Historical and phylogenetic constraints on the incidence of entire leaf margins: insights from a new South American model

Aim The relationship between the proportion of species with an entire leaf margin (pE) and mean annual temperature (MAT) is one of the most powerful tools for estimating palaeotemperatures. However, phylogenetic and phytogeographic constraints on this relationship have remained unexplored. Here we investigate the pE–MAT relationship for modern floristic assemblages from southern South American forests, assess its conformity to other models and test for the existence of historical constraints on pE–MAT models. Location South America. Methods We used samples from 30 sites located in Chile between 32° and 44° S to test for a pE–MAT relationship and compared it with four regional models. We assessed the reliability of these models for predicting MAT from instrumental records in eight modern temperate‐forest localities in Chile. Additionally, palaeotemperatures for Cenozoic fossil floras were estimated. To assess historical constraints in pE, we measured the phylogenetic signal in leaf margin type and the association between leaf margin and phytogeographic affiliation, defined by the distribution of genera. Results We found a significant pE–MAT relationship for Chilean forest species that differed from Australia and Northern Hemisphere models, but not from tropical South America (TSA). Temperatures for southern South American localities predicted from the new regional model – combining Chilean and TSA datasets – were more accurate than those from previous models. We also showed that leaf margin type has a strong phylogenetic signal, which was further confirmed by the highly significant effect of phytogeographic element on leaf margin type. Main conclusions Differences between the Chilean and other regional models are explained by historical legacy, as Chilean leaf margin types are strongly affected by phylogenetic closeness and phytogeographic elements. We highlight that leaf margin analyses should be conducted within the context of a flora with a shared history. Thus, we propose a new model for South America to estimate palaeotemperatures for regional fossil floras.