Phylogenetic patterns are not proxies of community assembly mechanisms (they are far better)

International audience1. The subdiscipline of ‘community phylogenetics’ is rapidly growing and influencing thinking regarding community assembly. In particular, phylogenetic dispersion of co-occurring species within a community is commonly used as a proxy to identify which community assembly processes may have structured a particular community: phylogenetic clustering as a proxy for abiotic assembly, that is habitat filtering, and phylogenetic overdispersion as a proxy for biotic assembly, notably competition.2. We challenge this approach by highlighting (typically) implicit assumptions that are, in reality, only weakly supported, including (i) phylogenetic dispersion reflects trait dispersion; (ii) a given ecological function can be performed only by a single trait state or combination of trait states; (iii)trait similarity causes enhanced competition; (iv) competition causes species exclusion; (v) communities are at equilibrium with processes of assembly having been completed; (vi) assembly through habitat filtering decreases in importance if assembly through competition increases, such that the relative balance of the two can be thus quantified by a single parameter; and (vii) observed phylogenetic dispersion is driven predominantly by local and present-day processes.3. Moreover, technical sophistication of the phylogenetic-patterns-as-proxy approach trades off against sophistication in alternative, potentially more pertinent approaches to directly observe or manipulate assembly processes.4. Despite concerns about using phylogenetic dispersion as a proxy for community assembly processes, we suggest there are underappreciated benefits of quantifying the phylogenetic structure of communities, including (i) understanding how coexistence leads to the macroevolutionary diversification of habitat lineage-pools (i.e. phylogenetic-patterns-as-result approach); and (ii) understanding the macroevolutionary contingency of habitat lineage-pools and how it affects present-day species coexistence in local communities (i.e. phylogeneticpatterns- as-cause approach).5. We conclude that phylogenetic patterns may be little useful as proxy of community assembly. However, such patterns can prove useful to identify and test novel hypotheses on (i) how local coexistence may control macroevolution of the habitat lineage-pool, for example through competition among close relatives triggering displacement and diversification of characters, and (ii) how macroevolution within the habitat lineage-pool may control local coexistence of related species, for example through origin of close relatives that can potentially enter in competition