The race for reaching mates by the time they are receptive, or sexual selection by scramble competition, has received little attention. We argue that smaller males are favored in species in which the male must climb to reach females located in high habitat patches. This new explanation we term the "gravity hypothesis" of sexual size dimorphism (SSD). We show that a simple biomechanical model of animal movement predicts that: (1) selection should favor a comparatively smaller size in the searching sex when searching involves climbing; and (2) this effect should be stronger in larger species than in smaller species. In reaching high habitats, smaller, faster searchers will be favored either through sexual selection by scramble competition and/or by escaping predation easier by running faster on vertical surfaces. Different spider species are found at a wide range of heights. We compiled a dataset of spider taxa and arranged their habitats according to four height categories, ranked from soil surface to trees. We show that, after controlling for phylogeny, both predictions of the gravity hypothesis of SSD are met. Thus, it appears that the constraint imposed by gravity on climbing males is a selective factor in determining male dwarfism.

This chapter uses data for 489 spider species from fifteen families to describe patterns of variation in sexual size dimorphism (SSD), and to evaluate hypotheses explaining these patterns. The direction and magnitude of SSD is found to depend strongly on the size measure chosen, and the use of carapace width is recommended because it is less affected by condition than body mass or length. Comparative analyses reveal that spiders do not exhibit allometry consistent with Rensch's rule. Instead, females appear to have diverged more than males over evolutionary time, and male and female body size show uncorrelated co-evolution, which is unusual for animals. Only two adaptive hypotheses — fecundity selection favouring large size in females and gravity selection favouring small size in males — have general explanatory power for patterns of SSD in spiders. However, processes may differ among species and comprehensive studies of selection within given species are needed.

Summary 1.Body condition (defined as the relative amount of energy reserves in the body) is an animal trait with strong ecological implications. In some animal taxa (e.g. arthropods), the external volume of the body part in which most nutrients are stored (e.g. abdomen) is used interchangeably with body mass to estimate body condition, making the implicit assumption that abdomen residual volume is a good surrogate of residual mass. However, the degree of correlation between these two measures should largely depend on the density of the nutrients stored. 2. We simulated two food-supplemented experimental groups of animals, each storing a slightly different amount of lipids either in their abdomens or in their entire bodies, and explored (i) how different estimates of condition were able to detect fixed differences between the groups; and (ii) how the amount of lipids stored could affect the outcome of non-intrusive measures of condition on a dichotomous variable (e.g. survival, mating success). We found that density body condition (body mass statistically controlled for structural body size and body volume) has much greater power to detect differences between experimental groups or effects on binary response variables than do classic mass/size or volume/size condition indices. 3. Using data on Lycosa tarantula (L.), a burrowing wolf spider, we report dramatic differences among these three indices in their ability to detect sex differences in the effect of feeding treatment on body condition at maturity. In particular, a plot of residual mass against residual volume reflecting nutrient density suggests that poorly fed spiders are nutritionally unbalanced, since well-fed spiders invest in nutrients of very different density. 4. Furthermore, using data on Scathophaga stercoraria (L.), the yellow dung fly, we found that an index of density condition was better at distinguishing condition differences among three populations than were mass or volume condition estimates alone. 5. We propose that including these three surrogates of condition (mass, volume and density) will substantially improve the accuracy of non-intrusive estimates of body condition, thus providing more powerful tools with direct application in a wide range of disciplines.

Summary 1.Evidence for reproductive senescence in invertebrate natural populations is scant probably because most groups are short-lived or because they lack natural markers of age. 2. Lycosa tarantula (L.) (the Mediterranean Tarantula) (Araneae, Lycosidae) is a slowly ageing burrowing wolf spider, in which females can reproduce for two consecutive seasons. Females in their first reproductive season (1Y) can easily be distinguished from females in their second reproductive season (2Y) because the lack of pilosity in the latter. 3. The diet of 1Y and 2Y females was supplemented and their reproductive performance was compared with that of control, non-food supplemented females. The predictions were that senescent 2Y females would show a worse reproductive performance than 1Y females, and that they would not be able to improve their performance relative to 1Y after food supplementation. 4. The predictions were met. Older females gained less mass, laid smaller egg sacs, produced fewer spiderlings and, if food supplemented, invested a smaller fraction of their mass in egg sacs. Although 2Y females foraged less actively than younger females, 2Y did not improve their performance relative to 1Y following food supplementation. This pattern of changes provides evidence for reproductive senescence in a natural spider population.

No abstract

Interactions among predators that prey on each other and are potential competitors for shared prey (intraguild [IG] predators) are widespread in terrestrial ecosystems and have the potential to strongly influence the dynamics of terrestrial food webs. Ants and spiders are abundant and ubiquitous terrestrial IG predators, yet the strength and consequences of interactions between them are largely unknown. In the leaf-litter food web of a deciduous forest in Kentucky (USA), we tested the direct and indirect effects of ants on spiders and a category of shared prey (Collembola) by experimentally subsidizing ants in open plots in two field experiments. In the first experiment, ant activity was increased, and the density of ants in the litter was doubled, by placing carbohydrate and protein baits in the center of each plot. Gnaphosa spiders were almost twice as abundant and Schizocosa spiders were half as abundant in baited plots relative to controls. There were more tomocerid Collembola in baited plots, suggesting possible indirect effects on Collembola caused by ant-spider interactions. The second experiment, in which screening of two mesh sizes selectively excluded large and small worker ants from a sugar bait, revealed that the large ants, primarily Camponotus, could alone induce similar effects on spiders. Gnaphosa biomass density was almost twice as high in the plots where large ants were more active, whereas Schizocosa biomass density was reduced by half in these plots. Although tomocerid densities did not differ between treatments, tomocerid numbers were negatively correlated with the activity of Formica, another large ant species. Path analysis failed to support the hypothesis that the ant Camponotus indirectly affected tomocerid Collembola through effects on densities of spiders. However, path analysis also revealed other indirect effects of Camponotus affecting tomocerids. These results illustrate the complexity of interactions between and within two major IG predator groups with disparate predatory behaviors, complexities that will have consequences for functioning of the forest-floor food web.

BackgroundSexual cannibalism may be a form of extreme sexual conflict in which females benefit more from feeding on males than mating with them, and males avoid aggressive, cannibalistic females in order to increase net fitness. A thorough understanding of the adaptive significance of sexual cannibalism is hindered by our ignorance of its prevalence in nature. Furthermore, there are serious doubts about the food value of males, probably because most studies that attempt to document benefits of sexual cannibalism to the female have been conducted in the laboratory with non-natural alternative prey. Thus, to understand more fully the ecology and evolution of sexual cannibalism, field experiments are needed to document the prevalence of sexual cannibalism and its benefits to females.Methodology/Principal FindingsWe conducted field experiments with the Mediterranean tarantula (Lycosa tarantula), a burrowing wolf spider, to address these issues. At natural rates of encounter with males, approximately a third of L. tarantula females cannibalized the male. The rate of sexual cannibalism increased with male availability, and females were more likely to kill and consume an approaching male if they had previously mated with another male. We show that females benefit from feeding on a male by breeding earlier, producing 30% more offspring per egg sac, and producing progeny of higher body condition. Offspring of sexually cannibalistic females dispersed earlier and were larger later in the season than spiderlings of non-cannibalistic females.Conclusions/SignificanceIn nature a substantial fraction of female L. tarantula kill and consume approaching males instead of mating with them. This behaviour is more likely to occur if the female has mated previously. Cannibalistic females have higher rates of reproduction, and produce higher-quality offspring, than non-cannibalistic females. Our findings further suggest that female L. tarantula are nutrient-limited in nature and that males are high-quality prey. The results of these field experiments support the hypothesis that sexual cannibalism is adaptive to females.

Field experiments to test the hypothesis that a cannibalistic species is territorial are rare. We conducted two field experiments to test the hypothesis that adult females of the Mediterranean tarantula, Lycosa tarentula (L.) are territorial. In a relatively longterm experiment we placed a female intruder in an artificial burrow within the hypothesized territory of a resident. The intruders disappeared from their burrows at 7ϫ the rate of spiders in a reference and two control treatments. Residency status, not relative size, determined whether the intruder or the resident remained, and evidence suggests that the winner frequently cannibalized the loser. We also conducted a short-term field experiment in which we induced encounters between females. The results were consistent with territorial defense because escalation was more likely if spiders were similar in size, and cannibalism, the outcome of one-third of the encounters, occurred only after escalation. Thus, adult females of the Mediterranean tarantula are territorial, and cannibalism may be a consequence of territorial defense.