The use of sighting records to infer species extinctions: an evaluation of different methods

In the absence of long-term monitoring data, inferences about extinctions of species and populations are generally based on past observations about the presence of a particular species at specified places and times (sightings). Several methods have been developed to estimate the probability and timing of extinctions from records of such sightings, but they differ in their computational complexity and assumptions about the nature of the sighting record. Here we use simulations to evaluate the performance of seven methods proposed to estimate the upper confidence limit on extinction times under different extinction and sampling scenarios. Our results show that the ability of existing methods to correctly estimate the timing of extinctions varies with the type of extinction (sudden vs. gradual) and the nature of sampling effort over time. When the probability of sampling a species declines over time, many of the methods perform poorly. On the other hand, the simulation results also suggest that as long as the choice of the method is determined by the nature of the underlying sighting data, existing methods should provide reliable inferences about the timing of past extinctions.