Climate affects geographic variation in host-plant but not mating preferences of Timema cristinae stick-insect populations

Background: The behaviour of individuals within a species often varies geographically and this can affect reproductive isolation between populations. Published data show that natural selection stemming from both host-plant use and reinforcement affect reproductive isolation in the stick insect Timema cristinae, by generating heritable population differences in feeding and mating preferences (i.e. habitat and sexual reproductive isolation, respectively). Genomic divergence in T. cristinae is correlated to climate, but the effects of climate on feeding and mating preferences have yet to be tested. Questions: How do multiple sources of natural selection combine, interact or counteract one another to affect geographic variation in T. cristinae behaviour? How does this affect reproductive isolation between populations? Organisms: Twelve populations of T. cristinae. Each is found on one of two host plants (Ceanothus or Adenostoma) and populations also differ in local climatic conditions. Eight of these populations undergo maladaptive hybridization with an adjacent population and thus exhibit the potential for reinforcement. But the other four populations are geographically isolated. Methods: We combine the published data with new data and analyses on climatic variability among populations. We employ univariate and multiple regression analyses to examine the relationship between mean population-level host-plant feeding and mating preferences and three factors: host-plant use, reinforcement, and climate. Results and conclusions: Climate was correlated with population-level host preferences: mean preference for the nutritionally superior host (i.e. Ceanothus) increased among populations with increasing temperature and aridity. This pattern occurred for populations on both hosts and thus climate neither increased nor decreased habitat isolation between populations on different hosts. We propose selection to prefer increases in Ceanothus under hotter and drier climatic conditions and does so for populations on both hosts. Climate was not correlated with population-level mating preferences because sexual isolation is largely driven by reinforcement between adjacent populations that are most similar in climate. Thus, host-plant use, reinforcement, and climate, coupled with patterns of gene flow across the landscape, combine to determine geographic variation in the behaviour of T. cristinae.