A single sample from living vegetation provides a measure of the diversity and abundance of plant species in the sample area at the time of sampling. Due to spatial and temporal averaging, a single paleobotanical sample provides only an estimate of the diversity and abundance of species that lived within the source area during the deposition of sediments. However, to estimate the patchiness of the vegetation across both modern and ancient landscapes, multiple samples must be used to identify variation in the distribution of species. Here I present a method to identify whether observed variation among paleobotanical samples is an artifact of random sampling or whether that variation could represent patchy vegetation or different taphonomic regimes. Three replicates were collected for each of four samples across a mudstone lens preserving a floodplain pond at the Allon locality (Crawford Co., GA). The lens is about 16m across with a maximum thickness of 0.8m. Samples were processed by sieving to recover mesofossils (charcoalified and lignified plant debris) and all identifiable specimens (approximately 11,000) were assigned to one of 46 morphotypes and counted. Samples drawn from the same original source population should differ only due to the effects of random sampling. The method compares variation within a sample (measured as the distribution of dissimilarities between replicates) to variation among samples after correcting for differences in sample size. If samples are more dissimilar to each other than expected from dissimilarity among replicates, then the samples reflect the patchy distribution of fossils across the outcrop. The differences among samples are therefore due to patchiness in the original vegetation or to a non-random bias in the preservation of fossils. The possibility of preservation biases can be addressed using additional (e.g., sedimentological) evidence.

Key words: heterogeneity, methods, paleobotany, paleoecology, patchiness, sampling