1. Spiders are the most sexually size dimorphic terrestrial animals and the evolution of this dimorphism is controversial. Patterns of sexual size dimorphism (SSD) in spiders have been related to individual performance and size. In 2002 Moya-Laraño, Halaj & Wise proposed the ‘gravity hypothesis’ to explain patterns of sexual size dimorphism in spiders whereby species building webs high in the vegetation are predicted to show greater SSD than those that build lower down. They advocated an advantage in climbing speed in smaller males searching for females in high places. The gravity hypothesis predicts a negative relationship between male size and climbing speed. In 2007 Brandt & Andrade questioned this interpretation and proposed that the pattern of SSD in spiders is better explained by an advantage for larger males of low-dwelling species to run faster along the ground.
2. We induced male spiders to run a standard distance up vertical poles of different diameters to examine the predicted relationship between size and climbing speed. We tested two species of extremely size-dimorphic orb-web spiders, Argiope keyserlingi and Nephila plumipes, that differ in the height at which females tend to build webs, and one species of jumping spider, Jacksonoides queenslandica, with low levels of size dimorphism. We also examined morphological determinants of horizontal motility by inducing males to run along a raceway.
3. Substrate diameter was consistently found to influence climbing performance. In N. plumipes, climbing speed was slowest on the widest diameter substrate. In A. keyserlingi, size-adjusted leg length and substrate diameter interacted to determine climbing speed, while in J. queenslandica, there was an interaction between body size and substrate diameter on climbing speed. In the effect of substrate diameter, we have identified a potential bias in previous tests of the gravity hypothesis.
4. Our results do not support the prediction of the gravity hypothesis. There was no evidence of a negative relationship between body size and climbing speed in the two orb-web species with high levels of SSD. Our results are also not consistent with a recent modification of the gravity hypothesis that suggests a curvilinear relationship between climbing speed and size.
5. Body size was positively associated with maximum running speed only in the cursorial hunter J. queenslandica. For this spider, results are more consistent with Brandt & Andrade’s explanation for variation in SSD in spiders, that larger males are selected for superior running ability in low-dwelling species, rather than selection for smaller size for climbing to females in high-dwelling species.
Moya-Laraño, Halaj & Wise (2002) identified a pattern of greater female-biased SSD in spider species occupying vertical structures such as trees, compared to those inhabiting shrubs and the ground layer. To explain this they proposed a biomechanical model that states that size (or mass) is inversely proportional to the speed that males can achieve on vertical surfaces (i.e. a negative relationship between size and vertical speed). It suggests that small males should be favoured during mate searching in those species in which males are required to travel in a three-dimensional habitat and climb to high habitats to reach females. Furthermore, a recent revision suggests a curvilinear rather than the original linear relation between climbing speed and male size (Moya-Laraño et al. 2009). While referred to as the ‘gravity hypothesis’ its basic tenets relate directly to motility in mate searching, postulating that males with smaller body size are more mobile and have superior performance in a situation of scramble competition. Some unrelated studies provide logical support. For example, weight reduction through self-inflicted amputation of a reproductive organ (pedipalp) in males of the spider Tidarren sisyphoides is known to result in increased locomotor performance in terms of maximum speed and endurance (Ramos, Irschick & Christenson 2004).
The gravity hypothesis has recently been sharply disputed by Brandt & Andrade (2007a,b), and stoutly defended by its proposers (Moya-Laraño et al. 2007a). Brandt & Andrade (2007a) proposed an alternative motility-based explanation of the patterns of SSD observed by Moya-Laraño, Halaj & Wise (2002). They hypothesized that the trend of greater female-biased SSD in tree-dwelling spiders arose not through motility advantages for small climbing males in tree-dwelling spiders but instead through motility advantages for large males of low-dwelling spiders. Brandt & Andrade (2007a) supported their argument with evidence from a study on male black-widow spiders (Latrodectus hesperus) that live close to the ground. In this species, larger males ran faster than small males on horizontal surfaces over short distances, whereas there was no difference in the speed that large and small males climbed up short sticks.
In the present study, we investigate morphological determinants of vertical motility in male spiders over a standard distance on three different diameters of wooden dowels. We can, therefore, also evaluate the effect of substrate diameter on climbing performance in males. Substrate diameter is known to influence locomotor performance in other animals (e.g. Losos & Sinervo 1989; Losos & Irschick 1996) but this possibility appears not to have been studied previously in spiders. An interaction between climbing speed and dowel diameter would indicate the possibility of biases in previous studies that have employed single substrates of different diameter. We also examine horizontal motility in a standardized raceway. We examine two Australian species of orb-web spiders that exhibit high degrees of female-biased SSD but have different tendencies in the heights at which females build webs, and one Australian jumping spider species that is a cursorial hunter with relatively low SSD. Thus, we seek to evaluate whether and how vertical and horizontal motility in male spiders is influenced by their morphology in species with different degrees of SSD.
Materials and methods
Collection and maintenance of specimens
We collected males of Argiope keyserlingi as adults from low bushes (Lomandra longifolia) bordering Shrimpton’s Creek, adjacent to Wilga Park and the Macquarie University campus in North Ryde (New South Wales, Australia) between December 2007 and January 2008. Adult males of Nephila plumipes were collected from Bicentenial Park, Pymble (New South Wales) between February and March 2008. Jacksonoides queenslandica were collected from various sites near Cairns (Queensland) in April 2008, where they build nests and hunt on the surface of riverside rocks and boulders. All specimens were maintained in the laboratory under controlled temperature (24°–26 °C) and relative humidity (65–75%). Light was supplied via overhead fluorescent strips on a 12 h light : 12 h dark cycle. Spiders were held in individual 1L, ventilated, plastic containers, constantly supplied with water from a reservoir in the base via a dental wick, and fed a diet of 3–4 Queensland fruit flies (Bactrocera tryoni) weekly, supplemented with house flies (Musca domestica) on alternate weeks.
Vertical motility trials
The performance of A. keyserlingi, N. plumipes and J. queenslandica males was tested in climbing trials. Individual males were removed from their holding cages and gently positioned at the bottom end of a vertical wooden dowel on a 100 mm long staging area, with the aid of a soft sable hair paintbrush. Dowels were coated with a textured Spray Stone paint (White Knight Paints, Australia) to provide traction. Males were then induced to climb up a 250 mm section of dowel three times. Trials were initiated by touching the male lightly on its hind legs with the soft-haired brush. Each male was tested on dowels of diameter 6, 16 and 25 mm. Males were not allowed to pause during the trials. Any male pausing was lightly brushed on the hind legs to encourage it to continue to the top of the dowel. The order in which each male encountered the three different diameter dowels was randomized. We recorded the time taken to complete each climb (from crossing the zero line at the end of the staging area, to contacting the top of the dowel with the first pair of legs) using a digital stopwatch. Climbing speed for each male over the total distance of 250 mm was calculated by dividing the distance travelled by the time taken to complete the climb. We used the fastest speed recorded for each male in statistical analyses in order to minimize the negative impact of differences in motivation to climb and sub-maximally performing individuals (see Losos, Creer & Schulte 2002).
Horizontal motility trials
We investigated performance of male spiders in running trials in specially constructed raceways. Horizontal raceways were consistent with those used by Brandt & Andrade (2007a). They were 700 mm long, 50 mm wide, with sides 50 mm high, fabricated from white foam board with a fine sandpaper base for traction. Males were removed from their individual cages as for the vertical motility trials, placed in a 50 mm staging area at one end of the raceway and gently brushed to induce them to run along the raceway to an identical staging area at the opposite end. Again, pausing was not permitted and constant running was encouraged by following the male with a soft paintbrush. Any males that paused where gently induced to run by touching on the hind legs. We recorded the time taken for A. keyserlingi and J. queenslandica males to complete each run of 600 mm (from the time they crossed a line marked on the substrate indicating the end of the staging area until their first pair of legs crossed a line marked on the substrate to denote the beginning of the staging area at the opposite end of the raceway) using a digital stopwatch. We used the fastest (speed) of three runs for each male in statistical analyses (as above). In consecutive trials, males were run in different directions along the raceway. In preliminary observations, N. plumipes males were considerably more reluctant to run on the raceway than A. keyserlingi and J. queenslandica males. As a result, we recorded running times in N. plumipes over a shorter distance of 250 mm.
Measurement of specimens
Males were weighed to the nearest 0·1 mg using a Shimadzu (Shimadzu Corporation, Kyoto, Japan; Model N595, Type AX200) electronic balance at the end of each trial. Each specimen was digitally photographed using a ProgResC10 digital camera focused through an Olympus SZX12 stereo-microscope and using proprietary software (Jenoptik L. O. S. GmbH, Germany). Body dimensions were calculated from digital images using ImageJ 1.30v (National Institutes of Health, Bethesda, MD, USA). To facilitate photography, males were subdued on the surface of a small (60 mm dia.), up-turned, petri dish with clear plastic film (Glad Products, Padstow, NSW, Australia). For each male, we measured mass, cephalothorax width and length, and the tibia-patella length on the first leg.
To assess the repeatability of performance measurements we calculated interclass correlation coefficients (one-way random effect model) between the fastest and next fastest trials for each individual in each of the vertical and horizontal motility assays (e.g. Brandt & Andrade 2007a).
We entered cephalothorax width and length into a principal components analysis (PCA) and extracted a single PC that estimates fixed body size (e.g. Foellmer & Fairbairn 2005a; Brandt & Andrade 2007a). We then obtained residuals from simple regressions of patella-tibia length (size-adjusted leg length) and body mass (condition) on PC1. This allowed us to examine the effects of leg length and condition on male locomotor performance separate from fixed body size. Individuals with relatively long legs may have a performance advantage (e.g. Sinervo & Losos 1991) and those in better condition with greater energy stores and larger opisthosoma (relative to fixed size/prosoma size) potentially suffer a disadvantage (Brandt & Andrade 2007a). Thus, for examination of climbing performance in males, PC1 scores, size-adjusted leg length and body mass residuals (condition) were entered into a linear mixed model with repeated measures on the diameter of the wooden dowel. Individual identity was entered as a random factor in the model. Model selection was based on removal of non-significant interaction terms according to the best subset criterion with Akaike’s Information Criterion to obtain the final model. The predictions of the gravity hypothesis about climbing speed and male morphology should hold for any proxy of body size (Moya-Laraño et al. 2009). However, as weight, or more specifically the cube root of weight, is predicted to be the underlying determinant of climbing performance in males (e.g. Foellmer & Moya-Laraño 2007), we also examined the effect of the cubic root of weight on climbing speed by substituting it into models in place of our measure of body condition (residual body mass).
Backwards stepwise multiple regression analysis (with standard values for inclusion (0·05) and exclusion (0·1) of variables, as performed by Brandt & Andrade 2007a) was used to investigate running performance in males, with PC1 scores, size-adjusted leg length and body mass residuals (condition) entered as predictors of running speed. Measures of body mass are not included in these models as it does not feature in the relevant explanation of SSD in spiders. Statistical analyses were carried out using SPSS version 16.0 for Macintosh OS X (SPSS Inc., Chicago, IL, USA), SAS version 9.1 (SAS Inst. Inc., Cary, NC, USA) and JMP version 5.0 (SAS Inst. Inc., Cary, NC, USA). Partial residual plots were generated in R version 2.9.
Species natural history and web-building characteristics
Both A. keyserlingi and N. plumipes are extremely sexually size dimorphic orb-web spiders. At the collection sites, females typically build their webs in shrubs and bushes. N. plumipes females build large orb-webs, either in aggregations or alone (Elgar 1989), at varying heights above the ground and as high as above 10 m at our field site. Typically, N. plumipes webs are built higher in the vegetation than those of A. keyserlingi. Jacksonoides queenslandica is a cursorial hunting spider, typically found on the surface of rocks and boulders.
Variation in linear morphological measures, body mass, vertical and horizontal locomotor performance in adult male A. keyserlingi, N. plumipes and J. queenslandica was considerable (Table 1). The repeatability of the two fastest climbs for individual males was high on all three diameters of wooden dowel for each species of spider examined (Table 2). Repeatability in running trials was also high (Table 2).
Table 1. Summary of morphological and performance data for spiders in vertical and horizontal motility trials. A. k., N. p. and J. q. refer to the species Argiope keyserlingi, Nephila plumipes and Jacksonoides queenslandica respectively. Leg length refers to the patella-tibia length of the first leg
All values are means ± standard error.
*Sample sizes in vertical trials: A. k. (n = 35), N.p. (n = 38), J.q. (n = 32). **Sample sizes in horizontal trials: A. k (n = 35), N. p. (n = 61), J. q. (n = 31).
18·43 ± 0·77
24·68 ± 1·19
24·00 ± 1·01
Leg length (mm)
3·33 ± 0·05
5·56 ± 0·10
3·59 ± 0·08
Prosoma width (mm)
2·49 ± 0·04
2·20 ± 0·04
2·31 ± 0·04
Max. climbing speed (cm s−1)*
6 mm dowel
9·26 ± 0·62
4·36 ± 0·27
5·70 ± 0·30
16 mm dowel
7·47 ± 0·57
4·71 ± 0·36
5·69 ± 0·40
25 mm dowel
6·25 ± 0·46
3·91 ± 0·31
5·94 ± 0·44
Max. running speed (cm s−1)**
33·60 ± 2·45
4·95 ± 0·19
3·98 ± 0·45
Table 2. Repeatability of measures of vertical and horizontal locomotor performance. Values presented are the interclass correlation coefficient between the two fastest trials for males climbing vertically on 6, 16, and 25 mm diameter dowels and running horizontally along a raceway
Vertical motility trials
There were no significant interactions between dowel diameter and either size (PC1: F2,62 = 1·20, P = 0·309) or mass residuals (F2,62 = 1·73, P = 0·186), but there was a significant interaction between diameter and size-adjusted leg length (F2,62 = 3·28, P = 0·044). Consequently, the interactions between diameter and PC1 scores and diameter and residual mass were removed from the model. The final model (χ27 = 34·90, P < 0·001) had dowel diameter and the interaction between diameter and size-adjusted leg length as significant factors (Table 3). We found a significant effect of dowel diameter on the maximum speed of A. keyserlingi males, with speed generally decreasing (increased time to climb) as the diameter of the dowel increased. None of the other variables in the model predicted climbing performance. Size, size-adjusted leg length and residual mass did not affect climbing speed in male A. keyserlingi. Furthermore, when the cube root of body mass was included in the model in place of residual body mass, it also did not influence climbing behaviour significantly. There was a significant interaction between the effects of dowel diameter and size-adjusted leg length on the maximum speed that A. keyserlingi males climbed wooden dowels (Fig. 1). Pairwise comparisons of the slopes of the relationships between size-adjusted leg length and maximum climbing speed on the three different diameter substrates indicated that the slopes differed between 6 mm and 16 mm (P = 0·032) and 6 mm and 25 mm (P = 0·028), but not between 16 mm and 25 mm (P = 0·95). Males with relatively shorter legs for their size experienced a performance advantage when climbing on the smallest diameter substrate. However, on wider diameter substrates, males with longer legs relative to body size experienced a performance advantage in climbing.
Table 3. Summary of statistical model of determinants of climbing speed in male A. keyserlingi (type 3 fixed effects). The effect of the cube root of body mass was also estimated by including it in the model instead of residual mass (see text)
Size-adj. leg length
Diameter × size-adj. leg length
No significant interaction terms were observed between dowel diameter and the predictor variables so they were removed from the final model (diameter × PC1: F2,68 = 1·055, P = 0·345; diameter × size-adjusted leg length: F2,68 = 1·646, P = 0·200; diameter × residual mass: F2,68 = 0·710, P = 0·495). In the final model (χ25 = 28·2, P < 0·001) we found a significant effect of dowel diameter on the speed of N. plumipes males climbing on wooden dowels (Table 4). As with A. keyserlingi, there was a general pattern of speed decreasing as the diameter of the dowel increased to 25 mm (Fig. 2). Within-subjects pairwise comparisons indicated differences in climbing speed between males on 16 mm compared to 25 mm dowels (P < 0·001) and between males climbing on 6 mm and 25 mm diameter dowels (P = 0·038), but not between 6 mm and 16 mm dowels (P = 0·493). N. plumipes males climbed significantly slower on the widest dowel (25 mm).
Table 4. Summary of statistical model of determinants of climbing speed in male N. plumipes (type 3 fixed effects). The effect of the cube root of body mass was also estimated by including it in the model instead of residual mass (see text)
Size-adj. leg length
Male size (PC1) did not affect climbing performance in N. plumipes however, the residual mass of males did (Table 4), with males that were heavier for a given body size (in better condition) completing the climb faster than males that were light for their size. There was also a negative relationship between size-adjusted leg length and climbing performance; males with relatively longer legs were slower to complete the climb. When we substituted cube root of body mass for residual mass in the model, we found that cube root body mass was positively related to climbing speed (Table 4). So heavier males climbed quicker. This alternative model also included a significant negative effect of body size (PC1) on climbing speed (F1,34 = 12·24, P = 0·0013, ß = −0·272) that was not evident in the previous model.
We obtained a final model (χ27 = 23·0, P < 0·01) in which the only interaction retained was between substrate diameter and body size (PC1) (diameter x size-adjusted leg length: F2,56 = 0·870, P = 0·425; diameter x residual mass: F2,56 = 0·365, P = 0·696). Unlike the other species tested, we did not find evidence of a significant effect of dowel diameter on the speed that J. queenslandica males climbed wooden dowels (Table 5). There was, however, a significant interaction between body size (PC1) and dowel diameter (Fig. 3). Pairwise comparisons of the slopes for the interaction of diameter and fixed body size (PC1) on climbing performance in J. queenslandica males indicated a significant difference in the effect of size on performance at 6 mm and 16 mm (P = 0·004). The slopes for 16 mm compared to 25 mm (P = 0·099) and 6 mm and 25 mm (P = 0·179) were not significantly different. At a substrate diameter of 16 mm, climbing performance increases with increasing fixed body size, such that small males are disadvantaged in climbing compared to larger males (Fig. 3). This effect is not as evident for males climbing on the wider diameter substrate (25 mm). Smaller males had a climbing advantage on smaller diameter (6 mm) dowels (Fig. 3). The positive relationship between size-adjusted leg length and climbing speed (Table 5) indicates that males with relatively long legs for their body size climbed faster. Residual body mass did not affect climbing performance in male J. queenslandica (Table 5). Furthermore, we also failed to find an effect of cube root of body mass on climbing speed in J. queenslandica, when it replaced residual mass in the model (Table 5).
Table 5. Summary of statistical model of determinants of climbing speed in male J. queenslandica (type 3 fixed effects). The effect of the cube root of body mass was also estimated by including it in the model instead of residual mass (see text)
Size-adj. leg length
Diameter × PC1
Horizontal motility trials
Stepwise multiple regression analysis of maximum running performance in male A. keyserlingi and N. plumipes did not reveal any significant predictors of running speed. We obtained a statistically significant model in J. queenslandica (F1,30 = 4·775, R2 = 0·141, P = 0·037, ß = 0·158) that included only male fixed body size (PC1) as a significant (positive) predictor of running performance (t = −2·164, P = 0·039). Running speed increased with male size (Fig. 4).
We observed high repeatability in the performance of males in vertical and horizontal motility trials. These results indicate that observed relationships between performance and morphological measures are robust, that there are consistent differences in locomotor performance between individuals, and that the majority of variability in both vertical and horizontal performance was due to individual differences between spiders.
Our data on climbing behaviour in males of three different spiders are not consistent with the key prediction of the original gravity hypothesis for SSD in spiders - that smaller males climb faster (Moya-Laraño, Halaj & Wise 2002). We failed to detect evidence for the expected negative relationship between male body size or weight and climbing speed in the two spider species in which males must climb up to the webs of females for matings (Tables 3 and 4). These findings are in line with those of Brandt & Andrade (2007a) who found male body size did not predict climbing performance in the highly dimorphic web spider Latrodectus hesperus.
Recently, in a departure from the original concept, the gravity hypothesis has been modified to meet the requirements of an observed curvilinear rather than the original purely negative relation between male size and climbing speed (Foellmer & Moya-Laraño 2007; Moya-Laraño et al. 2009). This arises from a phylogenetically controlled comparative analysis that indicated that the climbing speed – body mass relationship conformed better to a quadradic than a linear function. This latest incarnation of the gravity hypothesis suggests an optimal body size for determining maximal climbing performance approximately equivalent to a body mass of 43 mg and suggests that SSD evolves when females exceed this optimum body mass for climbing. It also indicates that, below this critical threshold, climbing speed and body size are in fact positively related. The negative relationship between climbing speed and male size predicted by the original gravity hypothesis is said only to occur when size exceeds the optimum body mass for climbing. As is the case for most spiders, males of all three species tested here fall below this optimal weight for climbing (although females in the two extremely size dimorphic orb-weavers exceed it). Our data also do not support the curvilinear gravity hypothesis. Neither fixed body size nor body mass were positively related to climbing speed in the highly size dimorphic A. keyserlingi. In N. plumipes we observed conflicting results that are largely inconsistent with the curvilinear gravity hypothesis. Body size did not affect climbing speed in the basic analysis. In accordance with the curvilinear gravity hypothesis, body mass (cubic root) was positively related to climbing performance, however, the negative relationship with body size that went along with this (after residual body mass was replaced by cubic root mass) is incompatible with the curvilinear hypothesis. We consider the observed mass effect to reflect a positive influence of body condition, rather than body mass, on performance as indicated by the positive relation between residual body mass and climbing speed (Table 4).
Male body size and climbing speed were positively related in the jumping spider J. queenslandica (Table 5). Again, this is strictly in accordance with the new curvilinear gravity hypothesis. However, this species exhibits low levels of size dimorphism and would not be expected to be the subject of selection for increased climbing speed to facilitate mate searching, as females build their nests under rocks and males wander over the rocks to find them. As body mass (cubic root) did not affect climbing speed in J. queenslandica, this relationship probably owes its explanation to the running advantage observed in larger males translating into other forms of locomotion rather than any specific influence of gravity on climbing performance. Overall, the consistent lack of evidence supporting the key predicted relationships of both the original and the new curvilinear gravity hypotheses in the two orb-web building spiders with extreme sexual dimorphism and variation in the height at which females are available to males, combined with a lack of evidence from other studies (Foellmer & Fairbairn 2005a; Brandt & Andrade 2007a), weighs heavily against the gravity explanation for extreme SSD in spiders.
We found no evidence of a positive relationship between male body size and running performance in either of the two highly dimorphic orb-weavers examined. The pattern of male performance in vertical movement trials in J. queenslandica is consistent with the finding that larger males also ran faster in horizontal movement trials (Fig. 4). This pattern of size-related locomotor performance is more compatible with Brandt & Andrade’s (2007a) explanation of the patterns of SSD than that offered by Moya-Laraño, Halaj & Wise (2002). Brandt & Andrade (2007a) tested climbing and running performance in the Western Black Widow, Latrodectus hesperus, and found no relationship between size and climbing speed whereas running speed was positively related to size. Amaya, Klawinski & Formanowicz (2001) also observed a positive relationship between male size and running speed in the cursorially hunting wolf spider Schizocosa ocreata. Our results for J. queenslandica provide tentative support for the idea that the patterns of SSD in the spiders are not due to selection for vertical climbing speed in mate searching males via the gravity hypothesis. However, Brandt & Andrade’s (2007a) alternative explanation also requires evidence that selection for protandry in males generates extreme SSD. Regardless of other criticisms of this model (Moya-Laraño et al. 2007a), evidence of selection for protandry in spiders is generally lacking (Foellmer & Moya-Laraño 2007).
Substrate, leg length and performance
A simple laboratory bioassay has been developed to test the gravity hypothesis in which speed in climbing bursts is assessed (Foellmer & Moya-Laraño 2007; Brandt & Andrade 2007a; Moya-Laraño et al. 2007b, Moya-Laraño et al. 2009). However, there is little consistency in general methodology, distances of trials and test substrate characteristics. In the present study, substrate diameter consistently influenced climbing performance in three spider species (Figs. 1–3), although the direction and strength of the effect varied among species. The effect of substrate diameter on climbing performance demonstrates a need for standardization of tests and suggests that previous lack of methodological consistency may limit the general applicability of earlier findings. Furthermore, variation in the effect of substrate diameter on climbing speed among males of the same species and among the three species examined, demonstrates the utility of performing tests over a range of substrates to assess the influence of habitat characteristics on individual locomotor performance in spiders. Variation in substrate diameter is known to affect locomotor performance in other animals, notably in lizards (e.g. Losos & Sinervo 1989; Losos & Irschick 1996), that may share generalities in climbing dynamics with other climbing species, e.g. cockroaches (Goldman et al. 2006). In arboreal lizards, surface diameter affects running speed. The extent of this effect is known to vary with morphology (leg length) and microhabitat features.
Highly size dimorphic spiders also show different patterns of leg dimorphism, with adult males exhibiting relatively longer legs than females, especially in orb-web spiders (Elgar, Ghaffar & Read 1990; Prenter, Montgomery & Elwood 1995; Foelix 1996). Leg length differences have been ascribed to morphological adaptations to mate searching in adult males (Prenter, Montgomery & Elwood 1995; Foelix 1996; Le Grand & Morse 2000; Foellmer & Fairbairn 2005a) and also the mechanics of pendulum motion, where spiders move while hanging up-side down (Moya-Laraño et al. 2008). Locomotor performance is a potentially important determinant of fitness especially in situations of scramble competition. However, there is a paucity of information on the impact of scramble competition on spider systems that may limit the understanding of processes underlying the evolution of SSD in spiders, especially for motility and mate-search based explanations. Leg length is known to affect the locomotor performance of other animals (e.g. Barbosa & Moreno 1999; Irschick & Garland 2001). In the strongly size dimorphic giant weta, Deinacrida rugosa (Orthoptera: Anostostomatidae), males with longer legs and smaller bodies are more mobile and have greater mating success (Kelly, Bussière & Gwynne 2008). Here we found an effect of leg length on climbing performance in A. keyserlingi (Fig. 1). Our data indicating an effect of leg length on performance in A. keyserlingi are consistent with the findings for the related A. aurantia where net selection for longer legs in males has been supported, at least in one population (Foellmer & Fairbairn 2005a). Male A. keyserlingi with relatively short legs had a performance advantage when climbing on substrates of small diameter, whereas males with relatively long legs for their body size performed better on wider diameter substrates (Fig. 1). As vegetational structure often varies from wider to thinner diameter with increasing height, there may be an increasing advantage of relatively shorter legs as males climb higher in the vegetation in search of females. Whether disadvantages at one height (e.g. wider diameter at base) are compensated for at another (e.g. smallest diameter at outer tips) remains to be tested, however our results suggest a morphology-performance trade-off with respect to vegetational structure.
Moya-Laraño et al. (2009) indicate that some species of spiders will be better adapted to move on inclined surfaces. They suggest that shape differences among spider species, especially those related to leg dimensions and the position of the legs relative to the body, could explain performance differences in climbing ability. This is the most likely explanation for the variation of individual performance with substrate diameter observed in the current study, with varying leg dimensions among the three species contributing to variable performance on the different diameter dowels (Table 1). Different leg dimensions will be better suited to facilitate attachment for climbing. The smallest of the three species, A. keyserlingi, with the shortest legs performed best when climbing on the smallest diameter substrate. It is interesting to note that males of N. plumipes, the species with the longest albeit relatively thin legs, did not achieve the fastest running speeds as a result of their relatively larger stride length. While leg movements were not specifically examined here, A. keyserlingi males appear to compensate for their small leg size by an increased stride frequency compared to the other spiders examined here. Jacksonoides queenslandica climbed best on the widest substrate that more closely resembled (of the three examined) the flat surfaces they run over in nature.
In addition to climbing, Moya-Laraño, Halaj & Wise (2002) also posited a predatory avoidance advantage of smaller males over larger males and this prediction of the gravity hypothesis has not been directly tested. It was proposed that faster moving, smaller males are better able to escape predatory attacks and reach females first in situations of scramble competition. There may also be an added advantage to being small in avoiding predation by aerial predators by being less conspicuous (see Moya-Laraño et al. 2008). Tests of forced bursts of locomotor activity, like those so far used to test climbing and running performance in spiders, may address more the predator avoidance than the climbing speed prediction of the gravity hypothesis. Standard tests of locomotor performance routinely examine maximum performance and while spiders may generally move at sub-maximal speeds in nature, maximum attainable speed is the only realistic metric on which to base comparison of performance ability. Thus, in practice it may not be feasible to distinguish between the anti-predator and mate searching aspects of the gravity hypothesis in tests of performance. However, a recent study of antipredator responses in jumping spiders (Stankowich 2009) indicates size dependency in antipredator behaviour. Smaller males retreated further than larger males before adopting a defensive stance. This study also suggested a trade-off between body size and physiological state (energy stores) when spiders were deciding whether to flee from a threat or turn and defend themselves.
In summary, our data on climbing performance in A. keyserlingi, N. plumipes and J. queenslandica do not provide support for the gravity hypothesis that female-biased SSD in spiders arises through selection for enhanced climbing performance in small males (Moya-Laraño, Halaj & Wise 2002). We have, identified an effect of substrate diameter that has potentially important implications for studies of climbing performance in orb-web spiders. Our data on J. queenslandica provide tentative support for the hypothesis that observed patterns of SSD in spiders may be explained by directional selection for large size in males allowing them to run faster on horizontal surfaces (Brandt & Andrade 2007a). However, this hypothesis also requires selection for protandry in males to generate SSD, evidence of which is currently lacking. A general adaptive explanation of small size in males and the uncoupling of the relationship between male and female size in extremely dimorphic spiders, therefore, remain elusive goals.
We thank Alan Taylor and Ben Fanson for statistical advice and assistance, Maria Castillo-Pando for assistance in maintaining spiders in the laboratory and Marie Herberstein for advice on collecting spiders. We are indebted to Dinesh Rao, Frank Messina, Jordi Moya-Laraño, Guadalupe Corcobado and two anonymous reviewers for comments on the manuscript.