Predicting extinction from fossil trajectories of geographical ranges in benthic marine molluscs
Correspondence: Melanie Tietje, Museum für Naturkunde, Leibniz Institute for Research on Evolution and Biodiversity at the Humboldt University Berlin, Invalidenstrasse 43, 10115 Berlin, Germany.
The geographical ranges of marine taxa wax and wane in a nearly symmetrical fashion over their durations. We use this pattern as well as the age distribution of genera to predict the time of extinction of extant genera of benthic marine molluscs.
A global occurrence database (Paleobiology Database) of fossil marine molluscs from the Mesozoic and Cenozoic eras was used for this study.
For the geographical range method, we determined the average trajectories of geographical range size for extinct genera of bivalves and gastropods. The maxima of these trajectories were used as landmarks for predicting the time of extinction for extant genera. Prediction was done by cross-multiplication, using the time elapsed until the maximum was reached and the position of the landmark. For the duration–frequency method, we calculated the time of extinction based on probabilities derived from the frequency distribution of stratigraphic ranges (durations) of all extinct genera. The accuracy and precision of predictions were tested with extinct genera and then applied to extant genera.
On average, bivalve and gastropod extinct genera reached their maximum geographical range size at 62% and 50% of their total duration. The accuracy of predictions ranged from −3.1 to 11.0 million years (Myr) for both groups, and the precision ranged from 13.6 to 19.5 Myr. Only a small subset of extant genera was suitable for prediction with the geographical range method, mainly because the geographical ranges of most extant genera are still rising on geological time-scales. The second method provided a similar prediction quality, but showed a substantial decrease in quality when predicting long-duration (> 100 Myr) genera.
The mean probability for an accurate prediction using the geographical range approach is 57% with a precision of ± 15 Myr. Prediction is possible only for genera whose peak geographical range has already elapsed (17% of extant bivalves, 16% of extant gastropods). The duration–frequency approach does not have this prerequisite and is therefore applicable to many more genera, but prediction quality declines with duration. The two methods therefore complement each other and can supplement the various efforts used to estimate extinction risk.