What is the matter with the gravity hypothesis?

Abstract

Although gravity undoubtedly exerts substantial effects on locomotor performance, we dispute the arguments made in defence of the Gravity Hypothesis (GH) in Moya-Laraño et al.'s Forum contribution. Moya-Laraño et al. presented a list of supposed errors in rebuttal of our paper (Brandt & Andrade 2007), yet failed to address core questions we raised as objections to the GH for sexual size dimorphism (SSD): (i) Why should muscle cross-sectional area be proportional to power? (ii) Why do so many studies fail to find a size-dependent decline in mass specific power? (iii) Why are there no studies showing an inverse relationship between size and climbing speed? In the absence of satisfactory answers to these questions, we remain unconvinced that the GH is a good explanation for patterns of sexual size dimorphism in spiders. Below, we counter some of the major points in their rebuttal. First, Moya-Laraño et al. argue that because muscles can activate a variable percentage of their fibres, our assumption that force should be proportional to muscle cross-sectional area is not realistic. However, the assumption that power output limits climbing speed, an assumption shared by both models, implies maximal muscle performance, which requires activation of all muscle fibres. The scaling of maximum force to muscle cross-sectional area arises from the parallel arrangement of muscle fibres and the serial arrangement of sarcomeres, as outlined in any modern physiology textbook (e.g. Randall, Burggren & French 2002, pp. 368–369). Moya-Laraño et al. are aware of this, since they write: ‘the force produced per muscle cross section has been shown to be constant across a wide range of body sizes and animal taxa (Medler 2002)’. Second, they argue that muscle power output should show an inverse relationship with size, as predicted by the GH. Medler's (2002) meta-analysis found that muscle contraction velocity is inversely related to size. Given that power is a product of force and contraction velocity, Moya-Laraño et al. proposed that power output should decline with size. However, both force and velocity change dynamically during the cyclical contractions typical of active locomotor muscles, complicating predictions about the scaling of power output. Importantly, power output is restricted to the period of active muscle contraction (Josephson 1993). As animal size decreases, muscle contraction frequencies increase, and the active contraction phase occupies an ever smaller proportion of the contraction cycle, thereby preventing power output from increasing as size decreases (Schilder & Marden 2004). Regardless of predictive models, empirical studies repeatedly find that, contrary to the GH, mass-specific mechanical power output does not vary with size (Marden 1987; Johnson et al. 1993; Full 1997; Schilder & Marden 2004). Third, Moya-Laraño et al. argue that their model is supported by a recent multi-species study in which climbing speed exhibits a curvilinear relationship with size, with a maximum at 42·5 mg (Foellmer & Moya-Laraño 2007). However, these results are not consistent with the inverse relationship predicted by the GH. Moreover, this analysis included uncontrolled variables likely to influence climbing speed, such as variation in reproductive status, sex, instar and phylogenetic relationships. For example, the heaviest individuals in their sample comprised gravid females, which are likely to be slow climbers due to the weight of eggs they carry. If we accept the suggestion that optimal climbing speed arises at 42·5 mg, then we are left with no explanation for SSD in species in which the females are larger than the optimal climbing size and males are considerably smaller than the optimum, as seen in many high habitat species, (Foellmer & Moya-Laraño 2007). Similarly, in species with females smaller than the optimal climbing size, males should be selected to be larger than females. Such male-biased SSD in spiders is very rare (Foellmer & Moya-Laraño 2007). Finally, the authors suggest that a lack of sufficient variation could account for our failure to detect an inverse relationship between body size and climbing speed. However, this claim ignores the substantial size variation in our two samples: CVbody mass = 32% and 52%, variation that allowed us to detect a significant positive effect of size on horizontal running speed. The GH (Moya-Laraño, Halaj & Wise 2002) is based on an erroneous biomechanical model, runs counter to numerous empirical studies in which mass-specific mechanical power output was found to be independent of size, and its prediction that climbing speed should be inversely related to size has no empirical support. We therefore maintain that selection for vertical climbing speed in males cannot account for patterns of SSD in spiders.