Coral Reefs

, Volume 32, Issue 2, pp 527–538

Effects of season, sex and body size on the feeding ecology of turtle-headed sea snakes (Emydocephalus annulatus) on IndoPacific inshore coral reefs

Authors

  • C. Goiran
    • Laboratoire LIVE and Laboratoire d’excellence CORAILUniversité de la Nouvelle-Calédonie
  • S. Dubey
    • Department of Ecology and EvolutionUniversity of Lausanne
    • School of Biological Sciences A08University of Sydney
Report

DOI: 10.1007/s00338-012-1008-7

Cite this article as:
Goiran, C., Dubey, S. & Shine, R. Coral Reefs (2013) 32: 527. doi:10.1007/s00338-012-1008-7

Abstract

In terrestrial snakes, many cases of intraspecific shifts in dietary habits as a function of predator sex and body size are driven by gape limitation and hence are most common in species that feed on relatively large prey and exhibit a wide body-size range. Our data on sea snakes reveal an alternative mechanism for intraspecific niche partitioning, based on sex-specific seasonal anorexia induced by reproductive activities. Turtle-headed sea snakes (Emydocephalus annulatus) on coral reefs in the New Caledonian Lagoon feed entirely on the eggs of demersal-spawning fishes. DNA sequence data (cytochrome b gene) on eggs that we palpated from stomachs of 37 snakes showed that despite this ontogenetic stage specialization, the prey comes from a taxonomically diverse array of species including damselfish (41 % of samples, at least 5 species), blennies (41 %, 4 species) and gobies (19 %, 5 species). The composition of snake diets shifted seasonally (with damselfish dominating in winter but not summer), presumably reflecting seasonality of fish reproduction. That seasonal shift affects male and female snakes differently, because reproduction is incompatible with foraging. Adult female sea snakes ceased feeding when they became heavily distended with developing embryos in late summer, and males ceased feeding while they were mate searching in winter. The sex divergence in foraging habits may be amplified by sexual size dimorphism; females grow larger than males, and larger snakes (of both sexes) feed more on damselfish (which often lay their eggs in exposed sites) than on blennies and gobies (whose eggs are hidden within narrow crevices). Specific features of reproductive biology of coral reef fish (seasonality and nest type) have generated intraspecific niche partitioning in these sea snakes, by mechanisms different from those that apply to terrestrial snakes.

Keywords

Cost of reproductionDietary specialistOophagyPredationSexual dimorphism

Introduction

One of the most fundamental issues for the ecology and management of any population is the kind of resources upon which it depends. Knowing the types of habitat that individuals require, and the kinds of food they eat, is central to understanding the functioning of that population, and how and why population demography changes through time (e.g., Caughley 1977; Aars and Ims 2002; Oro et al. 2004). One common complication in studying dietary composition is that any given population contains individuals with a wide range of phenotypes (e.g., sex, size, locomotor ability) and is exposed to variable environmental conditions (e.g., seasonally or annually) that influence the availability of food and the ability of individuals to locate and ingest that food (Daltry et al. 1998; Aplin and Cockburn 2012). In combination, those sources of variation may generate intraspecific niche partitioning, with the diet of an individual animal dependent upon the organism’s phenotype as well as the environmental conditions under which its diet is sampled (Mushinsky 1987; Shine 1989; Shine and Wall 2005; Vincent and Herrel 2007). For example, reproductive activities often may curtail foraging opportunities (Sherry et al. 1980; Shaffer et al. 2003).

Within-population variation in foraging biology is of interest for several reasons. For example, such variation increases the number of resources used by the population: an ontogenetic shift in diets may mean that the organisms cannot persist in an area that lacks suitable food resources for juveniles, even if abundant food is available for adults. At the same time, such intraspecific niche variation offers powerful opportunities for research into the determinants of dietary composition. If sympatric males and females differ in food habits, for example, we can explore the reasons for that divergence without having to deal with the many confounding factors that complicate analyses of dietary variation through space or among species (Aplin and Cockburn 2012). As a result, niche divergence between age classes and sexes within populations has attracted extensive research in organisms ranging from albatross to elephants (Ruckstuhl and Neuhaus 2005).

A recent review of intraspecific niche partitioning in squamate reptiles showed that diets differed more as a function of sex and body size within snakes than within lizards and suggested reasons for that difference (Shine and Wall 2007). More specifically, Shine and Wall’s (2007) analysis of published data concluded that intraspecific shifts in prey type were most common in snake species with large adult body sizes, small offspring sizes, high relative prey mass (ratio of prey size to predator size) and a diet composed at least partly of endotherms (mammals and birds). These attributes are uncommon among aquatic snakes, suggesting that intraspecific niche partitioning may be uncommon also in such taxa (Shine and Wall 2007). In the present paper, we test that prediction with data on a sea snake species that does not exhibit any of the attributes identified by Shine and Wall (2007) as facilitating intraspecific niche partitioning in terrestrial snakes. Nonetheless, we found divergences in feeding rates and dietary composition as a function of snake sex and body size. Such exceptions from general patterns (or perhaps, differences between aquatic and terrestrial species) can clarify the ecological pressures that result in intraspecific niche divergence.

Materials and methods

Study species and area

Turtle-headed sea snakes (Emydocephalus annulatus Krefft 1869) are small (to 1 m total length) hydrophiine sea snakes belonging to the Aipysurus lineage within the proteroglyphous family Elapidae (Heatwole 1999; Lukoschek and Keogh 2006; Sanders et al. 2008). Females grow larger than males (mean adult snout-vent lengths of 54 and 51 cm, mean body masses 150 vs. 120 g, respectively; Shine et al. 2003, 2012) and have slightly larger heads relative to body length (Shine et al. 2012). Individuals of this species have been reported to feed only upon the eggs of demersal-spawning gobies and blennies (based on egg morphology; Voris 1966; McCosker 1975), although Guinea (1996) reported that the eggs of at least one damselfish (pomacentrid) species were consumed also. The unusual diet of E. annulatus is accompanied by morphological features such as atrophied fangs and venom glands (Gopalakrishnakone and Kochva 1990), modified throat musculature (McCarthy 1987) and an enlarged supralabial scale (Guinea 1996).

Although Emydocephalus annulatus occurs widely through the IndoPacific, available ecological data on the species (including the present study) come primarily from two adjacent bays of the Great Lagoon of New Caledonia, close to the city of Noumea (Anse Vata 22°18′15″S, 166°26′12″E and Baie des Citrons 22°18′05″S, 166°26′13″E). Previous work has described basic foraging and reproductive behaviors of snakes within these bays (Shine et al. 2004; Shine 2005) as well as habitat use (Shine et al. 2003), social groupings (Shine et al. 2005), effects of salinity variation (Brischoux et al. 2012) and the interactions between snake morphology and ecology (Avolio et al. 2006; Shine et al. 2010). Courtship and mating occur through the winter months, with females ovulating in late spring and giving birth in autumn (Shine et al. 2004; Shine 2005). Mark-recapture records and genetic analyses have revealed that snakes rarely move between these two adjacent bays (Lukoschek and Shine 2012).

Sampling and identification of prey from field-collected snakes

In the course of our mark-recapture studies, we collected snakes by hand and returned them to a nearby laboratory to measure snout-vent length (SVL) and record sex (based upon tail length and scale rugosity, see Avolio et al. 2006). We also conducted standardized surveys at our study sites from September 2011 to June 2012 to record feeding rates (based on proportions of animals found feeding [or not] during observation periods). The surveys for feeding rates consisted of an observer snorkeling for 90–120 min (depending on water temperature), over the entire study site, recording size, sex and activity (foraging, feeding or sleeping) of all snakes seen. The feeding behavior of Emydocephalus cannot be mistaken for any other behavior, as the snake stays for several minutes (5–40 min) at the same place, with its head in a hole or between coral branches scraping the substrate with its mouth (see Electronic Supplementary Material, ESM, for video sequences of this behavior). The snake’s throat, when visible, can be seen to move as the snake swallows food. In contrast, snakes searching for prey swim at approximately 1 m min−1 (Shine et al. 2004).

In a subset of the surveys during 2011, we palpated all of the animals that we captured, to check for recently ingested food (fish eggs). We obtained identifiable prey records from the stomachs of 37 snakes (23 in summer [January 2011] and 14 in winter [July–September 2011]). Many other snakes were collected over the same periods, but had empty stomachs (see below). Snakes were unharmed by the procedure and released at their sites of capture immediately thereafter. Fish eggs (the sole prey type recovered) were preserved in ethanol for subsequent genetic analysis. Snakes either had empty or virtually empty stomachs (sometimes containing fluids that likely were the remnants of earlier meals) or had hundreds of virtually undigested eggs, all of the same approximate size and color. We thus interpret these as single meals, obtained shortly before the snake was collected. In two cases, a snake regurgitated eggs that appeared to comprise two types (based on color) but our analyses of those replicate samples showed that in each case, the two subsamples were genetically identical to each other and thus probably came from a single meal. However, more than one egg mass may be involved; in some species of Gobiidae and Blenniidae living in our study area, several females may spawn successively in the same nest (Hernaman and Munday 2007; Awata et al. 2010).

Over the same period, we collected as many demersal-spawning fish species as possible from the same bays, to provide tissue samples whose DNA could be compared with that of the regurgitated eggs. We obtained specimens of 19 species (1 or 2 individuals per species). Fish were killed with eugenol, photographed for reference purposes, and then the left pectoral fin was removed for use in genetic tests. The remaining carcass was frozen as a voucher specimen for later analysis. We also obtained GenBank data on demersal-spawning fish taxa from this location (based on a species list that we prepared) as well as species from other locations (Table 1).
Table 1

Fish species for which we obtained genetic data for use in our analyses to identify the fish eggs consumed by turtle-headed sea snakes

Species

Collection code

Family

GenBank accession number

Acanthurus lineatus

 

Acanthuridae

EU273284

Amia calva

 

Amiidae

AB042952

Apogon maculatus

 

Apogonidae

AY722182

Rhinecanthus aculeatus

 

Balistidae

AP009210

Petroscirtes breviceps

 

Blenniidae

AP004550

Salarias fasciatus

 

Blenniidae

AP004551

Salarias fasciatus

CGUNC5

Blenniidae

JX001660

Seriola lalandi

 

Carangidae

AB517557

Chaetodon auripes

 

Chaetodontidae

AP006004

Heniochus diphreutes

 

Chaetodontidae

AP006005

Heterostichus rostratus

 

Clinidae

AY973062

Platax orbicularis

 

Ephippidae

AP006825

Amblygobius phalaena

CGUNC4

Gobiidae

JX001658

Amblygobius phalaena

CGUNC17

Gobiidae

JX001663

Amblygobius sphynx

CGUNC18

Gobiidae

JX001664

Asterropteryx semipunctata

 

Gobiidae

EU380948

Gobiodon citrinus

 

Gobiidae

FJ617108

Istigobius decoratus

 

Gobiidae

JN575315

Istigobius decoratus

CGUNC14

Gobiidae

JX001661

Istigobius decoratus

NewCal_117

Gobiidae

JX001667

Ptereleotris microlepis

CGUNC1

Gobiidae

JX001658

Valenciennea longipinnis

CGUNC7

Gobiidae

JX001653

Valenciennea longipinnis

CGUNC12

Gobiidae

JX001657

Valenciennea puellaris

CGUNC3

Gobiidae

JX001652

Myripristis berndti

 

Holocentridae

AP0028940

Cheilinus undulatus

 

Labridae

GU296101

Lutjanus argentimaculatus

 

Lutjanidae

JN182927

Pseudalutarius nasicornis

 

Monacanthidae

AP009226

Gymnothorax kidako

 

Muraenidae

AP002976

Plotosus lineatus

 

Plotosidae

DQ119351

Centropyge loricula

 

Pomacanthidae

AP006006

Pomacanthus maculosus

 

Pomacanthidae

JN604380

Abudefduf margariteus

 

Pomacentridae

F457873

Abudefduf sexfasciatus

 

Pomacentridae

AY208555

Abudefduf vaigiensis

 

Pomacentridae

JF457889

Abudefduf whitleyi

 

Pomacentridae

AY208562

Amblyglyphidodon curacao

 

Pomacentridae

AY208564

Amblyglyphidodon indicus

 

Pomacentridae

JF457893

Amblyglyphidodon leucogaster

 

Pomacentridae

AY208565

Amphiprion akindynos

 

Pomacentridae

DQ343945

Amphiprion melanopus

 

Pomacentridae

DQ343955

Chromis viridis

 

Pomacentridae

JF458067

Chrysiptera rollandi

 

Pomacentridae

AY208573

Chrysiptera taupou

CGUNC10

Pomacentridae

JX001656

Dascyllus aruanus

 

Pomacentridae

JF458113

Hemiglyphidodon plagiometopon

 

Pomacentridae

AY208577

Neoglyphidodon melas

 

Pomacentridae

JF458177

Neopomacentrus cyanomos

 

Pomacentridae

JF458184

Neopomacentrus nemurus

 

Pomacentridae

AY208586

Pomacentrus adelus

 

Pomacentridae

AY208591

Pomacentrus bankanensis

 

Pomacentridae

AY208593

Pomacentrus moluccensis

 

Pomacentridae

AY208601

Pomacentrus aurifrons

CGUNC8

Pomacentridae

JX001654

Pomacentrus aurifrons

CGUNC9

Pomacentridae

JX001655

Pomacentrus chrysurus

CGUNC22

Pomacentridae

JX001665

Pomacentrus moluccensis

NewCal_118

Pomacentridae

JX001666

Pomacentrus nagasakiensis

CGUNC15

Pomacentridae

JX001662

Stegastes nigricans

 

Pomacentridae

JF458265

Heteropriacanthus cruentatus

 

Priacanthidae

DQ197957

Pterois volitans

 

Scorpaenidae

DQ482606

Epinephelus areolatus

 

Serranidae

AY786721

Siganus unimaculatus

 

Siganidae

AP006031

Genetic analyses

Total cellular DNA was isolated from whole eggs and fins. Then, tissues were placed in 200 μL of 5 % Chelex containing 0.2 mg/mL of proteinase K, incubated overnight at 56 °C, boiled at 100 °C for 10 min and finally centrifuged at 13,200g for 10 min. Then, the supernatant containing purified DNA was collected and stored at −20 °C. Double-stranded DNA amplifications of the cytochrome b gene (cyt-b) were performed with the primer pairs Gludg-L/H15915, Gludg-L/H16460 and primers F/R, see Table 2 for more details. Amplification conditions included a hot start denaturation of 95 °C for 3 min, followed by 35 cycles of 95 °C for 60 s, 55 °C annealing temperature for 60 s, 72 °C for 120 s and a final extension of 72 °C for 7 min.
Table 2

Primers used to identify the eggs of fishes that were obtained from stomachs of turtle-headed sea snakes, Emydocephalus annulutus

Primer

Sequence

Reference

Gludg-L

5′-TGACTTGAARAACCAYCGTTG-3′

Palumbi et al. (1991)

H15915

5′-AACTGCAGTCATCTCCGGTTTACAAGAC-3′

Irwin et al. (1991)

H16460

5′-CGAYCTTCGGATTACAAGACCG-3′

Quenouille et al. (2004)

F

5′-GTGATCTGAAAAACCACCGTTG-3′

Song (1994)

R

5′-AATAGGAAGTATCATTGCGGTTTGATG-3′

Taberlet et al. (1992)

The sequences were aligned by eye (GenBank accession numbers: JX001645–JX001667). Tests were conducted on the total fragment (697 bp); all codon positions were used. The tree was rooted using a sequence of Amia calva (Amiidae, Holostei). In addition, we included previously published sequences of the following Families in our dataset: Acanthuridae, Apogonidae, Balistidae, Blenniidae, Carangidae, Chaetodontidae, Clinidae, Ephippidae, Gobiidae, Holocentridae, Labridae, Lutjanidae, Monacanthidae, Muraenidae, Plotosidae, Pomacanthidae, Pomacentridae, Priacanthidae, Serranidae and Siganidae (see Table 1 for more details). For maximum likelihood (ML) analyses, we used jModelTest 0.1.1 (Guindon and Gascuel 2003; Posada 2008) to select the model of DNA substitution. The TIM + I + G (Posada 2003) model best fitted the dataset with Akaike’s information criterion (AIC), and ML heuristic searches were performed using phyml (Guindon and Gascuel 2003).

Statistical analyses

We used logistic regression (in the statistical package JMP 9.0; SAS Institute, Cary, NC) to examine effects of snake sex, body size (SVL) and season (summer vs. winter) on (a) whether or not a snake contained freshly ingested prey in its stomach when palpated and (b) the family to which identifiable prey belonged.

Results

Identity of prey species

Genetic analysis revealed that the eggs we palpated from snake stomachs came from a diverse array of fish taxa (Fig. 1). In no case did we find more than a single prey species in the stomach of a given snake (based on egg size, color, state of digestion and—in two cases where this was ambiguous—on genetic analyses). In some cases, identification of prey was straightforward because DNA samples from locally collected fish matched the DNA in eggs (e.g., Salarias fasciatus,Chrysiptera rollandi) or were nested within DNA sequences for a taxon collected from other areas (e.g., Istigobius decoratus). Problems in delimiting species mean that in other cases, we could only assign our egg samples to a clade of related species (e.g., Pomacentrus moluccensis-chrysurus-bankanensis). One group of taxa (winter samples in the uppermost part of Fig. 1) lay outside the sequence data that we gathered on local species, but presumably are damselfish (when we blast these sequences in GenBank, the first match is the damselfish genus Abudefduf, with a genetic distance of 14 %).
https://static-content.springer.com/image/art%3A10.1007%2Fs00338-012-1008-7/MediaObjects/338_2012_1008_Fig1_HTML.gif
Fig. 1

Species identities of fish eggs recovered from the stomachs of turtle-headed sea snakes (Emydocephalus annulatus) and of demersal-spawning fishes in the Noumea region of the New Caledonian Lagoon where the snakes were sampled. The phylogram is based on a maximum likelihood analysis of DNA sequence data (cytochrome b gene)

Feeding rates as a function of snake sex, size and season

Within the sample of snakes from which we obtained prey, females averaged significantly larger than males (means = 52.1 vs. 58.0 cm SVL, F1,35 = 5.90, P < 0.025). Overall, 35 % (86 of 244) of snakes contained prey items (fish eggs) in their stomachs when palpated. Of these, we successfully identified 37 samples. About half of the females contained prey in both summer and winter samples, whereas males often contained food in summer, but almost never in winter (Fig. 2). Logistic regression on whether or not a snake contained prey in its stomach thus revealed a significant interaction between sex and season (likelihood ratio \( \chi^{2}_{1} = 44.80 \), P < 0.0001). During summer when both sexes were feeding, the proportion of snakes with food was higher in males than females (63 % of 40 vs. 40 % of 42, respectively; \( \chi^{2}_{1} = 4.01 \), P = 0.04). Snake body length (SVL) did not significantly affect feeding frequency (\( \chi^{2}_{1} = 0.06 \), P = 0.81). In summer when many but not all females were gravid, a female’s reproductive state did not significantly affect her probability of containing freshly ingested food (29 % of 14 gravid females contained food vs. 42 % of 24 non-gravid females; \( \chi^{2}_{1} = 0.66 \), P = 0.42).
https://static-content.springer.com/image/art%3A10.1007%2Fs00338-012-1008-7/MediaObjects/338_2012_1008_Fig2_HTML.gif
Fig. 2

Effects of season and snake sex on the feeding rates and dietary composition of turtle-headed sea snakes (Emydocephalus annulatus) from inshore reefs in the Noumea region of the New Caledonian Lagoon. During winter, male snakes did not feed, and females fed primarily on the eggs of pomacentrid fishes

Our standardized behavioral surveys provided additional and more detailed data on seasonal shifts in feeding frequency as a function of snake sex (Table 3). The adult sex ratio is close to 50:50 in this population, based on longterm mark-recapture data (e.g., Shine et al. 2010), and the snakes are highly philopatric, so sex ratios do not change seasonally (Lukoschek and Shine 2012). Thus, shifts in the relative numbers of male versus female snakes seen mostly reflect changing activity patterns. For example, monthly observations show an abrupt decrease in the frequency of foraging (swimming about actively, tongue-flicking the substrate) and feeding (stationary, with heads down burrows or inside coral matrices; see ESM videos) in female snakes partway through pregnancy (N = 13 records in February, 0 in March). In contrast, male snakes continued to feed frequently over this period (N = 25, 30 over the same 2 months: comparing the two time periods in males versus females, \( \chi^{2}_{1} = 10.57 \), P < 0.002). High numbers of males recorded as “foraging” midyear (May to August) reflect mate searching rather than feeding, because few males were actually found feeding over this period (Table 3). Sample sizes are lower for latter parts of the year, but show a general similarity in foraging and feeding rates between the sexes (Table 3). A functional link to reproduction is supported by the fact that females commence feeding very soon after they give birth; for example, in 2012, we first recorded neonates on 15 May and saw an adult female foraging (for the first time since late February) only 2 days later.
Table 3

Numbers of turtle-headed sea snakes recorded as foraging, feeding or inactive during standardized surveys in the Baie des Citron, New Caledonia

Month

# of inactive snakes

# of foraging snakes

# of feeding snakes

Female

Male

Female

Male

Female

Male

February

21

16

9

19

13

25

March

54

15

5

51

0

30

April

18

13

0

9

0

1

May

11

1

1

10

0

2

June

11

2

2

28

0

0

July

3

1

0

1

0

0

August

5

0

4

18

0

0

September

1

0

6

9

8

1

October

0

1

8

7

3

4

“Foraging” includes any cases where snakes were actively swimming about tongue-flicking the substrate and thus includes mate searching by males as well as foraging. “Feeding” is restricted to cases where the snake was stationary for at least 1 min, with its head and forebody inserted into a crevice. “Inactive” refers to snakes that were immobile when sighted; most of these animals were partly hidden beneath coral. All data were collected during 2012; these snakes were not used for gut-contents analyses. Variation in total sample sizes among months reflects variation in sampling effort

Composition of the diet as a function of snake sex, size and season

The taxonomic composition of the diet shifted between summer and winter, with the eggs of pomacentrids (damselfish) dominating the diet of snakes in winter (>80 % of records) but not summer (<20 % of records: see Fig. 2b). The number of records of snakes with blenny eggs in their stomachs decreased from summer to winter, and the numbers that were found to have consumed goby eggs showed an even stronger trend, comprising about one-third of all prey items in summer, but never recorded in winter (Fig. 2c, d). Male and female snakes had similar diets (Fig. 2). Logistic regression showed that dietary composition did not differ significantly between the sexes within summer (\( \chi^{2}_{2} = 1.32 \), P = 0.51) nor if both seasons were included (\( \chi^{2}_{2} = 1.36 \), P = 0.50). Looking only within female snakes, diets differed strongly between seasons (\( \chi^{2}_{2} = 16.79 \), P = 0.0002). Dietary composition also was affected by a snake’s body size, with larger snakes consuming damselfish eggs more often than did smaller conspecifics (Fig. 3). Logistic regression on these data shows that the taxa of fish eggs consumed were affected by a snake’s body size (\( \chi^{2}_{2} = 12.79 \), P < 0.002) but not its sex (\( \chi^{2}_{2} = 0.21 \), P = 0.90). The effect of body size on diet did not differ significantly between the sexes (interaction sex*size \( \chi^{2}_{2} = 2.06 \), P = 0.36; Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs00338-012-1008-7/MediaObjects/338_2012_1008_Fig3_HTML.gif
Fig. 3

Body sizes (snout-vent lengths) of turtle-headed sea snakes (Emydocephalus annulatus) in the Noumea region of the New Caledonian Lagoon that consumed the eggs of fishes belonging to three Families. The graph shows mean values and associated standard errors of snake snout-vent lengths. In both sexes, the snakes consuming pomacentrid eggs were larger than those consuming the eggs of other fish families

Discussion

Intuition suggests that a predator species with a highly specialized diet is unlikely to show much intrapopulation variation in foraging biology. Contrary to that prediction, turtle-headed sea snakes show intraspecific variation in feeding rates and dietary composition not only seasonally but also with respect to body size, and those patterns impinge differentially on male and female snakes. Although Emydocephalus annulatus is highly specialized in terms of the ontogenetic stage of prey that it consumes (eggs), it is relatively generalized in terms of the taxonomic diversity of its prey (at least 14 species from three Families: Fig. 1). By comparison, previous surveys of sea snake diets have found that many of these snakes feed primarily upon a few types of prey (e.g., a single prey species often comprises >50 % of the snake’s diet, Glodek and Voris 1982). Turtle-headed sea snakes are thus specialized in one respect (ontogenetic stage of prey consumed), but generalized in another (range of prey taxa consumed).

Seasonality of feeding is the rule in cool-climate snake species (which typically are inactive over winter; but see Mori and Toda (2011) for a counter example). Even in tropical areas, food resources often are available (or most abundant) only at restricted times of the year (Martins et al. 2002). However, such constraints generally fall in a similar way on both sexes. In contrast, both sexes of turtle-headed sea snakes feed during summer, but females ceased feeding as parturition approached (whereas males continued to feed at this time), and males ceased feeding during winter (whereas females continue to feed, Fig. 2a).

Why does reproduction conflict with foraging in this species? Males are actively engaged in mate searching during winter and move about more rapidly and extensively than is true for either sex at other times of year (Shine et al. 2003; Avolio et al. 2006). Such an increase in movements during the mating season is common in male snakes (Gregory et al. 1987), but does not inevitably lead to a decrease in feeding rates by males. For example, in wide-foraging snake species that rely upon chemosensory cues both to locate mates and to locate prey, a male searching suitable crevices may be able to consume any prey items that he finds in the course of his mate-searching activities (Shine 1977). In contrast, active mate searching is incompatible with the crypsis and immobility required for ambush predation, so that males of ambush-foraging species may need to forego feeding during the mating season (e.g., viperids—King and Duvall 1990; Bea et al. 1992; Madsen and Shine 1993; Aldridge and Brown 1995; Olsson et al. 1997; Daltry et al. 1998; pythons—Madsen and Shine 2000). Turtle-headed sea snakes are active foragers that rely upon chemical cues to locate their prey (Shine et al. 2004), suggesting that feeding and mate searching would be compatible activities. However, the difficulty of detecting substrate-deposited pheromones in water has forced male E. annulatus to rely upon vision not scent in their search for mates (based upon the males’ vigorous responses to experimentally presented visual cues from females), and thus, mate-searching males swim rapidly (average of about 3 m min−1, vs. 1 m min−1 in foraging snakes, Avolio et al. 2006) in midwater scanning for females rather than tongue-flicking crevices on the seafloor as is typical of foraging snakes (see also Shine et al. 2004; Shine 2005). Hence, marine life has enforced a shift in the sensory modalities used for mate location (from chemosensory to visual), rendering mate searching and feeding incompatible in this system (see also Shine et al. 2003).

In contrast to the present study, most reports of seasonal anorexia associated with reproduction in terrestrial snakes involve females not males. Widespread cessation of feeding by reproductive females presumably reflects the locomotor burden of the enlarging litter or clutch, such that foraging becomes energetically expensive, and perhaps unproductive and even dangerous. Thus, heavily gravid females of many snake species forego feeding until after they give birth or lay their eggs (for reviews see Shine 1980; Gregory et al. 1999; Brischoux et al. 2011). More generally, pregnancy reduces feeding rates in a wide variety of animal taxa (Weeks 1996). Our data show this pattern, but only toward the end of pregnancy when the offspring are very large. In the earlier phases of pregnancy, female turtle-headed sea snakes continued to feed at similar rates during summer (when many females were gravid) and winter (when they were not), and in midsummer, the proportion of snakes containing prey was not significantly lower in gravid females than in non-gravid conspecifics. Continued foraging (despite the obvious bodily distension caused by pregnancy) may reflect both the nature of the prey resource and the medium in which foraging occurs. First, demersal fish eggs are immobile, so that foraging snakes can travel at a sedate pace (approximately 1 m min−1, Shine et al. 2003, 2004): any reduced locomotor speed is unlikely to curtail foraging effectiveness. Second, swimming is less energy expensive than terrestrial locomotion (for studies on locomotion in snakes, see Jayne 1985; Cundall 1987; Webb 2004; Winne and Hopkins 2006), plausibly reducing the energy disadvantages of a distended hindbody. A scarcity of predators in the shallow-water reef systems inhabited by E. annulatus (Shine et al. 2003) may further reduce the costs of foraging and the importance of rapid locomotion. Thus, the evolutionary shift to marine habitats has reduced the costs of reproduction for females, enabling them to feed for most of pregnancy. However, the switch to visually based mate-searching tactics (above) has rendered reproduction incompatible with foraging for males.

Even if (as we suggest) the seasonal anorexia of male snakes is due to conflicts with mate searching, the adaptive process responsible for that link remains ambiguous. That is, has the timing of male reproduction evolved in response to seasonal patterns in prey availability (such that males forego feeding at a time when foraging is relatively unproductive anyway), or are seasonal schedules of feeding a result (not cause) of seasonality in reproductive schedules? Several authors have speculated that in tropical reptiles, the seasonal timing of reproduction has evolved to maximize prey availability for the newborn young (Vitt and Pianka 1994; Brown and Shine 2006). We doubt this interpretation for our study system, because of strong phylogenetic conservatism in breeding seasonality among species related to turtle-headed sea snakes. Among snakes in seasonal climates, embryonic development generally occurs in the warmest part of the year (as it does in turtle-headed sea snakes), probably reflecting the benefits of high and stable temperatures for development of the offspring’s phenotype (Lourdais et al. 2004). For example, midsummer pregnancies are the rule in most terrestrial Australian elapid snakes, the lineage from which viviparous sea snakes evolved (Shine 1985).

A divergence between the sexes in the seasonality of feeding will translate into a sex difference in diets only if prey resources also shift seasonally. This is clearly the case in our study system, with damselfish comprising a far higher proportion of prey in winter than in summer. The most parsimonious interpretation of this seasonal shift is that different fish species have different reproductive schedules, with (for example) gobies reproducing only in warmer months of the year. We lack specific data on breeding biology of fishes in the Noumea region to test this hypothesis, but previous work has revealed strong species-level and family-level variation in breeding seasonality of coral reef fishes (Thresher 1984; Robertson 1991). Published data suggest that damselfish from other regions spawn throughout the year, but (at least in cooler climates) with a peak in warmer months (Doherty 1983; Asoh and Yoshikawa 2002; Hilder and Pankhurst 2003; Srinivasan and Jones 2006). Reproductive seasonality of the pomacentrids in our study area may differ from that of damselfish species from other parts of the world, or (more likely) reproduction is simply less strictly seasonal in damselfish than in gobies and blennies (see Awata et al. 2010 for data on the breeding season of Salarias fasciatus in Japan). Thus, damselfish eggs may be the only ones that are available to snakes in winter. An alternative explanation for the seasonal shifts in snake diets would be that snake diets shift because of changes in prey choice, but the broad dietary composition of snakes in summer renders this possibility unlikely. Thus, the seasonal shift in diets most plausibly reflects seasonal differences in available prey resources.

A sea snake’s sex may influence its diet not only because of sex-specific effects of reproduction on feeding rates, but also because female turtle-headed sea snakes grow larger than males (the sex divergence in mean adult body size was statistically significant among our sample of snakes with prey, but the sex difference in diet composition was not significant after SVL effects were removed). In both sexes, larger snakes tend to consume the eggs of damselfish (pomacentrids) rather than gobies or blennies (Fig. 3). Effects of body size on prey type are widespread both in terrestrial and in aquatic snakes, and generally result from gape limitation (Mushinsky 1987; Shine 1991; Arnold 1993; Shine et al. 1998; Brischoux et al. 2007; Vincent and Herrel 2007). That is, because snakes must consume prey items entire, and often take prey items that are very large compared to predator size, a small snake cannot physically ingest many of the prey species consumed by conspecific adults. This explanation cannot apply to turtle-headed sea snakes, because even the smallest snake can easily ingest the tiny fish eggs (<2 mm diameter: Shine et al. 2004) that are their sole prey. Why, then, does the composition of the diet shift with snake body size (Fig. 3)? Foraging sea snakes often struggle to penetrate deeply enough into nesting burrows to reach the eggs (R. Shine, pers. obs.). Plausibly, larger snakes are unable to penetrate the narrow crevices and burrows that gobies and blennies use as nesting sites, but these large snakes can more readily access the eggs of pomacentrids (which often are laid in relatively open areas, rather than inside burrows, Thresher 1984; Ormond et al. 1996; Neat and Lengkeek 2009; Herler et al. 2011). The same constraint of body size on crevice use generates correlations between prey size and predator size in crevice-foraging terrestrial snake species (Shine 1991). Alternatively or additionally, larger sea snakes tend to forage in deeper water (Shine et al. 2003), and damselfish nests may be disproportionately located in such areas rather than in the extreme shallows often used by gobies and blennies (C. Goiran, pers. obs.).

Links between attributes of the predator (sex and body size) and its foraging ecology (what it eats, when it eats and where it eats) thus are forged by several processes and may differ between aquatic and terrestrial systems. Although terrestrial snakes exhibit many superficially similar examples of intraspecific niche partitioning (e.g., Greene 1983; Mushinsky 1987; Luiselli and Angelici 1998), the invasion of marine habitats by proteroglyphous snakes has altered parameters that influence both foraging ecology and reproduction (Shine 1988). For example, the eggs of demersal fishes offer an abundant food resource, such that Emydocephalus consumes more and smaller prey items than do any terrestrial snakes except for colubrids that consume amphibian eggs (Warkentin 1995; Greene 1997) and the phylogenetically basal scolecophidians that feed upon insect eggs and larvae (Shine and Webb 1990; Webb et al. 2000). The immobility and lack of escape behavior of this food resource have resulted in the snakes adopting a foraging mode more similar to that of herbivorous ungulates (browsing or grazing) than to foraging modes reported in terrestrial snakes (Shine et al. 2004). Seasonal variation in the availability of alternative prey types is widespread in terrestrial as well as aquatic systems and may often generate seasonal variation in predator feeding rates and dietary composition (e.g., Arnold and Wassersug 1978; Sun et al. 2001). If reproductive activities are incompatible with foraging, seasonal shift in trophic resources may result in sex differences in feeding rates and prey types (Shine and Wall 2005). Nonetheless, influences of the marine environment on locomotor costs and the effectiveness of alternative sensory modalities for mate searching have resulted in reproduction-associated cessation of feeding being stronger in males than in females, contrary to the usual pattern in terrestrial snakes (Brischoux et al. 2011). Thus, the phylogenetic shift to marine habits may have imposed powerful new forces on the foraging ecology of snakes.

Acknowledgements

We thank Richard Farman and Philippe LeBlanc of the Aquarium de Noumea for logistical assistance, and our volunteer divers (Terri, Mac and Ben Shine, Heather Zimmerman and Marielle Dunaj) for assistance in collecting snakes. The work was funded by the Australian Research Council and the Swiss National Science Foundation. Samples were collected under the New Caledonia Southern Province permit number 930-2011/ARR/DENV.

Supplementary material

View video

Supplementary material 1 (MOV 28618 kb)

View video

Video sequences of turtle-headed sea snakes foraging and feeding in the Baie des Citrons, New Caledonia. Photography by Claire Goiran. Supplementary material 2 (AVI 15978 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2013