Deviation from the brownian motion expectation implies episodic adaptive divergences in traits of lithocarpus species in Taiwan

Adaptive trait divergence is usually episodic rather than universal in a phylogeny. To determine whether trait divergence reaches an extreme level (i.e. deviation from neutrality), the evolutionary rate of the species and intraspecific variation must be taken into account. Accordingly, we attempt to explore the trend of trait divergence over divergence times conditioned on the genetic distance. The Brownian motion (BM) model, a commonly used random-walk process for describing the neutral evolution of traits, is used to simulate the distribution of trait divergence under neutrality. The observed trait divergences are then compared to the BM expectations to detect outliers, which are considered to be selected. We assessed the ability of this method to characterize trait divergence under selection among 14 Lithocarpus species in Taiwan. The results are consistent with the previous inference of phylogenetic constraint based on tests of the phylogenetic signal, but further signals of adaptive and conserved trait divergences are detected between some lineages derived from particular nodes (common ancestors), indicating differential investments in growth rate and chemical defense among some endemic species in Taiwan. These results show that the adaptive divergence of environment-related traits is transient, episodic, and punctuated. Under strong selection pressure, the trait divergence after standard deviation correction will be more significant because the trait variance within the population decreases. However, it tends to be an overestimation of conservative evolution in older common ancestors due to the larger standard deviation produced from diverse descendants. Nevertheless, this study still provides a simple approach to detect adaptive divergence in a phylogenetic framework.