RECONSTRUCTING A MODERN TROPICAL RAINFOREST USING SOIL PHYTOLITH ASSEMBLAGES: IMPLICATIONS FOR PHYTOLITH STUDIES IN PALEOECOLOGY

Phytolith analysis has traditionally been used in archaeobotany and Quaternary paleoecology, but during the last 15 years it has also emerged as a powerful tool for deep-time paleoecology and evolutionary biology studies. As the use of phytoliths in paleontology is expanding, it is crucial to refine this tool by establishing more rigorous, precise, and reliable protocols allowing application and comparisons among diverse studies.

In phytolith assemblage analysis, different phytolith morphotypes are typically classified into plant functional groups, and their proportions are used to infer habitat structure (ratio of forest:open habitat indicators). However, each of these broad categories includes tens of different phytolith morphotypes, and identical results (e.g., a phytolith assemblage dominated by forest indicators indicating a closed habitat) can be obtained through multiple morphotype combinations. Our work tries to take advantage of these differences to gain a more detailed understanding of how well phytoliths reflect spatial heterogeneity in forest type and, more broadly, habitat structure, across different modern Neotropical ecosystems (dry forest, rainforest, and savanna). Here, we present results from a Costa Rican rainforest, where we collected and analyzed phytolith soil assemblages from 15 quadrants along a 1.5 km transect. In each quadrant, we identified and collected all plant species present, and measured their dbh (diameter at breast height); we are also processing modern plant material from all species present in the quadrants to identify diagnostic phytoliths.

Results show that the phytolith signal is extremely local in closed habitats such as rainforests. A comparison of the phytolith assemblages and plant species composition suggests that variation in plant species abundance and biomass/age (from dbh) between quadrants influence the abundance of diagnostic morphotypes in phytolith assemblages, which therefore should provide additional information about forest structure. We aim to place these data in a multidimensional space where we can describe and predict the phytolith composition of different sites within a forest characterized by different plant species, and use comparison with phytolith assemblages from the modern plant material to validate our predictions.