types used in this analysis

Alethinophidians (N = 181) Scolecophidians (N = 14)

Number Percent Number Percent of species of clade of species of clade

Lizards 100 55.25 0 0 Mammals 83 45.86 0 0 Anurans 72 39.78 0 0 Birds 48 26.52 0 0 Fish 46 25.41 0 0 Snakes 37 20.44 0 0 Other Arthropods 27 14.92 14 100 Amph. eggs 14 7.74 0 0 Reptile eggs 12 6.63 0 0 Bird eggs 12 6.63 0 0 Crustaceans 9 4.97 0 0 Gastropods 8 4.42 0 0 Annelids 8 4.42 0 0 Caecilians 6 3.32 0 0 Chilopods 6 3.32 2 14.29 Salamanders 4 2.22 0 0 Amphisbaenians 4 2.21 0 0 Carrion 3 1.66 0 0 Tortoises 2 1.11 0 0 Crocodilians 2 1.10 0 0 Fish eggs 2 1.11 0 0 Invert. eggs 2 1.11 13 92.86 Diplopods 1 0.55 2 14.29 Coleopterans 1 0.55 1 7.14 Neuropterans 1 0.55 2 14.29 Arachnids 0 0.00 4 28.57 Unidentiﬁable Hexapods 0 0.00 12 85.71 Isopterans 0 0.00 8 57.14 Dermapterans 0 0.00 2 14.29 Lepidopterans 0 0.00 2 14.29 Dipterans 0 0.00 2 14.29 Hemipterans 0 0.00 2 14.29 Orthopterans 0 0.00 2 14.29 Hymenopterans 0 0.00 14 100

Percent of species is the percent with respect to the total number of species in each clade. The number of snake species represented in each clade is indicated in parentheses. Prey types are ranked by their frequency of use in alethinophidians.

(Shine, 1991; Greene, 1997). We ﬁrst discuss dietary squamate reptiles typically referred to as ‘lizards’ divergences detected in our analysis. We then discuss (Vitt et al., 2003). Blind snakes eat small inverte- the relevant morphological, behavioural, and ecologi- brates, most of which are social insects and their cal correlates of dietary shifts in snakes. Finally, larvae (Webb & Shine, 1993; Webb et al., 2000). These we discuss our results in light of the deep history snakes inhabit nests of social insects, usually but hypothesis. not always underground, and occasionally eat other arthropods, probably within social insect nests. They do not eat a variety of other invertebrates that are DIETARY DIVERGENCES common underground but usually not associated with Our analysis of snake diets indicates that signiﬁcant social insect nests (e.g. earthworms, cicada larvae, dietary shifts occurred deep within snake evolution- and scarabeid beetle larvae). By contrast, advanced ary history, a pattern similar to that described for snakes eat a wider range of prey types varying from

Uromacer frenatus Antillophis parvifrons Erythrolamprus aesculapii Liophis breviceps Xenodon rhabdocephalus Xenodopholis scalaris Xenoxybelis argenteus Hydrodynastes gigas Helicops angulatus Helicops hagmanni Helicops leopardinus Philodryas olfersii Philodryas patagoniensis Phylodryas viridissimus Clelia clelia Drepanoides anomalus Pseudoboa coronata Taeniophallus brevirostris TachymeniniTachymenini Taeniophallus nicagus Oxyrhopus formosus Tripanurgos compressus 6 Tomodon dorsatus Imantodes cenchoa Dipsas indica Atractus alphonsehogei Atractus latifrons Thamnophis couchii Thamnophis elegans Thamnophis marcianus Thanmophis melanogaster Thamnophis sirtalis Storeria occipitomaculata Storeria dekayi Regina grahamii Nerodia clarkii Nerodia cyclopion Nerodia erythrogaster Nerodia fasciata Nerodia harteri Nerodia rhombifer Nerodia sipedon Natrix natrix 3 Natrix maura Natrix tesselata Afronatrix anoscopus Phyllorhynchus decurtatus Phyllorhynchus browni Spilotes pullatus Tantilla gracilis Tantilla melanocephala NatricinaeNatricinae Oxybelis aeneus Oxybelis fulgidus Oligodon formosanus Chironius fuscus Chironius multiventris Pseustes poecilonotus Leptophis ahaetulla Boiga irregularis Boiga blandingi Dasypeltis scabra Lycodon aulicus Symphimus mayae Dendrophidion dendrophis Mastigodryas boddaerti Drymoluber dichrous Gyalopion quadrangulare Rhinobothryum lentiginosum Arizona elegans Rhinocheilus lecontei Pantherophis obsoleta Pituouphis catenifer Coronella austriaca Zamenis lineatus Elaphe quatorlineata Coluber hippocrepis Grayia smythii Tropidechis carinatus Notechis scutatus Hoplocephalus bungaroides Aipysurus eydouxii Emydocephalus annulatus Lapemis curtis Oxyuranus microlepidotus Oxyuranus scutellatus Vermicella annulata Unechis dwyeri Unechis spectbilis Unechis nigrostriatus Unechis nigriceps Unechis monachus Unechis gouldii Unechis flagellum Unechis carpentariae Suta suta Laticauda crockeri Laticauda colubrina Micrurus averyi Micrurus surinamensis Micrurus spixii Micrurus lemniscatus Micrurus hemprichii Naja nigrircollis Naja nigricincta Naja mossambica Naja anchietae Naja annulifera Naja nivea Naja melanoleuca Dendroaspis jamesoni Hemachatus haemachatus Aspidelaps lubricus Psammophis phillipsi Psammophis trinasalis AparallactinaeAparallactinae Psammophis mossambicus Psammophis notostictus Psammophis namibensis Psammophis jallae Psammophis breviostris Mehelya capensis Atractaspis aterrima 5 Polemon acanthias Enhydris bocourti Enhydris sieboldii Enhydris innominata Enhydris polylepis Enhydris plumbea Enhydris enhydris Enhydris doriae Enhydris chinensis Myron richardsonii HomalopsinaeHomalopsinae Gerarda prevostiana Fordonia leucobalia Cerberus rynchops Cantoria violacea Bitia hydroides 4 Homalopsis buccata Crotalus horridus Crotalus lepidus Crotalus o. concolor Sistrurus catenatus Agkistrodon piscivorus Lachesis muta Bothrops moojeni Bothrops atrox Gloydius shedaoensis Calloselasma rhodostoma Atheris squamiger Echis coloratus Bitis nasicornis PareatinaePareatinae Bitis gabonica Vipera ammodytes Vipera ursinii 7 Pareas carinatus Acrochordus granulatus Eryx colubrinus Eryx tataricus Eryx johnii Eryx jayakari Eryx jaculus Eryx elegans Eryx conicus Eryx miliaris Charina bottae Charina trivirgata Candoia aspera Candoia carinata Candoia bibroni Corallus enydris Corallus hortulanus Corallus caninus Epicates cenchria Eunectes murinus Boa constrictor Calabaria reinhardti Python sebae Python regius Morelia spilota Loocemus bicolor Xenopeltis unicolor Cylindrophis rufus Anilius scytale Tropidophis melanurus Ramphotyphlops australis Ramphotyphlops pinguis Ramphotyphlops nigrescens AlethinophidiaAlethinophidia Ramphotyphlops bituberculatus Rhinotyphlops lalandei 1 Rhinotyphlops s. petersii Rhinotyphlops mucruso Typhlops bibronii LeptotyphlopidaeLeptotyphlopidae Typhlops fornasinii Typhlophis squamosus Leptotyphlops humilis 2 Leptotyphlops dulcis ScolecophidiaScolecophidia Leptotyphlops scutifrons

Figure 1. Phylogenetic hypothesis for 196 snake species used in the present study. Numbers indicate the seven clades that explain most of the variance in the diet matrix (Table 2).

Table 2. Results of a phylogenetic ordination analysis based on a canonical correspondence analysis for diets of 196 snake species representing all major clades in all major biomes of the world

Clade Variation Variation% FP

Scolecophiida/Alethinophidia 0.74 24.75 21.26 < 0.01 Leptotyphlopidae 0.32 10.73 9.61 < 0.01 Homalopsinae 0.29 9.73 9.52 < 0.01 Natricinae 0.29 9.73 9.10 < 0.01 Aparallactinae 0.16 5.43 5.44 0.02 Tachymenini 0.13 4.16 4.30 0.03 Pareatinae 0.12 4.13 4.20 0.04 Aniliidae 0.09 2.86 2.99 0.08 Atractaspididae 0.08 2.73 2.90 0.10 Viperidae 0.07 2.40 2.54 0.01 Pythoninae 0.07 2.30 2.47 0.06 Colubrinae 0.07 2.20 2.39 < 0.01 Dipsadinae 0.07 2.20 2.39 0.03 Boidae 0.06 1.97 2.16 0.01 Boodontinae 0.05 1.80 1.99 0.09 Elapinae 0.05 1.60 1.78 0.04 Elapidae 0.04 1.33 1.50 0.10 Colubridae 0.04 1.33 1.50 0.16 Psammophiinae 0.04 1.27 1.40 0.17 Cylindrophiidae 0.04 1.23 1.37 0.16 Laticaudaudinae 0.04 1.20 1.34 0.23 Anomalepididae 0.03 1.07 1.18 0.19 Crotalinae 0.03 1.00 1.11 0.31 Caenophidia 0.03 0.87 0.96 0.40 Henophidia 0.02 0.73 0.82 0.30 Boinae 0.02 0.53 0.58 0.84 Xenopeltidae 0.01 0.33 0.37 0.77 Xenodontinae 0.01 0.23 0.27 0.80

Clades are ranked by the amount of variation explained at each node. Although signiﬁcance drops off after the seventh clade, a few clades lower in the ranking attain signiﬁcance. However, the portion of total variance explained by each of these is minimal (less than 2.5% for each one).

invertebrates to large vertebrates. None specialize on (blind snakes) to large prey types (advanced snakes) adults, larvae or pupae of social insects. Advanced is associated with a wide range of changes in mor- snakes that do specialize on invertebrates (small prey phology. Blind snakes have relatively rigid nonki- relative to head size), such as the Aparallactinae, netic skulls, whereas advanced snakes have toothed tend to specialize on invertebrates such as centipedes, palatopterygoid jaw arches and a highly kinetic lower although some species also eat amphisbaenians and jaw (Kley, 2006; Vincent et al., 2006). Rigid nonkinetic earthworms (Gower & Rasmussen, 2004). The Apar- skulls of blind snakes limit the size of prey that can allactinae is located closest to the origin in Figure 2, be consumed. The highly kinetic skull of advanced suggesting that its dietary preference is not strongly snakes, along with independence of lower jaws result- associated with a particular group of prey. ing from a lack of an anterior symphysis, allows them to feed on much larger prey (relative to head size) using a ratcheting motion of the lower jaws, effec- MORPHOLOGICAL CORRELATES OF tively pulling the prey down the snake’s throat. Alter- DIETARY DIVERGENCE natively, recent studies have suggested that the The present study shows that dietary divergence morphology of blind snakes might comprise a highly within a clade may be an evolutionary ﬁrst response derived specialization for fossoriality and/or a diet to morphological adaptations. The transition from based on ants and termites (Kley, 2006; Rieppel, Kley small (relative to body size) invertebrate prey types & Maisano, 2009).

Gastropods Crustaceans Annelids Prey category Homalopsinae Amph. Eggs Clades Natricinae

Scolecophidia Fish Alethinophidia

Salamanders Aquatic Prey Natricinae Homalopsinae Hymenopterans Trachymeninae Isopterans Pareatinae Hexapods Anurans Inverts (unid) Alethinophidia Arthropods (unid) CCA Axis 2 Scolecophidia

Aparallactinae Crocodilians Coleopterans Arachnids (unid)

CCA Asix 2 Tortoises Chilopods Leptotyphlopidae Aves Neuropterans Snakes Lizards Diplopods Mammals Lepidopterans Bird Eggs Hemipterans Arthropodprey Carrion Dermapterans (excluding aquatics) Caecilians Amphisbaenians Orthopterans Fish Eggs Dipterans Reptiles (misc.) -1.5 2.5 -1.0CCA Axis 1 3.0 Vertebrate Prey Figure 2. Biplot from a phylogenetic ordination based on canonical correspondence analysis (CCA) relating snake phylogeny (arrows) to snake diets (triangles). Canonical axes represent linear combinations of diet to snake phy- -2.0 4.0 logeny. Arrows represent correlations of phylogeny with -1.5CCA Axis 1 2.5 axes and arrow length represents the strength of the relationship. Snake diet items are weighted averages of Figure 3. Bipolot showing blind snakes (red circles) and each species scores. The first canonical axis accounts for advanced snakes (all other symbols) in the first two 32.5% and the second canonical axis accounts for 31.3% of canonical correspondence axes of dietary niche space with total variation. Diet items (triangles) close to the graph advanced snakes broken down to show that all natricines centre (0, 0) indicate either low association with any snake and homalopsines (two clades of advanced snakes) have clade (arrows) or a positive association with a specific taken advantage of aquatic prey with all other advanced combination of all snake clades. Diet items displayed in snakes restricted primarily to other vertebrate prey types. the periphery of the graph indicate either high association CCA, canonical correspondence analysis. with a specific snake clade or an occasional association, particularly for those clades with low occurrence (Clark, 2002, 2004). Other advanced snakes such (Ter Braak & Smilauer, 2002). as racers (e.g. Coluber, Masticophis, and Chironius), which employ an active foraging mode, use vision to locate prey and chemical cues to discriminate prey BEHAVIOURAL CORRELATES OF DIETARY DIVERGENCE (Greene, 1997; Mullin & Cooper, 1998). Blind snakes An obvious behavioural consequence of dietary diver- have reduced eyes without the ability to simply locate gence comes from the fact that, within advanced prey visually (Greene, 1997) and lack infra-red snakes, several different modes of prey detection and sensory systems (Greene, 1997). capture have evolved, including possibly multiple origins of venom delivery systems (Fry et al., 2008). Chemosensory systems have been shown to evolve to ECOLOGICAL CORRELATES OF DIETARY DIVERGENCE detect specific prey types and are most effective at The most obvious ecological correlate of the dietary detecting current prey (Cooper, 2008). Even though shift from subterranean insects to largely terrestrial all snake species utilize chemosensory systems to and aquatic vertebrates is the shift from a fossorial find prey (Halpern, 1992), the diversity of prey eaten lifestyle in blind snakes compared to the much more by advanced snakes suggests that chemosensory diverse range of lifestyles found within advanced systems have likely diverged in response to historical snakes. Blind snakes occupy a relatively narrow niche dietary shifts. Some advanced snakes, such as pit- globally, with all species living within subterranean vipers (e.g. Crotalus, Bothrops, and Trimeresurus), (and occasionally, arboreal) social insect nests. The which employ a sit-and-wait foraging mode, use results obtained in the present study clearly reveal chemosensory cues to detect ambush sites and then that snakes in the Scolecophidia almost exclusively use infra-red sensory systems to detect passing prey exploit a resource not used by advanced snakes

(Fig. 3). Most snakes in the Alethinophidia are ter- overdispersion). The interplay of these two factors restrial, arboreal, aquatic or marine, although some will determine the relative contribution of the two have independently evolved subterranean lifestyles. processes on shaping community structure (Webb Among the latter, some species live in sand (e.g. et al., 2002; Cooper, Rodriguez & Purvis, 2008). For Chionactis and Chilomeniscus) or under leaf litter example, in groups that show niche conservatism if and within humus (e.g. Atractus) but none live within the local community is assembled as a subset of social insect nests. As a consequence of this diverse closely-related species (phylogenetic clustering), this use of microhabitats, diets of advanced snakes are not would be evidence for habitat filtering playing a role only diverse (Fig. 3) but have diverged repeatedly in structuring the local assemblage, whereas, if the during the evolutionary history of snakes. Another local community is assembled by a subset of distant evident ecological correlate of diet occurs in aquatic related species, this would be evidence that competi- lineages, especially Natricinae and Homalopsinae, tion played a role in structuring the local assemblage both of which feed mainly on aquatic vertebrates (Webb et al., 2002). (frogs, salamanders, and fishes). These lineages have A recent study showed that phylogenetic overdis- independently specialized on aquatic prey: the persion is a common tendency for mammalian com- Natricinae primarily in freshwater habitats in the munities (Cooper et al., 2008). However, that same New World, and the Homalopsinae primarily in fresh- study did not assess whether any ecological trait of water habitats the Old World, including Australia. coexisting species was related to the phylogeny (e.g. phylogenetic overdispersion would not be an evidence of competition if a given niche characteristic were not LOCAL COMMUNITY STRUCTURE AND THE related to phylogeny). Overall, phylogenetic structure DEEP HISTORY HYPOTHESIS and ecological interactions may both act to determine The results of the present study demonstrate that community structure (Kraft, Valencia & Ackerly, major dietary shifts occurred early in snake evolu- 2008; Werneck et al., 2009). The challenge remains in tionary history, therefore supporting the deep history identifying the relative contribution of these two hypothesis. However, these results do not rule out the major forces in different organisms, regions, and spe- potential role of competition and predation in the cific traits (Kembel, 2009). Our analysis reveals that structure of local communities. Examination of local snake diets are associated with ancient events in the communities alone has little exploratory power in evolutionary history of the group. Future studies ecology and evolutionary biology (Ricklefs, 2008). The should focus on whether or not local snake commu- field of phylogenetic community ecology provides a nities around the world are composed of random framework to study the relative contribution of com- subsets (i.e. not clustered or overdispersed) of the petition and habitat filtering in community structure regional species pool. (Webb et al., 2002; Chazdon et al., 2003; Anderson, Lachance & Starmer, 2004). A necessary first step is to examine the distribution of ecological characters CAVEAT among species on an independently derived phylog- Similar to recent studies on nonsnake squamates eny. Three possible scenarios exist: (1) ecological char- (lizards), the present analysis shows that a major acters evolved randomly in the group (i.e Brownian ecological trait of snakes, the food they eat, has shifted motion), in this case sister species will tend to share during the evolutionary history of the clade, and that some degree of ecological similarity (phylogenetic many present-day species simply eat what their ances- inertia); (2) ecological characters will tend to be more tors ate, regardless of where they now live. Unlike similar than expected under a random walk model studies on lizards, in which quantitative and compa- (niche conservatism), in this case sister species are rable data were available on relative proportions of very similar to each other ecologically; and (3) eco- prey categories in the diets of many species (Vitt & logical characters are more different than expected by Pianka, 2005, 2007), the present study was con- a random walk model, in this case sister species will strained to presence–absence data in constructing our have very different ecological strategies (niche evolu- prey matrix for analyses. Effectively, this renders a tion). With this information in hand, it is possible to specialist that might occasionally eat something dif- explore how the local community composition reflects ferent as a generalist, relatively speaking. As a conse- the regional species pool (Webb et al., 2002). Again, quence, our analysis misses many potential dietary Three possibilities exist: (1) the local community is a shifts that would be detected with quantitative data random subset of the regional species pool; (2) the and is thus conservative. In addition, reconstruction of local community is a subset of closely related species the evolutionary relationships of squamates, including (phylogenetic clustering); and (3) the local community snakes, is in a state of flux, as new techniques, is a subset of distant related species (phylogenetic additional genes, better analyses, and better taxon

© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, ••, ••–•• SNAKE DIETARY SHIFTS 9 sampling are applied. The phylogeny used in the Cooper N, Rodriguez J, Purvis A. 2008. A common ten- present study was based on current available pub- dency for phylogenetic overdispersion in mammalian assem- lished analyses and is likely to change as more data blages. Proceedings of the Royal Society of London Series B, and better analyses are used. Nevertheless, we are Biological Sciences 275: 2031–2037. confident that the major dietary shifts identified in the Cooper WE Jr. 2008. Tandem evolution of diet and present study will hold. chemosensory responses in snakes. Amphibia-Reptilia 29: 393–398. França FGR, Araújo AFB. 2007. Are there co-occurence ACKNOWLEDGEMENTS patterns that structure snake communities in central Brazil? Brazilian Journal of Biology 67: 33–40. We thank the University of Oklahoma Zoology França FGR, Mesquita DO, Nogueira CC, Araújo AFB. Department and the Sam Noble Oklahoma Museum 2008. Phylogeny and ecology determine morphological of Natural History for financial support. We gra- structure in a snake assemblage in the central Brazilian ciously thank the many snake natural historians who Cerrado. Copeia 23–38. collected diet data used in our analyses. G.C.C. was Fry BG, Scheib H, van der Weerd L, Young B, McNaugh- supported by a Fulbright/CAPES PhD fellowship tan J, Ramjan SFR, Vidal N, Poelmann RE, Norman (#15053155-2018/04-7). JA. 2008. Evolution of an arsenal: structural and functional diversification of the venom system in the advanced snakes REFERENCES (Caenophidia). Molecular and Cellular Proteomics 7: 215– 246. Anderson TM, Lachance MA, Starmer WT. 2004. Fryer G, Iles TD. 1972. The cichlid fishes of the great lakes The relationship of phylogeny to community structure: of Africa. Their biology and evolution. Edinburgh: Oliver the cactus yeast community. American Naturalist 164: 709– Boyd. 721. Giannini NP. 2003. Canonical phylogenetic ordination. Andrewartha HG, Birch LC. 1954. The distribution and Systematic Biology 52: 684–695. abundance of animals. Chicago, IL: University of Chicago Glor RE, Kolbe JJ, Powell R, Larson A, Losos JB. 2003. Press. Phylogenetic analysis of ecological and morphological diver- Burbrink FT, Lawson R, Slowinski JB. 2000. Mitochon- sification in hispaniolan trunk-ground anoles (Anolis cybotes drial DNA phylogeography of the polytypic North American group). Evolution 57: 2283–2397. rat snake (Elaphe obsoleta): a critique of the subspecies Gower DJ, Rasmussen JB. 2004. The caecilian amphibian concept. Evolution 54: 2107–2118. Scolecomorphus kirkii Boulenger as prey of the burrowing Burbrink FT, Pyron RA. 2008. The taming of the skew: asp Atractaspis aterrima Gunther: trophic relationships of estimating proper confidence intervals for divergence dates. fossorial vertebrates. African Journal of Ecology 42: 83– Systematic Biology 57: 317–328. 87. Cadle JE, Greene HW. 1993. Phylogenetic patterns, bioge- Greene HW. 1997. Snakes: the evolution of mystery in nature. ography, and the ecological structure of Neotropical snake Berkley, CA: University of California Press. assemblages. In: Ricklefs RE, Schluter D, eds. Species diver- Halpern M. 1992. Nasal chemical senses in reptiles: sity in ecological communities: historical and geographical structure and function. In: Gans C, Crews D, eds. Biology perspectives. Chicago, IL: University of Chicago Press, 281– of Reptilia. Chicago, IL: University of Chicago Press. 293. Heise PJ, Maxson LR, Dowling HG, Hedges SB. 1995. Caldwell MW, Lee MSY. 1997. A snake with legs from the Higher-level snake phylogeny inferred from mitochondrial marine Cretaceous of the Middle East. Nature 386: 705– DNA sequences of 12S rRNA and 16s rRNA genes. Molecu- 709. lar Biology and Evolution 12: 259–265. Caprette CL, Lee MSY, Shine R, Mokany A, Downhower Henderson RW, Dixon J, Soini P. 1979. Resource parti- JF. 2004. The origin of snakes (Serpentes) as seen through tioning in Amazonian snake communities. Milwaukee Public eye anatomy. Biological Journal of the Linnean Society 81: Museum. Contributions in Biology and Geology 22: 1–11. 469–482. Hutchinson GE. 1959. Homage to Santa Rosalia or why are Chazdon RL, Careaga S, Webb C, Vargas O. 2003. Com- there so many kinds of animals? American Naturalist 93: munity and phylogenetic structure of reproductive traits of 145–159. woody species in wet tropical forests. Ecological Mono- Jiang ZJ, Castoe TA, Austin CC, Burbrink FT, Herron graphs 73: 331–348. MD, McGuire JA, Parkinson CL, Pollock DD. 2007. Clark RW. 2002. Diet of the Timber Rattlesnake, Crotalus Comparative mitochondrial genomics of snakes: extaordi- horridus. Journal of Herpetology 36: 494–499. nary substitution rate dynamics and functionality of the Clark RW. 2004. Timber rattlesnakes (Crotalus horridus) use duplicate control region. BMC Evolutionary Biology 7: 1–14. chemical cues to select ambush sites. Journal of Chemical Kembel SW. 2009. Disentangling niche and neutral influ- Ecology 30: 607–617. ences on community assembly: assessing the performance of Cody ML. 1974. Competition and the structure of bird com- community phylogenetic structure tests. Ecology Letters 12: munities. Princeton, NJ: Princeton University Press. 949–960.

Kley NJ. 2006. Morphology of the lower jaw and suspen- Smith JNM, Grant PR, Grant BR, Abbott IJ, Abbott LK. sorium in the Texas blindsnake, Leptotyphlops dulcis 1978. Seasonal variation in feeding habits of Darwin’s (Scolecophidia: Leptotyphlopidae). Journal of Morphology ground finches. Ecology 59: 1137–1150. 267: 494–515. Ter Braak CJF. 1986. Canonical correspondence analysis: a Kraft NJB, Valencia R, Ackerly DD. 2008. Functional new eigenvector technique for multivariate direct gradient traits and niche-based tree community assembly in an ama- analysis. Ecology 67: 1167–1179. zonian forest. Science 322: 580–582. Ter Braak CJF, Smilauer P. 2002. Canoco reference manual Kraus F, Brown WM. 1998. Phylogenetic relationships of and canodraw for Window’s user guide: Software for cano- colubroid snakes based on mitochondrial DNA sequences. nical community ordination, Version 4.5. Ithaca, NY: Zoological Journal of the Linnean Society 122: 455– Microcomputer Power. 487. Vidal N, Hedges SB. 2005. The phylogeny of squamate Lawson R, Slowinski JB, Crother BI, Burbrink FT. 2005. reptiles (lizards, snakes, and amphisbaenians) inferred from Phylogeny of the Colubroidea (Serpentes): new evidence nine nuclear protein-coding genes. Comptes Rendus Biolo- from mitochondrial and nuclear genes. Molecular Phyloge- gies 328: 1000–1008. netics and Evolution 37: 581–601. Vidal N, Kindl SG, Wong A, Hedges SB. 2000. Phylogenetic Losos JB. 1994. Historical contingency and lizard community relationships of xenodontine snakes inferred from 12S and ecology. In: Vitt LJ, Pianka ER, eds. Lizard ecology: histori- 16S ribosomal RNA sequences. Molecular Phylogenetics and cal and experimental perspectives. Princeton, NJ: Princeton Evolution 14: 389–402. University Press, 319–333. Vidal N, Delmas AS, David P, Cruaud C, Coujoux A, Losos JB, Jackman TR, Larson A, de Queiroz K, Hedges SB. 2007. The phylogeny and classification of cae- Rodriguez-Schettino L. 1998. Contingency and determin- nophidian snakes inferred from seven nuclear protein- ism in replicated adaptive radiations of island lizards. coding genes. Comptes Rendus Biologies 330: 182–187. Science 279: 2115–2118. Vincent SE, Dang PD, Herrel A, Kley NJ. 2006. Morpho- Martins M, Oliveira ME. 1998. Natural history of snakes in logical integration and adaptation in the snake feeding forests of the Manaus region, Central Amazonia, Brazil. system: a comparative phylogenetic study. Journal of Evo- Herpetological Natural History 6: 78–150. lutionary Biology 19: 1545–1554. Melville J, Schulte JA, Larson A. 2001. A molecular phy- Vitt LJ, Caldwell JP. 2008. Herpetology: an introductory logenetic study of ecological diversification in the Australian biology of amphibians and reptiles. San Diego, CA: Aca- lizard genus. Ctenophorus. Journal of Experimental Zoology demic Press. 291: 339–353. Vitt LJ, Pianka ER. 2005. Deep history impacts present day Morin PJ. 1983. Predation, competition, and the composition ecology and biodiversity. Proceedings of the National of larval anuran guilds. Ecological Monographs 53: 119– Academy of Science of the United States of America 102: 138. 7877–7881. Mullin SJ, Cooper RJ. 1998. The foraging ecology of Vitt LJ, Pianka ER. 2007. Feeding ecology in the natural the gray rat snake (Elaphe obsolete spiloides) – visual world. In: Reilly SM, McBrayer LD, Miles DB, eds. Lizard stimuli facilitate location of arboreal prey. American ecology. Cambridge, NY: Cambridge University Press, 141– Midland Naturalist 140: 397–401. 172. Pianka ER. 1973. The structure of lizard communities. Vitt LJ, Zani PA, Espósito MC. 1999. Historical ecology of Annual Review of Ecology and Systematics 4: 53–74. Amazonian lizards: implications for community ecology. Pough FH. 2001. Herpetology. Upper Saddle River, NJ: Oikos 87: 286–294. Prentice Hall. Vitt LJ, Pianka ER, Cooper WE Jr, Schwenk K. 2003. Ricklefs RE. 2008. Disintegration of the Ecological History and the global ecology of squamate reptiles. Ameri- Community. American Naturalist 172: 741–750. can Naturalist 162: 44–60. Rieppel O, Kley NJ, Maisano JA. 2009. Morphology of the Webb CO, Ackerly DD, McPeek MA, Donoghue MJ. 2002. skull of the white-nosed blindsnake, Liotyphlops albirostris Phylogenies and community ecology. Annual Review of (Scolecophidia: Anomalepididae). Journal of Morphology Ecology and Systematics 33: 475–505. 270: 536–557. Webb JK, Shine R. 1993. Dietary habits of Australian blind- Schluter D, Grant PR. 1984. Determinants of morpholo- snakes. Copeia 762–770. gical patterns in communities of Darwin finches. American Webb JK, Shine R, Branch WR, Harlow PS. 2000. Life- Naturalist 123: 175–196. history strategies in basal snakes: reproduction and dietary Schoener TW. 1974. Resource partitioning in ecological habits of the African thread snake Leptotyphlops scutifrons communities. Science 185: 27–39. (Serpentes: Leptotyphlopidae). Journal of Zoology 250: 321– Shine R. 1991. Australian snakes: a natural history. Ithaca, 327. NY: Cornell University Press. Werneck FP, Colli GR, Vitt LJ. 2009. Determinants of Shine R, Thomas J. 2005. Do lizards and snakes really assemblage structure in Neotropical dry forest lizards. differ in their ability to take large prey? A study of relative Austral Ecology 34: 97–115. prey mass and feeding tactics in lizards. Oecologia 144: Wilcox TP, Zwickl DJ, Heath TA, Hillis DM. 2002. 492–498. Phylogenetic relationships of dwarf boas and a comparison

of Bayesian and bootstrap measures of phylogenetic advanced snakes (Serpentes, Caenophidia) with an empha- support. Molecular Pylogenetics and Evolution 25: 361–371. sis on South American Xenodontines: a revised classiﬁcation Zaher H, Grazziotin FG, Cadle JE, Murphy RW, Moura- and descriptions of new taxa. Papéis Avulsos de Zoologia 49: Leite JC, Bonatto SL. 2009. Molecular phylogeny of 115–153.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1. References used for diet data. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.