Community-level traits are a useful starting point linking environmental properties and traits of organisms even when we cannot easily sample individuals in nature, as is the case for microorganisms. Here, I present a simplified community assembly scenario showing how traits × environment correlations emerge at the community level under an environmental filtering / species sorting paradigm. Assume an environmental gradient consisting of 5 distinct environmental patches with contrasted conditions for an environmental parameter x(i.e., the red-blue scale in the large squares in the figure). Also assume a pool of 5 species with given values for a trait y , exemplified in the figure using a similar red-blue color scale, such that low values for trait y translates into high reproductive success in environments with a low value for parameter x . We start (t = 1, first row of panels in the figure) by randomly drawing 100 individuals from the species pool and placing them in each local community. Then, we simulate the extinction of a given number of individuals (here, 5 per time step) and replace them based on the reproductive success of the remaining species in the local community, as opposed to a zero-sum ecological drift assumed by a neutral model. This is rather in line with a species sorting paradigm where local community dynamics are assumed to reach an equilibrium faster than the environmental properties change, and with little/no impact of dispersal from neighbouring communities (i.e., source-sink dynamics) (Leibold et al., 2004). If, at each time step, we compute a community-weighted mean (CWM) trait y for each local community, and correlate CWM and local environmental property x , we see that community-levels traits rapidly mirror local environmental conditions (right panel).