Fig. 2.

Adaptive manipulation to induce nematomorph-infected mantids to enter the perennial water bodies (A–C) and maladaptive attraction of these mantids to asphalt roads (D–F). A) The box plots of the DOP reflected from the water surfaces of perennial waters and intermittent waters and from asphalt roads. Whiskers are expressed as 1.5 times the interquartile range. B) Fitted logistic regression (with 95% CI) of the probability of polarized third selection based on the DOP. Transparent points represent the raw data. C) The number of infected mantids that entered the pools with different DOP. D) The prevalence of nematomorph (C. formosanus) infections on mantids (H. patellifera in TKY-A and OSK; T. formosana in TCC; and H. chinensis in TKY-H) in asphalt roads and natural arboreal habitats in four different sites in Taiwan (TW) and Japan (JPN). Numbers on the bars represent the sample sizes of the mantids. ND, no data available for the category. E) The number of walking events of the infected mantids on roads with different DOP. F) Photos of infected mantids walking and/or road-killed on asphalt roads. In (A) and (E), different letters within graphs indicate statistically significant differences.