Southeastern Naturalist 503 A. Collier and J.Y.S. Hodgson 22001177 SOUTHEASTERN NATURALIST 1V6o(4l.) :1560,3 N–5o1. 54 A Shift in Escape Strategy by Grasshopper Prey in Response to Repeated Pursuit Alex Collier1,* and Jay Y.S. Hodgson1 Abstract - We used grasshoppers as a model organism to examine the response of prey to repeated approach from a persistent predator (human observer). We randomly assigned adult Chortophaga australior (Southern Greenstriped Grasshopper) to either low- or highrisk treatments. For both groups, we approached each grasshopper during 15 consecutive encounters and recorded the distance fled (DF) and overall flight path. We approached grasshoppers assigned to the low-risk treatment after a 30-second delay upon landing between each escape flight. Those in the high-risk treatment were approached immediately upon landing and given no opportunity to recover. Grasshoppers assigned to the low-risk treatment exhibited an erratic, or protean flight path to evade detection and traveled shorter distances across consecutive encounters. Those in the high-risk treatment exhibited longer escape flights that were more commonly oriented directly away from the approach of the observer. The results of our study provide additional evidence that prey may shift escape strategies in response to real-time assessment of predation risk. Introduction There are many outcomes that can occur when prey first detect the presence of an approaching predator. Ydenberg and Dill (1986) were the first to develop a graphical model that considered the economic costs of escape compared to predation risk. In their model, prey should not immediately flee upon detection of the predator; rather, it should only initiate an escape decision, quantified as flight initiation distance (FID), when the predator reaches a minimum distance at which the costs of escape, such as the loss of foraging or reproductive success, equals the costs of remaining (i.e., risk of injury or death). Stankowich and Blumstein (2005) performed a meta-analysis that examined FID in both aquatic and terrestrial systems. They found that prey perception of risk increased in response to predator size, approach direction, and/or speed. Another escape variable is the distance prey travel after initiating escape. This distance fled (DF) may also reflect a balance between risk and cost (Cooper 2006a). Indeed, various studies have shown that the DF is influenced by the same risk factors described above for the FID (Cooper 1997, 2006b; Martín and López 1996; Stone et al. 1994). Regardless of which metric is used to assess prey risk, the majority of past empirical studies have treated the interaction between predator and prey as “static”, when in reality, prey continually update their assessment of risk during pursuit and adjust their escape behaviors accordingly to increase their chance of survival (Cooper 2006b). 1Department of Biology, Armstrong State University, Savannah, GA 31419. *Corresponding author - alex.collier@armstrong.edu. Manuscript Editor: JoVonn Hill Southeastern Naturalist A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 504 To date, relatively few studies have examined how prey respond to repeated approach from a persistent predator (summarized by Bateman and Fleming 2014). Papers on this topic have suggested that repeated approach can affect both FID and DF, but that the particular response varies based on the escape strategy of individual prey species. For example, some prey, such as Podacris lilfordi Günther (Mediterranean Wall Lizard) and Dissoteira carolina L. (Carolina Grasshopper), respond to repeated approach by increasing both their FID and DF (Cooper 2006a, Cooper et al. 2009). Others, including Psinidia fenestralis Serville (Longhorn Band-wing Grasshopper), exhibit greater FID but a declining DF across consecutive encounters, possibly due to a drop in their energetic reserves (Bateman and Fleming 2014). Still others, like Schistocerca alutacea Harris (Leather-colored Bird Grasshopper) exhibit no change in FID, yet their DF increased following repeated approach (Bateman and Fleming 2014). Additional escape behaviors in response to persistent pursuit include greater reliance on the use of refuge (Bateman and Fleming 2014, Cooper 1997, Cooper et al. 2009, Martín and López 1999) and a hesitancy to emerge from cover and resume normal activity (Cooper 1998, 2012; Cooper and Avalos 2010; Martín and López 1999; Rodríquez-Prieto and Fernández-Juricic 2005). Bateman and Fleming (2014) stressed the importance of incorporating repeated pursuit in behavioral studies to highlight the adaptive range of escape strategies that a particular prey species may employ. In recent years, there has been more attention placed on variation in escape behavior between individuals in a given population (Blumstein et al. 2010, Jones and Godin 2010). Historically, the most common behavioral response to an encounter with a predator was considered the optimal strategy for that population of prey. Individuals that behaved unpredictably were often viewed as maladapted (Dall et al. 2004, Jones et al. 2011, Reale et al. 2007). However, prey populations with higher levels of inter-individual variance may limit a predator’s ability to develop an effective counter-strategy (Humphries and Driver 1967). In addition, individuals that respond erratically to pursuit may further disorient the predator and increase their own likelihood of survival (Jones et al. 2011). Unpredictable, or protean, escape behavior is practiced by many animals including rabbits, minnows (Driver and Humphries 1988), and grasshoppers (Bateman and Fleming 2014, Cooper 2006a). Jones et al. (2011) described this overall change in focus from an emphasis on optimality toward a greater appreciation of behavioral variability as a paradigm shift in behavioral ecology and the study of predator–prey interactions in particular. In this study, we used grasshoppers as a model organism to determine if they would adopt different escape strategies in response to repeated approach from a simulated predator (human observer). We focused primarily on the DF as a metric to measure prey stress and randomly assigned grasshoppers to 1 of 2 treatment groups. Grasshoppers assigned to the “low-risk” treatment were approached repeatedly but only after a 30-second delay following landing from a previous escape flight. In the “high-risk” treatment, grasshoppers were approached immediately upon landing and had no opportunity to recover between consecutive encounters. We predicted Southeastern Naturalist 505 A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 that grasshoppers assigned to the high-risk treatment would exhibit a longer average escape flight (DF) from each repeated approach and travel an overall longer total distance (sum of jump vectors) than those approached after a brief delay. In addition, we expected that the final landing position for high-risk prey would lie further from the original point of release (as a single vector) compared to grasshoppers assigned to the low-risk treatment. Field-Site Description We conducted the study in July 2014 on a privately owned parcel of land located in Bryan County, GA. The study site consisted of roughly a 4046-m2 (1 acre) field surrounded on all sides by dense stands of 12–14-year-old Pinus taeda L. (Loblolly Pine). The field was cleared of pines in 2012, and the groundcover at the time of the study was characterized by a variety of native perennials. These included Aristida beyrichiana Trin. and Rupr. (Southern Wiregrass), Galactia allottii Nutt. (Elliott’s Milkpea), Cnidoscolus urens var. stimulosus Michx. (Spurge Nettle), and sparse patches of Eupatorium capillifolium (Lam.) Small (Dogfennel) and the invasive Paspalum notatum Alain ex Flüggé (Bahiagrass). A small number of hardwood trees that were not removed when the plot was cleared, including Quercus nigra L. (Water Oak) and Liquidambar stryaciflua L. (Sweet Gum), were also scattered throughout the study site. Methods Study species and sampling methodology We divided the field-study site described above into 5-m2 quadrats marked in an equidistant grid with surveyor flags. We labeled the rows and columns that subdivided the field alphabetically (rows) or numerically (columns) so that grasshopper flight patterns could be spatially mapped (see Fig. 1 for the grid layout). We removed the bottom of a 3.8-L, 7.5-cm diameter clear plastic collection-jar and attached an eyehook to the lid. During the study, we placed individual grasshoppers within this open-ended chamber that rested on the ground in the center quadrat (G6) of the subdivided field. A rope tied to the eyehook was routed through a pulley secured to a 2-m tall shepherd hook positioned directly above the chamber. At the start of each experiment, the observer stood 5 m directly south of the chamber and pulled on the opposite end of the rope to lift the jar and release the grasshopper. We selected adult Chortophaga australior Rehn and Hebard (Southern Greenstriped Grasshopper) for this study because of its relative abundance at the study site and because they fly readily upon approach. In addition, males typically produce sound (crepitation) during flight, making them relatively easy to track (Capinera et al. 1999). We collected adult (18–28 mm) grasshoppers on-site and housed them in grass-filled collection jars overnight prior to the morning field study. We collected all behavioral data on sunny to partly cloudy mornings between 8:30 am and 11:00 am in temperatures between 27 oC and 32 oC. We randomly chose each grasshopper, Southeastern Naturalist A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 506 Figure 1. Frequency of landing position by quadrat across c o n s e c u t i v e encounters for grasshoppers assigned to (A) low-risk, and (B) highrisk treatments. Southeastern Naturalist 507 A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 removed it from the collection jar, and immediately placed it on the ground under the plastic chamber described above. The observer retreated to the starting position and removed the chamber 30 seconds later by pulling on the rope. As in past investigations, we relied on a human observer to simulate a persistent predator (Bateman and Fleming 2014, Cooper 2006a). On each sampling day, we randomly assigned grasshoppers to either a low- or high-risk treatment. Grasshoppers assigned to the low-risk treatment were not approached until 30 seconds after the chamber had been raised. At this time, the observer approached the animal at a practiced speed of ~80 m/min and visually tracked its escape flight path and subsequent landing position. During longer escape flights, the observer followed directly behind but came to an immediate halt each time the grasshopper landed and remained motionless at a minimum distance of 2–3 m away. After visually observing the animal for 30 seconds the observer again approached the grasshopper and inserted a numbered surveyor flag into the ground at the approximate landing position from the previous flight. We repeated this for 15 consecutive escape flights after release from the chamber. The numbered surveyor flags (1–15) provided a spatial map of the landing spots between each escape flight. We immediately collected each grasshopper upon landing after their last escape attempt. We recorded morphological measurements including the total body length (head–abdomen) and femur, tibia, and tarsus lengths to the nearest 0.1 mm using a vernier caliper. We also followed the methods of Capinera et al. (1999) to determine the sex of each individual. We later released the grasshoppers at a different location to minimize the chance that any individual would be collected on more than 1 occasion. In total, 22 grasshoppers were assigned to the high-risk treatment and 15 were assigned to the low-risk treatment. We examined the flight path of each grasshopper by recording the location of each landing site within a particular quadrat. We measured to the nearest 0.1 m the DF for each escape flight using a rolling measuring wheel. In addition, we recorded the total distance traveled across 15 consecutive escape flights (sum of jump vectors), the average DF, and the distance from the final landing position to the original release site in quadrat G6 (as a single vector). We collected the flight-path data and morphological measurements for grasshoppers assigned to the high-risk treatment in an identical fashion with one exception: we immediately approached the grasshoppers assigned to this group at the same gait speed as soon as the chamber was lifted at the start of the experiment. Upon landing after each escape attempt, we approached these animals without delay, giving them little to no recovery period before the next approach. We repeated this procedure for all 15 encounters. Regardless of treatment, we commonly lost sight of the focal prey or had difficulty distinguishing it from other nearby individuals. In these instances, we terminated the experiment and discarded the collected data. Statistical methodology We conducted all analyses in NCSS 2007 statistical software using α = 0.05 unless specifically mentioned otherwise. To test for morphological differences Southeastern Naturalist A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 508 between individuals randomly assigned to the high- and low-risk treatments, we compared the variables measured using the 2-sample Hotelling’s T 2 test. A preliminary Box’s M Test confirmed that all assumptions for this test were met. We used separate Aspin–Welch t-tests to compare the total distance traveled across 15 jumps (sum of jump vectors) and the distance from the final landing site to the original starting location (a single vector) between low- and high-risk treatments. We adopted this non-parametric alternative to the two-tailed t-test because the variances between groups were unequal (McDonald 2014). We employed the Bonferroni correction to select the proper confidence interval (α = 0.025) to reduce the chance of committing a Type-I error. To determine if there were differences in the average escape distance grasshoppers traveled across consecutive encounters, we compiled the raw jump-length data for all individuals assigned to both low- and high-risk treatments. We used a repeated-measures ANOVA with the Geisser–Greenhouse adjustment to correct for sphericity to compare these data. We then conducted a Tukey–Kramer post-hoc test because significant between-treatment differences were detected. In addition, we employed principal component analysis (PCA) of the jump vectors of each grasshopper to generate separate simplified spatial-distribution maps for the high-risk and low-risk treatments. Performing PCA on map data is a commonly used technique (Chavez and Kwarteng 1989) because it reduces complicated spatial variability in raw data (Euclidean distances) by generating a normalized, uncorrelated, and lower-dimensional pattern (eigenvectors) with minimal loss of information (Adler and Golany 2001, Horel 1981, Overland and Preisendorfer 1981). These considerations are important to our study because the cardinal direction of each successive jump is determined by the first jump. For example, if the observer initially approached the grasshopper from the south, the first jump may have been due west and the second jump after the follow-up approach may have been due north. Alternatively, depending on the initial body position of the grasshopper as the observer approached, the jumps may have been due north and east, respectively. In both cases, the shared escape behaviors were clockwise and orthogonal, but the Euclidean measurements would be very different based solely on the randomness of the body position prior to the initial jump. Some of this variability is shown in the contour plots in Figure 1. By using PCA, we reduced this random effect and would be able to determine, if it existed, a directional pattern of shared escape behaviors among grasshoppers within a treatment. We performed PCA using the correlation matrix of each treatment. The first 2 PCA axes extracted 80.9% of the variance in the high-risk treatment and 69.1% of the variance in the low-risk treatment, demonstrating that PCA robustly interpreted the prevailing flight paths. Results There were no statistical differences in any of the morphological measurements examined between high- and low-risk treatment groups (df1 = 4, df2 = 35, T2 = 8.68, P = 0.12). The total distance traveled across 15 consecutive jumps statistically Southeastern Naturalist 509 A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 differed between the low- and high-risk treatments (df = 35, t = 2.43, P = 0.0205). Grasshoppers assigned to the high-risk treatment traveled an average of 37.0 m compared to 29.7 m for those in the low-risk group. The frequency of landing positions by quadrat illustrate differences in the escape flight pattern between lowand high-risk treatments (Fig. 1). In addition, the distance from the final landing position (jump 15) to the original release point (quadrat G6) statistically differed between treatment groups (df = 35, t = 4.85, P < 0.0001). High-risk grasshoppers landed an average of 16.8 m away from the release point compared to 7.9 m for low-risk prey. The DF across consecutive jumps also differed between low- and high-risk treatments (df = 1, F = 4.87, P = 0.034). Post-hoc analysis revealed that grasshoppers assigned to the high-risk treatment traveled a greater average distance across consecutive encounters. These grasshoppers traveled an average of 2.5 m (min–max = 2.1–4.4 m) per jump compared with 1.9 m (min–max = 1.6–2.3 m) for those assigned to the low-risk treatment (Fig. 2). PCA revealed differences in the flight path between grasshoppers assigned to the 2 treatments. The flight path for the low-risk treatment (Fig. 3A) was more random than for the high-risk treatment, as indicated by the reduction in the variance described (69.1% to 80.9%). Conversely, the flight path for the high-risk treatment was predominantly a semicircle in the clockwise orientation (Fig. 3B). The first 4 jumps were in a near-linear line away from the observer before developing a prevailing arc over the subsequent 11 jumps. Figure 2. Average distance fled (DF) across consecutive encounters for grasshoppers assigned to low- and high-risk treatments. Southeastern Naturalist A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 510 Figure 3. Flight paths extracted by PCA for grasshoppers assigned to (A) low-risk, and (B) high-risk treatments. The numbers (1–15) indicate successive jumps. Southeastern Naturalist 511 A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 Discussion Various studies have investigated prey response to repeated approach, with many focusing on lizards (Cooper 1997, 1998, 2012; Cooper and Alvalos 2010; Cooper et al. 2009; Martín and López 1999, 2003), frogs (Rodríquez-Prieto and Fernández-Juricic 2005), and some on grasshoppers (Bateman and Fleming 2014, Cooper, 2006a). Our grasshopper study is novel because it exposed the animals to more repeated approaches (15) than others (cf. Cooper 2006a, Bateman and Flemming 2014). Furthermore, none of these studies had examined the potential relationship between the morphology of individual prey and their particular escape strategy. When randomly collecting grasshoppers for our study, we were interested in assessing any potential differences in escape behaviors related to morphological measurements and sex because females of the Southern Greenstriped Grasshopper tend to be larger than adult males (Capinera et al. 1999) and may exhibit different flight capabilities or escape strategies. However, because we found no significant differences in individual morphology between low- and high-risk treatments (Hotelling’s T 2 test) and our samples had a male bias (30 males: 7 females), we are unable to determine any potential relationships. The remainder of our analyses presented here focus solely on variations in escape behavior between treatments. Future studies could include an intentional blocking design using males and females of different sizes; potential intersexual differences in escape behavior among grasshoppers and other prey remains largely unexplored. Our predictions that the total escape flight distance and the average DF would be greater for prey assigned to the high-risk treatment than the low-risk treatment were supported by the data. These predictions were based in part on the findings of Cooper (2006a), which investigated Carolina Grasshopper, another species of bandwinged grasshopper in the Subfamily Oedipodinae, in response to human observers. In that study, the observer flushed the prey for the first time by walking through the study habitat. The prey were then visually tracked and approached again across a series of encounters. It found that both the FID and DF for grasshoppers approached repeatedly increased although prey were only approached a maximum of 3 times. Cooper (2006a) reasoned that the increased FID reduced the risk of prey being overtaken during each escape attempt while the greater DF made it more difficult for the predator to track its escape flight and locate its landing position. We also used an observer to simulate a persistent predator, but relied on a different study design in which prey were initially collected and placed within the release chamber positioned in the center (G6) quadrat of our marked field. We did not measure FID but instead used DF as a metric to assess prey perception of risk. Unlike Cooper (2006a), we did not observe that the DF increased or decreased in a predictable fashion over all jumps. Instead, the average DF oscillated slowly in response to repeated approach for both high- and low-risk treatment groups (Fig. 2). For both treatments, the average DF was greatest following the first approach (Fig. 2). This potential stress-response was not surprising because the grasshopper had been transferred from a collection jar and positioned inside the clear chamber just prior to its release and the first approach of the observer . Southeastern Naturalist A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 512 Bateman and Fleming (2014) repeatedly approached another Oedipodine grasshopper, Longhorn Band-wing Grasshopper, an average of 5.9 times and observed a similar decline in DF. They speculated that this result might reflect a drop in the prey’s energetic resources to power escape, which may not have been observed by Cooper (2006a) because prey were typically approached only 2–3 times. Our data suggest that, although prey may become metabolically taxed in response to repeated approach, they still retain enough energetic reserves to power occasional bursts of escape flight that carry them across longer distances. This possibility is best illustrated by observing the spike in DF among high-risk grasshoppers during their final few escape flights (Fig. 2). Prey approached immediately upon landing clearly adjusted their strategy and responded by consistently traveling greater distances compared to those approached after a 30-second delay. Grasshoppers belonging to the Subfamily Oedipodinae, including Carolina Grasshopper and Longhorn Band-wing Grasshopper, typically possess colorful yellow to orange hind wings with a broad black band that crosses near the center (Capinera et al. 1999). The Southern Greenstriped Grasshopper, used in this study, belongs the same subfamily but its wing coloration is often less pronounced and the transverse band is poorly defined (Capinera et al. 1999). Regardless of coloration, the hind wings are only visible during flight. Otherwise, these grasshoppers are cryptically colored and blend in seamlessly as the wings are withdrawn upon landing in the sunny, open habitats they prefer (Capinera et al. 1999). The flash of color during flight may disorient a potential predator especially if the grasshopper flies perpendicular to the predator’s approach and lands to the side in its peripheral field of vision (Bateman and Fleming 2014, Cooper 2006a). The prey’s escape angle is typically measured on a 180 o scale in which the prey either flees directly toward (0o), away from (180o), or laterally at right angles (90o) from the approach of the observer (Bateman and Fleming 2014, Cooper 2006a). Lateral, or perpendicular, escape flight appears to be an adaptive response used by bandwinged grasshoppers to relocate outside of the direction of the predator’s approach. Upon landing they immediately rely on crypsis through body coloration and general immobility to further mask their location (Bateman and Fleming 2014, Cooper 2006a). In our study, we did not measure the escape angle but instead mapped out the flight path for each grasshopper using surveyor flags and recorded the landing position within the grid of the subdivided field. Each landing position corresponded to a particular quadrat and we analyzed these data for both low- and high-risk treatments using PCA. PCA revealed that 20–30% of the variance observed for both treatments could not be explained with the extraction of new variables. Carrete and Telle (2010) note that such variance could provide diversity upon which natural selection may act. Overall, the differences observed between the flight path of grasshoppers assigned to low- and high-risk treatments appear to reflect a shift in escape strategy (Fig. 3). The random flight path for low-risk grasshoppers as extracted by PCA reflects numerous lateral escape flights mixed with those that carried the prey directly away from the observer (Fig. 3A). This strategy has also been observed for both Longhorn Band-wing Grasshopper and Carolina Grasshopper, which responded to repeated approach by relying on a similar mix of escape-flight directions (Bateman Southeastern Naturalist 513 A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 and Fleming 2014, Cooper 2006a). Across encounters, this unpredictable or protean pattern of escape behavior may further disorient a persistent predator (Bateman and Fleming 2014, Humphries and Driver 1970). In contrast, the flight path for highrisk individuals illustrates that most escape flights took the prey directly away or at slight angles from the approach of the observer (Fig. 3B). As mentioned previously, the average DF was also greater among the high-risk treatment. As a result, highrisk grasshoppers landed with greater frequency across a larger number of quadrats that subdivided the field (Fig. 1B). Bateman and Fleming (2014) observed a similar shift in strategy following repeated approach in Leather-colored Bird Grasshopper, a grasshopper with strong flight capabilities belonging to the Subfamily Cyrtacanthacridinae. These grasshoppers shifted from perpendicular patterns of escape to longer flights that carried them directly away from the observer’s approach (Bateman and Fleming 2014). However, these long flights made them difficult to track and they were, on average, approached only 2.9 times. We tracked high-risk prey across 15 consecutive encounters; thus, we were able to determine their overall flight path. After an initial series of relatively linear escape flights, there was a characteristic clockwise rotation that began to emerge over the course of 4–5 jumps (Fig. 3b). Had the grasshoppers not engaged in this banking maneuver and continued on a linear trajectory, they would have drawn closer to the dense stand of loblolly pines that surrounded the study field. High-risk prey in this study appeared to adopt the same escape strategy as that of Leathercolored Bird Grasshopper, with the exception that they made corrections in their overall flight path in response to repeated pursuit. Overall, this curve in flight path may have helped prevent these grasshoppers from leaving their preferred habitat. Although this strategy may place more distance between the prey and a potential predator, it is not without its own risk. Spending more time aloft in flight and traveling greater distances may expose the prey to increased risk from opportunistic visual predators like birds (Bateman and Fleming 2014). As predicted, high-risk prey at the end of their pursuit landed a greater distance away from the site of their original release (G6 quad). The results of our study provide additional evidence that grasshopper prey continually assess risk and shift their escape strategy in response to persistent pursuit (Bateman and Fleming 2014, Cooper 2006a). Bateman and Fleming (2014) described this as a behavioral switch from “Plan A” to “Plan B”. In this study, prey assigned to the low-risk treatment responded with Plan A. They traveled shorter distances with each escape flight and relied on crypsis through coloration and immobility upon landing. Over consecutive encounters they exhibited an erratic, or protean, flight path to evade detection. Prey assigned to the high-risk treatment adopted Plan B, traveling longer distances during each escape flight typically in the opposite direction from which they were approached. Acknowledgments This work was partially funded by the Department of Biology at Armstrong State University. We thank Matthew Draud, Austin Francis, Michele Cutwa, and Michele Guidone Southeastern Naturalist A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 514 for their assistance and support. We also appreciate the contribution of all involved in the review process who helped clarify and improve this manuscript. Literature Cited Adler, N., and B. Golany. 2001. Evaluation of deregulated airline networks using data envelopment analysis combined with principal component analysis with an application to Western Europe. European Journal of Operational Research 132:260 –273. Bateman, P.W., and P.A. Fleming. 2014. Switching to plan B: Changes in the escape tactics of two grasshopper species (Acrididae: Orthoptera) in response to repeated predatory approaches. Behavioral Ecology and Sociobiology 68:457–465. Blumstein, D.T. 2010. Flush early and avoid the rush: A general rule of antipredator behaviour? Behavioral Ecology and Sociobiology 21: 440–442. Capinera, J.L., C.W. Scherer, and J.M. Squitier. 1999. The Grasshoppers of Florida. Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. Available online at http://ufdc.ufl.edu/UF00066916/00001. Accessed 4 May 2014. Carrete, M., and J.L. Tella. 2010. Individual consistency in flight initiation distances in burrowing owls: A new hypothesis on disturbance-induced habitat selection. Biology Letters 6:167–170. Chavez, P.S., and A.W. Kwarteng. 1989. Extracting spectral contrast in landsat thematicmapper image-data using selective principal component analysis. Photogrammetric Engineering and Remote Sensing 55(3):339–348. Cooper, W.E., Jr. 1997. Threat factors affecting antipredatory behavior in the Broad-headed Skink (Eumeces laticeps): Repeated approach, change in predator path, and predator’s field of view. Copiea 3:613–619. Cooper, W.E., Jr. 1998. Effects of refuge and conspicuousness on escape behavior by the Broad-headed Skink (Eumeces laticeps). Canadian Journal of Zoology 75:943–947. Cooper, W.E., Jr. 2006a. Risk factors and escape strategy in the grasshopper Dissosteira Carolina. Behavior 143:1201–1218. Cooper, W.E. Jr. 2006b. Dynamic risk assessment: Prey rapidly adjust flight-initiation distance to changes in predator-approach speed. Ethology 112:858–864. Cooper, W.E. Jr. 2012. Risk, escape from ambush, and hiding time in the lizard Sceloporus virgatus. Herpetologica 68:505–513. Cooper, W.E., Jr., and A. Avalos. 2010. Predation risk, escape, and refuge use by Mountain Spiny Lizards (Sceloporus jarrovii). Amphibia–Reptilia 31:363–373. Cooper, W.E., Jr., D. Hawlena, and V. Perez-Mellado. 2009. Effects of predation-risk factors on escape behavior by Balearic Lizards (Podarcis lilfordi) in relation to optimalescape theory. Amphibia-Reptilia 30:99–110. Dall, S.R.X., A.I. Houston, and J.M. McNamara. 2004. The behavioral ecology of personality: Consistent individual differences from an adaptive perspective. Ecology Letters 7:734–739. Driver, P.M., and D.A. Humphries, 1988. Protean Behaviour. Clarendon Press, Oxford, UK. 339 pp. Horel, J.D. 1981. A rotated principal component analysis of the interannual variability of the northern hemisphere 500-mb height field. American Meteorological Society 109:2080–2092. Humphries, D.A., and P.M. Driver. 1967. Erratic display as a device against predators. Science 156:1767. Humphries, D.A., and P.M. Driver. 1970. Protean defense by prey animals. Oecologia 5:285–302. Southeastern Naturalist 515 A. Collier and J.Y.S. Hodgson 2017 Vol. 16, No. 4 Jones, K.A., and J.G.J. Godin. 2010. Are fast explorers slow reactors? Linking personality type and anti-predator behavior. Proceedings of the Royal Society B:Biological Sciences 277:625–632. Jones, K.A., A.L. Jackson, and G.D. Ruxton. 2011. Prey jitters: Protean behavior in grouped prey. Behavioral Ecology 22:831–836. Martín, J., and P. López. 1996. The escape response of juvenile Psammodromus lizards. Journal of Comparative Psychology 110:187–192. Martín, J., and P. López. 1999. When to come out from a refuge: Risk-sensitive and statedependent decisions in an alpine lizard. Behavioral Ecology 10: 487–492. Martín, J,. and P. López. 2003. Changes in the escape responses of the lizard Acanthodactylus erythrurus under persistent predatory attacks. Copeia 2003:408–413. McDonald, J.H. 2014. Handbook of Biological Statistics. 3rd Edition. Sparky House Publishing, Baltimore, MD. 305 pp. Overland, J.E., and R.W. Preisendorfer. 1981. A significance test for principal components applied to a cyclone climatology. Monthly Weather Review 110(10):1–4. Reale, D., S.M. Reader, D. Sol., P.T. McDougall, and N.J. Dingemanse. 2007. Integrating animal temperament within ecology and evolution. Biological Reviews of the Cambridge Philosophical Society 82:291–318. Rodríguez-Prieto, I., and E. Fernández-Juricic. 2005. Effects of direct human disturbance on the endemic Iberian Frog, Rana iberica, at individual and population levels. Biological Conservation 123:1–9. Stankowich, T., and D.T. Blumenstein. 2005. Fear in animals: A meta-analysis and review of risk assessment. Proceedings of the Royal Society B: Biological Sciences 272:2627–2634. Stone, P.A., H.L. Snell, and H.M. Snell. 1994. Behavioral diversity as biological diversity: Introduced cats and Lava Lizard wariness. Conservation Biology 8:569–573. Ydenberg, R.C., and L.M. Dill. 1986. The economics of fleeing from predators. Advances in the Study of Behavior 16:229–249.