Palms of the past: can morphometric phytolith analysis inform deep time evolution and palaeoecology of Arecaceae?

BACKGROUND AND AIMS

Palm fossils are often used as evidence for warm and wet palaeoenvironments, reflecting the affinities of most modern palms. However, several extant palm lineages tolerate cool and/or arid climates, making a clear understanding of the taxonomic composition of ancient palm communities important for reliable palaeoenvironmental inference. However, taxonomically identifiable palm fossils are rare and often confined to specific facies. Although the resolution of taxonomic information they provide remains unclear, phytoliths (microscopic silica bodies) provide a possible solution because of their high preservation potential under conditions where other plant fossils are scarce. We thus evaluate the taxonomic and palaeoenvironmental utility of palm phytoliths.

METHODS

We quantified phytolith morphology of 97 modern palm and other monocot species. Using this dataset, we tested the ability of five common discriminant methods to identify nine major palm clades. We then compiled a dataset of species' climate preferences and tested if they were correlated with phytolith morphology using a phylogenetic comparative approach. Finally, we reconstructed palm communities and palaeoenvironmental conditions at six fossil sites.

KEY RESULTS

Best-performing models correctly identified phytoliths to their clade of origin only 59 % of the time. Although palms were generally distinguished from non-palms, few palm clades were highly distinct, and phytolith morphology was weakly correlated with species' environmental preferences. Reconstructions at all fossil sites suggested that palm communities were dominated by Trachycarpeae and Areceae, with warm, equable climates and high, potentially seasonal rainfall. However, fossil site reconstructions had high uncertainty and often conflicted with other climate proxies.

CONCLUSIONS

While phytolith morphology provides some distinction among palm clades, caution is warranted. Unlike prior spatially restricted studies, our geographically and phylogenetically broad study indicates phytolith morphology may not reliably differentiate most palm taxa in deep time. Nevertheless, it reveals distinct clades, including some likely to be palaeoenvironmentally informative.