Arachnologische Mitteilungen 55

Effect of spider web design on wasp parasite 31 to 26.0 °C and the mean annual precipitation from 3500 to 4500 mm at these sites ( Martínez 2012 , Tirimbina Biological Reserve 2010). Spiders and their hosts were collected in matu- re and old second growth lowland rainforests. In the laboratory each spider with its parasite was placed in a 30 cm diameter plastic container with plants of diffe- rent families (e.g. Commelinaceae and Crassulaceae) which covered nearly the entire diameter of the container. We re- corded daily the larval stage and the design of the web of the spider, which was fed every other day with Lucila sp. flies (Calliphoridae). We regularly photographed each spider and its parasitoid larva (daily in one spider) and video-recorded the behaviour of two last-stage larvae. Results “Cocoon web”. In the laboratory the spiders constructed webs similar to those in nature (Fig. 1a). The second-instar parasitoid larva induced the host spider to build the cocoon web and during the night it killed the spider. However, diffe- rences between the cocoon web and normal web were nearly imperceptible.The only differences between cocoon and nor- mal webs were that the spider was induced to produce a hole through the centre of the hub and to add a few additional threads in the central section of the orb, to which the cocoon was then attached by the larva (Fig. 1b, N = 3 cocoon webs). We did not observe the spider breaking the silk threads of the hub, but in one case a last instar larva constructed its cocoon within an open space in the hub which was intact the previ- ous night. In the field we also observed a cocoon constructed within the space in the centre of the hub (Fig. 1d). To observe a cocoon web built from scratch, we carefully removed one spider with a second-instar larva attached to its opisthosoma from its web the night before it killed its host. We then destroyed the web, cleaned the plant, and returned the spider with its larva back to the same plant in the same container, and allowed the spider to build its cocoon web. In this case, the second-instar larva induced the spider to build a relatively sparse, horizontal tangle that did not have an orbi- cular design (Fig. 1c). Larval development and behaviour. Two of the host spiders were found with wasp eggs attached to them (Fig. 2a) and in both cases the larva emerged three days later in the lab. Both larvae passed through three stages, which lasted a total of nine days (Tab. 1). We determined the number of larval stages by examining the number of shed-skins that were attached to the “saddle” (the mass of coagulated spider hemolymph which adhered to the spider’s opisthosoma; Nielsen 1923, Eberhard 2000,Weng & Barrantes 2007); the last stage is characterized by the presence of dorsal tubercles. Fig. 1: Distinct aspects of the web of Kapogea cyrtophoroides . a. Natural web in the forest (the web in the picture is similar to at least other 20 webs we have seen in the field). b. The hub of a natural web. c. Cocoon web that the larva induced the spider to build after the spider was removed from its web; arrows point to cocoon web threads. d. Central section of the cocoon web and the wasp’s cocoon