Context-dependent running speed in funnel-web spiders from divergent populations


*Correspondence author. E-mail:


1. Locomotor performance can influence individual fitness through several ecological contexts, such as prey capture and predator escape. One means of determining which contexts act as significant selective forces on running speed is to quantify individual speed in each context. The underlying hypothesis is that animals will exhibit their highest speeds in contexts most crucial to fitness.

2. We measured running speeds in three ecological contexts (prey capture, fleeing predators and territory defence) in lab-reared offspring of the funnel-web spider Agelenopsis aperta collected from two arid grassland and two riparian populations. Arid populations experience little predation pressure, are prey limited, and are highly territorial; riparian populations experience high predation, have high prey availability, and are less territorial in nature.

3. The offspring of arid individuals exhibited their highest burst speeds in territory defence, and ran more slowly in response to predator threats. The offspring of riparian populations, however, ran fastest when responding to predatory threats and displayed lower velocities in prey capture and territory defence. Thus, our findings support the hypothesis that A. aperta are selected to exhibit their highest speeds in contexts most important to their fitness.

4. Contextual use of running speed can differ among conspecific populations experiencing differing selective forces on locomotion.


The importance of burst speed for individual fitness has become well-established in a variety of animal species (reviewed in Irschick et al. 2008). Superior burst-speed performance has been linked with many measures of fitness, including increased survival (Jayne & Bennett 1990; Swain 1992; Watkins 1996; Miles 2004; Husak 2006a; Pruitt unpublished data), success during agonistic encounters (Garland, Hankins, & Huey 1990; Robson & Miles 2000; Perry et al. 2004), male territory size (Husak 2006a; Peterson & Husak 2006), and reproductive success (Husak et al. 2006; Husak, Fox, & Van Den Bussche 2008). Speed can hypothetically influence an organism’s fitness through several ecological contexts: improved territory defence, enhanced ability to escape predators, and/or greater probability of capturing prey (Garland & Losos 1994; Irschick & Garland 2001). Because locomotion can serve many roles in accomplishing important ecological tasks of a species, it can be difficult to determine the contexts favouring increased burst speed. However, if animals tend to exert the most effort (i.e. their highest burst speeds) in contexts most important for fitness, quantifying to what extent burst speed is utilized in different contexts can help elucidate the selective pressures favouring higher running speeds (Irschick & Garland 2001). For instance, maximum burst speed has been shown in some species to be reserved for anti-predator situations (Domenici & Blake 1997; Irschick 2000a,b; Irschick & Garland 2001), suggesting that predation pressure is the most important selective force on running speed in these species.

The importance of various ecological contexts as selective pressures on burst speed might vary within a species, depending on the age class, sex or population. For example, within a population of collared lizards (Crotaphytus collaris), females exhibit their highest running speeds when escaping predators. Males, however, reserve their highest speeds for territorial encounters. Neither sex uses high speeds when attempting to capture a potential prey item (Husak & Fox 2006). Males and females experience differing selective pressures in this system, and this is evidenced by differences in how the sexes use speed. Males are highly territorial and suffer a potentially high cost to their fitness if they do not quickly respond to intruding rivals (Husak et al. 2006; Husak, Fox, & Van Den Bussche 2008), whereas females are not territorial in this population and have little selective pressure to respond intensively to rival females. Thus, the relative strengths of selection on locomotor performance across demographic groups within a single population are reflected by differences in use of locomotor performance.

Similarly, one might expect the importance of selective forces to vary between ecologically and/or behaviourally divergent populations, and therefore, individuals from different populations may exhibit their highest speed in different contexts. Many animal species exhibit ecotypic variation in behavioural characters reflecting local adaptation to their environment (e.g. Arnold 1981a,b; Riechert & Hall 2000; Scotti & Foster 2007), but rarely do studies examined population variation in performance (Garland & Adolph 1991). Since performance traits are typically thought to be the direct targets of selection, as opposed to the morphological and physiological traits that underlie them (Huey & Stevenson 1979; Arnold 1983; Bennett & Huey 1990), examining geographic variation in performance traits may elucidate the extent to which form and function may respond to differing selective pressures within a species’ range. By applying this intra-specific approach to performance characters in well-studied animal systems we can (i) assess to what extent performance characters can be shaped by local environmental pressures and (ii) help elucidate to what extent behavioural differences shape context-specific locomotor performance. To our knowledge, however, no such population level investigation has been performed.

In this study, we tested whether the offspring of individuals collected from different source populations vary in their usage of running speed. Specifically, we tested the hypothesis that offspring from populations experiencing divergent selection pressures will exhibit their highest speeds in contexts known to be important to individual fitness in the source populations. Furthermore, to determine whether individual performances are independent between contexts, we tested for correlations between performance among contexts (i.e. do fast escapers also run fast during territorial contexts?). Correlations in individual performance across contexts have to the potential to generate evolutionary conflict; if faster individuals apply higher speeds in every context, they might needlessly ‘overachieve’ in low-priority circumstances. Metabolic costs to ‘overachieving’ have been suggested in multiple studies (Pennycuick 1975; Hoyt & Taylor 1981; Perry et al. 1988; Kenagy & Hoyt 1989; Irschick et al. 2005), but few investigations have considered running performance in multiple ecological contexts to test for cross-contextual correlations (Husak 2006b).

We used the funnel-web spider Agelenopsis aperta (Araneae; Agelenidae) (Fig. 1) as the model species for this investigation, as locomotion is important to numerous aspects of its natural history. Previous behavioural investigations in A. aperta have identified ecotypic variation in aggressive behaviour (reviewed in Riechert, Singer, & Jones 2001). Individuals from arid populations are prey limited and experience little pressure from avian predation (Hammerstein & Riechert 1988; Riechert 1991). In contrast, spiders from riparian populations experience higher prey availability but suffer from a high incidence of avian predation (e.g. 40% of individuals are lost per week during bird-nesting periods) (Riechert & Hedrick 1990; Riechert 1993). Furthermore, due to low prey availability, the importance of defending established territories from intruding conspecifics is much greater for individuals in arid habitats (Riechert & Hedrick 1993). Thus, relative to individuals from riparian habitats, arid individuals are markedly more aggressive towards predators, prey, and conspecifics (e.g. arid individuals exhibit shorter latencies of attack towards prey, are more prone to escalate in territorial encounters and demand a larger territory) (see Riechert, Singer, & Jones 2001 for review). We test the prediction that the offspring of arid individuals will exhibit their highest speeds when capturing prey and defending their territories, contexts known to influence fitness in arid populations. Conversely, we predict the offspring of riparian individuals will exhibit their highest speeds in anti-predator situations and display lower running speeds when foraging and defending their territories.

Figure 1.

 A female funnel-web spider Agelenopsis aperta (Araneae; Agelenidae) residing at the entrance of its funnel.

Materials and methods

Study system

Agelenopsis aperta is a funnel-web spider which inhabits a variety of habitats throughout the southwestern U.S.A. and Mexico (Riechert & Tracy 1975). This species builds a permanent sheet-web either along the ground or in low-lying vegetation. The sheet-web is attached to a protective funnel of webbing which extends into a protective microhabitat (e.g. grass clump, under a rock, into the soil). This species utilizes high burst speeds in territory defence, prey capture and anti-predator contexts (Riechert & Hedrick 1993) and is capable of maneuvering quickly both on and off of webbing (Foelix 1996; Pruitt personal observations).

Collection and laboratory maintenance

Spiders were hand collected as early instar juveniles from two arid grassland (Palo Duro (NA): 34·57°N–101·40°W, New Mexico Lava Beds (NALB): 33·38°N–104·53°W) and two riparian [Perry Texas (RP): 31·33°N–94·74°W, Kerrville Texas (RK): 30·05°N–91·44°W] localities in June 2008. Individuals were collected by dropping a prey item on their sheet and capturing the spider as it initiated its capture sequence. Spiders were then transported back to laboratory at the University of Tennessee, Knoxville. They were housed individually, maintained between 24–25°C and fed a maintenance diet of 2-week-old crickets ad libitum twice weekly until maturity. Individuals were then randomly mated with an individual from the same source population. The resulting egg cases were isolated and misted with a spray bottle twice weekly. Upon hatching, spiderlings were isolated in 59 mL plastic containers and fed size-matched crickets ad libitum twice weekly until maturity. As the juveniles reached maturity, a single female was randomly selected from each of forty broods for each source population (total N = 160) and run through the locomotor assays described below. We limit our present investigation to mature females because adult male A. aperta exhibit a dramatically reduced behavioural repertoire. For instance, males commonly abandon their territories to search for females and refuse to forage after their final molt.

Assessing burst speed usage

Burst speed assays were performed four days after a routine feeding. To avoid exhausting an individual, trials were each separated by 1 week. All trials were completed on an individual within 4 weeks of their final molt. To minimize the effects of experience and the age of an individual, the trial sequence was randomized for each spider.

To assess burst speed, females were transferred to a new 960 mL clear plastic container lined with graphing paper with half-centimetre demarcations. Females were then given a 1 week acclimation period. All females constructed fully functional webs in this period and were kept on their normal feeding schedule. Each female’s running speed was then assessed once in each of the three functional contexts: anti-predator response, prey capture, and territory/web defence. Females were assessed for their anti-predator behaviour by disturbing their web with a puff of air (i.e. a simulated avian threat) (Riechert & Hedrick 1993). This puff of air was issued 6 cm distant from the female’s web using a medicine bulb while the female was positioned at the front of her funnel. Running speed during prey capture was assessed by dropping a size-matched cricket (25% of the focal female’s mass ±0·5%) 10 cm away from the female’s funnel. Similarly, running speed in territory defence was assessed by placing a size-matched female conspecific (80% of the focal female’s mass ±0·5%) onto the focal female’s web 10 cm away from the funnel. Intruder females were coaxed onto the web via an open tipped syringe (after Riechert 1978). Prey items were size matched to standardize resource value, and intruder conspecifics were size matched to ensure escalatory aggressive behaviour by the focal female.

We assessed burst speed as the time taken for a female to travel from rest across 10 cm of webbing, as measured from the time elapsed on a video recording (Sony HDR-HC3). Repeatability of laboratory running speed for Agelenopsis was assessed in a previous study using partitioning of variance into two components, within versus between individuals (after Boake 1989), and was determined to be highly repeatable (r = 0·78) (Pruitt unpublished data).

Statistical methods

Neither the maximum speed exhibited nor the proportion of maximum speed used in any context varied between representative populations of the arid or riparian ecotypes (Table 1). Thus, for the remainder of our analyses we pooled data within each of the arid and riparian ecotypes. We calculated the proportion of ‘maximal’ speed individuals used in each of the three contexts. Since we did not have ‘maximal’ capacity as determined in a laboratory (e.g. Husak & Fox 2006), we calculated proportions using two methods. First, we divided the speed exhibited in each context by the highest running speed displayed by each individual in any context. For comparisons with previous work in lizard systems (e.g. Irschick 2000a,b), we also analysed our data using a second method by dividing individuals’ speeds in each context by the average highest speed exhibited for their population (i.e. individuals’ speed/average highest speed of the source population). Throughout our results we refer to results from the first method as the ‘proportion of highest speed’ and results from the second method as the ‘proportion of average highest speed’. Because these response variables are proportions, the data were arcsine-transformed prior to analyses.

Table 1. T-tests comparing the burst speed exhibited and the percent of highest speed used in three ecological context between representative populations of the arid and riparian ecotype
 TraitT-scoreDFNA Mean (SD)NALB Mean (SD)P-value
 Maximum sprint speed exhibited−0·177825·75 cm s−1 (4·72)25·92 cm s−1 (4·69)0·57
 Burst speed: anti-predator1·177821·82 cm s−1 (5·67)20·46 cm s−1 (5·03)0·26
 Burst Speed: Foraging0·077822·45 cm s−1 (5·71)22·36 cm s−1 (6·46)0·95
 Burst speed: territory defence0·397823·39 cm s−1 (4·75)22·44 cm s−1 (5·69)0·69
 % Highest speed: anti-predator1·227884·31% (12·72)79·44% (15·59)0·11
 % Highest speed: foraging0·217887·05% (14·58)85·31% (16·03)0·42
 % Highest speed: territory defence0·537891·20% (11·34)88·67% (14·94)0·29
 % Average highest speed: anti-predator1·267884·70% (22·1)78·91% (19·4)0·21
 % Average highest speed: foraging0·177887·21% (22·2)86·33% (24·9)0·86
 % Average highest speed: territory defence0·527890·80% (18·5)88·54% (21·9)0·61
 TraitT-scoreDFRP Mean (SD)RK Mean (SD)P-value
  1. No significant differences were detected after Bonferroni correction between the representative populations of either ecotype. (a) Is the raw burst-speed exhibited, b and c represent the proportion of highest speed used using the two methods to calculate the proportion of highest speed (b = speed in each context/highest speed exhibited by that individual, c = speed in each context/average highest speed of their source population).

 Maximum sprint speed exhibited−0·457827·76 cm s−1 (4·22)25·29 cm s−1 (4·91)0·65
 Burst speed: anti-predator−0·877824·13 cm s−1 (4·66)23·04 cm s−1 (5·47)0·39
 Burst speed: foraging−0·517822·31 cm s−1 (5·01)21·73 cm s−1 (5·03)0·61
 Burst speed: territory defence−1·517821·46 cm s−1 (5·17)19·91 cm s−1 (4·07)0·14
 % Highest speed: anti-predator−0·627893·66% (8·84)90·50% (15·01)0·73
 % Highest Speed: Foraging−0·357886·77% (13·51)86·16% (12·04)0·63
 % Highest speed: territory defence−0·697883·20% (13·51)79·93% (14·93)0·75
 % Average highest speed: anti-predator−0·537893·72% (18·1)91·1% (25·66)0·59
 % Average highest speed: foraging−0·167886·60% (19·5)85·9% (19·92)0·87
 % Average highest speed: territory defence−1·147883·33% (20·1)78·7% (16·11)0·26

We first compared burst speeds used by individuals across contexts (i.e. anti-predator, foraging, territory defence) within each ecotype using a repeated-measures anova with post-hoc Tukey tests. We also compared the proportion of highest speed used and the proportion of average highest speed (separately) across contexts within each ecotype using a repeated-measures anova with post-hoc Tukey tests. To compare the two ecotypes’ use of burst speed in each context we used t-tests with a Bonferroni corrected α = 0·017. We did these analyses to compare burst speeds, as well as the proportion of highest speed used and the proportion of average highest speed.


We detected significant differences in burst speeds used across different contexts within the arid (F2,79 = 5·12, = 0·007) and riparian ecotypes (F2,79 = 12·22, < 0·0·001; Fig. 2). Similarly, we detected significant differences in the proportion of highest burst speed used across ecological contexts for both the arid (F2,79 = 6·09, = 0·003) and riparian (F2,79 = 12·01, < 0·001; Fig. 3) ecotypes and in the proportion of average highest speed (arid: F2,79 = 5·11, = 0·007; riparian: F2,79 = 12·19, < 0·001; Fig. 3). For all three analyses, post-hoc Tukey tests indicated that individuals from arid populations exhibit their highest burst speeds, and use a higher proportion of their highest speed, in territory defence and their lowest speeds in response to simulated predator threats. Arid individuals’ speed when attacking prey was intermediate and indistinguishable from their responses in other contexts. Riparian populations, however, exhibited their highest burst speeds, and higher proportions of their highest speed, in response to simulated predator threats, and displayed lower and statistically indistinguishable velocities in prey capture and territory defence (Figs 2 and 3).

Figure 2.

 Mean burst speeds (cm s−1) exhibited by the arid and riparian ecotypes in each of three ecological contexts: anti-predator, foraging and territory defence. Bars which do not share a letter significantly differ within an ecotype using post-hoc Tukey tests at α = 0·05.

Figure 3.

 Mean proportion of highest speed (± SE) used by the arid and riparian ecotypes in each of three ecological contexts: anti-predator, foraging and territory defence. Bars which do not share a letter significantly differ within an ecotype using post-hoc Tukey tests at α = 0·05. The upper panel represents the proportion of highest speed used, and the lower panel represents the proportion of average highest speed used (see text for details). Both methods revealed the same trend.

We confirmed the anova findings using a contingency table analysis for each ecotype, independently where we noted the context in which each individual exhibited their highest speeds and compared the observed distribution to the expected distribution assuming random use of speed among contexts (Table 2). These analyses agreed with the anova results and showed that individuals from riparian populations tended to exhibit their highest speeds in anti-predator situations, whereas individuals from arid populations non-randomly exhibited their highest speeds in territorial encounters.

Table 2.   The proportion (and number) of individuals from arid and riparian populations that exhibited their highest speed in each context
  1. Chi-squared statistics compared the observed distribution to the expected distribution if individuals exhibited their highest speeds randomly among contexts.

Anti-predator0·18 (14)0·54 (43)
Foraging0·37 (30)0·28 (22)
Territory defence0·45 (36)0·18 (15)

Arid and riparian ecotypes did not differ in their highest sprint speeds (t158 = −0·46, = 0·67; Table 3). However, pair-wise comparisons of burst speeds used in the three contexts revealed significant differences between ecotypes (Table 3). Similarly, the proportion of highest sprint speed used in each ecological context between ecotypes revealed significant differences (Table 3). Individuals from riparian populations used higher speeds in response to simulated predator threats compared to individuals from arid populations, whereas the ecotypes were indistinguishable in their speed usage when capturing prey. Individuals from arid populations used higher burst speeds in territory defence compared to individuals from riparian populations.

Table 3. T-tests comparing arid versus riparian ecotypes in highest burst speed, burst speeds in three ecological contexts, and the percent of highest speed used in three ecological contexts
  1. The two different methods to calculate the proportion of highest speed revealed similar trends (a = speed in each context/highest speed exhibited by that individual, b = speed in each context/average highest speed of their source population).

 Highest sprint speed0·421580·67
Anti-predator speed−3·111580·003
Foraging speed0·401580·69
Territory defence speed3·001580·004
(a)% Highest speed: anti-predator−5·47158<0·001
% Highest speed: foraging0·451580·33
% Highest speed: territory defence4·29158<0·001
(b)% Highest speed: anti-predator−3·111580·003
% Highest speed: foraging0·131580·897
% Highest speed: territory defence2·851580·005

We then tested whether individual differences in burst speed were correlated across contexts (i.e. whether fast individuals were consistently fast, regardless of ecological context). We found that running speeds were significantly and positively correlated across ecological contexts for both the arid and riparian ecotypes (Table 4).

Table 4.   Pearson’s correlation coefficients (r) comparing individual running speed across contexts: anti-predator, foraging and territory defence
  1. All correlation coefficients were significant (< 0·001).



Our population-level comparison revealed several novel findings about how selection operates on performance. First, even though the populations did not differ in their highest speeds, they did differ in how fast individuals ran in the three ecological contexts we examined. Second, the differential use of speed among populations in the three contexts matched a priori predictions which were based on previous knowledge of ecological differences between the populations. Finally, the speeds that individuals used in any given context were positively correlated, and this was consistent across populations. In other words, good performers in one context were good performers in the other two contexts, and this was true in both arid and riparian populations. Because the test spiders were offspring raised in a common garden, differences between ecotypes appear to reflect selected, genetic-based differences between ecotypes. However, this design is insufficient to completely remove field-derived parental effects.

The behavioural ecology of Agelenopsis aperta has been the subject of several prior investigations, and there is substantial pre-existing knowledge about the selective pressures shaping individual fitness in arid versus riparian habitats. Arid populations are prey limited and experience little predation pressure (Hammerstein & Riechert 1988; Riechert 1991). Furthermore, because prey is limited in arid habitats, females from arid sites are particularly aggressive when defending territories from intruding conspecifics (Maynard Smith & Riechert 1984; Riechert & Maynard Smith 1989; Riechert & Hedrick 1993). Thus, we predicted individuals from arid habitats would exhibit their highest speeds when defending territories and/or attacking prey. Our data generally support this hypothesis (Figs 2 and 3); arid individuals exhibited their highest burst speeds when defending territories, intermediate speeds when attacking prey and low speeds when fleeing predators. In contrast, riparian populations experience high prey availability, high instances of predation, and the pressure to defend established territories is significantly lower (Riechert & Hedrick 1990; Riechert 1993). Accordingly, riparian individuals exhibited their highest speeds when fleeing predators, and were significantly slower when reacting to prey and conspecifics on their webs (Figs 2 and 3). Future selection studies on the populations will determine to what extent the population differences in speed afford a fitness advantage to individuals. Nonetheless, our data are consistent with two key evolutionary hypotheses: (i) A. aperta exhibit their highest running speeds in ecological contexts most important for fitness, and (ii) in systems where the context(s) shaping relative fitness vary between populations, contextual use of running speed will covary accordingly.

There is a growing body of evidence that although individual performance is frequently associated with variation in physiological and morphological characters, performance is commonly mediated by behaviour (Garland & Losos 1994; Irschick & Garland 2001; Husak 2006b; Husak & Fox 2006). Our finding that individuals vary how they use speed in different contexts supports this hypothesis. The two ecotypes did not statistically differ in their top speeds, on average, but they modulated their speeds according to context (Figs 2 and 3, Table 2). Thus, our data suggest that there is apparently selection to not use top speeds in every situation. Interestingly, despite the fact that individuals varied their usage of their top speeds, individual differences in raw performance (cm s−1) do not appear independent (i.e. individual differences in performance were generally maintained across ecological contexts) (Table 4).

Our results revealed that individuals fast in one context were typically faster in all contexts (Table 4), whereas a similar study on collared lizards (Husak 2006b) found no correlations among field speeds in the same three contexts. That is, collared lizards that were fast foragers were not necessarily faster escapers. This is an intriguing difference, and the paucity of data on field speeds for the same individuals in natural populations makes it difficult to interpret this discrepancy in a general sense. One hypothesis that could explain our results is that faster speeds are not universally favoured in every context and fast individuals needlessly out-perform slower individuals in some low-priority situations. This would imply that faster individuals tend to ‘overachieve’ (Irschick et al. 2005) in some instances, using a higher proportion of their abilities than is necessary in some contexts. If there is a metabolic cost to such ‘overachieving’ (Pennycuick 1975; Hoyt & Taylor 1981; Perry et al. 1988; Kenagy & Hoyt 1989; Irschick et al. 2005), then there may be selection against excessive use of speed. Thus, the net result of selection for generally faster speeds, combined with opposing selection against overachieving, could lead to maintained variation in the population. The strength of selection in each direction, and temporal variation in selection intensity (Grant 1999), would determine the extent of variation that would be maintained in a population and constrain the distribution of locomotor performance upon which selection can operate. More comparative data on field speeds for the same individuals within species, and the fitness consequences of that variation, are necessary for these hypotheses to be tested. Comparative data will also allow generation of a priori predictions of when one should expect to see correlations among field speeds and when one should not.

A key finding of our study is that ecologically divergent populations can exhibit their highest burst speeds in different contexts. It is not uncommon that studies investigating burst speed measure performance in only a single context, most commonly as an intense anti-predator escape response (e.g. chasing an individual down a track in the laboratory). Our results suggest this protocol has the potential to provide an incomplete or even misleading impression of some systems. For instance, if we were to consider running speed in only an anti-predator context in A. aperta, we might conclude that riparian individuals were, on average, faster runners than arid individuals and that higher running speeds are associated with increased predation pressure. However, when we considered a suite of ecological contexts (e.g. anti-predator, foraging, territory defence) we revealed that the ecotypes do not differ significantly in their top speeds; they merely exhibit their peak performances in different contexts. We feel these data emphasize a needed caution when evaluating performance characters in lesser known systems: (1) some species might exhibit their fastest speeds in non-anti-predator contexts, and (2) the contexts in which individuals use their highest speeds can vary between populations, and thus, contextual use of running speed is best viewed as an age class, sex, and population-dependent phenomenon.


We would like to thank Jason Jones, Kyle Demes, Michael DeAngelis, and Darrin Hulsey for their comments on previous versions of this manuscript. We are particularly indebted to the editor, Susan Riechert and two anonymous reviewers, whose thoughtful comments have improved the clarity of our work. This research was supported by grant #0315901 awarded by the Population Biology panel of the National Science Foundation.