Performance of generation time approximations for extinction risk assessments.
Abstract
Generation time is a fundamental component of extinction risk assessments for the International Union for Conservation of Nature's Red List of Threatened Species. The calculation of generation time requires age-specific data on survival and fecundity rates and knowledge of population growth rates. These data are generally lacking for threatened species, so approximations including only partial demographic information need to be used. This leads to potential errors in generation time estimates. To quantify the magnitude of potential errors in generation time estimates, we compared seven approximations with exact generation time measures, calculated either from complete life tables available for 58 mammalian species or from simulated data. We tested the influence of these errors on conservation assessments conducting mock assessments for ten species. We also tested the commonly used prediction of generation time based on the allometric relationship with body mass using phylogenetic generalized least squares. Root mean square errors were largest in measures assuming constant fecundity with age, some of which are currently used in Red List assessments. We found that the measure of generation time that only ignores population growth rates performed relatively well, but tended to underestimate the generation time for decreasing populations, and over-estimate it for increasing populations. In the mock assessment, we found that we underestimated the threat level in 10% to 90% of the species depending on the generation time approximation we used. The predictive metric of generation time based on body mass is inaccurate. We propose an alternative predictive metric based on body mass, age at first reproduction, and reproductive life span. Synthesis and applications. Our results demonstrate potential errors that occur when estimating generation time in the absence of key demographic information. We offer practical recommendations for extinction risk assessments including more rigorous mathematical formulations of generation time, such as the measure of generation time that includes population growth rates and the appropriate age-specific vital rates. Furthermore, we recommend alerting risk assessors of the uncertainties in proxy measures, such as the underestimation of generation time resulting from the assumption of constant fecundity in future Red List assessments.