The eco-evolutionary consequences of phenotypic variance depend on the relationship between the mean phenotype and the environmental optimum
Ecologists are increasingly interested in the role of intraspecific variation in contemporary ecological processes. In particular, recent work has investigated the influence of variation per se, by manipulating levels of genotypic or phenotypic diversity within a local population. Many of these studies show positive effects of trait or genetic variation on a number of processes, including colonization success and disease resistance. However, most studies documenting positive effects of variation per se only find them when populations express ‘suboptimal’ performance in the local environment. When the environment is relatively neutral, the positive effects of variation tend to disappear, even in the same study. We generalize this pattern, hypothesizing that context-dependent effects of phenotypic variance arise from the relationship between a population’s mean phenotype and the local environmental optimum. Using the host Daphnia pulex and its parasite Metschnikowia bicuspidata, we factorially manipulated the mean and variance of fitness-related traits in replicated host polycultures with identical genotypic richness. We used these populations, and knowledge of fitness trade-offs in environments with and without parasites, to ask whether the ecological and evolutionary consequences of increasing phenotypic variance depend on the relationship between a population’s trait mean and the local environmental optimum.
When populations expressed suboptimal trait means, increasing trait variance benefitted host populations, augmenting host abundance by ~14% when parasites were absent, or reducing the size of disease epidemics by ~35% when parasites were present. On the other hand, increasing trait variance when trait means were near-optimal had negative effects, reducing host population size by ~10% or doubling the size of disease epidemics. Importantly, the consequences of altering phenotypic variance were consistent across environments, even though the optimal trait value differed. This indicates that it is the relationship between a mean phenotype and the local optimum that matters, not the trait mean itself. Furthermore, the two host populations suffering the highest disease burden showed significant increases in mean individual resistance throughout the experiment. But the reason for the high disease burden differed in each case; in one, epidemics were large because of low variance and a suboptimal trait mean. In the second, epidemics were large even with a near-optimal trait mean, due to high phenotypic variance. Thus, our hypothesized mechanism not only helps to synthesize previous results, but also provides a framework linking ecological performance, selection, and rapid evolutionary change.