Using biodiversity deconstruction to disentangle assembly and diversity dynamics of understorey plants along post‐fire succession in boreal forest

Aim  The study aims to decipher the co‐occurrence of understorey plant assemblages and, accordingly, to identify a set of species groups (diversity deconstruction) to better understand the multiple causal processes underlying post‐fire succession and diversity patterns in boreal forest. Location  North‐eastern Canadian boreal forest (49°07′–51°44′ N; 70°13′–65°15′ W). Methods  Data on understorey plant communities and habitat factors were collected from 1097 plots. Species co‐occurrence was analysed using null model analysis. We derive species groups (i.e. biodiversity deconstruction) using the strength of pairwise species co‐occurrences after accounting for random expectation under a null model and cluster analyses. We examine the influence of a set of spatiotemporal environmental variables (overstorey composition, time‐since‐fire, spatial location and topography) on richness of species groups using Bayesian model averaging, and their relative influence through hierarchical partitioning of variance. Results  Understorey plant assemblages were highly structured, with co‐occurrence‐based classification providing species groups that were coherently aggregated within, but variably segregated between, species groups. Group richness models indicate both common and distinct responses to factors affecting plant succession. For example, Group 2 (e.g. Rhododendron groenlandicum and Cladina rangiferina ) showed concurrent contrasting responses to overstorey composition and was strongly segregated from Groups 3 (e.g. Clintonia borealis and Maianthenum canadense ) and 4 (e.g. Epilobium angustifolium and Alnus rugosa ). Groups 3 and 4 showed partial similarity, but they differed in their response to time‐since‐fire, drainage and latitude, which were more important for Group 1 (e.g. Ptilium crista‐castrensis and Empetrum nigrum ). A single successional model based on total richness masked crucial group‐level relationships with factors that we examined, such as latitude. Main conclusions  By demonstrating the co‐occurrence structure and linking to causal factors, the results from this study characterize both common and distinct responses of understorey plants to biophysical attributes of sites, and potential interspecific interactions, behind non‐random assemblage structure during post‐fire succession. A biodiversity deconstruction approach could offer a concise and explicit framework to gain a better understanding of the complex assembly of ecological communities during succession.