The measurement of taxonomic evolution: preservational consequences

The accuracy of numerical summaries of data from the fossil record has been hotly contested in recent years. In this paper I present a computer simulation which mimics the process of preservation and the resultant loss of data in survivorship records. The simulation accepts as its input data a hypothetical “original” record and a time dependent model of preservational loss. These parameters are used to generate a “preserved” record, and the “original” and “preserved” data sets are then compared. Eleven hypothetical original records having different patterns of diversity and turnover were “preserved” in this manner. Indices of diversity, origination, extinction, turnover, and longevity were evaluated for each of five different models of preservational bias. All of the indices behaved in a single characteristic fashion. Taxonomic data that contained large fluctuations (of the order 100%) were preserved accurately. Similarly the large scale changes in preserved records matched original distributions. Records with small fluctuations (30%) were variably preserved and when such fluctuations were present in the preserved record they did not always correlate with events in the original record. Longevities were more accurately preserved than were other forms of taxonomic data. The results are believed to reflect the levels of accuracy obtainable from generic and familial data and suggest ways in which data sets warranting more detailed study may be singled out.