Phytolith Analysis In Cenozoic Paleoecology and the Integration with Palynology: Spatial, Temporal, and Taxonomic Scales In Vegetation Inference Using Plant Microfossils

Although considered standard in Quaternary paleoecology and archaeobotany, plant silica (phytolith) assemblage analysis gained widespread use as a paleoecological tool only in the 1970s. Its application to pre-Quaternary is even more recent. Nevertheless, a body of work on modern and fossil phytoliths has shown the vast potential of phytolith analysis for reconstructing Cenozoic vegetation types, aspects of climate, and taxonomic evolution within the grass family, Poaceae. Because plant silica can preserve in highly oxidized sediment where pollen and macro-fossils are rarely found, the use of phytoliths has also substantially increased the geographic areas and number of microhabitats from which paleoecological data can be collected.

Here I review the status of phytolith assemblage analysis as a tool for deep-time paleoecology. In particular, I focus on recent efforts attempting to compare palynomorphs and phytoliths at sites where both types of data are preserved, such as the early-middle Miocene flora from Shanwang, China. This work shows that analysis of pollen/spores and phytoliths often result in roughly comparable vegetation interpretations, but there are inherent differences in the information that the two records provide. These differences relate to varying production and preservation of pollen and phytoliths from ecologically important plants (e.g., conifers vs. grasses) and the diverse spatial scales recorded by palynomorphs and phytoliths, respectively. Another important discrepancy relates to the systematic specificity of the two types of data. For example, whereas the woody components of a flora produce diagnostic pollen types, their phytoliths are often less taxonomically distinct. In contrast, Poaceae pollen are ubiquitous, but grass phytoliths provide good resolution among Poaceae subclades. Thus, although future work should seek to improve the correlation between vegetation inferences based on pollen/spores and phytoliths, these sources of data are presently best regarded as complementary, rather than redundant.