Fossilisation and Vegetation Patterns: Another Study of Decay and its Implications

Following on from my recent post about the decay of chordate animals, I have encountered a related paper, this time from Quaternary Research and focusing on the preservation of plants in middens (rubbish dumps) constructed by woodrats. This paper, written by Nowak et al. (2000), explores the question of how well these middens represent the vegetation surrounding them, by developing a method which calculates the probability that species that are missing from the midden are actually not present in the landscape.

To do this, Nowak et al. carry out two surveys. In one, they examine 27 woodrat middens less than 200 years old and compare them with the vegetation that currently surrounds them, and in the other, they compare middens from the same location and of the same age. From the first study, they obtain a "best case" scenario, which allows them to estimate the upper limit of the probability that midden remains accurately represent the surrounding vegetation and the lower limit of the probability that the midden is not representative of vegetation (Nowak et al. 2000). From the second experiment, which is more realistic (because the authors do not actually know the exact nature of the vegetation surrounding the midden), they can calculate the lower bound of the probability of accurate representation and the upper bound of the probability of inaccuracy - a worst case scenario.

Their findings are promising for palaeontologists interested in plant fossils. Overall, the probability of a false interpretation of vegetation pattern based on midden composition is between 7 and 11%, and for some species, it is between 0 and 6%. The grasses are the obvious exception to this, however, with inaccurate representation of grass species potentially as high as 40% (Nowak et al. 2000), although for a minority of species the results were inconclusive because of difficulties identifying fossil specimens, random fluctuations in probabilities, small sample sizes or (possibly) selection against those species by the woodrats. At the same time as these results are promising for palaeontological investigations of vegetation patterns, and potentially extensible to other organisms (like animals), however, I cannot help but feel that for most species - particularly those which are now extinct - carrying out analyses to this level of detail will be impossible and making assumptions based on studies of extant relatives potentially risky; there is no way to know how much we can generalise from results like these. That said, any moves forward in the study of palaeontological data quality are highly valuable, and, if these studies are continued, they may prove useful across the board at least in providing a ballpark probability that absence from a site actually implies absence from the surrounding area. I will be watching out for further studies of this type focusing on mammals - woodrats do not range far from their middens, so the scientists only had to evaluate 100m circles of vegetation, but I imagine surveying the surroundings of recent mammal assemblages will be much more arduous.



References

Nowak, R. (2000). Probability That a Fossil Absent from a Sample Is Also Absent from the Paleolandscape Quaternary Research, 54 (1), 144-154 DOI: 10.1006/qres.2000.2143