Ecological communities are assembled in a spatial-temporal continuum. However, we still have a poor understanding of the relative importance of different mechanisms structuring community composition (i.e., beta-diversity) in space and time. In this study, we start by introducing a conceptual model that capitalizes upon the core-occasional species concept to predict that the assembly process in tropical mountains is driven by the deterministic turnover of core species in space via habitat sorting, but the turnover of occasional species through time via stochastic events of colonization and local extinctions. We then propose a general analytical framework that allows assessing these predictions by partitioning the total variance of a species-by-site-by-time matrix (i.e., total beta-diversity) among its purely spatial (variation in space independent of time), purely temporal (variation in time independent of space), and spatiotemporal (i.e., variation across different sites across different moments in time) components. Through simulation models, we provided theoretical support that the proposed analytical framework is suitable to test the predictions derived from our conceptual model. We then used this framework to identify general patterns and quantify the relative importance of processes underlying the spatial and temporal organization of ten distinct insect metacommunities along a tropical elevational gradient. As predicted, we found that, across taxa, spatial beta-diversity was mainly explained by environmental variation alone: a pattern that indicates the spatial turnover of core species. In contrast, temporal beta-diversity could not be distinguished from the expectation of null models where communities are simply represented by random draws from species pools: a pattern that indicates a temporal turnover of occasional species within communities. Taken together, our findings illustrate how our conceptual model and quantitative framework can articulate a better understanding of community assembly in space and time.

Sampling was carried out in the Serra dos Orgãos National Park (22°27′49″S; 43°01′50″W), Rio de Janeiro, Brazil. The park is one of the most preserved remaining fragments of the Atlantic Rainforest (Castro 2018), one of the hottest hotspots of global biodiversity (Myers et al. 2000). Along a complete elevational transect that ranges from 100 to 2130 meters above sea level (MASL), two flight interception Malaise traps were placed every interval of approximately 150 MASL, summing up to 15 sampled elevations. The collecting bottle of each Malaise trap was replaced every month from December 2014 to November 2015, summing up to 360 samples (2 traps x 15 elevations x 12 months). A data logger was placed next to each pair of traps to record hourly variation in climatic conditions (air temperature and relative humidity) over the entire sampling period. Insects captured in each sample were sorted at the species level by our team and external collaborators (see acknowledgments). We focused our study on ten relatively well-known families and subfamilies of beetles (Lampyridae, Cerambycidae, Carabidae, Phengodidae, Eumolpinae, Anthribidae) and wasps (Ichneumonidae: Metopiinae, Pimplinae; Braconidae: Mesostoinae, Meteorini/Meteorus ).

We considered only the adults of the focal taxonomic groups in the final species-by-time-by-site matrices (i.e., a total of 6996 individuals from 549 species across taxa). As such, this dataset considers only the months with the highest activity of adults across all taxonomic groups (i.e., from December to February and June to August).