How Behavior and Anatomy Affect Resource Use by Snakes

Home , Cantoria violacea, Homalopsidae, Hypsiscopus, Liodytes, Ophiophagy, Pseudoferania

A thesis submitted to the

Graduate School

of the University of Cincinnati

in partial fulfillment of the

requirements for the degree of

Master of Science

in the Department of Biological Sciences

of the College of Arts and Sciences

Noah D. Gripshover

B.S. Biology, University of Louisville, April 2017

Committee Chair: Bruce C. Jayne, Ph.D.

May 2020

ABSTRACT

The physical structure of animals and their environment are two obvious factors that can limit what animals do. However, the behaviors of animals and the choices they make can result in animals doing things that are only a small subset of what is physically possible. I used two systems to gain insights into the role of behavior in affecting resource use by snakes. First, I studied how varying the darkness, shape, and locations of artificial branches affected where snakes chose to go. Second, I studied two species of snakes to test how diet and feeding behavior were affected by the size of prey relative to the snakes’ anatomical constraints on prey size.

Variation in the environment can affect the mechanical demands of locomotion as well as influence where animals choose to go. Arboreal habitats facilitate studying path choice by animals because variable branch structure has known mechanical consequences and different branches create discrete choices. Recent studies found that arboreal snakes can use vision to select shapes and locations of destinations that mechanically facilitate bridging gaps. However, the extent to which the appearance of objects unrelated to biomechanical demands affects the choice of destinations remains poorly understood for most animal taxa including snakes. Hence,

I manipulated the intensity (black, gray or white), contrast, structure, and locations of destinations to test for their combined effects on perch choice during gap bridging of brown tree snakes and boa constrictors. The results presented herein provide a striking example of how visual cues unrelated to the physical structure of surfaces, such as contrast and intensity, can bias choice and, in some cases, supersede a preference for mechanically beneficial surfaces.

Snakes consume their prey whole. Consequently, variation in the anatomy of the trophic apparatus of snakes directly affects gape and limits maximal prey size. However, for the foraging ecology of snakes and other systems, scant data exist regarding how often maximal capacities are

ii taxed in nature. Hence, I quantified: 1) maximal gape, 2) the size of prey relative to maximal gape, and 3) how the type and relative size of prey affected behavior and prey handling times for two species of natricine snakes that primarily eat soft- (Regina septemvittata) or hard-shell

(Liodytes alleni) crayfish. Several of the differences between the crayfish-eating snakes including maximal gape, prey size, prey handling times and behavior resemble those between two phylogenetically distant species of homalopsid snakes that consume either hard- or soft-shell crabs. In both groups of crustacean-eating snakes, the decreased prey capture success in captivity and the rare consumption of relatively large hard-shell crustaceans in the field suggest that the ability to capture this type of prey constrains prey size more commonly than maximal gape.

Regina septemvittata was superior to the other species based on new metrics of potential feeding performance that integrated snake size and gape with the relative mass of prey.

iii

ACKNOWLEDMENTS

First and foremost, I would like to thank my advisor, Dr. Bruce Jayne, for the support, opportunities, and advice during my three years at the University of Cincinnati. I have learned so much about how to conduct high quality research in both laboratory and field studies.

I greatly appreciate all your support in my efforts to collect snakes in Kentucky and Florida. I will always remember biking through the everglades to collect pythons and wading through the mud and water hyacinths looking for Liodytes.

I thank my committee members, Dr. Takuya Konishi and Dr. Daniel Buchholz, for their insightful comments on my thesis and discussions during our committee meetings. Thank you for taking the time to meet with me and help me with my projects.

I thank all of the undergraduate students that have assisted me with my research and animal care. I especially thank Miranda Rodgers, Curran Bobbitt, Trey Saunders, and Brittany

Fithen who assisted with the gap bridging experiments and Lily Bischoff, Nicole Addison, and

Trey Saunders who assisted with the collection and experiments of the crayfish-eating snakes. I have enjoyed working with all of you and you have made my time here at the University of

Cincinnati much more enjoyable and fulfilling.

Last but certainly not least, I would like to thank my family for their continued support throughout my graduate career. Thank you for teaching me that hard work pays off and allowing me to follow my passion in studying the animals that I was never allowed to bring into the house.

TABLE OF CONTENTS

Abstract……………………………………………………………………………………..……ii

Acknowledgments………………………………………………………………………….……v

List of Tables and Figures……………………………………………………………….....…viii

Chapter 1: Visual Contrast and Intensity Affect Perch Choice of Brown Tree Snakes (Boiga irregularis) and Boa constrictors (Boa constrictor)………………………………………...….…1

Abstract…………………………………………………………………………...……….2

Introduction………………………………………………………………………………..3

Materials and Methods…………………………………………………………………….5

Results……………………………………………………………………………………10

Discussion………………………………………………………………………………..11

References……………………………………………………………………………….20

Chapter 2: Crayfish Eating in Snakes: A Model System for Testing How Anatomy, Behavior and Performance Affect Foraging Ecology……………………………………………………...31

Abstract…………………………………………………………………………………..32

Introduction………………………………………………………………………………33

Materials and Methods…………………………………………………………………...36

Results……………………………………………………………………………………40

Discussion………………………………………………………………………………..44

References………………………………………………………………………………..53

Supplemental Information.………………………………………………………………66

vii

LIST OF TABLES AND FIGURES

Chapter 1

Table 1.1. Mean + s.e.m. preference for destination perches……………………………………27

Figure 1.1. Overhead view of the experimental apparatus………………………………………29

Figure 2.1. Schematic views of the destinations showing the manipulations of destination intensity and structure……………………………………………………………………………30

Chapter 2

Figure 2.1. Simplified phylogeny of Natricine and Homalopsid snakes based on Figueroa et al.

(2016)…………………………………………………………………………………………….59

Figure 2.2. Scaling relationships for morphological data………………………………………..60

Figure 2.3. Computed tomography scans of R. septemvittata and L. alleni……………………..61

Figure 2.4. Relationships between the absolute and relative sizes of prey………………………62

Figure 2.5. Prey handling behaviors and prey handling times…………………………………...63

Figure 2.6. Constraints of gape on feeding performance for four species of crustacean-eating specialists………………………………………………………………………………………...64

Figure 2.7. Three-dimensional maximal feeding performance spaces for four species of crustacean-eating snakes…………………………………………………………………………65

Table S2.1. Regression analyses of morphology and handling time……………………….……66

Table S2.2. Summary of prey handling behaviors during laboratory trials..…………………….67

viii

Table S2.3. Mean values of prey handling times during laboratory trials.………………………68

Table S2.4. Final multiple regression models for total handling time and the occurrence of some behaviors....………………………………………………………………………………………69

Table S2.5. Univariate regressions for behavior and handling times……………………………70

Table S2.6. ANCOVA results for morphology and behaviors comparing the two species and sexes within a species……………………………………………………………………………72

CHAPTER ONE

Visual Contrast and Intensity Affect Perch Choice of Brown Tree Snakes (Boiga

irregularis) and Boa Constrictors (Boa constrictor)

Noah D. Gripshover and Bruce C. Jayne*

Department of Biological Sciences

University of Cincinnati

Cincinnati, OH, 45221 USA

*The work presented here, as well as the preparation and completion of the narrative for this manuscript, was principally composed by Noah D. Gripshover with the assistance of Bruce C.

Jayne.

The manuscript can be viewed in its published form, in the journal Zoology:

Gripshover, N. D., and B. C. Jayne. 2020. Visual Contrast and Intensity Affect Perch Choice of

Brown Tree Snakes (Boiga irregularis) and Boa Constrictors (Boa constrictor). Zoology 139:DOI

10.1016/y.zool.2020.125744

ABSTRACT Habitat structure can affect animal movement both by affecting the mechanical demands of locomotion and by influencing where animals choose to go. Arboreal habitats facilitate studying path choice by animals because variation in branch structure has known mechanical consequences and different branches create discrete choices. Recent laboratory studies have found that arboreal snakes can use vision to select shapes and locations of destinations that mechanically facilitate bridging gaps. However, the extent to which the appearance of objects unrelated to biomechanical demands affects the choice of destinations remains poorly understood for most animal taxa including snakes. Hence, we manipulated the intensity (black, gray or white), contrast, structure, and locations of destinations to test for their combined effects on perch choice during gap bridging of brown tree snakes and boa constrictors. For a white background and a given perch structure and location, both species had significant preferences for darker perches. The preference for darker destinations was strong enough to override or reduce some preferences for biomechanically advantageous destinations such as those having secondary branches or being located closer or along a straighter trajectory. These results provide a striking example of how visual cues unrelated to the physical structure of surfaces, such as contrast and intensity, can bias choice and, in some cases, supersede a preference for mechanically beneficial surfaces. Because these two species are so phylogenetically distant, some of their similar preferences suggest a sensory bias that may be widespread in snakes. The manipulation of surface color may facilitate management of invasive species such as the brown tree snakes by enhancing the efficiency of traps or making certain objects less attractive to them.

INTRODUCTION

Habitat structure can affect animal movement both by affecting the mechanical demands of locomotion and by influencing where animals choose to go. Animals as diverse as snails

(Atkinson 2003), spiders (Tarsitano 2006), ants (Graham and Collett 2006), fish (Bisazza et al.

1997), frogs (Munteanu et al. 2016), lizards (Lustig et al. 2013), and cats (Poucet et al. 1983) often plan routes rather than just moving randomly through the environment. Once the capacity for nonrandom movement is established, a deeper understanding of this capacity can be obtained by determining the cues that animals use to choose different routes. The ability of animals to plan and make a detour around an impassible obstacle has been an important and dominant method for gaining insights into path choice (Kabadayi et al. 2018). However, besides the dichotomy of passable versus impassable, a wide variety of physical structures in the environment creates a continuum in variation that can affect the speed and ease of animal locomotion as well as providing the cues that could be used to make choices.

Arboreal habitats are a model system for studying path choice by animals because variation in branch structure has known mechanical consequences, and different branches create discrete choices. The mechanical effects of habitat structure on movement can be affected by variation in behavior and morphology within a body plan as well as among different body plans.

For example, the decrease in the running speeds of anole lizards associated with decreased branch diameter is less severe for species with short compared to long limbs (Losos and Sinervo

1989), whereas decreased branch diameter commonly increases the locomotor performance of arboreal snakes (Jayne et al. 2015). The diameter of branches upon which anoles are found is also positively correlated with the relative limb length of different species (Schoener 1968,

Pounds 1988, Losos 1990), but species with different limb lengths have a remarkably uniform

3 preference for the biomechanically beneficial thicker branch when they encounter a branching point as they are running (Mattingly and Jayne 2005). This emphasizes the importance of using experimental approaches to directly test for the cues that affect path choice rather than relying solely on microhabitat preference as well as a likely primacy of mechanical attributes for predicting the paths that animals choose.

The paths that animals choose while they are moving can also be affected by the nature of the task. For example, the preference of snakes for larger rather than smaller diameter cylinders when bridging gaps seems counterintuitive at first glance because the locomotion of snakes on cylinders is impeded by increased diameter (Mansfield and Jayne 2011, Jayne et al. 2014).

However, a larger cylinder diameter increases target size, and this facilitates reaching tasks by requiring less precise motor control (Jayne et al. 2014). Thus, while bridging gaps, the preference of snakes for choosing destinations with a structure that facilitates making first contact may override factors that affect the subsequent locomotion on the surface.

In addition to possible mechanical consequences of choosing different destinations, the common preference of large objects by arboreal animals could be the result of larger objects simply being visually more conspicuous. Thus, a lingering problem is that many of the attributes that affect the mechanics of moving on and between branches, such as larger branch diameter, secondary branches or greater width, also seem likely to make a destination visually more conspicuous (Mansfield and Jayne 2011). Modifying the color or intensity of objects provides a convenient experimental method for altering how visually conspicuous objects are without an attendant effect on the mechanical attributes that would affect the ease of locomotion.

In this study we varied the intensity (white, gray, and black) of objects as well as the distance, location, and structure of artificial branches and used choice tests primarily to

4 determine how visual cues unrelated to the mechanical demands for animal movement affect the choice of destinations. We used snakes bridging gaps for our model system as the mechanical demands of this task and many effects of perch structure on gap bridging performance, locomotion, and perch choice are already well understood, including for our study species

(Byrnes and Jayne 2012, Jayne et al. 2014, Jayne et al. 2015, Mauro and Jayne 2016). We used two phylogenetically distant species (brown tree snakes and boa constrictors) to gain insights into the generality of any possible preferences. First, we tested whether these snakes preferred a particular destination intensity (i.e. shading was black, gray, or white) when all other factors were constant. Second, we tested whether the background intensity (black or white) affected the preference for the intensity of the destination. Finally, we investigated the interactions between different destination intensities, variation in perch structure, and perch location that affected the ease of bridging gaps. For a given perch structure, we expected the snakes to prefer intensities of destinations that made them more visually conspicuous by increasing their contrast relative to the background. We also expected that a preference for biomechanically advantageous destinations would supersede any preferences for intensity or contrast.

MATERIALS AND METHODS

Experimental subjects

Our two study species are phylogenetically very distinct. Brown tree snakes (Boiga irregularis) belong to the family Colubridae, which is nested deeply within a large clade of “advanced” snakes (infraorder Caenophidea), whereas boa constrictors (Boa constrictor) belong to the family

Boidae which is basal to all of the advanced snakes (Figueroa et al. 2016). The clade arising

5 from the most recent common ancestor of Boa and Boiga contains all but two of the 22

Alethinophidian families, and this clade contains more than 87% of the more than 3,500 of known extant species (Figueroa et al. 2016). The family Colubridae, to which Boiga belongs, arose more than 50 million years after its most recent common ancestor with Boidae (Sanders et al. 2010).

Brown tree snakes are highly arboreal (Rodda et al. 1999a), and they have a slender body and anatomical specializations in axial musculature (Hoefer and Jayne 2013) that have repeatedly evolved within several clades of arboreal caenophidian snakes (Jayne 1982).

Although boa constrictors lack any obvious morphological specializations for arboreality (Hoefer and Jayne 2013), they commonly climb trees, especially as juveniles (Greene 1983). Both of these species have strong nocturnal tendencies, but they are not exclusively nocturnal (Greene

1983, Rodda et al. 1999a).

The brown tree snakes that we used (n = 16) were collected from Guam in 2010 and

2011, whereas the boa constrictors (n = 14) were born and raised in captivity. The experiments with the brown tree snakes and boa constrictors were performed during 2017 and 2018–2019, respectively. The brown tree snakes (3 females and 13 males) had average values of snout-vent length (SVL) of 159 cm (range = 119–179 cm) and mass of 808 g (range = 540–1020 g), whereas the boa constrictors (9 females and 5 males) had average values of SVL of 133 cm

(range = 126–143 cm) and mass of 1154 g (range = 900–1555 g). We housed the snakes individually in cages with incandescent bulbs that allowed regulation of body temperature to 25–

33oC. We did not test any snakes within 5 days of being fed or when their eyes were cloudy as a result of ecdysis. The care of the animals and experimental procedures were approved by the

Institutional Animal Care and Use Committee at the University of Cincinnati (protocol number

07-01-08-01), and the brown tree snakes were captured and imported with permits from the U.S.

Fish and Wildlife Service (MA214902; MA3500A-0).

Experimental apparatus

The test apparatus had a cylindrical (length = 80 cm, diameter = 5 cm) starting perch and two destination perches in a horizontal plane 150 cm above the ground (Fig. 1.1). With the exception of two treatments (Fig. 1.1C; Table 1.1 tmt 19, 20), we arranged the perches in a Y- shape so that the trajectories to the alternative destinations from the end of the starting perch had a yaw angle of 45° (Fig. 1.1A,B; Table 1.1). In treatments 19 and 20 one of the destinations was along a straight trajectory, whereas the other required a 90° turn (Fig. 1.1C; Table 1.1).

With the exception of two treatments (Fig. 1.1B; Table 1.1 tmt 17, 18) the standard distances of the gaps between the ends of the starting perch and the destinations were 47 and 45 cm for the brown tree snakes and boa constrictors, respectively. These standard gap distances are approximately 70% of the maximal relative gap distance (%SVL) that our study species can bridge when following a straight horizontal trajectory (Jayne and Riley 2007, Hoefer and Jayne

2013). To test the effects of gap distance in treatments 17 and 18, one of the destinations was 10 cm closer than the alternative destination with the standard gap distance.

To create a uniform background, all of the walls of the 3.2 x 2.6 m and 2.4 m high experimental area were covered by canvas cloth, and the inner surface of the chamber wall behind destination perches was also covered either with overlapping 107 cm wide strips of white poster paper (Hewlett Packard durable banner paper with DuPont Tyvek) oriented vertically or a black cloth (only tmt 4, 7). The base and vertical pipe to which the destination perches were attached were placed behind the background material which had 5 cm diameter holes so that

7 each destination perch extended into the arena while the vertical support was hidden. Fluorescent light fixtures (with a total of 1828 W 3500K bulbs), suspended 45 cm below the ceiling, provided illumination, and the brightness of the incident light at the edge of the gap was approximately 600 lux.

For ten destinations, we varied the presence of pegs and whether the pegs, the sides, and the end of the large cylinders were black, gray, or white (Fig. 1.2). The pegs always had a diameter of 4 mm and a length of 10 cm. Whenever pegs were present, they were oriented 45° relative to a horizontal plane (Fig. 1.2). Each destination cylinder was covered with white, black, or gray Shurtape P-665® gaffer’s tape (Shurtape Technologies, LLC; Hickory, NC USA 28602) to provide a uniform color and a texture that reduced the slipping of snakes. We painted the pegs on the destination perches black, white, or gray. To facilitate locomotion and gap bridging, the starting perch was also covered with gray gaffer’s tape, and it had two parallel rows of pegs placed at 5 cm intervals along the length of the perch (Fig. 1.1).

Experimental procedures for choice tests

To reduce possible confounding effects of prior experience, we divided individuals of each species into two batches, and each batch of snakes within a particular species experienced the treatments in a different randomized order. For different treatments, we also randomized the order of testing individuals within a batch. Prior to testing, we placed the snakes in a heated container until their body temperatures were 29–31oC, which is within range of their field active body temperatures (Montgomery and Rand 1978, Anderson et al. 2005). At the beginning of each trial, we placed the snake on the end of the starting perch farthest from the gap, and we orientated the snake towards the destinations while standing directly behind the starting perch. If the snake did not move within 10 s, we gently tapped or lifted the snake’s tail to prompt it to

8 move. Each trial concluded when the snake crawled onto a destination. After testing each snake three times in rapid succession for a particular treatment, the snakes rested for at least two hours before another bout of three trials in which the left-right position of the same pair of perches was reversed.

Previous studies found that possible olfactory cues from snakes repetitively crawling on the destinations did not influence choice with a gap distance > 40 cm (Mansfield and Jayne 2011,

Hoefer and Jayne 2013). However, as an added precaution to minimize a possible effect of olfactory cues, we cleaned each destination perch with paper towels dampened with 70% ethanol after each snake completed its set of three trials. Furthermore, we replaced the tape covering the destinations after each batch of snakes had completed three trials for a particular treatment.

Data collection and analysis

For each individual and each treatment, we calculated preference as the percentage of the six trials when a snake selected a particular destination. For each species we then calculated the mean value of preference (Pref1) for a particular treatment (n = the number of individuals). We then used a one-tailed t test to determine if Pref1 for the destination that was chosen most often was significantly (p < 0.05) greater than the value expected for randomly choosing between the two alternative perches (50%). To facilitate evaluating the effects of making multiple comparisons, as suggested by (Moran 2003), we provide exact, uncorrected p values (Table 1.1).

However, we also performed a table-wide sequential Bonferroni correction (Rice 1989) that divided 0.05 by the rank order of the p values for all 40 comparisons.

RESULTS

When both destinations were black, the snakes of both species preferred the destination with attributes that enhanced the ease of crossing a gap. For example, for equal gap distance and yaw angles of 45°, the snakes had a significant preference for the destination with pegs rather than the one without pegs (Table 1.1, tmt 6). For yaw angles of 45°, the snakes also had a significant preference for the destination with a gap distance that was shorter by 10 cm (Fig.

1.1B; Table 1.1, tmt 17). Finally, the snakes also had a highly significant preference for the destination along a straight trajectory rather than the alternative with 90° yaw angle (Fig. 1.1C;

Table 1.1, tmt 19).

For the three treatments with a white background and for which each destination had pegs, a yaw angle of 45°, and a uniform intensity, snakes of both species had a significant preference for the darker destination (Table 1.1, tmt 1–3). When one destination was black and the other was gray (Table 1.1, tmt 2) the preference for the darker destination was not as strong as when one of the destinations was white (Table 1.1, tmt 1,3).

In four of the remaining eight treatments with a white background and both destinations having pegs and a yaw angle of 45° (Table 1.1, tmt 9–16), both species had significant and similar preferences. These uniform preferences included: 1) a black rather than a white cylinder

(tmt 9), 2) an all black destination rather than one with only a black end (tmt 11), 3) black pegs rather than white pegs when both cylinders were white (tmt 12), and 4) an all-black destination rather than one with only a black end and pegs (tmt 16). Thus, for destinations with combinations of black and white, the snakes commonly preferred the destination that had a greater amount of black.

When the background was black, the brown tree snakes unexpectedly continued to have a significant preference for an all-black rather than an all-white destination (Table 1.1, tmt 4).

However, the boa constrictors had a significant preference for the all-white rather than the all- black destination when the background was black.

In several cases the preference for a darker destination with a white background was so strong that it overrode or lessened the preference for a mechanically advantageous destination.

For example, the brown tree snakes retained a significant preference for a black destination even when it was farther away than a white destination (Table 1.1, tmt 18), and the boa constrictors in this treatment no longer preferred the closer destination when it was white. In the presence of a black destination without pegs, neither of the species continued to prefer the mechanically advantageous white destination with pegs (Table 1.1, tmt 5). Furthermore, in both species the presence of a black destination with a yaw angle of 90°, decreased the preference for the mechanically advantageous white destination that was straight ahead (tmt 20).

DISCUSSION

The ability of diverse animals to travel along non-random paths is widespread (Kabadayi et al. 2018), and it implies some ability to sense cues in the environment to facilitate the orientation of movements. However, the visual cues involved in path choice are minimally understood for most animals compared to the role of vision in systems such as mate choice, the interactions between animals, and interactions between pollinators and plants (van der Kooi et al.

2019). More specifically, understanding the role of vision for path choice of snakes remains rudimentary although numerous studies of snakes have documented the role of olfactory

(Burghardt 1967) and visual cues (Drummond 1985) for making choices during predatory behavior, especially for species with specialized infrared receptors (Buning 1983). Given that size, distance, orientation, color, and contrast can convey important information in many other contexts such as species and sex recognition (Agrawal and Dickinson 2019), dominance and territorial interactions (Chunco et al. 2007), and feeding of diverse animals (Ewert 1982, van der

Kooi et al. 2019), one could reasonably expect some of these same visual cues to be important for path choice, especially if they are correlated to the biomechanical demands for moving.

Snakes probably evolved from a burrowing ancestor in which a functional eye was nearly lost but subsequently re-elaborated as snakes diversified (Caprette et al. 2004), and this has led to a common assumption that the visual acuity of snakes is poor compared to that of lizards.

However, the limited experimental data indicate that the visual acuity of snakes is reasonably good considering the size of their eye (Rumpff 1979, Baker et al. 2007) although their ability to perceive colors is probably reduced compared to lizards (Simoes et al. 2015).

Visual cues, such as distance, location, size, and shape, can facilitate choosing the destination that is most mechanically beneficial when bridging gaps. For example, snakes bridging gaps prefer locations that are closer and along straighter trajectories, and these preferences reduce torques that tend to make the body of the snake buckle or topple sideways

(Mansfield and Jayne 2011, Hoefer and Jayne 2013). Snakes also prefer wider destination perches (Mansfield and Jayne 2011, Jayne et al. 2014), which create larger targets and thus require less precise motor control (Jayne et al. 2014). The preference of brown tree snakes for destinations with a V-shaped rather than an inverted V-shape formed by a pair of pegs at the end of the perch reduces the chance slipping off the side of the destination (Jayne et al. 2014). Brown tree snakes may also have a short-term ability to learn to avoid small diameter perches that are

12 extremely compliant compared to thick perches with negligible bending (Mauro and Jayne

2016).

Some additional preferences during gap bridging suggest a primacy for structural variation where snakes make first contact rather than structural variation that enhances the crawling of snakes once they are on the destination. For example, the preference for a perch with a V-shaped pair of pegs rather than a pegless cylinder decreases as the pegs are farther from the gap (Mansfield and Jayne 2011). The preference for a larger diameter pegless cylinder and the lack of a consistent preference for pegs along the top-center of the destination (Mansfield and

Jayne 2011, Jayne et al. 2014) further suggest the primacy of factors that facilitate making first contact without falling off.

A possible confounding factor regarding preferences for mechanically beneficial objects is that increased apparent size (as a result of being closer), increased cylinder diameter, the presence of pegs, and increased peg length also all probably make a destination visually more conspicuous. For example, in common with snakes bridging gaps, arboreal lizards running on connected networks of perches also prefer branches with larger diameter (Mattingly and Jayne

2005). One possible explanation for this preference of lizards is that their running speeds are enhanced by larger cylinder diameter (Losos and Irschick 1996), but of course a larger cylinder is probably also visually more conspicuous. Hence, manipulating the contrast of destinations can facilitate disentangling some of the potential effects of being visually more conspicuous from those that affect the mechanical demands.

When the structural attributes of both destinations were the same, the following four observed trends in preferences were consistent with both of our study species preferring to go to the object with higher contrast relative to the background. First, when the background was white

13 and each destination had uniform intensity, both species preferred the black rather than the white, the gray rather than the white, and the black rather than the gray destination. Second, the preference for the darker destination weakened when the alternatives were black and gray compared to black and white. Third, when the snakes had a significant preference between destinations with variable amounts of black, they preferred the greater amount of black. Fourth, the boa constrictors preferred the white rather than the black destination when the background was black.

An unexpected result was that, unlike the boa constrictors, the brown tree snakes preferred the black destination even when the background was black. This suggests that a lower intensity by itself may be sufficient to elicit a preference in some circumstances and that high contrast between an object and the background is not necessary to elicit a preference in brown tree snakes. Although this preference for the black rather than the white perch against a black background was significant, it was not as strong as for the black versus the white perch in front of a white background. Furthermore, brown tree snakes had a stronger preference for black versus white rather than for either black versus gray or for gray versus white perches.

Consequently, for brown tree snakes, increased contrast also may increase the strength of preferences even though contrast by itself may not be sufficient to elicit a preference.

Additional results suggest that the brown tree snakes had a stronger sensory bias for dark objects than the boa constrictors. First, the brown tree snakes had a significant preference for certain dark objects rather than a more mechanically beneficial white destination (tmt 18), whereas boa constrictors lacked a significant preference for either of these alternatives. Second, boa constrictors were not as responsive as brown tree snakes to variable proportions of black and white within a destination (tmt 10, 13).

As with previously documented preferences for mechanical factors, the strength of many of the observed preferences was proportional to the magnitude of the differences between destinations. For example, the strength of preference increases when the disparity between the destinations increases for: 1) lengths of pegs, 2) cylinder diameter, 3) gap distance, 4) location of the first pair of pegs, and 5) turning angle within a horizontal plane (Hoefer and Jayne, 2013;

Jayne et al., 2014; Mansfield and Jayne, 2011). Similarly, the snakes in our study had some stronger preferences for objects with uniform intensity and greater contrast as well as for destinations with greater amounts of amount of black.

Additional insights regarding the strength of preferences and whether preference for intensity trumps preference for mechanical factors, or vice versa, can be gained by further examining certain interactive effects. For example, when the destination with pegs was white and the pegless destination was black, the preference for a destination with pegs was eliminated (tmt

5). By contrast, the preference for going straight to a white perch rather than to a black perch requiring a right angle turn shows how a large mechanical disparity can supersede the bias for a dark, high-contrast object, but the preference for going straight rather than turning was less when the straight-ahead destination was white rather than black (tmt 20 vs. 19). Thus, an effect of intensity or high contrast also persisted, but it was not sufficient to completely override a stronger preference for the large mechanical benefits of the alternative. Because a right angle turns maximizes the mechanical demands of bridging gaps in horizontal plane (Byrnes and Jayne

2012), we would predict that at some point for turns less severe than 90° the snake would actually favor turning towards a darker perch rather than going straight ahead to a lighter destination.

The above precedent for a large difference in mechanical demands diminishing the importance of object intensity leads to some additional predictions. Our experiments only varied gap distance by 10 cm, and our standardized gap distances were only approximately 70% of the expected maximal gap bridging ability of our subjects. Hence, we would expect that the preference of both species for the closer object rather than the darker object could be restored either by using gap distances with much larger disparities or that are closer to the limit of gap bridging ability.

Although a more conspicuous object need not necessarily be an attractant to an animal, dark objects with high contrast were very strong attractants for both of our phylogenetically distant study species (Figueroa et al. 2016), and this may result from some combination of the following factors. Increasing contrast is a very general mechanism for eliciting more attention from the visual systems of diverse animal species (Ingle et al. 1978), and many of the preferences that we observed could result from this widespread property of animal visual systems that extends well beyond snakes and even beyond vertebrates. However, dark objects may also have additional importance to snakes for some of the following reasons. For example, snakes evolved from a burrowing ancestor (Caprette et al. 2004), and snakes are often quite secretive and very adept at finding and hiding under objects and in crevices or burrows (Greene

1997). Both of our study species are not exclusively nocturnal, but they do have strong nocturnal tendencies (Greene 1983, Rodda et al. 1999a), as do many other species of snakes. Hence, one interesting area of future work would be testing whether or not nocturnal species are more likely to be attracted to dark objects and dark places under well-lit conditions than diurnal snakes. It would also be interesting to determine whether or not a preference for either darker or contrasting objects would vary with different levels of light.

The time available for the snakes to make a decision in our experiments appears much longer than in some other systems. For example, the entire escape response and the determination of the escape trajectory in fishes commonly occurs in only a fraction of a second and without any pausing (Foreman and Eaton 1993). Tests of route choice of anole lizards in the laboratory often lasted less than 3 s from the release of the lizard until it reached a bifurcation after running 1 m, and if any pauses occurred, they were usually only a fraction of a second

(Mattingly and Jayne 2005). Although we did not systematically quantify the time of trials in our experiments, many trials were completed within 10 – 60 s, some lasted a few minutes, and pauses lasting more than a second were common. We assume that our presence behind the snake and touching it provided some averse stimulus. However, subjectively both species usually appeared to be moving well below their maximal speeds of arboreal locomotion (Jayne et al.

2015). Hence, our study system appears to be very far from an all-or-none, maximal speed escape response. Perhaps future work detailing the time course and path traveled by the head could provide more insights into whether increased contrast decreases the time course of events and increases the directness of the path traveled.

An additional interesting but unresolved issue is why boa constrictors preferred the white perch and brown tree snakes preferred the black perch when the background was dark. Testing more species could quickly resolve whether this observed difference reflects some fundamental difference between Henophidian and Caenophidian snakes. Both of our study species have strong nocturnal tendencies (Greene 1983, Rodda et al. 1999a) and vertically elliptic pupils, but nuances such as the exact amounts of time they spend in the dark or bright light are not well known. Despite young boa constrictors often occurring in trees (Greene 1983), boa constrictors are not considered an arboreal specialist, and they are rather stout and have relatively short tails

17 compared to highly arboreal relatives in the genus Corallus (Pizzatto et al. 2007). By contrast, brown tree snakes are considered arboreal specialists based on their ecology (Rodda et al. 1999a) and anatomical specializations including a slender build and muscles with long tendons, both of which have evolved convergently many times in arboreal Caenophidian snakes (Jayne 1982).

Nonetheless, it is presently difficult to imagine how any of these differences between our study species might be causally related to the relatively few differences that we did observe in preferences for different destinations.

Although we expected that the intensity and contrast of destinations could influence where snakes choose to go, an overriding preference for intensity leading to a mechanically maladaptive choice was not anticipated. However, apparently maladaptive preferences are a recurrent finding for studies that have investigated the existence and consequences of sensory bias. Two classic examples of this are female preferences for mate choice in certain fish (Basolo

1990) and frogs (Ryan et al. 1990). To the best of our knowledge, our results are the first to show how a sensory bias for color or contrast can result in counterintuitive choices that could impede movement (by increasing slipping or requiring greater muscular force to oppose greater buckling forces associated with larger gap distances).

The conspicuous sensory bias of brown tree snakes for darker objects may have useful implications for the management of this highly invasive species, which has had devastating ecological and economic consequences for the island of Guam (Rodda et al. 1999b). For example, trapping this species might be facilitated if traps and objects providing access to the traps are dark in order to attract the snakes. Alternatively, light colors might enhance the effectiveness of structures intended to be barriers or reduce the attractiveness of infrastructure associated with nest boxes and electrical systems.

ACKNOWLEDGEMENTS

Collection of snakes in Guam was possible thanks to the assistance of G. Rodda and B. Lardner of the USGS Brown Treesnake project, J. Schwagerl and the staff of the Guam National Wildlife

Refuge, and C. Clark and the staff of USDA–APHIS Wildlife Services Guam. C. Bobbitt, T.

Saunders, and B. Fithen assisted with the experiments, and B. Fetsko provided helpful comments on a draft of the manuscript. This work was partially supported by a grant from the National

Science Foundation [IOS 0843197 to B.C.J.].

REFERENCES

Agrawal, S., and M. H. Dickinson. 2019. The effects of target contrast on Drosophila courtship.

Journal of Experimental Biology 222:jeb203414 doi: 203410.201242/jeb.203414.

Anderson, N. L., T. E. Hetherington, B. Coupe, G. Perry, J. B. Williams, and J. Lehman. 2005.

Thermoregulation in a nocturnal, tropical, arboreal snake. Journal of Herpetology 39:82-

90.

Atkinson, J. W. 2003. Foraging strategy switch in detour behavior of the land snail Anguispira

alternata (Say). Invertebrate Biology 122:326-333.

Baker, R. A., T. J. Gawne, M. S. Loop, and S. Pullman. 2007. Visual acuity of the midland

banded water snake estimated from evoked telencephalic potentials. Journal of

Comparative Physiology A 193:865-870.

Basolo, A. L. 1990. Female preference predates the evolution of the sword in swordtail fish.

Science 250:808-810.

Bisazza, A., R. Pignatti, and G. Vallortigara. 1997. Laterality in detour behaviour: interspecific

variation in poeciliid fish. Animal Behaviour 54:1273-1281.

Buning, T. D. 1983. Thermal sensitivity as a specialization for prey capture and feeding in

snakes. American Zoologist 23:363-375.

Burghardt, G. M. 1967. Chemical-cue preference of experienced snakes: comparative aspects.

Science 157:718-721.

Byrnes, G., and B. C. Jayne. 2012. Three-dimensional trajectories affect gap bridging

performance and behavior of brown tree snakes (Boiga irregularis). Journal of

Experimental Biology 215:2611-2260.

Caprette, C. L., M. S. Y. Lee, R. Shine, and A. Mokany, J. F. Downhower. 2004. The origin of

snakes (Serpentes) as seen through eye anatomy. Biological Journal of the Linnean

Society 81:469-482.

Chunco, A. J., J. S. McKinnon, and M. R. Servedio. 2007. Microhabitat variation and sexual

selection can maintain male color polymorphisms. Evolution 61:2504-2515.

Drummond, H. 1985. The role of vision in the predatory behaviour of natricine snakes. Animal

Behaviour 33:206-215.

Ewert, J. P. 1982. Neuronal basis of configurational prey selection in the common toad, in: Ingle,

D. J., M. A Goodale, and R. J. W. Mansfield (Eds.), Analysis of Visual Behavior. MIT

Press, Cambridge, MA, pp. 7-45.

Figueroa, A., A. D. McKelvy, L. L. Grismer, C. D. Bell, and S. P. Lailvaux. 2016. A species-

level phylogeny of extant snakes with description of a new colubrid subfamily and genus.

Plos One 11:DOI: 10.1371/journal.pone.0161070.

Foreman, M. B., and R. C. Eaton. 1993. The direction change concept for reticulospinal control

of goldfish escape. Journal of Neuroscience 13:4101-4113.

Graham, P., and T. S. Collett. 2006. Bi-directional route learning in wood ants. Journal of

Experimental Biology 209:3677-3684.

Greene, H. W. 1983. Boa constrictor (Boa, Bequer, Boa constrictor) in: Janzen, D. H. (Ed.),

Costa Rican Natural History. University of Chicago Press, Chicago, pp. 380-382.

Greene, H. W. 1997. Snakes the Evolution of Mystery in Nature. University of California Press,

Berkeley.

Hoefer, K. M., and B. C. Jayne. 2013. Three-dimensional locations of destinations have species-

dependent effects on the choice of paths and the gap-bridging performance of arboreal

snakes. Journal of Experimental Zoology A 319:124-137.

Ingle, D. J., M. A. Goodale, and R. J. W. Mansfield. 1978. Analysis of Visual Behavior. MIT

Press, Cambridge, Massachusetts, p. 834.

Jayne, B. C. 1982. Comparative morphology of the semispinalis-spinalis muscle of snakes and

correlations with locomotion and constriction. Journal of Morphology 172:83-96.

Jayne, B. C., A. M. Lehmkuhl, and M. A. Riley. 2014. Hit or miss: branch structure affects perch

choice, behaviour, distance and accuracy of brown tree snakes bridging gaps. Animal

Behaviour 88:233-241.

Jayne, B. C., S. J. Newman, M. M. Zentkovich, and H. M. Berns. 2015. Why arboreal snakes

should not be cylindrical: body shape, incline and surface roughness have interactive

effects on locomotion. Journal of Experimental Biology 218:3978-3986.

Jayne, B. C., and M. A. Riley. 2007. Scaling of the axial morphology and gap-bridging ability of

the brown tree snake (Boiga irregularis). Journal of Experimental Biology 210:1148-

1160.

Kabadayi, C., K. Bobrowicz, and M. Osvath. 2018. The detour paradigm in animal cognition.

Animal Cognition 21:21-35.

Losos, J. B. 1990. Ecomorphology, performance capability, and scaling of West Indian Anolis

lizards: An evolutionary analysis. Ecological Monographs 60:369-388.

Losos, J. B., and D. J. Irschick. 1996. The effect of perch diameter on escape behaviour of Anolis

lizards: laboratory predictions and field tests. Animal Behaviour 51:593-602.

Losos, J. B., and B. Sinervo. 1989. The effects of morphology and perch diameter on sprint

performance of Anolis lizards. Journal of Experimental Biology 145:23-30.

Lustig, A., H. Ketter-Katz, and G. Katzir. 2013. Lateralization of visually guided detour

behaviour in the common chameleon, Chamaeleo chameleon, a reptile with highly

independent eye movements. Behavioural Processes 100:110-115.

Mansfield, R. H., and B. C. Jayne. 2011. Arboreal habitat structure affects route choice by rat

snakes. Journal of Comparative Physiology A 197:119-129.

Mattingly, W. B., and B. C. Jayne. 2005. The choice of arboreal escape paths and its

consequences for the locomotor behaviour of four species of Anolis lizards. Animal

Behaviour 70:1239-1250.

Mauro, A. A., and B. C. Jayne. 2016. Perch compliance and experience affect destination choice

of brown tree snakes (Boiga irregularis). Zoology 119:113-118.

Montgomery, G. G., and A. S. Rand. 1978. Movements, body temperature and hunting strategy

of a Boa constrictor. Copeia 3:532-533.

Moran, M. D. 2003. Arguments for rejecting the sequential Bonferroni ecological studies. Oikos

100:403-405.

Munteanu, A. M., I. Starnberger, A. Pasukonis, T. Bugnyar, W. Hodl, and W. T. Fitch. 2016.

Take the long way home: Behaviour of a neotropical frog, Allobates femoralis, in a

detour task. Behavioural Processes 126:71-75.

Pizzatto, L., S. M. Almeida-Santos, and R. Shine. 2007. Life-history adaptations to arboreality in

snakes. Ecology 88:359-366.

Poucet, B., C. Thinusblanc, and N. Chapuis. 1983. Route planning in cats, in relation to the

visibility of the goal. Animal Behaviour 31:594-599.

Pounds, J. A. 1988. Ecomorphology, locomotion, and microhabitat structure in a tropical

mainland Anolis community. Ecological Monographs 58:299-320.

Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225.

Rodda, G. H., T. H. Fritts, M. J. McCoid, and E. W. I. Campbell. 1999a. An overview of the

biology of the Brown Treesnake (Boiga irregularis), a costly introduced pest on Pacific

islands, in: Rodda, G. H., Y. Sawai, D. Chiszar, and H. Tanaka (Eds.), Problem Snake

Management: The Habu and the Brown Treesnake. Cornell University Press, Ithaca.

Rodda, G. H., Y. Sawai, D. Chiszar, and H. Tanaka. 1999b. Problem Snake Management: The

Habu and the Brown Treesnake. Cornell University Press, Ithaca.

Rumpff, H. 1979. Experimental studies on vision in Indian snakes. Journal of the Bombay

Natural History Society 76:475-480.

Ryan, M. J., J. H. Fox, W. Wilczynski, and A. S. Rand. 1990. Sexual selection for sensory

exploitation in the frog Physalaemus pustulosus. Nature 343:66-67.

Sanders, K. L., Mumpuni, A. Hamidy, J. J. Head, and D. J. Gower. 2010. Phylogeny and

divergence times of filesnakes (Acrochordus): Inferences from morphology, fossils and

three molecular loci. Molecular Phylogenetics and Evolution 56:857-867.

Schoener, T. W. 1968. The Anolis lizards of Bimini: resource partitioning in a complex fauna.

Ecology 49:704-726.

Simoes, B. F., F. L. Sampaio, C. Jared, M. M. Antoniazzi, E. R. Loew, J. K. Bowmaker, A.

Rodriguez, N. S. Hart, D. M. Hunt, J. C. Partridge, and D. J. Gower. 2015. Visual system

evolution and the nature of the ancestral snake. Journal of Evolutionary Biology 28:1309-

1320.

Tarsitano, M. 2006. Route selection by a jumping spider (Portia labiata) during the locomotory

phase of a detour. Animal Behaviour 72:1437-1442. van der Kooi, C. J., A. G. Dyer, P. G. Kevan, and K. Lunau. 2019. Functional significance of the

optical properties of flowers for visual signalling. Annals of Botany 123:263-276.

FIGURE LEGENDS Fig 1.1. Overhead view of the experimental apparatus. All perches were 1.5 m above the ground, and the standard gap distances were 47 cm for B. irregularis or 45 cm for B. constrictor.

(A) For treatments 1–16 (Table 1.1) both destinations had a standard gap distance and a yaw angle of 45° from the end of the starting perch. (B) For treatments 17–18, one of the gap distances was 10 cm less than the standard distance. (C) For treatments 19–20, both destinations had a standard gap distance, but the yaw angles were 0° and 90°.

Fig. 1.2. Schematic views of the destinations showing the manipulations of destination intensity and structure. The text in parentheses indicates the abbreviations used in Table 1.1 for when the sides of the cylinder, the end of the cylinder, and the pegs had intensities of black (b), gray (g) and white (w) and when pegs were absent (-).

Table 1.1. Mean + s.e.m. preference for destination perches

Destination B. irregularis (n = 16) B. constrictor (n = 14) Tmt Perch 1 Perch 2 Bkg Pref1 (%) p Pref1 (%) p 1 A (b,b,b) B (w,w,w) w 84±4 <0.001* 85±5 <0.001* 2 A (b,b,b) C (g,g,g) w 63±6 0.024 67±4 0.001* 3 C (g,g,g) B (w,w,w) w 79±4 <0.001* 86±5 <0.001* 4 A (b,b,b) B (w,w,w) b 67±6 0.004 37±5 0.013 5 D (b,b,-) B (w,w,w) w 48±7 0.389 51±6 0.452 6 A (b,b,b) D (b,b,-) w 78±6 <0.001* 86±5 <0.001* 7 A (b,b,b) I (w,w,-) b 68±7 0.015 86±3 <0.001* 8 G (b,b,w) D (b,b,-) w 68±6 0.003 82±5 <0.001* 9 A (b,b,b) E (w,w,b) w 75±5 <0.001* 63±6 0.022 10 G (b,b,w) E (w,w,b) w 64±6 0.022 56±7 0.176 11 A (b,b,b) F (w,b,w) w 67±6 0.004 83±4 <0.001* 12 E (w,w,b) B (w,w,w) w 67±6 0.009 70±6 0.002* 13 A (b,b,b) G (b,b,w) w 66±4 0.001* 55±5 0.168 14 H (b,w,b) F (w,b,w) w 50±6 0.999 70±6 0.002* 15 F (w,b,w) E (w,w,b) w 53±6 0.314 58±7 0.125 16 A (b,b,b) J (w,b,b) w 60±4 0.014 69+5 0.001* 17 A (b,b,b)-10 A (b,b,b) w 69±5 <0.001* 71±7 0.003 cm 18 A (b,b,b) B (w,w,w) - w 74±5 <0.001* 58±6 0.102 10 cm 19 A (b,b,b) 0° A (b,b,b) w 89±4 <0.001* 92±3 <0.001* 90° 20 B (w,w,w) 0° A (b,b,b) w 71±5 <0.001* 61±6 0.041 90° Notes: Mean + s.e.m. preference for perch 1 (Pref1) of the two destinations (Fig. 1.1) placed 47 cm (B. irregularis) or 45 cm (B. constrictor) and oriented 45o from the starting perch, unless noted otherwise. As in Fig. 1.2, text indicates surface intensities of white (w), gray (g) or black

(b), for the destinations (sides, end, pegs) and background (Bkg). P-values are uncorrected for

27 multiple comparisons, but * indicates significant after a table-wide sequential Bonferroni correction.

Fig 1.1. Overhead view of the experimental apparatus A Destinations

45o45o

End-on view

m 80 cm c 0 1

5 cm Start B

Gap Gap -10 cm

Fig. 1.2. Schematic views of the destinations showing the manipulations of destination intensity and structure

A (b,b,b) B (w,w,w)

C (g,g,g) D (b,b,-)

E (w,w,b) F (w,b,w)

G (b,b,w) H (b,w,b)

I (w,w,-) J (w,b,b)

CHAPTER TWO

Crayfish Eating in Snakes: A Model System for Testing How Anatomy, Behavior and

Performance Affect Foraging Ecology

Noah D. Gripshover

Department of Biological Sciences

University of Cincinnati

Cincinnati, OH, 45221 USA

ABSTRACT

Quantifying the performance of animals is a powerful methodology for determining the functional consequences of morphological variation. For example, snakes consume their prey whole, and variation in the anatomy of their trophic apparatus directly affects gape and limits maximal prey size. However, for the foraging ecology of snakes and other systems, scant data exist regarding how often maximal capacities are taxed in nature. Hence, we quantified: 1) maximal gape, 2) the size of prey relative to maximal gape, and 3) how the type and relative size of prey affected behavior and prey handling times for two species of natricine snakes that primarily eat soft- (Regina septemvittata) or hard-shell (Liodytes alleni) crayfish. Liodytes alleni had significantly larger maximal gape than R. septemvittata with equal snout-vent length. The percentages of large prey (> 60% maximal gape area) consumed in the field were low in both R. septemvittata (22%) and L. alleni (2%). However, R. septemvittata, especially juveniles, ate relatively larger prey than L. alleni. Strategies for dealing with the seasonal scarcity of small crayfish differed as juvenile R. septemvittata commonly removed and ate only chelipeds from crayfish too large to swallow whole, whereas juvenile L. alleni ate many small odonate nymphs.

During laboratory trials, unlike R. septemvittata, L. alleni usually used its body to restrain prey with behaviors that depended on relative prey size and prey hardness. Liodytes alleni consumed soft-shell crayfish significantly faster than R. septemvittata and significantly faster than hard- shelled crayfish. Several of the differences in gape, prey size and prey handling times and behavior between the crayfish-eating snakes resemble those between two phylogenetically distant species of homalopsid snakes that consume either hard- or soft-shell crabs. In both groups of crustacean-eating snakes, the decreased capture success in captivity and the rare consumption of relatively large hard-shell crustaceans in the field suggest that the ability to capture this type

32 of prey constrains prey size more commonly than maximal gape. Regina septemvittata was superior to the other species based on new metrics of potential feeding performance that integrated snake size and gape with the relative mass of prey.

INTRODUCTION

Quantifying the performance of animals is widely recognized as a powerful methodology for determining the functional consequences of morphological variation. Although laboratory tests have been useful and common, quantifying the performance of animals in nature is usually difficult and hence much less common (Irschick and Higham 2016). Consequently, the question of whether or not animals are “olympians” that routinely tax their maximal capacities in nature

(Hertz et al. 1988) has rarely been addressed with empirical data. As a result of their environment and their behavior, the performance of some animals in nature may be considerably less than their maximal capabilities (Irschick and Higham 2016) just as realized niches are often small compared to the fundamental niches of animals. However, some species have higher performance in nature than under presumably optimal laboratory conditions (Jayne and Ellis

1998). This primacy of behavior for affecting ecological performance is emphasized further by some systems in which animals with inferior laboratory performance outperform other animals with higher laboratory performance as a result of using a greater fraction of their maximal capacities when in the field (Husak and Fox 2006).

The feeding of snakes has several attributes well suited for employing the performance testing paradigm to gain fundamental insights into the roles of morphology and behavior for affecting foraging ecology. For example, the maximal gape of snakes has a rather straight- forward relationship to the dimensions of the relevant bones and the distensibility of soft tissues

(Cundall 2019), whereas traits such as maximal locomotor speed have effects of morphology that are mediated by muscle physiology, neural control, and the motivation of the test subject

(Irschick and Higham 2016). Consequently, the gape of snakes was a prominent example in a seminal paper that developed the morphology-performance-fitness paradigm (Arnold 1983).

Ironically, maximal gape has only recently been measured directly in less than ten of more than

3,500 species of snakes with diverse ecology and morphology (Hampton and Moon 2013,

Hampton and Kalmus 2014, Hampton 2018, Jayne et al. 2018). Consequently, relating the size of prey to the maximal gape and foraging ecology of snakes remains in its infancy.

Additional advantages of studying foraging ecology using snakes are that determining the size and identity of prey is facilitated because snakes consume their prey whole (Voris and Voris

1983, Mushinsky 1987), and large samples of stomach contents often can be obtained with non- destructive sampling (Mushinsky et al. 1982, Glaudas et al. 2019). Consequently, many studies of snakes have documented the types, shapes and sizes of prey and how they are related to the lengths and masses of snakes (Glaudas et al. 2019). Overall, larger snakes eat larger prey, and larger individuals commonly do not consume prey as small as those of smaller conspecifics

(Mushinsky 1987, Arnold 1993). However, such previous analyses of the sizes of predators and prey do not directly address the role of anatomical constraints in affecting prey size.

Additional metrics of feeding performance besides prey size include the ease of capture and prey handling time, which are affected by the type of prey (Arnold 1993, Willson and

Hopkins 2011). For example, many snakes with the generalized morphology of slender, pointed teeth adeptly eat prey such as earthworms, amphibians, fishes and mammals, whereas snakes that eat hard-bodied prey often have reduced tooth length and sharpness (Savitzky 1983), suggesting that such prey are difficult for more generalized snakes to exploit. Their hard and spiny

34 exoskeletons and powerful chelipeds make many crustaceans formidable prey for snakes.

Nonetheless, crustacean-eating specialists have evolved independently in the New World natricine and southeast Asian homalopsid snakes (Fig. 2.1) with some genera specializing in consuming either freshly molted (Gerarda, Regina) or hard-shelled crustaceans (Fordonia,

Liodytes) (Gibbons and Dorcas 2004, Jayne et al. 2018). This presents an unusual opportunity to examine the extent to which convergent ecology is correlated with convergent evolution of morphology, feeding behavior, and performance.

We studied two species of crayfish-eating snakes, Regina septemvittata and Liodytes alleni to test how an anatomical constraint (maximum gape) and prey choice interact with prey handling behaviors to affect diet and feeding performance. We quantified the scaling relationships between maximum gape and overall snake size, and for snakes captured in the field we determined the size of prey relative to maximal gape. Determining gape allowed us to test the extent to which snakes in the field use their maximal capacity for swallowing prey and how relative prey size affects behavior and prey handling times (observed in laboratory experiments).

The reduced hardness and mobility of soft-shelled crayfish and previous findings for crustacean- eating homalopsid snakes (Jayne et al. 2018), led us to hypothesize that R. septemvittata would consume soft-shelled crayfish more rapidly and select prey with larger relative size in the field than L. alleni. Finally, we compared crustacean-eating species of snakes using new metrics of feeding performance that combine relative prey mass and relative prey area.

MATERIALS AND METHODS

Study species and field observations

We collected all snakes in our study from wild populations. Between April and October in 2018 and 2019, we collected R. septemvittata and co-occurring crayfish, Orconectes rusticus, by overturning rocks along a stream in Kenton Co., Kentucky (KDFWR permit SC1711317 and

SC1911197). Similar to (Godley 1980), during May 2019 in Gilchrist and Levy Counties in

Florida, we collected L. alleni and co-occurring crayfish, Procambarus fallax, by seining mats of aquatic plants (Limnobium spongia) (FWC permit LSSC-18-00055A). We retained some gravid

L. alleni to obtain neonatal snakes for measuring maximal gape. All of the husbandry and experimental methods were in compliance with the IACUC of the University of Cincinnati.

To collect stomach contents from snakes captured in the field, we were able to gently palpate the stomachs of R. septemvittata to force regurgitation. After measuring snout-vent length (SVL) and mass, we released these snakes at their site of capture. We used palpation to detect stomach contents in L. alleni (n = 17), but their hard and spiny prey required euthanizing the snakes to recover stomach contents (Godley 1980). We then pooled our data with those obtained from 17 snakes collected by Godley (1980) that are in the collection of the National

Museum of Natural History (Suitland, Maryland, USA).

Gape and prey size

We measured gape for the following samples that ranged from neonates to large adults.

The 15 male and 12 female R. septemvittata had overall means+SE of SVL and mass of

317+24.5 mm (range = 165–586 mm), and 21.9+4.6 g (range = 2.6–133 g), respectively. The 6 male and 24 female L. alleni had had overall mean values of SVL and mass of 243+20.3 mm

(range = 120–490 mm) and 26.7+6.3 g (range = 2.2–142 g), respectively. To measure gape we

36 euthanized snakes with sodium pentobarbital. To allow insertion and removal of gape probes proceeding from anterior to posterior, we then made a transverse incision through all of the skin and structures ventral to the vertebrae approximately two skull lengths posterior to the skull.

The probes for measuring gape were 3D-printed. The probes had a cylindrical base and a hemispherical end, and we lubricated them with KY Jelly. The successive incremental increases in probe diameter were 0.5 and 1.0 mm for probes with diameters less than or greater than 11 mm, respectively. Upon encountering appreciable resistance, we waited 10 min before inserting the next probe. We inserted larger probes until they did not fit or until we observed tissue failure

(such as torn skin), after which we reinserted the next smaller probe (= maximal gape).

We converted the maximal cross-sectional areas and intact masses of all prey to values of relative prey area (RPA) and relative prey mass (RPM) that were percentages of the maximal cross-sectional area of snake gape and snake mass, respectively. For intact reference specimens

(Orconectes rusticus, n = 66; Procambarus fallax, n = 44), we estimated maximal cross-sectional area as the area of an ellipse at the portion of the carapace where maximal height and width of the carapace formed the major and minor axes, respectively. To estimate the intact size of stomach contents, we used various regressions that related carapace and chela dimensions from the reference collection to maximal cross-sectional area. The crayfish recovered from stomachs were only classified as either soft- or hard-shelled rather than making finer distinctions, such as pre-molt versus intermolt as in Godley (1980).

Laboratory experiments

We housed and tested snakes individually in 10-gallon glass aquariums with 25-watt incandescent light bulbs that allowed the snakes to thermoregulate body temperatures from

37 approximately 25–30ºC, which is within the range of active field body temperatures. For example, 233 R. septemvittata that we captured had a mean body temperature of 27.6+0.2 ºC

(range = 15.1–34.5 ºC), and (Godley 1980) found prey in R. alleni captured where water temperature ranged from 15–27ºC. The aquariums were tilted so that a pool of deionized water in the lower half of the container had a maximum depth of 3 cm. The lower half of the tank was covered with a layer of smooth pebbles (Imagitarium River Rock Shallow Creek Gravel, Petco,

San Diego, CA, USA), whereas the upper half was covered with artificial turf.

For captive feedings many of the crayfish molted at unknown times overnight as much as

12 hr before they were used for an experiment. We had two categories for these recently molted crayfish: 1) soft (incapable of walking outside of water) and 2) medium (softer than the intermolt condition but capable of walking outside of water). Although immobile on land, the soft crayfish could walk in water and perform a tail flip escape response.

From HD (1920x1080 pixels) videos of captive feedings, we determined total handling time (HT) as the sum of the durations of the following six behaviors. 1) Attack extended from the first strike until the final strike that succeeded in capturing the crayfish. 2) Jaw holding occurred immediately after the successful strike as snakes grasped prey with immobile jaws. 3)

Lateral jaw movements repositioned the prey prior to the beginning of swallowing. 4) Pauses occurred during lateral jaw walking. 5) Swallowing ended as soon as jaw movements ceased and the prey item was no longer visible. 6) Pauses occurred during swallowing.

We noted the following additional events. We scored the number of successful and unsuccessful strikes and the number of escapes by the crayfish and the snake behaviors during which the escapes occurred. We scored the following four locations of the final strike that succeeded in capturing and consuming the crayfish: 1) the abdomen only, 2) the joint between

38 the abdomen and carapace, 3) the carapace only, and 4) the cheliped. Immediately after a successful strike, we recorded whether the back of the mouth of the snake (the joint between the quadrate and lower jaw) was located on the lateral, dorsal, or ventral surface of the crayfish. We also recorded the direction of swallowing prey (tail first [0] or head first [1]) and the surface of the prey that was oriented towards the palate (lateral [1], dorsal [2], ventral [3]). For L. alleni, we recorded the type and total duration of body restraint used by the snakes.

Statistical analysis

All morphological data and handling times were log10 transformed before they were analyzed. In both of our study species, females attain larger overall size than males (Gibbons and

Dorcas 2004). Unless stated otherwise, we found no significant differences between males and females within each species, and the least-squares scaling regressions and other analyses were performed for a combined sample of males and females. We used analyses of covariance

(ANCOVA) to compare species, and unless stated otherwise, these comparisons lacked significant heterogeneity of slopes. To compare the cumulative frequency distributions of RPA of stomach contents, we used a two-sample Kolmogorov-Smirnov Test. We used forced-fit multiple regressions to relate variables such as RPA, strike location, swallowing direction and prey orientation to HT in laboratory trials. Our final choice of a model was one for which all independent variables were individually significant and the highest R2 value was attained. Means are reported +SE. Tabular summaries of descriptive statistics of behavior, regression analyses, and ANCOVAs are provided in the supplemental materials S2.

RESULTS

Morphology

For a given SVL, L. alleni was significantly heavier (Fig. 2.2A) and had significantly larger maximal gape area (Fig. 2.2B) than R. septemvittata. For a given mass, the gape of R. septemvittata was significantly larger than that of L. alleni (Fig. 2.2C). In both species, maximal gape diameter and area scaled with negative allometry relative to SVL and mass. The only significant differences between sexes within a species were for gape diameter and gape area as a function of mass for R. septemvittata. The CT scans of 9 individuals per species revealed that the mean (n = 9 per species) percent contributions to maximal gape area of skull width, quadrate length, lower jaw length and soft tissue between the lower jaws for R. septemvittata and L. alleni were 13.9+0.3%, 11.3+0.4%, 57.9+1.3% and 16.9+1.1%, and 13.4+0.4%, 14.2+0.5%,

57.5+1.5% and 15.0+1.6%, respectively (Fig. 2.3).

Diet and prey size

In 55% of the total of 340 R. septemvittata that were captured, we recovered parts from a total of 194 freshly molted crayfish (Orconectes rusticus), and of these stomach contents 25 were only a cheliped. For the crayfish that were consumed whole, we were able to estimate intact size,

RPA and RPM for 155 individuals. Regina septemvittata ingested 63% of the whole crayfish abdomen first, and all of the detached chelipeds were ingested from distal to proximal. Values of

RPA of crayfish consumed whole ranged from 11–123% (mean = 47.2±1.88%) (Fig. 2.4A), and

RPM ranged from 2–85% (mean = 17.4±1.2%). For the separated chelipeds by themselves, the values of RPA and RPM ranged from 3–81% (mean =25.4±3.4%) and from 0.3–37% (mean =

7.1+1.6%), respectively. The estimated sizes of the intact crayfish for these chelipeds had RPA ranging from 23–226% (mean = 92.6±9.6%) and RPM ranging from 4–236% (mean =

67.8±14.5%). As SVL increased the ranges in absolute prey diameters were quite similar (Fig.

2.4E), whereas smaller snakes had larger ranges of RPA and were more likely to consume relatively large prey (Fig. 2.4G).

Of the 35 L. alleni that we collected plus the 68 L. alleni collected by (Godley 1980), 44 snakes (43%) contained 54 identifiable prey items consisting of 55.6% crayfish, Procambarus fallax) (8 pre-molt, 1 soft, 21 intermolt), 42.6% odonate nymphs (Miathyria marcella), and 1.9% grass shrimp (Palaemonetes paludosus), and we estimated intact size for 43 of these items

(63.6% crayfish, 34.1% odonates, 2.3%shrimp) from 33 snakes. All crayfish were consumed whole and tail first, and their values of RPA and RPM ranged from 10–66% (mean = 35.7 ±

2.4%) and 1.3–18% (mean = 6 ± 0.7%), respectively (Fig. 2.4B). Values of RPA and RPM of the odonate nymphs ranged from 4–22% (mean = 12 ± 1.5%) and 0.3–3.4% (mean = 1.4 ± 0.3%), respectively. The one grass shrimp had RPA = 17% and RPM = 0.3%. Hence, most of the prey with RPA < 20% were consumed by small snakes and were odonates (Fig. 2.4H).

The frequency distributions of RPA for whole crayfish did not differ significantly between R. septemvittata (n = 155) and L. alleni (n = 28) (Fig. 2.4A,B; D = 0.21, P = 0.227), or when the estimated intact sizes of the entire crayfish were included for when only chelipeds were consumed by R. septemvittata (n = 180) (D = 0.26, P = 0.072). However, the frequency distributions of RPA for the combined samples of all whole prey consumed by L. alleni (n = 43) and R. septemvittata (n = 155) had highly significant differences (D = 0.33, P = 0.002) as result of L. alleni consuming more small prey (Fig. 2.4B).

Feeding behavior and handling time

We fed 8 R. septemvittata 118 freshly molted crayfish (O. rusticus) with RPA (Fig. 2.5D) and RPM ranging 4–154% and 1–124%, respectively. Regina septemvittata usually struck the

41 carapace of the crayfish, and 26% of the trials required more than one strike to capture the crayfish (Table S2.2). Tail flips allowed the crayfish to escape strikes at least once in 27% of the trials. Jaw holding behavior occurred in 82% of the trials. In 33% of the trials during which the crayfish flipped their tails after being captured, they escaped. In 61% of the trials, the crayfish were swallowed with their lateral surface contacting the palate of the snake.

In 19 trials R. septemvittata removed and ate a cheliped, and these chelipeds were always swallowed from distal to proximal. In three of these cases, snakes only consumed the cheliped

(whole crayfish RPA = 26, 65, and 154%), whereas in 16 cases after eating a cheliped, the snake subsequently consumed the entire crayfish (RPA = 7–64%).

Despite not having a fully hardened exoskeleton, during the initial attack by R. septemvittata, the crayfish displayed the following defensive behaviors: 1) raising the chelipeds to fend off the snake, 2) using a tail flip to flee from the snake, or 3) pinching the snake. After capture, however, pinching rarely resulted in an escape.

Prey handling times of R. septemvittata ranged from 6–2376 s and increased significantly in univariate regressions with increased values of RPA (R2 = 0.57, Fig. 2.5D) or RPM (R2 =

0.55). For the 118 feedings when the entire crayfish was consumed, a multiple regression revealed that HT increased significantly (overall R2 = 0.66, P < 0.001) with increases in RPA (P

< 0.001), prey hardness (P < 0.001), strike location (P = 0.007), and the number of crayfish pinches (P = 0.025). Another regression revealed that the presence of the holding behavior (1= present, 0 = absent) was significantly more likely to occur (overall R2 = 0.42, p < 0.001) with increases in RPA (P < 0.001), prey hardness (P < 0.001), and strike location (P = 0.007).

We fed 15 L. alleni 66 freshly molted (RPA = 13 – 83%, RPM = 1 – 36%) and 61 hard- shell (RPA = 7 – 62%, RPM = 1 – 23%) Procambarus fallax (Fig. 2.5E). The final strike hit the

42 abdomen, carapace, and the carapace-abdomen joint in 38%, 27% and 35% of the trials, respectively. In 13% of the trials more than one strike was required to capture the crayfish. Less than 10% of the post-capture tail flips allowed the crayfish to escape. We never observed L. alleni striking or removing a cheliped. The snakes displayed jaw holding in only 11% of the trials. Snakes used their body to restrain prey in 69% of the trials. In 97% and 90% of the trials prey were swallowed tail first and with their lateral surface contacting the palate, respectively.

Liodytes alleni used the following categories of increasingly greater amounts of body restraint: 0) no body restraint, 1) u-loop during which the snake pushed the crayfish against a concave side of its body that partially encircled crayfish (Fig. 2.5A), 2) body pinning during which the ventral surface of the snake pushed down on the crayfish (Fig. 2.5B), or 3) coiling, during which the snake completely encircled the crayfish (Fig. 2.5C). RPA better predicted the amount of prey restraint for hard-shell (n = 61, R2 = 0.19, P < 0.001) than for soft-shell crayfish

(n = 66, R2 = 0.09, P = 0.013). For the combined sample of soft- and hard-shell crayfish L. alleni used increasingly greater categories of body restraint with increased RPA (n = 127, R2 = 0.06, P

= 0.005), but more than three times as much variance (overall R2 = 0.21, P < 0.001) in the categories of body restraint was accounted for in a multiple regression that included RPA (P <

0.001), prey hardness (P < 0.001) and the number of pinches (P = 0.02) as independent variables.

In univariate regressions performed separately for hard- and soft-shelled crayfish consumed by L. alleni (Table S2.1), handling time increased significantly both with increased

RPA (Fig. 2.5E) and RPM, but the R2 for RPA exceeded that of RPM. In a multiple regression, the handling times pooled for hard- and soft-shell crayfish increased significantly (n = 127, R2 =

0.71, P < 0.001) with increases in RPA (P < 0.001), body restraint of prey (P = 0.001), prey hardness (P < 0.001), and number of pinches after capture (P = 0.007).

DISCUSSION

Convergent evolution of dietary specialists

The ability of predators to exploit prey is affected by attributes of the prey such as rarity, absolute size, mobility, strength, defensive behaviors, and resistance to physical injury and death and attributes of the predator such as the size and the abilities to detect, subdue, consume and digest prey (Endler 1991, Arnold 1993). Many snakes species with generalized dentition and no conspicuous specializations in behavior eat fishes and amphibians, and this is the most parsimoniously inferred diet of the most recent common ancestors of the crayfish-eating natricines (Liodytes, Regina) and the crustacean-eating homalopsids (Cantoria, Fordonia,

Gerarda) (Gibbons and Dorcas 2004, Jayne et al. 2018) (Fig. 2.1). Recent phylogenies (Figueroa et al. 2016) also imply that within natricines the specialized diet of primarily crayfish evolved independently three times, and the diet of soft-shelled crayfish evolved twice. The two crab- eating genera (Fordonia [hard-shell]; Gerarda [soft-shell]) are monophyletic (Fig. 2.1), and their sister taxon, Cantoria, eats hard-shelled snapping shrimp and crabs (Ghodke et al. 2018, Jayne et al. 2018). Hence, in both natricines and homalopsids no evidence suggests that the diet of more readily consumed soft-shelled crustaceans preceded a diet of mostly hard-shelled crustaceans.

Several similar and convergent trends occur within natricines and homalopsids that eat hard- versus soft-shelled crustaceans. Compared to those that prey on soft-shelled crustaceans, the genera that eat hard-shelled prey: 1) have blunter teeth, 2) thickened stomachs, 3) larger gape relative to SVL, 4) larger mass relative to SVL, and 5) consume prey with smaller values of RPA and RPM (Savitzky 1983, Jayne et al. 2018).

The use of the body to restrain or constrict prey has evolved repeatedly within snakes

(Greene and Burghardt 1978), and this behavioral specialization seems likely to facilitate

44 handling prey that are large or difficult to subdue. However, using the body to restrain prey is uncommon in both natricines (Gregory et al. 1980) and homalopsids (Jayne et al. 2018).

Nonetheless, body restraint has evolved in the most recent common ancestors of both the natricines that eat hard-shelled crayfish and of the crustacean-eating homalopsids. With increased RPA, the extent of encircling prey with the body of L. alleni (Fig. 2.5) increases significantly as does the amount of body restraint used by Gerarda (Jayne et al. 2018). However,

Fordonia is more stereotyped in using its body to restrain crabs of nearly all sizes although the body postures that are used can be quite variable (Jayne et al. 2018). The faster prey handling times (Fig. 2.5F) and the reduced probability of prey escaping in L. alleni versus R. septemvittata provide compelling support for the benefits of snakes using their body to restrain prey.

Another specialized and very rare behavior in snakes is consuming prey in pieces, and doing so circumvents the limitations of gape on the size of prey that can be exploited (Jayne et al. 2018). We observed this in both laboratory trials and for the prey consumed in the field by R. septemvittata, and this was significantly more likely to occur for snakes in the field as RPA increased (n = 180, R2 = 0.17, P < 0.001). Cantoria, Fordonia, and Gerarda also all break off appendages of their crustacean prey (Ghodke et al. 2018, Jayne et al. 2018). Gerarda also combines body restraint with ripping apart the carapace of crabs to eat crabs that are too large to be swallowed whole, and Gerarda has remarkably fast prey handling times for a given RPA compared to all of the other crustacean-eating species (Fig. 2.5F). Thus, consuming pieces of prey (usually appendages) is reasonably common in snakes that specialize in crustaceans.

Furthermore, the specialized behaviors of R. septemvittata and G. prevostiana allow both of them to exploit prey with intact sizes that are too large to be consumed whole.

Given the formidable body plans of crustaceans, it is interesting to consider why snakes would evolve this prey preference. Their abundance may be the primary advantage of crustaceans as prey. Where L. alleni occurs the densities of crayfish can be extremely high (61 m-2) (Godley 1980), and in mangroves where crustacean-eating homalopsids are found, the biomass of crabs may be as high as 80% of the entire macrofauna with densities as high as 80–90 m-2 (Macintosh 1988). Despite the high abundance of these crustaceans, their energetic content per unit wet mass (Godley 1980, Dupreez and McLachlan 1983) is lower than that of some fish and amphibians consumed by natricine snakes (Willson and Hopkins 2011). However, the energetic content of crustaceans can increase after molting (Stein and Murphy 1976, Godley

1980), and both the natricine and homalopsid snake species that exploit this prey resource are adept at using olfactory cues to distinguish freshly molted from hard-shelled individuals (Jackrel and Reinert 2011, Jayne et al. 2018).

Effects of predator and prey size on feeding performance

An advantage of studying the foraging ecology of snakes is that identifying prey species and quantifying their size is facilitated by a lack of mastication, and the maximal gape of snakes usually limits maximal prey size. Patterns of predator and prey size within a species of snake have been analyzed most commonly by plotting a single measure of prey size versus a single measure of snake size (as in Fig. 2.4C–F). As reviewed in (Arnold 1993), two recurrent findings from such bivariate plots are with increased snake size: 1) the mean prey size increases and 2) the variance in prey size increases. In addition, larger snakes of many species often do not continue to eat small prey; hence, the cloud of points has a wedge-like shape. By contrast, such data for a lesser number of species form a right triangle with an upward sloping hypotenuse forming the upper boundary. What determines the lower limit of prey size in snakes remains

46 unclear (Arnold 1993), but behavior may be one factor. The upper boundary is constrained by the maximal gape of snakes, but if snakes simply choose not to consume large prey, then the upper boundary could be well below that imposed by the constraint of gape. However, no previous study has directly measured gape and integrated this with plots of prey and snake size.

The patterns of prey and snake size that we observed differed substantially depending on the metrics of size. The data for prey mass versus snake mass resembled a right triangle (Fig.

2.4C,D), whereas those for prey diameter versus SVL resembled a parallelogram with unexpectedly similar ranges in prey diameter over a wide range in SVL (Fig. 2.4E,F). Values of

RPA versus SVL, especially for R. septemvittata, resembled a right triangle with a downward sloping hypotenuse (Fig. 2.4G). However, L. alleni (Fig. 2.4H) lacked this conspicuous contraction in RPA with increased SVL. If a line were fit through the upper points of the data for prey diameter versus SVL, one could mistakenly conclude that over a wide range of SVL the snakes performed at similar fraction of their maximal capacity.

All metrics of prey and snake size indicate that R. septemvittata consumed prey that were larger relative to their size more commonly than L. alleni, but this difference between species was most pronounced for the juveniles (Fig. 2.4). The timing of the reproduction of the crayfish and snakes causes a seasonal scarcity of small crayfish for the young of the year for both R. septemvittata (Prins 1968) and L. alleni (Godley 1980), but these two species cope with this in different ways. Besides eating whole crayfish with larger relative size, R septemvittata eats chelipeds from crayfish larger than can be swallowed whole, whereas L. alleni eats many small odonate nymphs (Fig. 2.4H). Hence, many young R. septemvittata could be viewed as

“overachievers” or “olympians” that use a high fraction of their maximal capacity (Hertz et al.

1988). This resembles other systems where animals in the field compensate for an inferior maximal capacity by using a larger proportion of their maximal capacity (Husak and Fox 2006).

A good choice of currency for understanding foraging ecology is the mass of the prey relative to that of the predator as the former is proportional to the energy consumed and the latter is proportional to the energetic need. Thus, relative prey mass (RPM) has long been used to quantify prey size for hundreds of species of snakes (Voris and Moffett 1981, Glaudas et al.

2019). However, RPM is not sufficient to determine the size of prey consumed relative to the maximal size that is possible. Hence, we integrated our scaling data and direct measurements of snake gape to define various performance spaces that included both RPA and RPM (Fig. 2.6).

For the natural prey of four species of crustacean-eating snakes, we plotted RPM versus

RPA to gain insights into the benefits for RPM from eating prey with different RPA (Fig.

2.6A,B). The three species (R. septemvittata, L. alleni and G. prevostiana) with negative allometry of gape area and SVL (slope < 1.67) had the following two trends. First, for a given

RPA and species, RPM decreases with increased snake size (Fig. 2.6A vs. 2.6B). Second, a nonlinear relationship causes the benefits in RPM to increase with increased RPA (Fig. 2.6A,B).

For example, for a R. septemvittata with SVL = 150 mm, as RPA increases from 20 to 30 %, and from 80 to 90%, RPM increases from 7 to 12% and 52 to 62%, respectively. In addition, for a given RPA, the crayfish-eating species can consume prey with larger RPM than the two crab- eating species (Fig. 2.6A,B). As a result of positive allometry between gape area and SVL for F. leucobalia, the relationship between RPM and RPA was nearly unaffected by SVL (Fig. 2.6A vs.

2.6B).

We also used the areas under the curves of RPM versus overall snake size (when prey cross-sectional area equals maximal gape area) to define the potential performance spaces that

48 are possible in light of the constraints on prey size imposed by gape (Fig. 2.6C,D). Regardless of whether SVL or mass was used for overall snake size, over nearly the entire range of snake size the rank order from greatest to smallest values of RPM was R. septemvittata, L. alleni, F. leucobalia and G. prevostiana (Fig. 2.6C,D). Consequently, the areas under these curves differed substantially (Fig. 2.6C, for SVL = 150 mm to 550 mm) as values for L. alleni, F. leucobalia and

G. prevostiana were 68%, 44% and 37%, respectively, of that for R. septemvittata. The realized performance spaces of RPM versus SVL for R. septemvittata and L. alleni were approximately

75% and 26% of the potential performance space, respectively (Fig. 2.6E,F). Hence, besides L. alleni having a smaller potential feeding performance space, unexpectedly it also used a much smaller fraction of its potential than R. septemvittata.

One can also delineate a volume of a three-dimensional performance space in which

RPM is a function of SVL and RPA (Fig. 2.7). For a range of SVL in common to the four crustacean-eating species (150 mm to 550 mm), the rank order of species based on their feeding performance volumes was the same as those based on the areas defined by curves of RPM versus snake size when RPA equaled 100% (Fig. 2.6C,D).

Most of the prey consumed by our study species were much smaller than the limit imposed by gape (Fig. 2.6E,F), and this was especially true for L. alleni. Similarly, in the field F. leucobalia consumes hard-shelled crabs with small RPA (Jayne et al. 2018). Thus, the foraging strategies of L. alleni and F. leucobalia run counter to the intuitive expectation that species with a small performance space might use a greater faction of it to accrue benefits for increased RPM.

By contrast, G. prevostiana has a very small performance space (Fig. 2.6C,D), and it does conform to this expectation as it often tears apart whole prey larger than its maximal gape and some of these pieces actually approach the limits of its gape (Jayne et al. 2018).

For the species that eat hard-shelled crustaceans, laboratory observations combined with the small RPA of prey consumed in the field suggest that successful capture is difficult, and we suspect this is a major factor contributing to the rarity of prey with large RPA. The snakes also may simply choose to not attack large and formidable prey. Nevertheless, both of these factors emphasize the extremely important roles of behavior in determining the patterns of resource use that occur even after accounting for the constraints imposed by anatomy.

One could use similar approaches to those described herein to provide further insights into the potential effects of consuming species of prey that differ in shape as well as size.

Theoretically, for a given gape area, gape-limited predators could attain the same values of RPM from eating either stout prey or elongate prey that have much smaller RPA. Indeed, many elongate species of prey are consumed by snakes such as eels, salamanders and snakes (Voris and Voris 1983, Jackson et al. 2004, Willson and Hopkins 2011, Sherratt et al. 2019). However, some species of sea snakes that specialize on eels and elongate fishes have such small heads that it seems likely that their gape is small (Voris and Voris 1983, Sherratt et al. 2019). Although consuming elongate prey could allow some species to exploit prey with high RPM by using only a small fraction of their maximal gape capacity, but some species with small heads and likely small gape may actually consume prey that are elongate and have high values of RPA. Direct measurements of gape and prey size could resolve these interesting possibilities.

Many species of prey consumed by snakes including crayfish have nearly circular cross- sectional areas, but prey such as laterally compressed species of fish depart substantially from this shape (Willson and Hopkins 2011). Close relatives of the crayfish-eating snakes in the genus

Nerodia have relatively large heads, and their diet includes laterally compressed species of fishes such as the genus Lepomis (Mushinsky et al. 1982, Willson and Hopkins 2011). The larger heads

50 of Nerodia seem likely to have larger gape, which can facilitate consuming prey either with large nearly circular cross-sectional areas or with modest cross-sectional areas but large ratios of height to width. Perhaps, having large gape relative to prey cross-sectional area could also facilitate consuming prey that have irregular shapes or are difficult to handle such as hard-shelled crayfish and hard-shelled crabs. However, many groups of snakes including Liodytes (Gibbons and Dorcas 2004) and the crustacean-eating homalopsids burrow (Jayne et al. 2018), and the microcephalic sea snakes probe narrow spaces when they forage (Voris and Voris 1983). Having small heads to facilitate these tasks may thus create tradeoffs with the morphologies that enhance gape (Fabre et al. 2016) and facilitate consuming non-circular prey.

Size and anatomy have profound effects on all aspects of animal biology including ecology. Our study illustrates how using functional morphology is a powerful first step for defining what is possible (maximal performance), and our field data tested the fraction of maximal performance that was realized in nature. Laboratory observations illuminated the importance of behavior for determining the success and difficulties in predator-prey interactions.

Despite the small number of snake species for which gape has been measured directly, some strikingly different foraging strategies are already becoming apparent. For example, we identified convergent patterns in two taxa (Regina and Gerarda) that consume relatively large and helpless prey such as recently molted crustaceans, whereas two others (Liodytes and

Fordonia) eat formidable prey (hard-shelled crustaceans) with smaller sizes relative to their gape. Relative prey size predicted the largest fraction of the variance in prey handling times, but prey hardness, prey defensive behaviors and the use of body restraint by the snakes also affected handling times and the success of the predator. Similar future integrative studies expanding the

51 comparative data that are available should facilitate understanding the relative importance of anatomy, behavior and phylogeny for large-scale patterns in variation in foraging ecology.

ACKNOWLEDGMENTS

Collection of L. alleni was made possible thanks to the help of J.S. Godley and P.E. Moler. T.

Saunders, L. Bischoff, and N. Addison, helped collect snakes and assisted with experiments.

Partial support was provided by the University of Cincinnati Wiemann Wendall Benedict

Graduate Fellowship.

REFERENCES

Arnold, S. J. 1983. Morphology, performance and fitness. American Zoologist 23:347-361.

Arnold, S. J. 1993. Foraging theory and prey-size-predator-size relations in snakes. Pages 87-115

in R. A. Seigel and J. T. Collins, editors. Snakes Ecology and Behavior. McGraw-Hill, Inc.,

New York.

Cundall, D. 2019. A few puzzles in the evolution of feeding mechanisms in snakes.

Herpetologica 75:99-107.

Dupreez, H. H., and A. McLachlan. 1983. Seasonal-changes in biochemical-composition and

energy content of the 3-spot swimming crab Ovalipes punctatus (Dehaan) (Crustacea,

Brachyura). Journal of Experimental Marine Biology and Ecology 72:189-198.

Endler, J. A. 1991. Interactions between predators and prey. Pages 169-196 in J. R. Krebs and N.

B. Davies, editors. Behavioral Ecology An Evolutionary Approach. Blackwell Scientific

Publications, Oxford.

Fabre, A. C., D. Bickford, M. Segall, and A. Herrel. 2016. The impact of diet, habitat use, and

behaviour on head shape evolution in homalopsid snakes. Biological Journal of the Linnean

Society 118:634-647.

Figueroa, A., A. D. McKelvy, L. L. Grismer, C. D. Bell, and S. P. Lailvaux. 2016. A species-

level phylogeny of extant snakes with description of a new colubrid subfamily and genus.

Plos One 11:DOI: 10.1371/journal.pone.0161070.

Ghodke, S., M. Chandi, and V. Patankar. 2018. Yellow-banded mangrove snakes (Cantoria

violacea) consume hard-shelled orange signaler crabs (Metaplax elegans). IRCF Reptiles and

Amphibians: Conservation and Natural History 25:50-51.

Gibbons, J. W., and M. E. Dorcas. 2004. North American Watersnakes a Natural History.

University of Oklahoma Press, Norman.

Glaudas, X., K. L. Glennon, M. Martins, L. Luiselli, S. Fearn, D. F. Trembath, D. Jelic, and G. J.

Alexander. 2019. Foraging mode, relative prey size and diet breadth: A phylogenetically

explicit analysis of snake feeding ecology. Journal of Animal Ecology 88:757-767.

Godley, J. S. 1980. Foraging ecology of the striped swamp snake, Regina alleni, in southern

Florida. Ecological Monographs 50:411-436.

Greene, H. W., and G. M. Burghardt. 1978. Behavior and phylogeny: constriction in ancient and

modern snakes. Science 200:74-77.

Gregory, P. T., J. M. Macartney, and D. H. Rivard. 1980. Small mammal predation and prey

handling behavior by the garter snake Thamnophis elegans. Herpetologica 36:87-93.

Hampton, P. 2018. Morphological indicators of gape size for red-tailed pipe snkes (Cylindrophis

ruffus). Journal of Herpetology 52:425-429.

Hampton, P., and T. Kalmus. 2014. The allometry of cranial morphology and gape size in red-

bellied mudsnakes (Farancia abacura). Herpetologica 70:290-297.

Hampton, P. M., and B. R. Moon. 2013. Gape size, its morphological basis, and the validity of

gape indices in western diamond-backed rattlesnakes (Crotalus atrox). Journal of Morphology

274:194-202.

Hertz, P. E., R. B. Huey, and T. Garland. 1988. Time budgets, thermoregulation, and maximal

locomotor performance - are reptiles olympians or boy scouts. American Zoologist 28:927-

938.

Husak, J. F., and S. F. Fox. 2006. Field use of maximal sprint speed by collared lizards

(Crotaphytus collaris): Compensation and sexual selection. Evolution 60:1888-1895.

Irschick, D. J., and T. E. Higham. 2016. Animal Athletes. Oxford University Press, Oxford.

Jackrel, S. L., and H. K. Reinert. 2011. Behavioral responses of a dietary specialist, the queen

snake (Regina septemvittata), to potential chemoattractants released by its prey. Journal of

Herpetology 45:272-276.

Jackson, K., N. J. Kley, and E. L. Brainerd. 2004. How snakes eat snakes: the biomechancal

challenges of ophiophagy for the California kingsnake, Lampropeltis getula californiae

(Serpentes: Colubridae). Zoology 107:191-200.

Jayne, B. C., and R. V. Ellis. 1998. How inclines affect the escape behaviour of a dune dwelling

lizard, Uma scoparia. Animal Behaviour 55:1115-1130.

Jayne, B. C., H. K. Voris, and P. K. L. Ng. 2018. How big is too big? Using crustacean-eating

snakes (Homalopsidae) to test how anatomy and behavior affect prey size and feeding

performance Biological Journal of the Linnean Society 123:636–650.

Macintosh, D. J. 1988. The ecology and physiology of decapods of mangrove swamps. Pages

315–341 in A. A. Fincham and P. S. Rainbow, editors. Aspects of Decapod Crustacean

Biology, Symposium Zoological Society London. Zoological Society of London, London.

Mushinsky, H. R. 1987. Foraging ecology. Pages 302-334 in R. A. Seigel, J. T. Collins, and S. S.

Novak, editors. Snakes Ecological and Evolutionary Biology. MacMillan Publishing

Company, New York.

Mushinsky, H. R., J. J. Hebrard, and D. S. Vodopich. 1982. Ontogeny of water snake foraging

ecology. Ecology 63:1624-1629.

Prins, R. 1968. Comparative ecology of the crayfishes Oconectes rusticus rusticus and

Cambarus tenebrosus Doe Run, Meade County, Kentucky. International Review of

Hydrobiology 53:667-714.

Savitzky, A. H. 1983. Coadapted character complexes among snakes: fossoriality, piscivory, and

durophagy. American Zoologist 23:397-409.

Sherratt, E., K. L. Sanders, A. Watson, M. N. Hutchinson, M. S. Y. Lee, and A. Palci. 2019.

Heterochronic shifts mediate ecomorphological convergence in skull shape of microcephalic

sea snakes. Integrative and Comparative Biology 59:616-624.

Stein, R. A., and M. L. Murphy. 1976. Changes in proximate composition of crayfish Orconectes

propinquus with size, sex, and life stage. Journal of the Fisheries Research Board of Canada

33:2450-2458.

Voris, H. K., and M. W. Moffett. 1981. Size and proportion relationships between the beaked sea

snake and its prey. Biotropica 13:15-19.

Voris, H. K., and H. H. Voris. 1983. Feeding strategies in marine snakes: an analysis of

evolutionary, morphological, behavioral and ecological relationships. American Zoologist

23:411-425.

Willson, J. D., and W. A. Hopkins. 2011. Prey morphology constrains the feeding ecology of an

aquatic generalist predator. Ecology 92:744-754.

FIGURE LEGENDS

Fig. 2.1. Simplified phylogeny of Natricine and Homalopsid snakes based on Figueroa et al.

(2016). Branch lengths are arbitrary. Blue and red symbols represent crustacean specialist that fed on soft or hard-shelled crustaceans, respectively. Asterisks and bold-faced type indicate the two study species.

Fig. 2.2. Scaling relationships for morphological data. For a given SVL L. alleni (n = 30) had greater mass and larger gape than R. septemvittata (n = 27), but for a given mass R. septemvittata had larger maximal gape.

Fig. 2.3. Computed tomography scans of R. septemvittata (A) and L. alleni (B). Snakes preserved at maximum gape and shown in the anterior view. The relative contributions to maximal gape area are shown for: skull width (SW), quadrate (QD), lower jaw (LJ), and the skin and intermandibular ligament between the lower jaws (SI).

Fig. 2.4. Relationships between the absolute and relative sizes of prey consumed in the field and the morphology of R. septemvittata (n = 180) and L. alleni (n = 43). Black, white and grey fills indicate a whole crayfish, only one cheliped from a crayfish, and prey other than crayfish, respectively. The red lines in panels C-F show the constraints arising from gape.

Fig. 2.5. Prey handling behaviors and prey handling times. Liodytes alleni used the following three behaviors to restrain crayfish with its body: (A) U-loop , (B) pinning , and (C) coiling.

Total prey handling time versus RPA for (D) R. septemvittata (n = 118) and (E) L. alleni (n =

127). (F) Regressions of handling time versus RPA for five species of crustacean-eating specialists consuming either hard- or soft shelled prey. Abbreviations: R.s., R. septemvittata

(crayfish); L.a., L. alleni (crayfish); F.l., F. leucobalia (crabs swallowed side to side); G.p.,

Gerarda prevostiana (crabs swallowed front to back); C.v., Cantoria violacea (snapping shrimp). The data for homalopsid species are from (Jayne et al. 2018).

Fig. 2.6. Constraints of gape on feeding performance for four species of crustacean-eating specialists. RMA versus the RPA for snakes with SVLof 150mm (A) and 450 mm (B). RMA versus snake SVL (C) and snake mass (D). Homalopsid data are from (Jayne et al. 2018). Sizes of whole prey (black: crayfish, grey: odonate nymphs and shrimp) consumed in the field for R. septemvittata (E) and L. alleni (F) with curves indicating the constraints of maximal gape. The area below the line represents the potential feeding performance space available to the snakes, and the grey area represent the realized feeding performance spaces.

Fig. 2.7. Three-dimensional maximal feeding performance spaces for four species of crustacean-eating snakes. The shaded surface shows RMA predicted from RPA and snake

SVL. Overall, crayfish-eating snakes (A and B) have a larger performance space than crab-eating snakes (C and D), and juvenile snakes can usually consume prey with larger RMA for a given

RPA.

Fig. 2.1. Simplified phylogeny of Natricine and Homalopsid snakes based on Figueroa et al. (2016)

Natricines Homalopsids

Dieurostus wallacei Liodytes pygaea Pseudoferania polylepis Liodytes rigida Myron richardsonii Liodytes alleni* Phytolopsis punctata

Clonophis kirtlandii Subsessor bocourti Virginia valeriae Homalopsis buccata Haldea striatula Cerberus (3 sp.) Storeria (3 sp.) Bitia hydroides Tropidoclonion lineatum Erpeton tentaculatum Regina grahamii Cantoria violacea Nerodia cyclopion Gerarda prevostiana Nerodia floridana Fordonia leucobalia R. septemvittata* Hypsiscopus (3 sp.) Nerodia (8 sp.) Myrrophis chinensis

Thamnophis (30 sp.) Enhydris (5 sp.)

Soft invertebrates Fish Amphibians Snapping Shrimp Crayfish Crabs Hard

Molt

Fig. 2.2. Scaling relationships for morphological data

A 2.8 B 2.8 C R. septemvittata

2.6 2.6 )

2 L. alleni )

m m

m 2.4

) 2.4

(

s e

r r

s 2.2 2.2

e e

m 1

g a

a 2.0 2.0

o l l 1.8 1.8

0 1.6 1.6 2.0 2.2 2.4 2.6 2.8 2.0 2.2 2.4 2.6 2.8 0.0 0.6 1.2 1.8 2.4 log SVL (mm) log SVL (mm) log Mass (g)

Fig. 2.3. Computed tomography scans of R. septemvittata (A) and L. alleni (B)

SW QD QD

LJ LJ

SI 1 cm B

SW QD QD

LJ LJ

Fig. 2.4. Relationships between the absolute and relative sizes of prey consumed in the field and the morphology of R. septemvittata (n = 180) and L. alleni (n = 43)

R. septemvittata L. alleni

25 A 25 B )

) 20 crayfish 20

% %

cheliped (

(

other y

y c

c 15 15

u q

q 10 10

F F 5 5

0 0 0 20 40 60 80 100 120 >140 0 20 40 60 80 100 120 >140 RPA (%) RPA (%)

20 20 )

) C D

g g

( (

s s s

s 15 15

a a

m m

y y

e e r

r 10 10

p p

t t

c c

a a t

t 5 5

n n

I I

0 0 0 25 50 75 100 125 0 25 50 75 100 125 Snake mass (g) Snake mass (g) 20 ) 20

) F m

E m

(

15 r 15

a a

i 10

e r 5 r

P 5 P

0 0 100 200 300 400 500 600 100 200 300 400 500 600 Snake SVL (mm) Snake SVL (mm) G H

150 150

)

100

x 100

(

50 50

R R 0 0 100 200 300 400 500 600 100 200 300 400 500 600 Snake SVL (mm) Snake SVL (mm)

Fig. 2.5. Prey handling behaviors and prey handling times A B C

D R. septemvittata E L. alleni F hard soft soft medium 3 3 hard 3 F.l. L.a.

C.v. R.s.

)

) )

s s

s G.p.

(

( (

T T T

H H

2 2 2

g g

o o o

l l

1 1 1

0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0 log RPA (% max gape) log RPA (% max gape) log RPA (% max gape)

Fig. 2.6. Constraints of gape on feeding performance for four species of crustacean-eating specialists

75 A SVL = 150 mm 75 B SVL = 450 mm

)

s s

s R. septemvittata

a a

L. alleni e

e 50 50 k

k F. leucolbalia

a a

n s

s G. prevostiana

( (

25 25

P P

R R 0 0 0 20 40 60 80 100 0 20 40 60 80 100 RPA (% max gape) RPA (% max gape)

75 C RPA = 100% 75 D RPA = 100%

)

s s

a a

50 e 50

k k

a a

s s

% % (

25 (

M M

P P

R R 0 0 200 300 400 500 50 100 150 SVL (mm) Snake mass (g)

E R. septemvittata F L. alleni )

80 ) 80

s s

a a

m m

60 60

e e

k k

a a

n n s

s 40 40

% %

( (

20 20

M M

P P R R 0 0 200 300 400 500 600 200 300 400 500 600 SVL (mm) SVL (mm)

Fig. 2.7. Three-dimensional maximal feeding performance spaces for four species of crustacean-eating snakes

A Regina septemvittata B Liodytes alleni 70 70

60 60

)

s a

50 a 50

m m

k a

40 a 40

(

30 30

P R 20 R 20

10 10

100 100 0 80 0 80 200 ) 200 ) 60 pe 60 pe 300 40 ga 300 ga S 400 ax S 40 ax VL m VL 400 m (mm 20 (% (m 20 % ) 500 PA m) 500 A ( 0 R 0 RP

C Gerarda prevostinia D Fordonia leucobalia 70 70

60 60

) )

s a

a 50 50

m m

e e

k a

a 40 40

n n

(

30 30

P P R R 20 20

10 10

0 100 100 80 0 80 200 60 ) 200 ) 300 ape 60 pe 40 g 300 ga SV 400 ax S 400 40 ax L (m 20 m VL ( 20 m m) 500 (% mm 500 (% 0 PA ) 0 PA R R

SUPPLEMENTAL INFORMATION Table S2.1. Regression analyses of morphology and handling time

Dependent Independent Slope ± Intercept ± n R2 Variable variable 95% CL 95% CL R. septemvittata log Mass (g) log SVL (mm) 2.930 ± 0.127 -6.156 ± 0.314 27 0.99 log Diam (mm) log SVL (mm) 0.815 ± 0.054 -0.874 ± 0.133 27 0.97 log Area (mm2) log SVL (mm) 1.630 ± 0.108 -1.854 ± 0.266 27 0.97 log Diam (mm) log Mass (g) 0.274 ± 0.024 0.843 ± 0.028 27 0.96 log Area (mm2) log Mass (g) 0.548 ± 0.048 1.581 ± 0.056 27 0.96 log HT (s) log RPA (%) (all) 1.329 ± 0.211 0.444 ± 0.306 118 0.57 log HT (s) log RPA (%) (soft) 1.689 + 0.293 -0.239 + 0.436 38 0.79 log HT (s) log RPA (%) (medium) 1.134 + 0.257 0.795 + 0.368 80 0.49 log HT (s) log RPM (%) (all) 0.826 ± 0.135 1.629 ± 0.129 118 0.55 log HT (s) log RPM (%) (soft) 1.080 + 0.184 1.238 + 0.188 38 0.79 log HT (s) log RPM (%) (medium) 0.692 + 0.165 1.817 + 0.151 80 0.46 L. alleni log Mass (g) log SVL (mm) 3.007 ± 0.150 -5.970 ± 0.354 30 0.98 log Diam (mm) log SVL (mm) 0.796 ± 0.045 -0.758 ± 0.106 30 0.98 log Area (mm2) log SVL (mm) 1.593 ± 0.090 -1.620 ± 0.212 30 0.98 log Diam (mm) log Mass (g) 0.261 ± 0.018 0.827 ± 0.022 30 0.97 log Area (mm2) log Mass (g) 0.522 ± 0.036 1.549 ± 0.044 30 0.97 log HT (s) log RPA (%) (all) 1.262 ± 0.237 0.480 ± 0.364 127 0.47 log HT (s) log RPA (%) (molted) 1.676 ± 0.297 -0.339 ± 0.479 66 0.66 log HT (s) log RPA (%) (hard) 1.501 ± 0.296 0.308 ± 0.429 61 0.63 log HT (s) log RPM (%) (all) 0.546 ± 0.142 1.984 ± 0.123 127 0.31 log HT (s) log RPM (%) (molted) 0.719 ± 0.190 1.708 ± 0.182 66 0.46 log HT (s) log RPM (%) (hard) 0.581 ± 0.203 2.094 ± 0.154 61 0.35

Notes: All regressions had P < 0.001. Prey types in analyses for L. alleni log HT were: all = soft, medium and hard-shell prey, molted = soft-shell and medium prey hardness, and hard = hard- shell prey only.

Table S2.2. Summary of prey handling behaviors during laboratory trials

Behavior Percent of trails with behavior R. septemvittata L. alleni (n = 118) (n = 127) Crayfish Pre-capture tail flip + escape 27% R 13% R Pre-capture pinching 5% 0% Post-capture tail flip 36% R, H 46% R, H Post-capture tail flip + escape 12% 4% R, H Post-capture pinch 19% H 48% H Post-capture pinch + escape 2% 1% Snake Cheliped removal 16%† 0% >1 unsuccessful strike 26% R, H 13% Strike location Cheliped 2% 0% Carapace 49% 27% Carapace-abdomen joint 18% 35% Abdomen 31% 38% R Hold prey before swallowing 82% R, H 11% Body restraint 0% 69% U-loop 0% 18%†† Pin 0% 28%†† Coil 0% 43%†† Crayfish orientation at swallow Dorsal 24% 1% Lateral 61% 90% Ventral 14% 9% Direction Swallow Head 22% 2% Mid-body 3% 1% Tail 73% H 97% R Chela 3% 0%

Notes: R and H indicate that when the presence (1) or absence (0) of a behavior in the left column was a dependent variable in a univariate regression, its presence was more likely with increased RPA and prey hardness, respectively (Table S2.5). †, total n = 120 including 2 trials in which only the cheliped was removed; †† More than one type of body restraint behavior could occur within a single trial.

Table S2.3. Mean values of prey handling times during laboratory trials

Behavior Mean values ± SE (range) R. septemvittata (n = 118) L. alleni (n = 127) Attack time (s) 57.7 ± 12.9 (0–711) 14.4 ± 3.9 (0–326) Holding (s) 79.1 ± 7.1 (0–488) R, H 2.4 ±0.7 (0–49) Lateral Jaw Walking (s) 82.1 ± 8.6 (0–402) R 50.1 ± 4.9 (1–293) R Pre-swallow pause (s) 20.6 ± 4.5 (0–283) R 10.0 ± 3.0 (0–203) Swallowing (s) 137 ± 19.3 (2–1584) R 214 ± 17.0 (12–1076) R Swallow pause (s) 23.0 ± 4.7 (0–350) R 93.1 ± 10.8 (0–671) R, H Handling time, HT (s) 342 ± 31.7 (6–2376) R, H 369 ± 28.9 (15–1823) R, H

Notes: R and H indicate that when the duration of a behavior in the left column was a dependent variable in a univariate regression, it changed significantly with increased RPA and prey hardness, respectively (Table S2.5).

Table S2.4. Final multiple regression models for total handling time and the occurrence of some behaviors

Model Dependent Variable Independent Variable Coefficient P R. septemvittata (n = 118) 1 log HT (s) log RPA 1.342 <0.001 2 (R = 0.66) Prey Hardness 0.198 <0.001 Strike Location 0.078 0.007 Num. of Pinches 0.094 0.025 Constant 0.102 0.517

2 Presence of holding log RPA 0.770 <0.001 2 (R = 0.42) Prey Hardness 0.312 <0.001 Strike Location 0.084 0.007 Constant -0.665 <0.001 L. alleni (n = 127)

3 log HT (s) log RPA 1.387 <0.001 2 (R = 0.71) Prey Hardness 0.087 <0.001 Num. of Pinches 0.039 0.007 Prey Restraint 0.057 0.001 Constant -0.049 0.769

4 Type prey restraint log RPA 1.983 <0.001 2 (R = 0.21) Prey Hardness 0.254 <0.001 Num. of Pinches 0.174 0.017 Constant -2.057 0.016

Notes: RPA = crayfish area/ snake maximal gape area; prey hardness: 0 = soft, 1 = medium soft,

2 = hard; strike location: 1 = tail, 2 = junction of tail and abdomen, 3 = carapace; Num. of

Pinches = number of pinches after crayfish capture; type of prey restraint: 0 = none, 1 = U-loop,

2 = pinning, 3 = coil.

Table S2.5. Univariate regressions for behavior and handling times

Model Independent Dependent Variable Slope ± Intercept ± R2 P Variable 95% CL 95% CL R. septemvittata (n=118) 1 log RPA log HT 1.329 ± 0.211 0.444 ± 0.306 0.57 <0.001 1 Prey Hardness log HT 0.172 ± 0.179 2.220 ± 0.147 0.02 0.059 1 Num. Pinches log HT 0.168 ± 0.131 2.293 ± 0.089 0.04 0.012 1 Location of Strike log HT 0.094 ± 0.095 2.139 ± 0.218 0.02 0.054 2 log RPA Presence of Holding 0.729 ± 0.232 -0.216 ± 0.336 0.24 <0.001 2 Prey Hardness Presence of Holding 0.281 ± 0.141 0.632 ± 0.116 0.11 <0.001 2 Strike Location Presence of Holding 0.094 ± 0.079 0.623 ± 0.180 0.04 0.019 log RPA Num. Unsuccessful Strikes -0.788 ± 0.646 1.589 ± 0.936 0.04 0.017 Prey Hardness Num. Unsuccessful Strikes -0.399 ± 0.365 0.737 ± 0.301 0.03 0.032 Prey Hardness Direction swallow 0.338 ± 0.164 0.500 ± 0.135 0.12 <0.001 log RPA Hold Time 111.9 ± 49.94 -80.33 ± 72.31 0.15 <0.001 Prey Hardness Hold Time 51.20 ± 28.83 44.34 ± 23.74 0.10 <0.001 log RPA Pre-Swallow Pause 19.81 ± 18.48 -22.49 ± 26.76 0.03 0.036 log RPA Lateral Jaw Walk Time 172.3 ± 86.51 -142.7 ± 125.3 0.12 <0.001 log RPA Swallow Pause 80.05 ± 33.02 -91.02 ± 47.82 0.16 <0.001 log RPA Swallow Time 541.9 ± 128.7 -612.1 ± 186.3 0.37 <0.001 log RPA Presence of Biting 0.618 ± 0.263 -0.092 ± 0.381 0.15 <0.001 Prey Hardness Presence of Biting 0.309 ± 0.150 0.580 ± 0.124 0.12 <0.001 log RPA Tail flip+escape (PreCap) -0.318 ± 0.306 0.724 ± 0.443 0.03 0.042 Prey Hardness Tail flip (PostCap) 0.188 ± 0.186 0.237 ± 0.153 0.03 0.048 Prey Hardness Pinch (PostCap) 0.236 ± 0.147 0.026 ± 0.121 0.08 0.002 L. alleni (hard, n=61) log RPA Prey Restraint 2.428 + 1.895 -1.448 + 1.307 0.19 <0.001 L. alleni (soft, n=66) log RPA Prey Restraint 2.027 + 2.563 -1.802 + 1.592 0.09 0.013 L. alleni (all, n=127) 3 log RPA log HT 1.262 ± 0.237 0.480 ± 0.364 0.47 <0.001 3 Prey Hardness log HT 0.038 ± 0.043 2.312 ± 0.121 0.02 0.082 3 Num. Punches log HT 0.102 ± 0.045 2.287 ± 0.083 0.13 <0.001 3 Type prey restraint log HT 0.158 ± 0.048 2.127 ± 0.105 0.24 <0.001 4 log RPA Prey Restraint 1.437 ± 0.998 -0.460 ± 1.533 0.05 0.005 4 Prey Hardness Prey Restraint 0.189 ± 0.132 1.293 ± 0.375 0.04 0.005 4 Num. of Pinches Prey Restraint 0.271 ±0 .146 1.430 ± 0.269 0.09 <0.001 log RPA Strike Location 0.802 ± 0.710 0.719 ± 1.091 0.03 0.027 log RPA Lateral Jaw Walk Time 117.8 ± 57.69 -118.9 ± 88.61 0.11 <0.001 log RPA Direction Swallow -0.217 ± 0.182 1.322 ± 0.280 0.03 0.020 log RPA Swallow Pause (s) 226.6 ± 88.06 -251.2 ± 135.3 0.17 <0.001 Prey Hardness Swallow Pause (s) 16.33 ± 12.47 55.81 ± 12.47 0.04 0.011 log RPA Swallow Time 736.7 ± 188.5 -812.7 ± 289.6 0.32 <0.001 Prey Hardness Presence of Biting 0.057 ± 0.047 0.161 ± 0.133 0.04 0.017

log RPA Tail flip+escape (PreCap) -0.292 ± 0.266 0.578 ± 0.408 0.03 0.031 log RPA Tail flip (PostCap) 0.889 ± 0.364 -0.887 ± 0.558 0.15 <0.001 Prey Hardness Tail flip (PostCap) -0.052 ± 0.052 0.590 ± 0.146 0.03 0.037 log RPA Tail flip+escape (PostCap) 0.163 ± 0.152 -0.209 ± 0.233 0.03 0.035 Prey Hardness Tail flip+escape (PostCap) -0.026 ± 0.020 0.099 ± 0.057 0.04 0.010 Prey Hardness Pinch (PostCap) 0.088 ± 0.050 0.279 ± 0.142 0.08 0.001

Notes: Independent variables that were also significant in multiple regressions are labelled with the number of the multiple regressions that are in Table S2.4. Variable abbreviations include:

RPA, crayfish area/ snake maximal gape area; prey hardness ( 0 = soft, 1 = medium soft, 2 = hard); HT = handling time; presence of holding ( 0 = no holding, 1 = holding); prey restraint ( 0

= no body restraint, 1 = u-loop, 2 = body pin, 3 = coil); presence of biting (0 = no biting, 1 = > 1 bite); direction swallow ( 0 = head first, 1 = tail first); PreCap = crayfish behavior pre-capture;

PostCap = crayfish behavior post-capture.

Table S2.6. ANCOVA results for morphology and behaviors comparing the two species and sexes within a species

Factor Covariate Dependent Effect Factor Covariate Variable FDF (P) X Factor Interaction FDF (P)

Species log Mass (g) log SVL (mm) La > Rs 402.711,56 (<0.001) 0.621,56 (0.436) 2 Species log Gape (mm ) log SVL (mm) La > Rs 123.151,56 (<0.001) 0.311,56 (0.580) 2 Species log Gape (mm ) log Mass (g) Rs > La 15.021.56 (<0.001) 0.821,36 (0.371)

Sex (Rs) log Mass (g) log SVL (mm) M > F 5.711,26 (0.025) 0.421,26 (0.522) 2 Sex (Rs) log Gape (mm ) log SVL (mm) F = M 1.711,26 (0.203) 0.071,26 (0.797) 2 Sex (Rs) log Gape (mm ) log Mass (g) F > M 5.451,26 (0.028) 1.691,26 (0.207)

Sex (La) log Mass (g) log SVL (mm) F = M 0.7801,29 (0.385) 1.421,29 (0.245) 2 Sex (La) log Gape (mm ) log SVL (mm) n/a n/a 4.701,29 (0.040) 2 Sex (La) log Gape (mm ) log Mass (g) F = M 0.011,29 (0.915) 0.461,29 (0.504)

Species Log RPA (all) Attack Time (s) Rs > La 13.081,244 (<0.001) 2.671,241 (0.103)

Species log RPA (all) Num. Missed Strikes Rs > La 6.301,244 (0.013) 2.421,244 (0.121)

Species log RPA (all) Strike location Rs = La 3.521,244 (0.062) 2.411,244 (0.122)

Species log RPA (all) Num. of Bites Rs > La 5.101,244 (0.025) 0.6951,244 (0.405)

Species log RPA (all) Hold Time (s) n/a n/a 18.21,244 (<0.001)

Species log RPA (all) Lat Jaw Walk (s) Rs > La 18.561,244 (<0.001) 1.031,244 (0.311)

Species log RPA (all) Orientation Swallow n/a n/a 4.951,244 (0.027)

Species log RPA (all) Direction Swallow La > Rs 33.601,244 (<0.001) 2.781,244 (0.097)

Species log RPA (all) Pre-Swallow Pause Rs > La 8.111,244 (0.005) 3.511,244 (0.062)

Species log RPA (all) Swallow Pause n/a n/a 10.081,244 (0.002)

Species log RPA (all) Swallow Time(s) La > Rs 9.641,244 (0.002) 2.921,244 (0.089)

Species log RPA (all) log HT (s) Rs = La 2.591,244 (0.109) 0.1741,244 (0.677)

Species log RPA (molt) log HT (s) Rs > La 27.81,83 (<0.001) 2.881,244 (0.091)

Species log RPA (hard/molt) log HT (s) La > Rs 5.8141,178 (0.017) 0.771,178 (0.381)

Species log RPM (all) log HT (s) n/a n/a 7.971,244 (0.005)

Notes: Factors include: species (La = L. alleni, Rs = R. septemvittata); and sex (F = female, M =

male). Covariates include: Gape Area and Diam = maximum gape area and diameter of both

species; RPA = crayfish area/ snake maximal gape area; RPM = crayfish mass/snake mass;

parenthetical notation next to RPA and RPM indicate the prey hardness of (La/Rs): all = soft,

medium, and hard-shell prey, soft = soft and medium prey, hard = hard-shell prey only.

Dependent variables include: strike location ( 1 = tail, 2 = junction of tail and abdomen, 3 =

72 carapace); Orientation swallow ( 1 = lateral, 2 = dorsal, 3 = ventral); direction swallow ( 0 = head first, 1 = tail first; HT = handling time). The effect column indicates which sex or species had larger values of the dependent variable n/a = equality of slopes failed and could not test for the effect of species or sex.