Prey or predator? Body size of an approaching animal affects decisions to attack or escape

Behavioral Ecology doi:10.1093/beheco/arq142 Advance Access publication 21 September 2010 Prey or predator? Body size of an approaching animal affects decisions to attack or escape William E. Cooper Jr a and Theodore Stankowich b a Department of Biology, Indiana University Purdue University Fort Wayne, 2101 East Coliseum Boulevard, Fort Wayne, IN 46835, USA and b Program in Organismic and Evolutionary Biology, Department of Biology, Morrill Science Center South, 611 North Pleasant Street, University of Massachusetts, Amherst, MA 01003, USA An animal must assess an approaching heterospecific to determine whether it is prey or predator. Because prey tend to be smaller than their predators, foraging responses decrease and escape responses are predicted to increase over a range of approacher size. Sufficiently, small approachers may be readily identified as prey and large ones as predators. However, more time and closer approach may be required to assess approachers of intermediate size, leading to predictions that predator/prey evaluations may differ among individuals, that attacks on the approacher begin from shorter distances than for smaller approachers and that escape begins at shorter distance than for larger approachers. In a field experiment, we dragged tethered model orthopterans of 3 sizes toward 2 lizard species that differ in body size. As model size increased, lizards were more likely to flee and had longer flight initiation distance. As model size decreased, lizards were more likely to advance toward and bite models and began advances from longer distances. At the intermediate model size, some individuals fled and others attacked intermediate models. Sceloporus jarrovii, the larger species, was more likely to attack and less likely to flee than S. virgatus at intermediate model size. The prediction of optimal escape theory that flight initiation distance increases with predator size due to greater risk was verified. Our evidence indicates that as approacher size increases, successive transitions occur between rapid assessment of the approacher as prey, uncertainty requiring closer inspection, and rapid identification of a potential predator. Key words: body size, escape behavior, flight initiation distance, foraging behavior, Squamata. [Behav Ecol 21:1278 1284 (2010)] While Although being approached, an animal may evaluate the approacher to determine whether it is food, a predator, or neither. The body size of the approaching animal relative to its own is one factor likely to affect an animal s assessment (Dickman 1988; Costa 2009). Some approachers may be too small to be worth attacking as food, but as approacher size increases, the approacher may be large enough to be included in the diet (Costa et al. 2008; Costa 2009). With further size increase, it may become too large to eat (Owen- Smith 2008) and may be more likely to be a predator (Robinson and Wellborn 1987). Ontogenetic changes in body size may reverse predator prey relationships between species (Woodward and Hildrew 2002). Thus, decisions about feeding and escaping both must be affected by the size of the approacher. For a prey, predation risk increases as predator size increases because larger predators are likely to be faster and more lethal. Because greater speed and lethality increase the threat posed by an approaching predator while it is at any particular distance from the prey, they may strongly influence escape decisions. Optimal escape theory predicts flight initiation distance, the distance separating an approaching predator from prey when escape begins (Cooper and Frederick 2007). In the scenario of optimal escape theory, a prey detects a predator, monitors its approach, and flees when the predator reaches the optimal flight initiation distance, where the prey s expected postencounter fitness is maximized. Escape decisions are affected by the prey s fitness when the encounter begins, energetic Address correspondence to W.E. Cooper. E-mail: cooperw@ipfw. edu. Received 10 February 2010; revised 9 May 2010; accepted 2 July 2010. Ó The Author 2010. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. Downloaded from For permissions, https://academic.oup.com/beheco/article-abstract/21/6/1278/332995 please e-mail: journals.permissions@oxfordjournals.org and opportunity costs of fleeing, and predation risk (Cooper and Frederick 2007). Optimal flight initiation distance increases as predation risk, which is affected by many factors (Stankowich and Blumstein 2005), increases. Predator approach speed is a major risk factor (reviewed by Stankowich and Blumstein 2005; lizard prey: Cooper et al. 2003; Cooper 2009a). Faster potential closing speed by larger predators might be taken into account by prey assessing risk. If so, flight initiation distance is predicted to be longer for larger predators. Optimal flight initiation distance also increases as lethality if the prey is overtaken increases (Cooper and Frederick 2010). This prediction is supported by longer flight initiation distance in response to larger predators (Stankowich and Blumstein 2005) and by prey with less body armor (McLean and Godin 1989; Abrahams 1995). Probability of fleeing when approached by a predator increases as risk increases, as shown by many studies of the effect of directness of approach by predators (reviewed by Stankowich and Blumstein 2005). If a dangerous predator approaches directly, all prey are expected to flee, as occurs in lizard studies (Cooper 2003; Cooper et al. 2003; Cooper and Whiting 2007). When predators approach along paths that pass near the prey without contact, probability of fleeing decreases as the minimum distance on the path from the prey s location increases. This suggests that probability of fleeing decreases as perceived risk decreases. In a range of predator size that must be determined empirically, predation risk increases with predator size. Therefore, we predicted that probability of fleeing increases as predator size increases. Prey size influences diet selection. Gape-limited predators cannot swallow prey bigger than some maximum size (Webb and Shine 1993; Forsman and Shine 1997). As size of a potential prey item increases, nutritive benefits increase, but energetic costs of subduing the prey and the risk of being injured increase. Thus, starting at very small prey size, the probability

Cooper and Stankowich Body size affects foraging and escape decisions 1279 of selecting the prey may increase with prey size, then reach a maximum (Curio 1976). As prey size increases beyond the maximum, the probability of attacking the prey is expected to decline as profitability decreases. At and above some size, potential prey may become unprofitable and should not be attacked (Curio 1976; Forsman 1996). When an approaching animal is close to the typical prey size or large enough that it is very likely to be a predator, animals can rapidly assess its status. If it is between these sizes, animals being approached initially may be uncertain whether the approacher is a prey item, a predator, or neither. To obtain additional information before assessing the nature of the approacher, an animal may require more information. A recent study of escape by the salticid spider Phidippus princeps (Stankowich 2009) provides a possible example. Typically, flight initiation distance increases with predator approach speed, but flight initiation distance by the spiders was unaffected by approach speed of a small black model that was large enough to be a potential predator but small enough to be a potential prey item (Stankowich 2009). Approach speed may have been irrelevant to risk assessment if the spiders were uncertain whether the model was a prey item or a predator (Stankowich 2009). Attack distance is expected to be longest when the approaching animal is large enough to be prey but too small to be a predator. As approacher size increases, attack distance should become shorter because prey require more information to determine whether the approacher is prey or predator, reaching zero when the approacher is too large to be prey. Based on uncertainty and need for more information to assess the status of approacher of intermediate size, we predicted that animals may allow closer approach before attacking, fleeing, or neither when the approacher s size is in this range. We examined effects of size of an approaching model on attack and escape decisions by 2 ambush foraging lizards, Sceloporus virgatus and S. jarrovii, the latter being somewhat larger (maximum SVL ¼ snout-vent length 106 mm vs. 71 mm; Stebbins 2003). As model size increases, we predicted that 1) probabilities of attacking and biting the model and distance from which attack was initiated decreases and 2) probability of fleeing and flight initiation distance increase. We also predicted that the larger species 1) would attack a higher percentage of larger models and initiate attacks from a greater distance, 2) would be less likely to flee from smaller models, and 3) would exhibit shorter flight initiation distance to models of intermediate size. METHODS Study site, animals, and pretrial procedure The study site located in the Chiricahua Mountains of southeastern Arizona in June 2009 at an altitude range of 1600 1800 m. The lizards were found in dry creek beds and adjacent open forest. During much of the activity season, S. virgatus occupies ground more frequently and logs less frequently than S. jarrovii, but both species are often observed on rocks, trees, and ground (Smith 1996). Smith (1996) observed few S. virgatus and S. jarrovii on trees. However, about one fourth of S. virgatus were on trees and nearly 70% were on rocks or ground at the same time of year as the present study (Cooper and Wilson 2007). The difference may reflect seasonal and habitat differences as well as restriction of data to adults in the latter study. Trees are important escape destinations for both species (Cooper and Wilson 2007). Observations were limited to warm, sunny days at times between 9:00 AM and 3:30 PM when lizards had completed their postemergence thermoregulatory basking. Lizards were located by searching visually while walking very slowly through the study area. Response was tested only for those sighted on ground or perched on trees, logs, or rocks no higher than 0.3 m above ground. We restricted observations to adult lizards: approximate SVLs were 50 65 mm for S. virgatus and 75 95 mm for S. jarrovii. In June, the smallest adult S. jarrovii have 68-mm SVL (Ruby and Dunham 1984), longer than all but a single S. virgatus in one of 3 years of a demographic study (Vinegar 1975) and larger than any observed in the present study. Models, their introduction, and behavioral variables Models presented to lizards were brown grasshoppers of 3 sizes: 25-, 50-, and 107-mm body length. The 50- and 107-mm models were made of plastic and were obtained from toy dealers. The 25-mm models were dead crickets that were soaked in polyurethane for several days and then dried. The plastic models were painted brown to match the color of the crickets. Thus, the models differed not only in size but also slightly in body form and color. The possible effect of polyurethane on detectability and responses to the smaller models is considered in the Discussion. Slight differences in body form between the crickets and model grasshoppers are very unlikely to have affected escape and attack decisions because both lizard species consume a wide range of invertebrate prey having diverse body sizes and shapes (Barbault et al. 1985; Phelan and Niessen 1989; Watters 2008). Dietary studies of S. jarrovii and S. virgatus reveal consumption of orthopterans (Barbault et al. 1985; Goldberg and Bursey 1990; Bursey and Goldberg 1993; Watters 2008), and both species readily consume crickets and grasshoppers (Phelan and Niessen 1989; Cooper WE, personal observations). To ensure that models were detectable by lizards throughout trials, to permit steady movement of models and to minimize noise, testing was done only on open ground devoid of vegetation or other obstructions. On detecting a testable lizard, the experimenter (W.E.C.) approached slowly enough to move into a position permitting testing at close range without first eliciting escape. Then he stopped 1.5 2.0 m from the lizard and stood motionless for 5 s before beginning a trial. For introduction to lizards, models were tied to the end of a 1.7-m rod with a 0.5-m string tether. At the outset of a trial, the experimenter lowered a model to the ground 2 m from a lizard s head and then dragged it toward the lizard as smoothly as possible. Dragging stopped when a lizard attacked or fled but continued until the model touched the lizard if the lizard neither attacked nor fled. If a lizard approached the model, the model stopped moving when the lizard reached it. If a lizard initially approached but stopped short of the model, the model continued moving until the lizard fled or the model reached it. Dragging speed was ;0.3 m/s. This speed is slightly faster than the 0.2 m/s recorded during escape runs by the cricket Gryllus bimaculatus, which has maximum body length of 23 mm (Gras and Hörner 1992). Because the model approached the lizard in all trials, we describe movement toward the model by lizards as advance rather than approach to avoid possible confusion. The experimenter recorded whether or not a lizard advanced toward the model, whether or not it bit the model, and whether or not it fled. It was possible for a lizard to advance without attacking, then stop, and sometimes flee. Whichever of these behaviors occurred were recorded. In addition to these binary variables, 2 continuous variables were recorded: attack distance and flight initiation distance. Attack distance is the distance between the model and the lizard when the lizard begins to advance toward the model. Distance variables were measured to the nearest 0.1 m.

1280 Behavioral Ecology Experimental design and analysis Each lizard was tested using all 3 model sizes using a testing sequence that was completely counterbalanced among individuals to preclude any bias due to testing order (n ¼ 24 for S. virgatus, n ¼ 12 for S. jarrovii). Thus, all possible sequences of model size were used with equal frequency for both species. No lizards entered refuge after fleeing or between trials. Between successive trials, the experimenter withdrew several meters, exchanged models, moved into position for the next trial, and conducted the next trial. The repeated measures design controlled effects of individual differences in variables, such as hunger and boldness. No individual was tested for more than one series of 3 trials, which was ensured by noting the position of the lizard after the trial and moving at least 10 m along a transect through the study area before selecting the next lizard for testing. This procedure was adequate to prevent retesting because none of the lizards fled far or to unknown locations after the trials. Each area only use only once during the study, a further precaution to avoid pseudoreplication. Differences in proportions of lizards that advanced, bit, and fled between model sizes were tested for significance in each species using Cochran Q tests. Where significant differences among models were detected within species, comparisons between pairs of means were made using a nonparametric multiple comparisons procedure (Zar 1996). Differences between species in these variables were tested separately for each model size using Fisher Exact probability tests. Differences among model sizes and species for attack distance and flight initiation distance were examined using 2 3 3 factorial analyses of variance with repeated measures on model size. Data were rank transformed prior to analysis because the assumption of normality was violated by the presence of many zero values, which indicate that the lizard did not move before being touched by the model. Newman Keuls tests were used for paired comparisons between species for the same model size and within species between model sizes. Significance tests were 2-tailed, with a ¼ 0.05 except where indicated otherwise and justified by directional prediction. A sequential Bonferroni procedure was used to determine the adjusted alpha level for each of the 3 Fisher Exact tests within species (Wright 1992). Effect sizes are reported as r equivalent for Fisher Exact tests (Rosenthal and Rubin 2003), g q 2 for ffiffiffiffiffiffiffiffi analyses of variance (Cohen 1992), and Cramer s V ¼ for Q, X 2 dfðnþ which follows a chi-squared distribution (Gliner et al. 2002). The latter is interpreted as r with df ¼ 1. Figure 1 The proportions of individuals Sceloporus jarrovii and S. virgatus that fled when approached by models of 3 sizes. Error bars represent 1.0 standard error. (P, 0.01) models (Figure 1). Interspecific comparisons show that probability of fleeing was similar for 25 mm (Fisher Exact P ¼ 0.54) and 107 mm models (Fisher P ¼ 1.00) but was significantly greater for S. virgatus than S. jarrovii for 50 mm models (Fisher P ¼ 0.0067, r equivalent ¼ 0.41; Figure 1). Lizards permitted close approach by all models (Figure 2). All effects in the species 3 model size analysis of variance of rank data were significant (species: F 1,102 ¼ 19.35, P ¼ 2.7 3 10 25, g 2 ¼ 0.04; model size: F 2,102 ¼ 181.36, P, 1.0 3 10 26, g 2 ¼ 0.74; species 3 model size: F 2,102 ¼ 4.10, P ¼ 0.019, g 2 ¼ 0.02). Although the interaction was significant, model size was by far the most important factor affecting flight initiation distance, as indicated by its very large effect size. In S. virgatus, flight initiation distance differed significantly between all pairs of model size (P, 0.00012 each; Figure 2). In S. jarrovii, flight initiation distance was slightly and significantly longer for the 50 mm than 25 mm model (P, 0.048), and the remaining paired comparisons were highly significant (P, 0.00012 each; Figure 2). Flight initiation distance did not differ between species for 25 mm models (P. 0.10) but was significantly longer for S. virgatus than S. jarrovii when responding to 50 mm (P, 0.00012) and 107 mm (P ¼ 0.020) models (Figure 2). Thus, the source of the interaction appears to have been a slightly greater flight initiation distances RESULTS Probability of fleeing and flight initiation distance For both species, the probability of fleeing was very low when approached by 25-mm models. It was higher for 50-mm models. All lizards of both species fled from 107-mm models (Figure 1). For S. virgatus, the probability of fleeing differed significantly among model sizes (Q 2 ¼ 35.57, P, 1.0 3 10 26 ; Figure 1), and the effect size (V ¼ 0.86) was very large. Lizards were significantly more likely to flee when approached by 107- (P, 0.001) or 50-mm (P, 0.0001) models than by 25-mm models and when approached by 107 mm than 50 mm models (P, 0.02). For S. jarrovii, the probability of fleeing differed significantly among model sizes (Q 2 ¼ 18.67, P ¼ 0.00018; Figure 1), and the effect size (V ¼ 0.88) was very large. The probability of fleeing did not differ significantly between approaches by 25 mm and 50 mm models (P. 0.10), but lizards were significantly more likely to flee when approached by 107 mm models than by 50 mm (P, 0.05) and 25 mm Figure 2 Flight initiation distances of Sceloporus jarrovii and S. virgatus when approached by models of 3 sizes. Error bars represent 1.0 standard error.

Cooper and Stankowich Body size affects foraging and escape decisions 1281 by S. virgatus than S. jarrovii in response to 50 mm and 107 mm models combined with equal flight initiation distance for the 2 species in response to 25 mm models. Probabilities of advancing and biting and attack distance The relationship between model size and the proportion of lizards that advanced showed a reverse trend to that for numbers that fled (Figure 3). For S. virgatus, the probability of advancing toward the model differed significantly among model sizes (Q 2 ¼ 31.14, P, 1.0 3 10 26 ; Figure 3), and the effect size (V ¼ 0.81) was very large. Lizards were significantly more likely to advance toward 25 mm models than 50 mm (P, 0.005) and 107 mm (P, 0.001) models and significantly more likely to advance toward 50 mm than 107 mm (P, 0.05) models. For S. jarrovii, the probability of advancing toward the model differed significantly among model sizes (Q 2 ¼ 22.00, P ¼ 3.4 3 10 25 ; Figure 3), and the effect size (V ¼ 0.96) was very large. Lizards were significantly more likely to advance toward 25 mm models and 50 mm models than 107 mm models (P, 0.05 each). All lizards advanced toward the 2 smaller models (P ¼ 1.00) but only one advanced toward a 107 mm model (and then fled). The probability of advance did not differ between species for 25 mm or 107 mm models (Fisher P. 0.10 each) but was significantly greater for S. jarrovii than S. virgatus for 50 mm models (Fisher P, 1.0 3 10 26, r equivalent. 0.70; Figure 3). Lizards were more likely to bite smaller models (Figure 4). In S. virgatus, the probability of biting the model differed significantly among models by size (Q 2 ¼ 18.50, P ¼ 0.00019; Figure 4), and the effect size (V ¼ 0.62) was very large. None of the 24 lizards bit the largest model, only one bit the model of intermediate size, and 11 bit the smallest models. Lizards were significantly more likely to bite the 25 mm models than the 50 mm and 107 mm models (P, 0.01 each). In S. jarrovii, the probability of biting the model differed significantly among models by size (Q 2 ¼ 18.67, P ¼ 0.00018; Figure 4). The effect size, V ¼ 0.88, is very large. Probability of biting was significantly greater for the 25 mm models than for the 50 mm (P, 0.05, 1-tailed) or 107 mm (P, 0.005) models and for the 50 mm than 107 mm models (P, 0.02; Figure 4). The proportion of individuals that bit models was significantly greater for S. jarrovii than S. virgatus for the 25 mm models (Fisher P ¼ 0.0022; r equivalent ¼ 0.46) and 50 mm models (Fisher P ¼ 0.00014, r equivalent ¼ 0.57); no lizard of either Figure 4 The proportion of individuals of Sceloporus jarrovii and S. virgatus that bit models of 3 sizes. Error bars represent 1.0 standard error. species bit the 107 mm models (Figure 4). In addition, 5 S. virgatus tongue-flicked, but did not bite, 25 mm models, suggesting chemosensory investigation of a potential food source. If these individuals are treated as if they viewed the 25 mm models as potential prey, the interspecific difference remains significant (Fisher P, 0.024, 1-tailed). All 3 effects of the 2 3 3 factorial analysis of rank data for attack distance were significant (species: F 1,102 ¼ 13.16, P ¼ 0.00045, g 2 ¼ 0.13; model size: F 2,102 ¼ 40.99, P, 1.0 3 10 26, g 2 ¼ 0.80; species 3 model size: F 2,102 ¼ 3.80, P ¼ 0.026, g 2 ¼ 0.07; Figure 5). Although the interaction was significant, model size was by far the most important factor affecting attack distance, as indicated by its very large effect size. In S. virgatus, attack distance differed significantly between all pairs of model size (25 vs. 50 mm: P ¼ 0.00067; 25 vs. 107 mm: P ¼ 0.00014; 50 vs. 107 mm: P ¼ 0.038; Figure 5). In S. jarrovii, attack distance did not differ significantly between the 25 and 50 mm model (P. 0.10) but were longer in response to both 25 and 50 mm models than to 107 mm models (P, 0.00014 each; Figure 5). Attack distance did not differ between species for 107 mm models due to absence or rarity of attacks (P. 0.10) or for 25 mm models (P. 0.10) but was significantly longer for S. jarrovii than S. virgatus when Figure 3 The proportion of individuals of Sceloporus jarrovii and S. virgatus that approached models of 3 sizes. Error bars represent 1.0 standard error. Figure 5 Attack distance, that is, the distance between the approaching model and the lizard when the lizards began to advance, decreased in Sceloporus jarrovii and S. virgatus with increase in size of the approaching model, with greater difference between species for models of intermediate size. Error bars represent 1.0 standard error.

1282 Behavioral Ecology responding to 50 mm models (P ¼ 0.00020; Figure 5). Thus, the source of the interaction for attack distance was that an interspecific difference was restricted to the 50 mm models. Interspecific differences in overall response Combinations of the attack, biting, and fleeing variables possible for each individual lizard were 1) advance, bite, and then flee; 2) advance and bite; 3) advance and then flee; 4) advance only; 5) bite without advancing and then flee; 6) bite only; 7) flee only; and 8) none of the behaviors. Responses by the 2 lizard species to 50 mm models differed dramatically. Six S. virgatus advanced only, 16 fled only, one advanced and bit, and one performed none of the behaviors. Eight S. jarrovii advanced and bit the models and 4 advanced and then fled. Thus, only one S. virgatus exhibited a response to 50 mm models similar to that of any individuals on S. jarrovii. If advance and bite responses are compared with all others pooled, frequencies of response combinations differ significantly (Fisher exact P ¼ 0.00013). As indicated by the proportions of individuals that bit or fled, 0.67 individuals of S. virgatus treated the 50 mm model as a predator, 0.04 treated a model as prey, and 0.29 exhibited neither predatory nor antipredatory responses. In contrast, 0.33 individuals of S. jarrovii treated 50 mm models as predators and 0.67 treated them as prey. Major differences in response of the 2 lizard species were restricted to the 50 mm models: all lizards of both species treated the 107 mm models as predators and all S. jarrovii and more than 0.80 of S. virgatus treated the 25 mm models as prey. Some individuals in each species exhibited responses indicating some uncertainty about 50 mm models or perhaps evaluated them as being irrelevant. These include the S. virgatus that advanced but then neither bit nor fled and the individual that performed none of the behaviors even when contacted by the model. That one S. jarrovii initially advanced but then fled suggests that these individuals may have advanced toward a potential prey but reevaluated the 50 mm models as potential predators on closer inspection. DISCUSSION Model size and escape behavior Intraspecific differences The predictions that both species of lizards would be increasingly likely to flee and have longer flight initiation distances as prey size increased were strongly verified by the pattern of significant differences between pairs of prey sizes and the large effect sizes for the main effect of model size. The probability of fleeing differed between successively larger model sizes in the smaller lizard species, S. virgatus. A similar pattern occurred in the larger species, S. jarrovii, but the difference in escape probability between the 2 smaller model sizes was not significant, presumably due to limited statistical power using a small sample size. Flight initiation distance was strongly affected by model size, increasing with model size in both species. It increased more rapidly with model size in the smaller species, accounting for a small interaction effect. It increased between all pairs of successively larger model sizes in both species. Thus, within species escape became more likely and flight initiation distance increased as model size increased. The increasing tendency to flee as approacher size increases is likely a widespread phenomenon because looming stimuli elicit escape by diverse vertebrate and invertebrate prey (Oliva et al. 2007) when their images on the retina subtend a certain size (Glantz 1974; Nalbach 1990). Retinal angle subtended may be a major cue to size-associated risk that affects optimal escape decisions. Flight initiation distances by both species were much shorter than is usually observed in studies of lizard escape behavior in which investigators simulate approach. In both species, flight initiation distance in such studies is often longer than 1 m even at slow approach speeds (S. virgatus: Smith 1996; Cooper 2007, 2008, 2009a, 2009b; Cooper and Wilson 2007; S. jarrovii: Cooper W, Avalos A, unpublished data). The approach speed of models was somewhat lower than in previous studies (ca. 0.5 2.0 m/s), which may account for some of the difference in flight initiation distance between studies. Flight initiation distance may not be a very appropriate term for the responses to models of intermediate size because most lizards that were recorded as fleeing moved very short distances (,0.10 m). Nevertheless, the decline in flight initiation distance as model size decreases suggests that some of the difference may be a consequence of the much large size of human-simulated predators. Interspecific differences We interpret interspecific differences in antipredatory responses (and in foraging responses) as being probable consequences of the difference in size between the 2 lizard species. Comparative research controlling for phylogenetic relations is needed to test these hypotheses rigorously. Probability of fleeing differed between species only for the 50 mm model size. Few lizards of either species fled from the small models, and all fled from the large ones. The proportion of the smaller species that fled from models of intermediate size was about double that of the larger species, further suggesting the importance of relative model and lizard sizes for escape decisions. That flight initiation distance was greater for S. virgatus than S. jarrovii during approaches by 50 mm models suggests that assessed risk was greater for the smaller species. These results for frequency of fleeing and flight initiation distance are as predicted by relative sizes of approachers and approachees (Robinson and Wellborn 1987; Owen-Smith 2008; Costa 2009) and corresponding differences in assessed risk (Stankowich and Blumstein 2005). The larger species has larger threshold model size for elicitation of escape. Escape theory The findings for flight initiation distance are consistent with the prediction of optimal escape theory that flight initiation distance increases as risk associated with increasing predator size increases and with previous empirical studies that have confirmed the prediction (theory: Cooper and Frederick 2007, 2010; fish: Dill 1974; Hurley and Hartline 1974; Karplus and Ben Tuvia 1979; Helfman and Winkelman 1997). Our study appears to be the first test of this prediction for a terrestrial predator although the lizard Ctenosaura similis has a longer flight initiation distance when an approaching predator has larger eye size (Burger et al. 1991). Our overall findings bear more directly on the transitions from perception of the approaching model as prey to uncertainty to perception that it is a predator. Nevertheless, for both species, the proportion of individuals that fled was substantial in trials with the 2 larger model sizes, and flight initiation distance was longer for the larger of those model sizes. These findings indicate that some lizards of both species perceived models of the 2 larger sizes as potential predators and increased their probabilities of fleeing and flight initiation distances accordingly. Model size and foraging decisions Intraspecific differences The relationship between model size and probability of advance was strongly negative, higher proportions of both species

Cooper and Stankowich Body size affects foraging and escape decisions 1283 advancing toward smaller than larger models. The probability of advance decreased between all pairs of successively larger model sizes in S. virgatus but declined only between the 50 and 107 mm models in the larger S. jarrovii. The negative relationships between model size and advancing suggest that the size range represented by progressively larger models spanned the size range between prey and animals too large to be eaten (Curio 1976; Forsman and Shine 1997; Owen- Smith 2008). Findings for proportions of lizards that bit models were similar to those for advancing, but proportions that bit were lower with the exceptions that all S. jarrovii advanced toward and bit the smallest models and none bit, but one advanced toward the largest models (Figures 3 4). We have assumed that lizards advancing toward models were motivated to investigate possible prey items. That high proportions of lizards bit models after advancing supports this view. Models were painted brown and crickets were coated in polyurethane, possibly increasing their detectability due to greater reflectivity or affecting their desirability as prey. If greater detectability increased attacks on the smallest models, its effect would have been similar to that predicted for attack distance. Greater detectability did not lead to any noticeable increase in probability of fleeing or flight initiation distance in trials with small models. It is possible that differences between crickets and grasshopper models affected lizard behaviors, but any such effects were more likely related to detectability than attack or escape. Moreover, because all models moved toward lizards, differences in detectability were minimized. Due to odors, polyurethane and paint were more likely to affect the decision to bite than detection. Any aversion to the more odorous polyurethane would have decreased the likelihood of biting cricket models. If such an effect occurred, it was too small to obscure effect of model size on biting. Attack distance was affected by species and model size interactively, but the effect size of model size was far greater than differences between species among model sizes. Attack distance decreased as model size increased in both lizard species, providing further confirmation of the prediction based on decreasing profitability of larger prey (Forsman 1996). Interspecific differences The only interspecific difference in proportion of lizards that advanced was the greater proportion of the larger species that advanced on model of intermediate size. The larger species also had a lower probability of fleeing from intermediate models. This suggests that the larger S. jarrovii was more likely to perceive the intermediate models as prey and less likely to perceive them as predators than the smaller S. virgatus. Becausepreybecomelessvulnerableastheirsizeincreases (Woodward and Hildrew 2002; Owen-Smith 2008), this difference suggests that intermediate models are near the maximum prey size for S. virgatus but remain well within thesizerangeofpreyforthelargers. jarrovii. Thatlower proportions of S. virgatus than S. jarrovii bit small and intermediate models and absence of attacks by either species on large models further emphasizes the importance of relative sizes of the predator and potential prey in feeding decisions. The decline in attack distance with increase in model size was more precipitous for S. virgatus than S. jarrovii, which is apparent from the smaller difference between attack distances for 25 and 50 mm models for S. jarrovii. This finding adds to the evidence that the larger lizards are more likely to perceive somewhat larger models as prey than are smaller lizards, consistent with the expected effect of the increase in prey size with increase predator body size (Forsman 1996; Owen-Smith 2008) on perception. Perceptual transition from prey to predator Assessment of approaching heterospecifics depends strongly on their body size. At intermediate approacher sizes, the approachee may be uncertain and require more information before deciding the approacher s status. This accounts for reduction in probabilities of advancing and biting and increase in probability of fleeing as model size increased from small to intermediate as well as mixed responses by lizards suggesting changes in perception of 50 mm models during approach. The sizes of the largest and smallest models allowed very consistent decisions about advancing, biting, and fleeing. Progressive shortening of attack distance and increase in flight initiation distance as model size increased may indicate that some individuals in each species viewed models of a given size as prey, whereas others viewed models of the same size as predators. The shorter attack distances by S. virgatus for 50 mm than 25 mm models and by S. jarrovii for 107 mm than 50 mm models suggest that lizards may have delayed attack as models became larger to permit better assessment of model suitability as prey. Increases in flight initiation distance from smaller to larger model sizes suggest that lizards may require closer inspection to assess the predatory nature of the model as model size decreases. Taken alone, these findings about flight initiation distance would be consistent with prey being able to assess the degree of decrease in level of threat due to decreasing body size (Stankowich and Blumstein 2005). However, relationships between model size and the foraging variables suggest that lizards were initially uncertain in some cases, especially for intermediate model size. Taken together, the findings for attack and flight initiation distances suggest that lizards may more closely inspect models of intermediate size to decide whether they are prey or predators. Interspecific differences in pattern of responses suggest that the model size requiring closer inspection is larger for the larger lizard species. Approaching animals that are small relative to the animal being approached are perceived as prey. As the relative size of the approacher increases, animals being approached may become uncertain whether the approacher is profitable as a prey item, is a predator, or is neither prey nor predator in some range of approacher size. In this size range, more prolonged evaluation and/or closer inspection of the approacher is required to assess its nature before the approachee selects appropriate actions. As approacher size increases further, the animal being approached may quickly perceive that the approacher is too big to be prey and may be a predator. This simple interpretation is consistent with the literatures on size relationships between predators and prey (e.g., Forsman 1996; Costa et al. 2008; Owen-Smith 2008; Costa 2009), transitions between them (Webb and Shine 1993; Woodward and Hildrew 2002; Owen-Smith 2008), and effects of risk on escape behavior (Stankowich and Blumstein 2005; Cooper and Frederick 2007, 2010). Escape responses increased and foraging responses decreased as size of the approaching model increased. Use of larger models would undoubtedly have been associated with longer flight initiation distances, as have been reported when lizards of the same species are approached by people (Smith 1996; Cooper 2007, 2008; Cooper W, Avalos A, unpublished data). Lizards ignore potential prey below a certain size (Cooper et al. 2007). As size of approaching animal increases from very small to very large, it is thus likely to be assessed successively as too small to be profitable prey, as prey, as having uncertain status, and as a predator. Future studies could examine size thresholds for assessment of the predatory or dietary nature of models of sizes intermediate to those in this study and test effects of approach speed on the assessments.

1284 Behavioral Ecology Body size is only one of many factors that may influence foraging and escape decisions. However, for a given pair of predator and prey species, body size may be unimportant due to restricted ranges of body sizes and/or potent defenses. Nevertheless, size relationships between prey and predators such as those reported here are presumably widespread. FUNDING Indiana University Purdue University Fort Wayne. W.E.C. thanks the staff of the American Museum of Natural History s Southwestern Research Station for hospitality and logistical assistance during the study. REFERENCES Abrahams MV. 1995. The interaction between antipredator behaviour and antipredator morphology: experiments with fathead minnows and brook sticklebacks. Can J Zool. 73:2209 2215. Barbault R, Ortega A, Maury ME. 1985. Food partitioning and community organization in a mountain lizard guild of northern Mexico. Oecologia. 65:550 554. Burger J, Gochfeld M, Murray BG Jr. 1991. The role of a predator s eye size in risk perception by basking black iguana, Ctenosaura similis. Anim Behav. 42:471 476. Bursey CR, Goldberg SR. 1993. Diet of neonatal Yarrow s spiny lizard, Sceloporus jarrovii jarrovii (Phrynosomatidae). Southwest Nat. 38: 381 383. Cohen J. 1992. A power primer. Psychol Bull. 112:155 159. Cooper WE Jr. 2003. Effect of risk on aspects of escape behavior by a lizard, Holbrookia propinqua, in relation to optimal escape theory. Ethology. 109:617 626. Cooper WE Jr. 2007. Compensatory changes in escape and refuge use following autotomy in the lizard Sceloporus virgatus. Can J Zool. 85:99 107. Cooper WE Jr. 2008. Visual monitoring of predators: occurrence, cost and benefit for escape. Anim Behav. 76:1365 1372. Cooper WE Jr. 2009a. Fleeing and hiding under simultaneous risks and costs. Behav Ecol. 20:665 671. Cooper WE Jr. 2009b. Flight initiation distance decreases during social activity in lizards (Sceloporus virgatus). Behav Ecol Sociobiol. 63: 1765 1771. Cooper WE Jr, Anderson RA, Frederick WG. 2007. Prey size selection under simultaneous choice by the broad-headed skink (Eumeces laticeps). Ethology. 113:417 425. Cooper WE Jr, Frederick WG. 2007. Optimal flight initiation distance. J Theor Biol. 244:59 67. Cooper WE Jr, Frederick WG. 2010. Predator lethality, optimal escape behavior, and autotomy. Behav Ecol. 21:91 96. Cooper WE Jr, Pérez-Mellado V, Baird T, Baird TA, Caldwell JP, Vitt LJ. 2003. Effects of risk, cost, and their interaction on optimal escape by nonrefuging Bonaire whiptail lizards, Cnemidophorus murinus. Behav Ecol. 14:288 293. Cooper WE Jr, Whiting MJ. 2007. Universal optimization of flight initiation distance and habitat-driven variation in escape tactics in a Namibian lizard assemblage. Ethology. 113:661 672. Cooper WE Jr, Wilson DS. 2007. Beyond optimal escape theory: microhabitats as well as predation risk affect escape and refuge use by the phrynosomatid lizard Sceloporus virgatus. Behaviour. 144:1235 1254. Costa GC. 2009. Predator size, prey size, and dietary niche breadth relationships in marine predators. Ecology. 90:2014 2019. Costa GC, Vitt LJ, Pianka ER, Mesquita DO. 2008. Optimal foraging constrains macroecological patterns: body size and dietary niche breadth in lizards. Glob Ecol Biogeogr. 17:670 677. Curio E. 1976. The ethology of predation. Berlin (Germany): Springer- Verlag. Dickman CR. 1988. Body size, prey size, and community structure in insectivorous mammals. Ecology. 69:569 580. Dill LM. 1974. The escape response of the zebra danio (Brachydanio rerio). I. The stimulus for escape. Anim Behav. 22:711 722. Forsman A. 1996. Body size and net energy gain in gape-limited predators: a model. J Herpetol. 30:307 319. Forsman A, Shine R. 1997. Rejection of non-adaptive hypotheses for intraspecific variation in trophic morphology in gape-limited predators. Biol J Linn Soc. 62:209 223. Glantz RM. 1974. Defense reflex and motion detector responsiveness to approaching targets: the motion detector trigger to the defense reflex pathway. J Comp Physiol. 95:297 314. Gliner JA, George AM, Harmon RJ. 2002. The chi-square test and accompanying effect size indices. J Am Acad Child Adolesc Psychiatry. 41:1510 1512. Goldberg SR, Bursey CR. 1990. Winter feeding in the mountain spiny lizard, Sceloporus jarrovi (Iguanidae). J Herpetol. 24:446 448. Gras H, Hörner M. 1992. Wind-evoked escape running of the cricket Gryllus bimaculatus. J Exp Biol. 171:189 214. Helfman GS, Winkelman DL. 1997. Threat sensitivity in bicolor damselfish: effects of sociality and body size. Ethology. 103: 369 383. Hurley AC, Hartline PH. 1974. Escape response in the damselfish Chromis cyanea (Pisces: Pomacentradae): a quantitative study. Anim Behav. 22:430 437. Karplus I, Ben Tuvia S. 1979. Warning signals of Cryptocentrus steinitzi (Pisces, Gobiida3) and predator models. Z Tierpsychol. 51: 225 232. McLean EB, Godin J-GJ. 1989. Distance to cover and fleeing from predators in fish with different amounts of defensive armour. Oikos. 55:281 290. Nalbach HO. 1990. Visually elicited escape in crabs. In: Wiese K, Krent WD, Tautz J, Reichert H, Mulloney B, editors. Frontiers in crustacean neurobiology. Basel (Switzerland): Birkhauser Verlag. p. 165 172. Oliva D, Medan V, Tomsi D. 2007. Escape behavior and neuronal responses to looming stimuli in the crab Chasmagnathus granulatus (Decapoda: Grapsidae). J Exp Biol. 210:865 880. Owen-Smith N. 2008. Predator-prey size relationships in an African large-mammal food web. J Anim Ecol. 77:173 183. Phelan JP, Niessen KG. 1989. Effect of satiation on activity patterns in Sceloporus virgatus. J Herpetol. 23:424 426. Robinson JV, Wellborn GA. 1987. Mutual predation in assembled communities of odonate species. Ecology. 68:921 927. Ruby DE, Dunham AE. 1984. A population analysis of the ovoviviparous lizard Sceloporus jarrovi in the Pinaleno mountains of southeastern Arizona. Herpetologica. 40:425 436. Rosenthal R, Rubin DB. 2003. r equivalent : a simple effect size indicator. Psychological Methods. 8:492 496. Smith GR. 1996. Correlates of approach distance in the striped plateau lizard (Sceloporus virgatus). Herpetol J. 6:56 58. Stankowich T. 2009. When predators become prey: flight decisions in jumping spiders. Behav Ecol. 20:318 327. Stankowich T, Blumstein DT. 2005. Fear in animals: a meta-analysis and review of risk assessment. Proc Roy Soc Lond Ser B Biol Sci. 272:2627 2634. Stebbins RC. 2003. A field guide to western reptiles and amphibians. 3rd ed. Boston: Houghton Mifflin. Vinegar MB. 1975. Demography of the striped plateau lizard Sceloporus virgatus. Ecology. 56:172 182. Watters JL. 2008. Comparison of two diet collection methods in the lizards, Sceloporus jarrovii and Sceloporus virgatus. Herpetol Rev. 39:307 310. Webb JK, Shine R. 1993. Prey-size selection gape limitation and predator vulnerability in Australian blindsnakes (Typhlopidae). Anim Behav. 45:1117 1126. Woodward G, Hildrew AG. 2002. Body-size determination of niche overlap and intraguild predation within a complex food web. J Anim Ecol. 71:1063 1074. Wright SP. 1992. Adjusted p-values for simultaneous inference. Biometrics. 48:1005 1013. Zar JH. 1996. Biostatistical analysis. 3rd ed.. Upper Saddle River (NJ): Prentice Hall.