Per-Simper, An Innovative Method for Identifying Community Assembly Processes Within Modern, Recent, and Deep-Time Paleontological Assemblages

How biological communities are assembled is an old but lively debate, especially today, as the efficiency of conservation policies depend on our capacity to correctly identify the assembly processes at play within the species assemblages we aim to protect. The wide range of assembly modes, once seen as mutually exclusive hypothesis are now seen as part of a continuum where the opposite ends correspond to niche- and dispersal-assembly perspectives. Niche-assembled communities are closed and balanced with a stable taxonomic composition depending on deterministic processes when dispersal-assembled communities are open, in non-equilibrium state with constantly changing compositions depending on historical process and continuous dispersion between communities. The former must be protected by increasing habitat diversity and heterogeneity when the latter needing high connectivity to ensure dispersal between habitats. Protecting modern at-risk communities will require an understanding of how assembly processes have been affected by past climate change (both recent and ancient) or by the historical exploitation of ecosystems by humans. However, few methods exist that are capable of identifying assembly processes within paleontological datasets and producing time series at multiple scales due to missing information in the vast majority of the fossil record (e.g., abundance, environmental information, highly-resolved phylogeny) or due to limited precision of compatible methods (i.e. random vs. non-random patterns). The new PER-SIMPER method (and its associated DNCImper R package) can be used to identify and quantify the respective roles of niche and dispersal processes from the distribution of occurrences of extinct and extant species. Based on three modes of permutation of the occurrence matrix, I will illustrate PER-SIMPER with deep-time (e.g., trilobites), ancient (e.g. Cenozoic mammals) and modern datasets (e.g. small mountain mammals, host-flea assemblages).