Blackwell Science, LtdOxford, UKBIJBiological Journal of the Linnean Society0024-4066The Linnean Society of London, 2003? 2003 80? 110 Original Article SEA SNAKE HABITAT USE R. SHINE Biological Journal of the Linnean Society, 2003, 80, 1–10. With 4 figures Intraspecific habitat partitioning by the sea snake Emydocephalus annulatus (Serpentes, Hydrophiidae): the effects of sex, body size, and colour pattern RICHARD SHINE*, TERRI SHINE and BENJAMIN SHINE Biological Sciences A08, University of Sydney, NSW 2006 Australia Received 7 October 2002; accepted for publication 29 January 2003 An individual’s sex, body size and colour pattern can influence its habitat use, and such partitioning can have impor- tant ecological and evolutionary consequences. We studied a system very different from those that have attracted previous research on this topic: sea snakes (Emydocephalus annulatus) in shallow-water coral-reef areas of New Caledonia. The snakes used habitats non-randomly in terms of substrate types and water depths, with frequent use of coral-rubble areas that also contained the nests of fish (damselfish and blennies) whose eggs are eaten by these snakes. Mate-searching adult male snakes were found across a broader range of habitat types than were foraging females and juveniles. Smaller snakes were found in shallower water. Colour polymorphism was evident (melanism increased with body size, and was more common in males than in females) but did not affect habitat use. The effects of colour morph on operative temperatures of physical models (evident in terrestrial situations) disappeared under water. Habitat use in this population is affected by a snake’s body size and sex, but not by colour. Studies of terres- trial snakes have emphasized thermal or camouflage benefits of colour polymorphism, but the superficially similar polymorphism in E. annulatus is not consistent with either of these hypotheses and thus challenges their generality. Similarly, there was no dietary difference between age or sex groups and thus dietary partitioning cannot explain the observed intraspecific habitat partitioning. © 2003 The Linnean Society of London, Biological Journal of the Lin- nean Society, 2003, 80, 1–10. ADDITIONAL KEYWORDS: dimorphism – Hydrophiidae – marine – niche – polymorphism. INTRODUCTION behaviour such that individuals of different pheno- types use available habitats differently (Kirkton & Frequently, a single population contains individuals Schultz, 2001; Telleria et al., 2001; Rolando, 2002; that encompass a wide range of phenotypes in terms Rossi et al., 2002). Thus, an animal’s phenotype can of factors such as sex, body size, colour, shape, loco- affect its probability of survival not only by direct pro- motor abilities, dietary specialization, antipredator cesses, but also by modifying the times and places of responses, and so forth (e.g. Brodie, 1989; Torre & fitness-relevant interactions. For example, an organ- Bosch, 1999; Kirkton & Schultz, 2001). Many environ- ism that matches its habitat selection to its phenotype ments are similarly heterogeneous, so that a given may thereby modify its exposure to food, predators phenotype confers different fitness in one habitat type and/or conspecifics. Among the most important pheno- than in another; for example, a particular colour or typic traits in this respect are: shape may provide effective camouflage against only 1 Body size – small animals may eat different types some kinds of background. In such situations, we and sizes of prey, be vulnerable to a wider range of expect natural selection to modify habitat-selection predators, and (because of higher surface area-to- volume ratio) heat, cool and exchange water with the surrounding environment more rapidly than do larger animals (Peters, 1983; Hatase et al., 2002). Small indi- *Corresponding author. E-mail: [email protected] viduals may fit more easily into small crevices inac- © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 1–10 1 2 R. SHINE ET AL. cessible to larger conspecifics, but social aggression MATERIAL AND METHODS may restrict smaller animals to suboptimal habitats (Law & Bradley, 1990; Golden, Smith & Rettig, 2001). STUDY SPECIES 2 Sex – males and females differ in many attributes Turtle-headed sea snakes, Emydocephalus annulatus, that can influence habitat use (Ruckstuhl & Neuhaus, are medium-sized (up to 1 m total length: Ineich & 2001; Mcdonald, 2002). Most obviously, they differ in Laboute, 2002) hydrophiid snakes, widely distributed behaviour, with males sometimes moving over much in shallow-water reef habitats across the Pacific larger areas than do females in the course of territo- (Cogger, 1975; Ineich & Laboute, 2002). Phylogenetic rial defence or mate-searching. Such movements are analyses suggest that Emydocephalus is close to the likely to influence the amounts of time spent in differ- base of the hydrophiid radiation (Rasmussen, 2002). ent habitat types (Andersson, 1994; Sims, Nash & Like all hydrophiids, Emydocephalus is viviparous, Morritt, 2001). Males and females also differ in body spends its entire life in the water, and is virtually inca- size in many taxa, and not infrequently in food habits, pable of terrestrial locomotion (Shine et al., 2002a). further favouring habitat subdivision within the pop- Emydocephalus annulatus feeds exclusively on the ulation (Mushinsky, Hebrard & Vodopich, 1982; Mush- eggs of small fish (Voris, 1966; McCosker, 1975; insky, 1987; Luiselli & Angelici, 1998; Bowyer et al., McCarthy, 1987), scraping the eggs from the substrate 2002; Lesage et al., 2002). with enlarged supralabial scales (Guinea, 1996). In 3 Colour – in a population with multiple colour mor- New Caledonia, these snakes consume the eggs of phs, individuals may select areas that provide effec- damselfish, blennies and gobies (Ineich & Laboute, tive camouflage (Endler, 1983; King, 1993a,b; Shine 2002). Our fieldwork in June–July 2002 coincided with et al., 1998), or facilitate the predator-escape behav- intense courtship activity by E. annulatus, and hence iours most effective for an animal of their phenotype with cessation of feeding by adult males. Juveniles (Brodie, 1989). Because colour can also modify rates and adult females often (15 of 27 such animals) regur- of thermal exchange, an animal’s colour can influence gitated prey items when handled, but no prey items the times and places that it is active (Peterson, Gibson were regurgitated by 38 adult males (R. Shine et al. & Dorcas, 1993). unpubl. data). Ectotherms offer excellent model systems to inves- tigate this topic. Typically, a single population of ecto- therms contains a much wider body-size range of STUDY AREA ecologically independent individuals than is the case We studied a population of E. annulatus in the Baie de for endotherms (e.g. Pough, 1980; Vitt, 2000). Conspe- Citrons, a tourist beach in Noumea, New Caledonia cific males and females often differ substantially in (22∞16¢S, 166∞26¢E). The study area (60 ¥ 60 m) had a mean body sizes, adding another level of complexity. mosaic substrate dominated by rocks, sand and coral Thermal exchange with the environment plays a rubble (0.8–1.5 m deep at high tide), extending to live major role for many ectotherms, and colour polymor- coral reef in deeper water (3–5 m deep at high tide; phism is common. Thus, we expect that a single pop- mean tidal range 0.9 m). The bay is protected from ulation of ectotherms may often contain a high degree winds and swell from most directions, and thus the of phenotypic variation attributable to body size, sex water is generally calm. and colour, and hence that intrapopulation variation in habitat use may be particularly important in such systems. SURVEYS Despite these advantages, most studies on the rela- Three people snorkelled over the site for 45–60 min tionship between phenotypic traits and habitat use observation sessions twice per day over the period 24 within populations have been based on endothermic June to 4 July 2002, except for two days when storms vertebrates (birds and mammals), typically in temper- decreased visibility and engendered a strong swell. ate-zone terrestrial habitats (e.g. Torre & Bosch, 1999; Observations were concentrated on periods approxi- Telleria et al., 2001; Bowyer et al., 2002; Mcdonald, mately 1 h before and 1 h following each high tide dur- 2002; McLoughlin et al., 2002; Rolando, 2002; Ruck- ing daylight hours. When we found a snake we stuhl & Neuhaus, 2002). We studied a system very dif- recorded the water depth and substrate characteris- ferent from these, involving sea snakes in tropical tics (% cover of coral rubble, live coral, rocks and sand) coral reefs. The species we studied displays a high of a 1-m2 quadrat centred on a point directly beneath degree of variation in body size and colour, with the the snake’s head at the time it was first seen. We also sexes differing in both of these attributes. Thus, it pro- scored the overall colour (lightness) of the substrate vides a very different situation in which to investigate within this quadrat on a five-point scale (0 = white, the ways in which an animal’s phenotype influences 5 = black). Some snakes were collected immediately its patterns of habitat use. after they were sighted, others were monitored for up © 2003 The Linnean Society of London, Biological Journal of the Linnean Society, 2003, 80, 1–10 SEA SNAKE HABITAT USE 3 to 10 min with habitat data taken every 60 s (in which 40 case we calculated average values for each snake over Juvenile the entire observation period) and then captured, and 30 Female some snakes were simply observed but not captured. Thus, our sample sizes for snake morphology differ Male from those for habitat records. 20 All captured snakes were measured (snout–vent length = SVL) and weighed. Sex could not be deter- 10 mined for neonatal snakes ( 35 cm), but for all other < No. of records snakes this was established based on relative tail 0 length, rugosity of dorsal scales and presence of a ros- 25 30 35 40 45 50 55 60 65 70 tral spine (both the latter traits characterize adult males: Guinea, 1996; pers.