Congenital feeding response to a novel prey in a Mexican gartersnake

Introduction

Intraspecific geographic differences in snake diets have been associated with behavioral microevolution within species, and geographical variation in selection on feeding (Arnold, 1992; De Queiroz, Henke & Smith, 2001; Tuttle & Gregory, 2009). Congenital, chemically based preferences for different prey taxa have been widely demonstrated in several Thamnophis species (Burghardt, 1970; Burghardt, 1975; Burghardt, 1992). Newborn snakes respond to species-characteristic prey stimulus with high tongue-flick incidence and predatory strikes (Burghardt, 1993), and these behaviours have a heritable basis (Brodie Jr & Garland, 1993) with low to moderate heritabilities (0.04 to 0.36; Arnold, 1981a; Arnold, 1981b), although they may show different developmental trends (Burghardt, Layne & Konigsberg, 2000). Adaptive plasticity was documented when the chemically based congenital preference of Mexican black-bellied garter snakes for a dangerous species of sympatric leech declined after experiencing its defences (Drummond & Macías-García, 1995).

Thamnophiini (Natricinae) snakes normally consume soft prey such as fish, amphibians, slugs, and earthworms (Arnold, 1993; Rossman, Ford & Seigel, 1996), so predation on hard items could be select for behavioral and morphological adaptation (Dwyer & Kaiser, 1997; Rossman, 1963). Crayfishes are a difficult prey for snakes because they have a hard exoskeleton which is soft for only a few hours after molting (Godley, McDiarmid & Rojas, 1984). Generally, snakes must capture and manipulate crayfishes only when they are a soft and vulnerable prey (Gibbons & Dorcas, 2004; Godley, 1980; Mushinsky, Hebrard & Vodopich, 1982). This is exemplified by variation in the predatory behavior of snakes of the closely related genera Regina and Lyodytes (previously Regina) on crayfish (Godley, 1980; Mushinsky, Hebrard & Vodopich, 1982; Gibbons & Dorcas, 2004). Regina septemvittata (Queen Snake) and R. grahamii (Graham’s Crayfish Snake) capture freshly molted individuals only (Hall, 1969; Mushinsky & Hebrard, 1977; Godley, McDiarmid & Rojas, 1984), seizing them by the abdomen or head (Hall, 1969; Godley, McDiarmid & Rojas, 1984) without a significant preference for the direction of ingestion. Lyodytes alleni (Striped Crayfish Snake) and L. rigida (Glossy Crayfish Snake) also feed on hard, unmolted crayfish (Franz, 1977), and they capture all crayfish by the abdomen and constrain or press them against the substrate to ingest, preferably starting with the abdomen (Godley, 1980; Godley, McDiarmid & Rojas, 1984; Myer, 1987). In addition, the two pairs of Regina and Lyodytes species differ in the shape of their dentition and skulls (Myer, 1987; Fontenot, Platt & Dwyer, 1993; Dwyer & Kaiser, 1997). Mexican Black-bellied Gartersnakes are within the well-defined monophyletic group of garter snakes, whereas Regina and Lyodytes are polyphyletic with respect to other thamnophiines (Alfaro & Arnold, 2001; De Queiroz, Lawson & Lemos-Espinal, 2002; Guo et al., 2012; McVay & Carstens, 2013). This suggests that crayfish ingestion has probably arisen independently among Regina and Lyodytes species via evolutionary convergence associated with the ingestion of soft versus hard crayfish.

Crayfish ingestion is rare in Thamnophis spp. Although foraging in and around aquatic habitats is common, extensive field studies of the 31 species failed to reveal any instance except in T. proximus. In these cases crayfish made up 0.8% of stomach contents. (Rossman, Ford & Seigel, 1996; Hampton & Ford, 2007). However, in a population of the Mexican Black-bellied Gartersnake (Thamnophis melanogaster), which forages on soft-bodied prey such as fish, tadpoles, and leeches (Drummond, 1983), 35% of prey were crayfish Cambarellus montezumae (Manjarrez, Macías García & Drummond, 2013). The gartersnake is extensively sympatric with the crayfish throughout its range, but eats it only in a 6,000 km² area straddling the back-to-back headwaters of the rivers Tula and Lerma, an area comprising only 3.0% of the area of sympatry (Manjarrez, Macías García & Drummond, 2013). This pattern has been interpreted as invasion of a niche novel to Thamnophis spp. and possibly in process of expansion through the Tula and Lerma basins, which provides an opportunity to study the early stages of microevolution toward the incorporation of a new prey in the diet (Arnold, 1981a). The 76 crayfish recovered from snake stomachs had been eaten within 24 h of molting, when the exoskeleton is still soft (Manjarrez, Macías García & Drummond, 2013), but chemosensory tests on wild-caught snakes suggested that freshly-molted and unmolted (hard) crayfish were both recognized as prey and readily attacked (Lozoya, 1988).

The Mexican black-bellied garter snake’s incorporation of a new prey into its diet in the headwaters of two adjacent basins is likely to involve two distinct processes whose relative importance in ecological divergence is the subject of much debate: microevolution (Arnold, 1981a) and phenotypic plasticity (West-Eberhardt, 2003). In garter snakes both are plausible.

In this study, we tested for a microevolutionary response of the crayfish-eating population of black-bellied garter snakes toward crayfish (C. montezumae). We compared inexperienced neonates born in captivity from crayfish-eating populations with neonates from non crayfish-eating populations for their chemosensory arousal, ingestion of crayfish tissue, and capture and manipulation of freshly-molted (soft) crayfish. We predicted neonates from the crayfish-eating populations would show tendencies in the direction of Regina species that specialize in eating soft crayfish. In relation to newly moulted C. montezumae, neonates from the crayfish-eating populations should show greater chemosensory and ingestive responsiveness, greater heritability of chemosensory responsiveness, ontogenetic stability or increase in ingestion of pieces, elevated predatory responses, and more efficient capture and manipulation. Presence of any of these effects would imply microevolution at an early stage of niche expansion, while absence of all of them could imply that niche expansion did not gain its initial impetus through microevolution.

Materials & Methods

Results

Chemosensory test

Separate repeated measures ANOVAs showed that there was no effect of sex and litter at each stimulus on tongue-flick attack score. However, there was an effect of sex only in control stimulus (Table S2).

In the neonates from NC localities, ln-TFAS for soft crayfish (0.91 ± 0.36) and hard crayfish (0.88 ± 0.35) were 1.3 and 1.2 times higher, respectively, than for the control water (0.69 ± 0.33; F_2,99 = 3.28, P = 0.04; Fig. 1). In neonates of C localities, the ln-TFAS for soft crayfish (1.05 ± 0.47.) and hard crayfish (0.90 ± 0.25) were similar to the control (0.91 ± 0.32; F_2,42 = 0.52, P = 0.60; Fig. 1A). The relative response for control was similar between C (0.14 ± 0.54) and NC neonates (0.21 ± 0.25; Student-t₄₇ = 0.53, P = 0.70; Fig. 1B). The relative response for hard crayfish was also similar between C (0.15 ± 0.33) and NC individuals (0.02 ± 0.25; Student-t₄₇ = 1.22, P = 0.11; Fig. 1B).

The number of attacks was low in both groups of neonates. Only 3.1% of neonates from C and 11.5% from NC attacked a swab. Of C localities, only two neonates attacked a swab with soft crayfish stimulus and did not attack other odors, while NC neonates attacked in similar percentages to soft and hard crayfish swabs (6.3% and 5.2%, respectively; X² = 0.31, df = 1, P = 0.58), but did not attack the control-water swabs. Attacks were only recorded towards hard and soft crayfish swabs. The number of attacks towards soft crayfish swabs (N = 328) was higher than hard crayfish swabs (N = 151; G = 5.34, df = 1, P = 0.02) in both T. melanogaster populations.

In C neonates, the heritability of chemosensory response to soft crayfish (1.00 ± 0.33) was greater than NC neonates (0.64 ± 0.28), while heritability to hard crayfish was lower (0.13 ± 0.48) than NC neonates (0.71 ± 0.35; Table 1). In the neonates from C localities, heritability to soft crayfish was higher than hard crayfish and control water (0.15 ± 0.40), while neonates from NC had heritability values that were similar for soft and hard crayfish (Table 1).

Table 1:

Heritability (h²) of chemosensory response (ln-TFAS) of neonates T. melanogaster to crayfish stimulus in C and NC neonates.

Heritability of the log-transformed TFAS for each stimulus estimated by intraclass correlation, obtained from a one-way analyses of variance as full siblings families.

Stimuli	C			NC
	15clutches, 65 neonates			34clutches, 253 neonates
	h2±SE	p	95% Confidence limit	h2±SE	p	95% Confidence limit
Soft crayfish	1.00 ± 0.33	<0.001	0.71–1.95	0.64 ± 0.28	>0.05	−0.01–0.95
Hard crayfish	0.13 ± 0.48	>0.05	−0.79–0.99	0.71 ± 0.35	>0.05	−0.02–1.17
Water	0.15 ± 0.40	>0.05	−0.67–0.83	0.95 ± 0.29	<0.001	0.89–1.91

DOI: 10.7717/peerj.8718/table-1

Ingestion test

A Generalized linear mixed model (ANOVA of separate repeated measures, showed that there was no effect of litter and sex (random factors) at each type of crayfish and fish on number of ingested pieces (Table S2).

Fifteen days old

The proportion of neonates who ingested only soft crayfish, only hard crayfish, both, or neither, was different between both T. melanogaster populations (X ² = 12.08, df = 3, P = 0.009; Fig. 2). Soft crayfish was ingested for a mayor proportion of C neonates (61%) than NC neonates (46%; X² = 4.5, df = 1, P = 0.03). Similarly, hard crayfish was ingested in greater proportion by C neonates (60%), than NC neonates (39%; X ² = 7.99, df = 1, P = 0.005). Mean number of ingested pieces per clutch of soft crayfish was similar between both T. melanogaster populations (Student-t₅₈ = 0.90, P = 0.18; Fig. 3). The preference score for soft crayfish also was similar between C (0.16 ± 0.51) and NC individuals (0.04 ± 0.60; Student-t₅₈ = 1.15, P = 0.87).

A difference in hunger level may explain the higher ingestion of both soft and hard crayfish by neonates from C localities (Fig. 2). In days 1 and 7, neonates from C populations ingested more fish pieces than neonates from NC (day 1, 79% vs 41%; X² = 26.6, df = 1, P = 0.0001), but on day 7 this difference did not reach significance (day 7, 60% vs 46%; X² = 3.44, df = 1, P = 0.07). The higher hunger level may also be due to the fact that C neonates were 15.5% heavier (2.02 ± 1.05 g) and 8.5% longer (15.4 ± 2.15 cm SVL) than NC (1.75 ± 0.40 g; Student-t₃₄₈ = 2.82, P = 0.004; 14.2 ± 1.59 cm SVL; Student-t₃₉₁ = 4.38, P = 0.0001).

The ingestion frequency of soft crayfish did not change during the five days of the test for C (F_4,83 = 0.35, P = 0.84), and NC neonates (F_4,300 = 0.92, P = 0.45; Fig. 4). The same was true for the hard crayfish pieces in C neonates (F_4,83 = 1.66, P = 0.16), but NC neonates showed significant variation between days (F_4,83 = 2.66, P = 0.03; Fig. 4). In all cases, the frequency distributions for number of crayfish pieces ingested during the five days of presentation presented unimodal distribution patterns, with mode of zero ingestions (Fig. 5).

Ninety days old

The proportion of neonates that ate only soft crayfish, only hard crayfish, both, or neither was similar between 15 and 90 days old for both T. melanogaster populations (C, X² = 3.3, df = 3, P = 0.32; NC, X² = 5.9, df = 3, P = 0.15).

For NC neonates, the mean number of ingestions per clutch of soft crayfish did not change between 15 (0.76 ± 0.11) and 90 days old (0.61 ± 0.10; Dependent Student-t₄₀ = 1.65, P = 0.11; Fig. 3), nor did it change for hard crayfish (15 d old, 0.60 ± 0.08; 90 d old, 0.58 ± 0.08; Student-t₄₀ = 0.93, P = 0.32; Fig. 3).

For C neonates, the preference to ingest pieces of crayfish changed between 15 and 90 days old. At 90 days old, neonates ingested half the soft crayfish (0.54 ± 0.12 pieces per clutch) that 15-day-olds ingested (1.17 ± 0.20; Student-t₁₈ = 3.8, P = 0.002; Fig. 3), and the same decrease applied to hard crayfish (15 d old, 1.02 ± 0.20; 90 d old, 0.50 ± 0.18; Student-t₁₈ = 3.1, P = 0.003; Fig. 3).

Discussion

Manjarrez, Macías García & Drummond (2013) reported a crayfish-eating novelty of Mexican Black-bellied Gartersnake from eats crayfish only in a 3.0% of the area extensively sympatric and suggested that this novelty was a result of inherent preferences rather than prey availability. Our study suggests that T. melanogaster are born with an innate chemosensory and ingestion interest in both freshly-molted (soft) and unmolted (hard) crayfish, because are both recognized as prey and readily attacked. In general, laboratory tests do not support the prediction of greater innate chemosensory, ingestion and predatory behavioral response to only freshly-molted crayfish (C. montezumae) by C snakes T. melanogaster. In addition, C and NC neonates had a similar congenital chemosensory and ingestion response to both soft and hard crayfish, although the response and heritability to odors of soft crayfish were higher in C neonates. The tests also failed to demonstrate the existence of variation in techniques to capture and manipulate the soft crayfish between either T. melanogaster populations.

Due to the low response behavioral response to crayfish stimuli by the neonates T. melanogaster, the lack of clear results is understandable. Although T. melanogaster is considered as highly defensive and prone to attack at birth (Herzog & Burghardt, 1986; Herzog & Burghardt, 1988; Herzog, Bowers & Burghardt, 1989a; Herzog, Bowers & Burghardt, 1989b; Herzog, Bowers & Burghardt, 1992), this propensity was not observed in our study. For example, even from the crayfish-eating population, 97% of the snakes did not attack a swab. This is reasonable, because T. melanogaster is less responsive to chemical cues, even from high-preferred prey such as fish, than the generalist T. sirtalis (Ford & Burghardt, 1993). On the other hand, the criteria to consider T. melanogaster as highly defensive and prone to attack, is based on studies developed from the defensive and antidepredatory responses of this snake (Herzog & Burghardt, 1986; Herzog & Burghardt, 1988; Herzog, Bowers & Burghardt, 1989a; Herzog, Bowers & Burghardt, 1989b; Herzog, Bowers & Burghardt, 1992), a different behavior from the behavioral feeding response recorded of this study.

It is possible there is no increased food behavioral response to soft crayfish by neonates of the populations that eat crayfish. We know of no physical, environmental or genetic barrier between the rivers Lerma and Tula, which completely isolates both T. melanogaster populations. So it is likely that migration and gene flow can occur between both locations, especially for snakes and other reptiles in the trans-Mexican volcanic belt (Sunny et al., 2015; Sunny, González-Fernández & D’Addario, 2017; González-Fernández et al., 2018), where T. melanogaster is present. Crayfish ingestion in only some locations may possibly be explained by environmental differences between localities (Arnold, 1981a); for example, spatial–temporal availability of crayfish or differences in use of microhabitats by T. melanogaster. For example, crayfish availability influences the innate chemosensory responses to crayfish in the snake, Lyodytes alleni (Waters & Burghardt, 2013). However, a sampling during the rainy season (June to October) indicated a greater abundance of crayfish in the ponds where snakes do not eat crayfish than where they do eat crayfish (mean of three drags = 2.08 ± 3.03 crayfish / haul network of 2.8 m, mesh five mm, in 8 NC localities, versus 0.66 ± 1.2, in 9 C localities), although Student-t did not detect a significant difference (P = 0.23).

Another explanation could be that when snakes depend on the crayfish, the other usual prey are less available. However, this possibility has little support because Lozoya (1988) found no evidence of a temporary preference in the ingestion of fish by T. melanogaster in the different locations of the Lerma and Tula basins. He also reports that snakes that eat crayfish also feed on other types of prey in a higher percentage (65%) than the crayfish, which suggests that crayfish availability and ingestion is independent of the availability of other prey.

Alternatively, there may be increased food behavioral response to crayfish by C neonates, but it was not detected by our test protocol due that this food behavioral difference is small in magnitude and hardly detectable. This is possible when soft crayfish is not a profitable prey or have foraging disadvantages to the snake. The decreased ingestion of soft and hard crayfish at 90 days old by neonates in C populations suggests a learning process to recognize the crayfish as a non-profitable prey compared to fish with which neonates were fed from 15 to 90 days old. Also, the results of the predatory behavioral test, suggest that freshly-molted crayfish is not fully recognized as profitable prey by either T. melanogaster dietary morph. A small behavioral divergence in magnitude between both populations may be due to a recent origin, so that behavioral divergence for recognizing the crayfish has not yet been expanded into the surrounding geographical areas.

The ancestor of Thamnophis has existed in North America for at least 13 million years and its divergence as a genus occurred possibly two million years ago in Mexico (Alfaro & Arnold, 2001; De Queiroz, Lawson & Lemos-Espinal, 2002). The last geological events that altered river systems where T. melanogaster eat crayfish (Barbour, 1973) occurred two million years ago. For T. elegans, Arnold (1981a) estimated that food divergence for ingesting slugs occurred about 8,000 years ago assuming low intensity selection of 1.0%. The appearance of crayfish ingestion in T. melanogaster is possibly more recent than 8,000 years ago considering that high heritability values to freshly-molted crayfish (Table 1) cause a rapid evolution of food preference.

Neonates of both T. melanogaster populations responded similarly to both crayfish odors, although in C neonates the heritability was higher for freshly-molted crayfish. This suggests three possibilities: (1) Crayfish ingestion in T. melanogaster was an ancestral behavioral character, and only remains a homologous behavior in populations with similar ecological conditions to ancestral populations. This possibility can be explored with a study of the phylogeographic structure of both dietary crayfish populations, which helps explain the patterns of gene flow between populations (Foster & Cameron, 1996). (2) Neonates of both T. melanogaster populations are unable to recognize the difference between soft and hard crayfish. This possibility is supported by the results of this study and also for R. grahamii, R. septemvittata, and fish-eating species such as Nerodia sipedon (Burghardt, 1968). (3) The high and significant heritability to freshly-molted crayfish by T. melanogaster neonates supports the existence of microevolution toward the incorporation of a new prey in the diet through the divergence between the two T. melanogaster populations; a process that may be recent, as previously discussed.

After birth, experience with fish did influence ingestion response to both crayfish by C snakes T. melanogaster, but not for NC snakes. C 90-day-old neonates ingested less toward both soft and hard crayfish, conversely NC neonates continued to respond both crayfish (see Fig. 1).

Why did only C snakes at 90-days-old decrease ingestion of crayfish pieces compared to 15-days-old? This decrease can be the result of different processes. First, the reduction or disappearance of hunger level in T. melanogaster neonates. Regina grahamii and R. septemvittata have been shown that hunger level differences affects the chemosensory response to crayfish (Waters & Burghardt, 2005). In T. melanogaster there may be a similar phenomenon in response to an alternative prey (soft crayfish) because in specialist-species hunger may influence predator behavior (Ford & Burghardt, 1993) through the choice of suboptimal or alternative prey that snakes can use, such as crayfish (Arnold, 1993). Decreased ingestion of crayfish pieces may also be the result of ontogenetic change in food response to prey. Ontogenetic changes in prey preferences between neonates and juveniles for several Thamnophis species have been widely demonstrated and associated with dietary experience and maturation; however few studies report these changes in times as short as those of this study; 15-90 days old (reviewed in Rossman, Ford & Seigel, 1996). For T. melanogaster, the analysis of his diet through the Tula and Lerma basins has not detected an ontogenetic change in crayfish ingestion (Manjarrez, Macías García & Drummond, 2013). Other possibility is that innate chemosensory responses to prey can be modified by prey availability and that they do not necessarily result from maturation alone, as was demonstrated in the innate chemosensory response to crayfish by Lyodytes alleni (Waters & Burghardt, 2013). Newborn R. alleni feeds on dragonfly larvae to survive the critical period after birth when suitable sizes of crayfish, their preferred prey are not available. In this context, T. melanogaster would develop an innate chemosensory preference for a prey when it becomes finally available, months after the birth of the offspring. Finally, the degree to which dietary experience influences chemosensory response to prey stimulus depends on the species being studied (Burghardt, 1993). Species fed alternative prey types often develop chemosensory preferences for initially less-preferred prey (Waters & Burghardt, 2005).

Ingestion of crayfish by T. melanogaster could be the beginning of an evolutionary convergence with the two species that specialize in soft crayfish, R. grahamii and R. septemvittata (Hall, 1969; Mushinsky & Hebrard, 1977). In regards to Regina spp., Dwyer & Kaiser (1997) suggest that ingestion of soft prey (such as freshly-molted crayfish), is a precondition to specialization towards the ingestion of hard prey (such as non-molted crayfish). This hypothesis supports the idea of invasion of a novel food niche and convergence with Regina and Lyodytes in C T. melanogaster populations. In addition to this food convergence, C T. melanogaster populations have dentition and morphological head features that also subtly trend toward convergence with R. grahamii and R. septemvittata (Manjarrez, Macías García & Drummond, 2017).

Conclusions

In conclusion, our analyses weakly support the prediction innate chemosensory, ingestion and predatory behavioral response by C snakes T. melanogaster associated with the ingestion of the freshly-molted crayfish, although the response and heritability to odors of soft crayfish were higher in C neonates. The weak tendency of behavioural feeding tendencies in the direction of Regina and Lyodytes could imply that niche expansion did not gain its initial impetus through microevolution possibly due to a recent origin without, so that behavioral divergence has not yet been expanded into the surrounding geographical areas.

Because the low response level to crayfish stimuli questions our results, a future exploration is required that includes within the protocol other prey stimuli known to be ingested by T. melanogaster, such as fish or tadpoles.

Supplemental Information