Arms race between weevil rostrum length and camellia pericarp thickness: Geographical cline and theory

The geographical cline of the coevolving traits of weevil rostrum (mouthpart) length and camellia pericarp (fruit coat) thickness provides an opportunity to test the arms race theory of defense (pericarp thickness) and countermeasure (rostrum length) between antagonistically interacting species. By extending the previous model for the coevolution of quantitative traits to introduce nonlinear costs for exaggerated traits, the generation overlap, and density-dependent regulation in the host, we studied the evolutionarily stable (ES) pericarp thickness in the Japanese camellia (Camellia japonica) and the ES rostrum length in the camellia–weevil (Curculio camelliae). The joint monomorphic ES system has a robust outcome with nonlinear costs, and we analyzed how the traits of both species at evolutionary equilibrium depend on demographic parameters. If camellia demographic parameters vary latitudinally, data collected over the geographical scale of rostrum length and pericarp thickness should lie on an approximately linear curve with the slope less than that of the equiprobability line A/B of boring success, where A and B are coefficients for the logistic regression of boring success to pericarp thickness and rostrum length, respectively. This is a robust prediction as long as the cost of rostrum length is nonlinear (accelerating). As a result, boring success should be lower in populations with longer rostrum length, as reported in the weevil–camellia system (Toju, H., and Sota, T., 2006a. Imbalance of predator and prey armament: Geographic clines in phenotypic interface and natural selection. American Naturalist 167, 105–117). The nonlinearity (exponent) for the cost of rostrum length estimated from the geographical cline data for the weevil–camellia system was 2.2, suggesting nonlinearity between quadratic and cubic forms.

► We studied coevolution of a seed-predator weevil and its host, the Japanese camellia.
► We found that spatial correlation of interacting traits arises along environmental cline.
► Cost nonlinearity of the predator has a significant role for trait correlation.
► Estimated nonlinearity of predator in cost coefficient is 2.2.