2005. The Journal of Arachnology 33:153±158

RELATIONSHIP BETWEEN ESCAPE SPEED AND FLIGHT DISTANCE IN A WOLF , (WALCKENAER 1805)

Matthew K. Nelson and Daniel R. Formanowicz Jr.: Department of Biology, University of Texas at Arlington, Arlington, Texas 76019. E-mail: [email protected]

ABSTRACT. The relationship between running speed and ¯ight distance is an important one in terms of escape from predators, especially in species that may have multiple defensive strategies. In the Hogna carolinensis, one important antipredator mechanism is ¯ight. We examined the relationship between sprint speed and ¯ight distance in wolf by measuring sprint speed on a running track and, in a separate set of experiments with the same individual spiders, measured the distance at which they ¯ed from an advancing model predator. Sprint speed was not signi®cantly correlated with mass, size, or sex of the spiders. Sprint speed was positively correlated with ¯ight distance. This correlation may be the result of a trade-off between two competing modes of antipredator mechanisms: escape and crypsis. In individuals with higher sprint speeds, escape may be the more advantageous option. Slower individuals may have a greater chance of surviving an encounter with a predator simply by remaining still and relying on crypsis. Keywords: Antipredator strategy, risk, ¯ee, sprint speed

Behavior patterns associated with predator made by Ydenberg & Dill's (1986) model is escape and avoidance are important to indi- that ``response'' does not necessarily equal vidual survival. These result in strong selec- ``detection,'' in that it can often be dif®cult to tive pressure favoring individuals that suc- assess whether or not the potential prey has cessfully avoid or escape from predators. detected a predator. In some cases, an individ- When an is approached by a potential ual may ignore an approaching predator until predator, it must evaluate the level of preda- it becomes necessary to initiate ¯ight. In other tion risk, and utilize the appropriate antipred- cases, certain ``alert behaviors'' may occur ator mechanism to neutralize the risk. The dis- that precede a ¯ight decision. The Ydenberg- tance from an approaching predator at which Dill model generally predicts that individuals an animal chooses to ¯ee has been referred to should delay ¯ight until the costs associated as ``¯ight distance'' (e.g. FernaÂndez-Juricic et with staying (e.g., increased predation risk) al. 2002), ``¯ight initiation distance'' (e.g. exceed the bene®ts associated with staying Bonenfant & Kramer 1996), ``approach dis- (e.g., time spent searching for food or mates). tance'' (e.g. Martin & Lopez 1999), and Several studies have examined the relation- ``¯ush distance'' (e.g. FernaÂndez-Juricic et al. ship between ¯ight distance and running abil- 2001). The latter two of these imply the per- ity. Rand (1964) found that body temperature spective of the predator. Since we will be dis- affected the distance at which Anolis lizards cussing the issue from the perspective of the ¯ed from approaching predators, attributing prey, ``¯ight distance'' seems the most appro- differences in escape distances among individ- priate and concise terminology. uals to lower body temperatures which re- Ydenberg & Dill (1986) discussed in detail duced sprint speeds. Cooler individuals tended the economics of escape from predators. They to ¯ee at greater distances because lower body suggested a cost-bene®t model of ¯ight dis- temperatures result in greater risk of capture. tance, incorporating the costs and bene®ts of Heatwole (1968) suggested that crypsis might continuing a particular behavior (such as for- also play a role in ¯ight distances. Cryptic aging) relative to the costs and bene®ts of species of Anolis may decrease their risk of ¯eeing. A concept critical to the predictions capture by remaining motionless and ¯eeing

153 154 THE JOURNAL OF ARACHNOLOGY at shorter distances than less cryptic species. usually ¯ee a meter or so, and then remain Species that rely on crypsis may not ¯ee from motionless (pers. obs.). an approaching predator until detection by the We examined the relationship between predator is certain. body size and running speed in male and fe- Schwarzkopf & Shine (1992) suggested male H. carolinensis, testing the hypothesis that ``vulnerability'' (risk of capture) of prey that larger individuals were faster. Since there should be evaluated in terms of the probability is a sexual dimorphism in body size in this of being detected by a predator. They found species, we also predicted that females and that gravid female water skinks, Eulamprus males should differ in running speed. Using tympanum, exhibited decreased running abil- the same spiders, we examined the distance at ity and shorter ¯ight distances relative to non- which they ¯ed from an approaching model gravid females. They suggested that gravid fe- predator. We used the data collected on run- males switched antipredator tactics from ning speed and ¯ight distance to test the fol- escape to crypsis because of the decrease in lowing hypothesis based on Ydenberg & running speed. They interpreted these results Dill's (1986) model: faster individuals would as evidence that escape speed may not always be expected to ¯ee at shorter distances from be the most important element involved in de- the predator. termining when to ¯ee. Formanowicz et al. METHODS (1990) investigated similar effects in the liz- ϭ ard Scincella lateralis with differences in Hogna carolinensis (n 77; 44 males, 33 sprint speed related to tail autotomy. Skinks females) used in this study were collected on with autotomized tails were found to exhibit 26 March and 11 April, 1997 at the Texas Na- slower running speeds and relatively shorter ture Conservancy's Independence Creek Pre- ¯ight distances. They suggested that individ- serve, approximately 37 km south of Shef- ®eld, Terrell County, Texas, on the uals that had lost their tails switched to a cryp- northeastern edge of the Chihuahuan desert. tic antipredator strategy to compensate for re- Voucher specimens have been deposited at the duction in sprint speed. Denver Museum of Nature & Science. Most In this study, we examined the relationship spiders were collected at night by using head- between running speed and ¯ight distance in lamps to produce eyeshine; a few were col- the wolf spider Hogna carolinensis (Walcken- lected by turning rocks during the day. Spiders aer 1805). Hogna carolinensis is a large, bur- were not found to be active during the day. rowing wolf spider that is active on the sur- Females were often found near the mouth of face of the ground from dusk until dawn and a burrow, and sometimes removed from a bur- is distributed from southernmost Maine and row. In almost every case, females were found Ontario throughout the southeastern U.S. and within a meter of the burrow. Males, however, west to Baja California (Dondale & Redner were only occasionally found near a burrow 1990). Very little has been published concern- but never in a burrow. Spiders were housed ing the life history of this species. Individuals individually, in clear plastic containers (18.5 of Hogna carolinensis construct a burrow ϫ 7.5 ϫ 9 cm) with a sand substrate (approx. with a turret of sticks and grass surrounding 1 cm deep). Each spider was fed one adult the mouth of the burrow. Or, in some cases, cricket/week, and water was available ad li- these spiders may inhabit a deserted rodent bitum. Temperatures in the housing and test- burrow (Shook 1978). The depth of the bur- ing area ranged from 25±26 ЊC. row likely varies between geographical re- Escape Speed.ÐSpider escape speeds were gions and possibly with the substrate. In west measured on a wooden track 9 cm wide and Texas, where this study was conducted, I have 2 m long, with sheet-metal side walls approx- found burrows as deep as 25 cm (pers. obs.). imately 21 cm high. All trials were conducted Likely predators include lizards, centipedes during daylight hours. A start box was sepa- (pers. obs.), scorpions, coyotes, owls and var- rated from the track by a removable metal di- ious predacious insects (Shook 1978). If a vider (21 ϫ 9 cm). A spider was placed in the burrow is near, individuals will retreat to a start box, allowed to acclimate for 15 minutes, burrow when disturbed (Kuenzler 1958); how- the divider was raised, and the spider was ever, if a burrow is not near, the animal will prodded on the posterior end of the abdomen NELSON & FORMANOWICZÐESCAPE DISTANCE OF WOLF SPIDER 155 with a ®berglass rod until it ran. Using a stop- 0.8640) indicating that the speed of the model watch, we recorded the time that the spider predator from trial to trial was not signi®- crossed each 50 cm segment of track. Each cantly different. Another set of trials was run spider was run twice with 24 hours between using only the motor without the model pred- trials, after which time its mass was recorded. ator to rule out the possibility of cues from Spider cephalothorax lengths were measured the sound and vibration of the motor. In these at the end of the study with a caliper. trials, none of the spiders responded to the The fastest 50 cm speed (cm/s) for each spi- activation of the motor (n ϭ 10). The response der was used for statistical analyses. T-tests of the spider to the approaching model pred- were used to determine whether mass, ceph- ator was viewed through the observation holes alothorax length and speed differed between in the wall of the spider chamber. Escape was the sexes. Normality was evaluated using Sha- operationally de®ned as a spider turning and piro-Wilk's W, and none of the groups violat- running in the opposite direction from the ed this assumption at 0.05 level. Pearson cor- model predator. When the spider exhibited an relation coef®cients were used to examine escape response, the motor was stopped and relationships between speed and mass, speed the distance was measured from the front end and cephalothorax length, and mass and ceph- of the model to the original position of the alothorax length. spider. Spiders were run only once, unless no Flight distances.ÐThe distances at which response occurred, in which case they were spiders ¯ed from an approaching model pred- given a second trial. Flight distances were de- ator were recorded using a wooden runway termined for thirty-eight of the individuals (19 apparatus 2.7 m long, and 28 cm wide with males, 19 females) whose escape speeds had black plastic walls approximately 48 cm high. been measured. The spider chamber (32 ϫ 28 ϫ 51 cm) was Flight distances were transformed using the located at one end of the runway, and sepa- natural log to alleviate normality issues. Pear- rated from the runway by a glass divider. The son correlation coef®cients were calculated to walls of the spider chamber were sheet metal examine the relationships between ¯ight dis- with two observation holes (0.5 cm diameter) tance and spider size (mass & cephalothorax that allowed the spider to be viewed with min- length) and sprint speed (as measured above). imal disturbance. The ¯oor of the spider Discrepancies in sample size between tests re- chamber had a sand substrate 11 cm deep. To sulted from specimen mortality and unrespon- minimize vibratory cues, the runway and the siveness of some individuals. All statistical spider chamber were separated from the coun- tests were carried out using SPSS 11.0.2. ter top by 5.5 cm of foam rubber, and sepa- rated from each other by a space of 2 cm. A RESULTS 15 cm tall green plastic lizard was used as a Escape Speed.ÐThis species shows some model predator (meant to represent a novel degree of sexual size dimorphism (cephalo- predator rather than a particular known pred- thorax length: females, n ϭ 43, mean Ϯ SE ator) to elicit escape behavior. Each spider ϭ 12.92 Ϯ 1.09 mm; males, n ϭ 33, mean Ϯ was placed in the spider chamber for 10 min. SE ϭ 11.94 Ϯ 0.74mm). Female H. caroli- to acclimate. The model predator was con- nensis are signi®cantly larger than males ϭ Ͻ cealed by a black plastic curtain at the end of (mass, t52 3.946, P 0.001 two-tailed; ϭ Ͻ the runway, opposite the spider chamber. Af- cephalothorax length, t72.947 4.641, P ter 10 min., the model predator was pulled 0.001 two-tailed [unequal variances]). Mass toward the spider using a length of ®shing line and cephalothorax length were signi®cantly connected to a spool that was turned by a correlated (Pearson's r ϭ 0.470, P Ͻ 0.001). small motor at a speed of approximately 33.9 However, when correlations were examined cm/s (mean ϭ 1.4747 Ϯ 0.0922 [seconds per separately for the two sexes, this relationship 50 cm segment]). We ran a set of 10 test trials only held true for the males (r ϭ 0.439, P ϭ where we measured four 50 cm segments of 0.017, males; r ϭ 0.209, P ϭ 0.326 females). track to test for consistency of speed of the Although males and females differed in both model predator. Repeated-measures ANOVA measures of size, their sprint speeds were not ϭ ϭ (using trial as the repeated measure) showed signi®cantly different (t55 1.439, P 0.156, ϭ ϭ no signi®cant effect (F9,27 0.4963; P two-tailed). Neither mass nor cephalothorax 156 THE JOURNAL OF ARACHNOLOGY

Figure 1.ÐScatterplot of the natural log of ¯ight distances and sprint speeds (r2 ϭ 0.241, P ϭ 0.013). length was signi®cantly correlated with fastest Differences in ¯ight distances for males and sprint speed (Pearson's r ϭϪ0.145, P ϭ females may be the result of different cost- 0.296, n ϭ 54; r ϭ 0.077, P ϭ 0.571, n ϭ 56, bene®t relationships for males and females. respectively). Males and females may have very different Flight distance.ÐThere was no signi®cant lifestyles that require different considerations. correlation between spider size and log ¯ight Shook (1978) suggested that females might distance (mass, Pearson's r ϭ 0.024, P ϭ tend to stray farther from the burrow than 0.916, n ϭ 22; cephalothorax, r ϭϪ0.314, P males. If this is true, they may have a different ϭ 0.055, n ϭ 38). However, males tended to escape strategy. For spiders associated with ¯ee from the model predator at greater dis- burrows, escape consists of only a short sprint ϭ ϭ tances than females (t37 2.663, P 0.011, to a burrow. However, for an individual that two-tailed). Furthermore, there was a positive is not near its burrow, escape may involve a correlation between fastest sprint speed and ϭ ϭ longer sprint, as well as an assessment of log ¯ight distance (Pearson's r 0.491, P available shelter. To date, we have not been 0.013, n ϭ 25, see Figure 1). successful in getting this species to occupy DISCUSSION burrows in the lab. Although females will oc- The results of our study indicate that sprint casionally inhabit a man-made burrow, none speeds and ¯ight distances of H. carolinensis of our spiders have excavated their own bur- are not affected by spider size. There were no rows in the lab. In the ®eld, males were rarely signi®cant differences in speed between males found near a burrow. It is possible that male and females and no signi®cant correlation be- H. carolinensis are not usually associated with tween sprint speed and either measure of size a burrow and, therefore, are more reliant upon (mass or cephalothorax length). Although size running to escape a predator. It would be in- does not affect sprint speed or ¯ight distance, teresting to determine whether ¯ight distance sex appears to affect the decision to ¯ee. is related to distance from the burrow in this Males ¯ed from the model predator at greater species as has been shown in other organisms distances than females. (squirrels, Dill & Houtman 1989; Cichlid ®sh, NELSON & FORMANOWICZÐESCAPE DISTANCE OF WOLF SPIDER 157

Dill 1990; woodchucks, Bonenfant & Kramer more likely to continue foraging, since the 1996; and skinks, Cooper 1997). risk of capture for any given distance is less Sexual size dimorphism is common in spi- for a faster individual than for a slower indi- ders. Although not as exaggerated in wander- vidual. ing spiders as it is in web-builders, size di- In the present study, the individuals were morphism is still present. In wolf spiders, not performing any speci®c task. We therefore females may have longer cephalothoraxes, need to consider what costs may be associated larger chelicerae, and larger abdomens than with ¯ight and what bene®ts may be associ- males (Walker & Rypstra 2001). The different ated with staying. For spiders, the energy ex- body shape of males and females may result pended during escape can be costly (Prestwich in different values for costs and bene®ts used 1988). Therefore, in cases where the risk of in decision-making. The lack of a correlation capture is low, it may not be worth the effort between mass and cephalothorax length for the individual to attempt to escape. This among females re¯ects the overall difference cost might be higher for females, since they in body shape between males and females. are larger than males and may have to expend The stouter build of the female in this species more energy when running. results in the size of the abdomen contributing Another, perhaps more important cost of more to overall mass than in males. Males ¯ight in some species involves cryptic anti- have a smaller abdomen relative to cephalo- predator mechanisms. In cryptic species, ¯ight thorax length. As a result, males may possess may actually increase the risk of capture lower energy stores, therefore placing a higher (Heatwole 1968; Regalado 1998; Cuadrado et value on foraging. Females are generally con- al. 2001). As a predator approaches a cryptic sidered to be more effective foragers than individual, the individual must decide whether males, since they often consume more prey it has been detected, making it necessary to items (e.g. Walker & Rypstra 2001). It is, ¯ee. If, however, the individual ¯ees before however, possible that males consume less the predator has detected it, the individual due to their smaller size, but are more reliant may draw attention to itself and increase its on regularity of foraging success than females. risk of capture. The individual may also at- In this case, a male that has not fed recently tract other potential predators. In this type of may be willing to risk predation in order to situation, the cost of ¯ight is related to the continue foraging. However, a male that has probability of detection by the potential pred- recently fed may ¯ee when an approaching ator. This is a function of the perceptual ®elds predator is farther away. It would be interest- of both the predator and prey species. If the ing to determine if feeding regime or the time predator has a larger perceptual ®eld than the since last feeding has an effect on ¯ight dis- prey, the prey would bene®t by ¯eeing while tance and if that effect is different for males the potential predator is still relatively far and females. away. When the predator has a smaller per- There was a positive correlation between ceptual ®eld than the prey, the prey would sprint speed and ¯ight distance. Spiders that bene®t by waiting to ¯ee until the predator is were faster ¯ed at greater distances from the closer, and the probability of detection is high- approaching model predator while slower spi- er (Heatwole 1968; MartõÂn&LoÂpez 1999; ders waited until the model predator was clos- Cuadrado et al. 2001). er before ¯eeing. This relationship between The wolf spiders used in this study are a sprint speed and ¯ight distance may seem light mottled brown color and blend in readily counter-intuitive. The cost-bene®t model of with the desert substrate where they are likely Ydenberg & Dill (1986) predicted that faster to be encountered. We believe, therefore, that individuals should wait until the predator was the results of this study can be explained upon closer before attempting to escape. When the the basis of crypsis. In faster individuals, it predator is still relatively far away, the cost of may be advantageous to ¯ee at farther dis- ¯ight (lost foraging time) would be higher tances, since there is a higher probability that than the risk of predation (risk of capture), the individual will survive entirely on the ba- resulting in an inverse relationship between sis of escape speed. In slower individuals with maximum sprint speed and ¯ight distance. Ac- less chance of escaping solely on the basis of cording to the model, faster individuals are speed, individuals may rely on crypsis to es- 158 THE JOURNAL OF ARACHNOLOGY cape detection. It may be advantageous for manuscript in its early development. I would slower individuals to remain still, relying on also like to extend a special thanks to Gail crypsis for survival rather than ¯eeing and be- Stratton for her thoughtful and detailed review coming more conspicuous to the predator. of this manuscript. We appreciate the help of Our results are similar to those obtained by all those who aided in the collection of spec- Formanowicz et al. (1990) in ground skinks, imens, including Chris Brown, Chris Amaya, Scincella lateralis. In individuals that had ex- Dan O'Connell, Marina Gerson, and Julia perienced tail loss 48 hours prior to testing, Long. sprint speeds were signi®cantly reduced, and LITERATURE CITED individuals exhibited shorter ¯ight distances. Bonenfant, M. & D.L. Kramer. 1996. The in¯uence They interpreted the shorter ¯ight distances in of distance to burrow on ¯ight initiation distance slower individuals to be the result of a behav- in the woodchuck, Marmota monax. Behavioral ioral compensation for tail loss. They sug- Ecology 7:299±303. gested that autotomized individuals compen- Cooper, W.E.,Jr. 1997. Escape by a refuging prey, sated for decreased speed by adopting a the broad-headed skink (Eumeces laticeps). Ca- cryptic anti-predator strategy. This interpre- nadian Journal of Zoology 75:943±947. tation was based on information involving the Dill, L.M. 1990. Distance-to-cover and the escape relationship between ¯ight distances and cryp- decisions of an African cichlid ®sh, Melanoch- romis chipokae. Environmental Biology of Fish- sis in lizards (Heatwole 1968; Bauwens & es 27:147±152. Theon 1981). Heatwole (1968) examined the Dill, L.M. & R. Houtman. 1989. The in¯uence of relationship between ¯ight distance and levels distance to refuge on ¯ight initiation distance in of crypsis in Anolis lizards. They found that the gray squirrel (Sciurus carolinensis). Canadi- cryptic species exhibited shorter ¯ight dis- an Journal of Zoology 67:233±235. tances than those that were less cryptic. Bau- FernaÂndez-Juricic, E., M.D. Jimenez & E. Lucas. wens and Theon (1981) found similar results 2001. Bird tolerance to human disturbance in ur- in gravid lizards, Lacerta vivipara. Gravid liz- ban parks of Madrid (Spain): Management im- ards compensated for decreased speed by plications. Avian ecology and conservation in an adopting a cryptic anti-predator strategy. urbanizing world., Kluwer Academic. 261±275. FernaÂndez-Juricic, E., M.D. Jimenez & E. Lucas. In summary, maximum sprint speed was 2002. Factors affecting intra- and inter-speci®c not signi®cantly different for males and fe- variations in the difference between alert dis- males, and maximum sprint speed was not tances and ¯ight distances for birds in forested signi®cantly affected by the size of the indi- habitats. Canadian Journal of Zoology 80:1212± vidual. Furthermore, ¯ight distance was not 1220. signi®cantly related to size, but males tended Kuenzler, E.J. 1958. Niche relations of three species to ¯ee at a greater distance from a model pred- of Lycosid spiders. Ecology 39:494±500. ator. Sprint speed and ¯ight distance were pos- Martin, J. & P. Lopez. 1999. Nuptial coloration and itively correlated. This positive correlation mate guarding affect escape decisions of male lizards Psammodromus algirus. Ethology 105: was considered to be the result of a trade-off 439±447. between two alternative modes of predator Shook, R.S. 1978. Ecology of the wolf spider, Ly- avoidance: escape and crypsis. cosa carolinensis Walckenaer (Araneae). Journal of Arachnology 6:53±64. ACKNOWLEDGMENTS Walker, S.E. & A.L. Rypstra. 2001. Sexual dimor- We would like to thank the Jobeth Holub phism in functional response and trophic mor- and the Nature Conservancy for use of the phology in Rabidosa rabida (Araneae: Lycosi- dae). American Midland Naturalist 146:161±170. Independence Creek Nature Preserve as a col- lection site. We also thank Jonathan Campbell Manuscript received 2 June 2003, revised 15 March and John Bacon for reviewing portions of this 2004.