The phylogenetic structure of communities (PSC) reveals how evolutionary history affects community assembly processes. However, there are important knowledge gaps on PSC patterns for annual communities and there is a need for studies along environmental gradients in dry ecosystems where several processes shape PSC. Here, we investigated the PSC of annual plants along an aridity gradient in Israel, including eight years, two spatial scales, the effects of shrubs on understory, and the phylogenetic signal of important traits. Increasing drought stress led to overdispersed PSC at the drier end of the gradient, indicating that species were less related than expected by chance. This was supported at a smaller spatial scale, where within the drier sites, communities in open – more arid – habitats were more overdispersed than those under nurse shrubs. Interestingly, some key traits related to drought resistance were not conserved in the phylogeny. Together, our findings suggested that while habitat filtering selected for drought resistance strategies, these strategies evolved independently along multiple contrasting evolutionary lineages. Our comprehensive PSC study provides strong evidence for the interacting effects of habitat filtering and plant–plant interactions, particularly highlighting that the conservative evolution of traits should not be assumed in future interpretations of PSC patterns.

The phylogenetic structure of communities (PSC) reveals how evolutionary history affects community assembly processes. However, there are important knowledge gaps on PSC patterns for annual communities and there is a need for studies along environmental gradients in dry ecosystems where several processes shape PSC. Here, we investigated the PSC of annual plants along an aridity gradient in Israel, including eight years, two spatial scales, the effects of shrubs on understory, and the phylogenetic signal of important traits. Increasing drought stress led to overdispersed PSC at the drier end of the gradient, indicating that species were less related than expected by chance. This was supported at a smaller spatial scale, where within the drier sites, communities in open – more arid – habitats were more overdispersed than those under nurse shrubs. Interestingly, some key traits related to drought resistance were not conserved in the phylogeny. Together, our findings suggested that while habitat filtering selected for drought resistance strategies, these strategies evolved independently along multiple contrasting evolutionary lineages. Our comprehensive PSC study provides strong evidence for the interacting effects of habitat filtering and plant–plant interactions, particularly highlighting that the conservative evolution of traits should not be assumed in future interpretations of PSC patterns.