Sprecher
Beschreibung
Estimating plant diversity through geological time is confounded by several biases. These can arise from the structure of databases as well as from the nature of plant fossils. Occurrence-based databases such as the Paleobiology Database (PBDB) record successful fossil sampling, and therefore cannot distinguish between true absence and false fossil absence caused by missing rock record. In addition, the database’s implicit proxy for rock availability, stratigraphic formations, is partly informed by subdivision of strata using faunal and floral turnover (i.e. biozones), causing bias towards higher formation number in fossiliferous vs non-fossiliferous strata. Plant fossils present an additional challenge because they are commonly represented by isolated organs that are non-uniformly preserved through time and space, and vary in informativeness and taxonomical resolution.
To investigate the extent of bias affecting our understanding of the plant fossil record, we used the Macrostrat database, which quantifies rock availability using lithological area and unit thickness (i.e. rock volume) independently of fossil occurrence data, thereby reducing occurrence and formation biases. We expanded Macrostrat’s South American coverage and constructed structural equation models integrating multiple rock-availability metrics derived from both databases with two curated, organ-specific diversity estimates through the Phanerozoic from the PBDB, namely leaves (1) and pollen & spores (2).
Our analyses reveal that sampling proxies exert heterogeneous effects on different organ records, with the rock volume showing a substantially stronger influence on leaf diversity than on pollen & spore diversity. We discuss the resulting implications for estimates of plant diversity through time and space.