Parasitoid pupa as potential prey. Out of six potential predators, five caused the mortality of the bare pupa, but with significantly different probabilities (GLM-b, χ25 = 24.1, P = 0.0002): both Pyrrhocoris and Lasius caused 100% mortality, followed by Acheta and Gryllus, which caused approximately 70% mortality, while Shelfordella caused about 20% mortality, and Platymeris did not consume pupa (Fig. 1A). Pupa mortality was not affected by the age of the pupa (GLM-b, χ21 = 0.2, P = 0.65) but declined slightly, though significantly, with the increasing body size of potential predators (GLM-b, χ21 = 10.4, P = 0.0155, Fig. 1B). The feeding latency was not affected by predator body size, predator species, or pupa age (GLM-g, F < 3.2, P > 0.17), and was, on average, 1737 s (SD = 1961).
Cocoon protection (the first defensive line). Out of six potential predators, four captured and consumed pupae protected by the cocoon, but with significantly different probabilities (GLM-b, χ25 = 22.4, P = 0.0004): the highest mortality was caused by Gryllus, followed by Acheta, Shelfordella, and Pyrrhocoris (Fig. 2A). Pyrrhocoris attacked only pupa with a soft wall (29%, N = 7). Generally, there was a significant difference in pupa mortality when comparing bare and protected pupa (GLM-b, χ25 = 21.6, P = 0.0006): Lasius, Pyrrhocoris, and Acheta caused lower mortality among protected pupa than among bare pupa.
The mortality of pupa protected by a cocoon was not affected by the cocoon type (GLM-b, χ21 = 0.1, P = 0.72) but increased slightly, though significantly, with the increasing body size of predators (GLM-b, χ21 = 4.0, P = 0.045, Fig. 2B). The mortality was significantly different for two mouthpart types (GLM-b, χ21 = 8.6, P = 0.003, Fig. 2C): predators with chewing mouthparts caused higher mortality than predators with sucking mouths. Scavengers caused significantly higher mortality than collectors (GLM-b, χ21 = 5.5, P = 0.019, Fig. 2D). Feeding latency was not affected either by cocoon type, predator size, or predator species (GLM-g, F < 15, P > 0.16) and was, on average, 27973 s (SD = 38729). The latency to attack was not affected significantly by predator size, foraging guild, or mouthpart type (GLM-g, F < 47, P > 0.05), but was affected significantly by cocoon type (GLM-g, F1,51 = 76.1, P = 0.0155): pupae in cocoons with a soft wall were attacked with a short latency (Fig. 3).
Web and cocoon location (the second defensive line). The mortality of the pupa was marginally significantly different between the locations of the web (GLM-b, χ21 = 3.5, P = 0.062): it was 5% for webs in the air compared to 20% for webs on the ground. Predators caused significantly higher mortality than collectors (GLM-b, χ21 = 9.8, P = 0.0018, Fig. 4A). Predators with chewing mouthparts caused significantly higher mortality than suckers (GLM-b, χ21 = 7.2, P = 0.0073, Fig. 4B).