Variation in signal-preference genetic correlations in Enchenopa treehoppers (Hemiptera: Membracidae)

Abstract

Fisherian selection is a within-population process that promotes signal-preference coevolution and speciation due to signal-preference genetic correlations. The importance of the contribution of Fisherian selection to speciation depends in part on the answer to two outstanding questions: What explains differences in the strength of signal-preference genetic correlations? And, how does the magnitude of within-species signal-preference covariation compare to species differences in signals and preferences? To address these questions, we tested for signal-preference genetic correlations in two members of the Enchenopa binotata complex, a clade of plant-feeding insects wherein speciation involves the colonization of novel host plants and signal-preference divergence. We used a full-sibling, split-family rearing experiment to estimate genetic correlations and to analyze the underlying patterns of variation in signals and preferences. Genetic correlations were weak or zero, but exploration of the underlying patterns of variation in signals and preferences revealed some full-sib families that varied by as much as 50% of the distance between similar species in the E. binotata complex. This result was stronger in the species that showed greater amounts of genetic variation in signals and preferences. We argue that some forms of weak signal-preference genetic correlation may have important evolutionary consequences.

Keywords: Assortative mating; Fisher’s runaway; male–female genetic covariance; preference function; vibrational communication.

Figure 1

Illustration of potential forms of genetic variation in the signal–preference relationship. In each panel, each line indicates signal and peak preference values for a genotype. (A) Strong signal–preference genetic correlation: among-genotype differences in the y-axis intercept, with all genotypes showing strong signal–preference correspondence (parallel lines), so that r = 1. (B) r is zero or very weak due to lack of overall genetic variation in signals and preferences. (C) r = 1 or nearly so. (D) 0 < r < 1 due to lower amounts of genetic variation in the preference, which also results in genotype-level signal–preference mismatch. (E) 0 < r < 1 due to genotype-level signal–preference mismatch. (F) r ≪ 1 due to genotype-level signal–preference mismatch, but there are still some genotypes (those with phenotypes at the extremes of the range) that remain distinct from others in their signal–preference relationship. (G) r ≪ 1 due to lack of genetic variation in the preference, which also results in genotype-level signal–preference. (H) r ≪ 0 due to strong genotype-level signal–preference mismatch. NB: Negative signal–preference genetic correlations have been documented (Bakker and Pomiankowski ; Greenfield et al. 2014). In cases of population-level correspondence between mean signal and preference values, stabilizing sexual selection due to mate choice is predicted (A, B, E–H). By contrast, in cases of population-level signal–preference mismatch, directional sexual selection is predicted (C, D).

Figure 2

Example of a female mate preference function for male signal frequency for one female individual from Enchenopa binotata “Ptelea.” The peak preference (arrow) is derived from the cubic spline (curved line) that fits the raw data (data points), and corresponds to the signal frequency to which this female had the strongest response.

Figure 3

Posterior distributions of heritability estimates for male signals and female preference of Enchenopa binotata “Viburnum” (A and B) and E. binotata “Ptelea” (C and D). Each distribution is a density curve of heritability estimates from 1000 iterations of an animal model run with 1,000,000 iterations with a burn-in of 500,000 iterations, and sampled every 500. All panels plotted on the same scale except (C).

Figure 4

The density of genetic correlation estimates sampled 1000 times across all iterations in the animal model. Posterior distributions of estimates of genetic correlations between male signals and female preference of Enchenopa binotata “Viburnum” (A) and E. binotata “Ptelea” (B). Each distribution is a density curve of heritability estimates from 1000 iterations of an animal model run with 1,000,000 iterations with a burn-in of 500,000 iterations, and sampled every 500.

Figure 5

Variation among full-sib families in male signal frequency and the peak of female preferences for signal frequency, in two members of the Enchenopa binotata complex. Each line shows one family’s mean values for signal frequency and female peak preference. The inclination of the lines indicates the degree of family-level signal–preference correspondence (perfect correspondence = horizontal line). Arrows indicate overall means pooling across families. The range of the y-axes indicates the overall range of phenotypic variation. Results differed between our two study species. (A) For E. binotata “Ptelea,” the signal–preference genetic correlation was negative (note line crossovers) but with CIs overlapping zero (Table2). However, some families (those with near parallel horizontal lines) remained distinct from some others in their signal–preference relationship (text; Table3). Note the population-level signal–preference correspondence (arrows). (B) For Enchenopa binotata “Viburnum,” lower overall genetic variation and fewer crossovers resulted in a positive signal–preference genetic correlation but with CIs overlapping zero (Table2). Note the population-level signal–preference mismatch (arrows).

Figure 6

Range of variation in signals and preferences among full-sib families in our two study species, contrasted with the magnitude of species differences in signals and preferences across the Enchenopa binotata complex. Red symbols and line: mean values for signal frequency and peak preferences for four sympatric species at the collecting site of one of our study species (from Rodríguez et al. with permission). Although over 11 species are known to exist in the E. binotata complex, these four span the known range of variation in signal frequency in the complex. The line indicates a one-to-one signal–preference relationship. Black symbols: family means for signals and peak preferences for our two study species. Note that in this figure, we distinguish two treehopper species that live on different Viburnum hosts in our two study sites (E. binotata “Viburnum rufidulum,” which is sympatric with E. binotata “Ptelea” in Missouri; and E. binotata “Viburnum lentago” in Wisconsin).