Climate change, human overkill, and the extinction of megafauna: a macroecological approach based on pattern-oriented modelling

Abstract

Background: The debate on Late Quaternary extinctions (LQE) has long been centred on the contrast between climatic and human-related explanations, with no consensus being reached. Here we propose a macroecological approach to study LQE that combines niche and demographic models to determine the ecological mechanisms behind the megafauna extinction. Question: Is the overkill hypothesis plausible and realistic when assessing the ecological mechanisms behind LQE? What is the range of population parameters for both human and now-extinct animals that is consistent with the actual extinction pattern? Features of model: Ecological niche models (ENMs) are used to assess climate impact on population viability, whereas density-dependent population models with reciprocal feedback between humans and their prey are used to simulate human hunting pressure. The feasibility of predicted extinction scenarios and explored parameter space is ordered based on a patternoriented modelling (POM) approach. We illustrate the usefulness of our framework using the extinct South American ground sloth Megatherium as an example. Methods: We built ENMs by using fossil occurrences records and palaeoclimatic simulations from the last glacial cycle and projected their predictions to the Holocene. Population modelling was based on 4000 random samples of ENM ensembles, from which prey density was estimated using a Gaussian central–peripheral abundance model. For each ensemble, a plausible range of demographic parameters (e.g. growth rate, carrying capacity, mortality of human population, meat consumption) was set across 100 random repetitions, giving 400,000 models simulating Megatherium’s extinction dynamics. Results: The macroecological approach highlighted many plausible mechanistic extinction scenarios capable of reproducing a wide range of hypotheses about the LQE. Most models (51%) simulating a vigorous human population with unrealistic growth rates (rh > 1.3% per year) and intense over-exploitation of prey (individual meat-consumption-rate – CI > 100 g per day from one prey only) produced scenarios of rapid extinction as predicted by the overkill hypothesis. However, such overkill scenarios unrealistically predicted earlier extinction times than that observed for Megatherium. Moreover, the high human population growth required to simulate overkill scenarios was attained only recently after the industrial revolution, specifically in the mid-1900s, and it is therefore not applicable for Pleistocene hunter-gatherers. Conversely, although less frequent across simulations, models that correctly predicted the observed extinction time (16%) revealed plausible and empirically acceptable demographic parameters, encompassing low growth rates (rh < 0.70%) and mortality for the human population (mo < 0.60%), combined with moderate values of individual meat consumption (CI > 70 g) and geographical range collapse in Megatherium, which is produced by climate change. Conclusions: These findings, based on POM reasoning, highlight that unique mechanisms such as the overkill explanation for LQE, although feasible from model simulations, only occur with implausible parameter combinations and predict unrealistic extinction dynamics