Quantifying the relative performance of two undetected-extinction models

Area of Science:

• Ecology
• Conservation Biology
• Biodiversity Science

Background:

• Undetected extinctions, the loss of undiscovered species, represent a significant but often overlooked component of global biodiversity loss.
• Accurate estimation of extinction rates is crucial for effective conservation strategies and understanding the true scope of biodiversity decline.

Purpose of the Study:

• To compare the performance of two models, Tedesco and SEUX, in estimating undetected extinctions using both simulated and real-world data.
• To assess the impact of detection rates varying with species abundance on the accuracy of extinction estimates.
• To evaluate the extent to which true extinction numbers exceed observed values in various taxa.

Main Methods:

• Simulated a birth-death process where less abundant species had higher extinction probabilities.
• Applied the Tedesco and SEUX models to simulated data with varying detection rates (dependent and independent of abundance).
• Validated model performance on eight real-world datasets, including Australian amphibians, Australian birds, and North American bivalves, using logistic regression to test for correlations between detection and extinction rates.

Main Results:

• Under conditions where detection rates were higher for common species, both models significantly underestimated undetected extinctions (up to 88.7%).
• When detection rates were independent of abundance, the SEUX model showed minimal bias (+3.1%), while the Tedesco model exhibited substantial underestimation (-62.3%).
• Real-world data suggested true extinctions could be 15% to 180% higher than observed; the SEUX model produced lower estimates than the Tedesco model for most datasets.
• Detection and extinction rates were correlated for three datasets, indicating potentially even higher extinction numbers.

Conclusions:

• Current models may significantly underestimate biodiversity loss when detection rates are biased towards common species.
• True extinction rates are likely substantially higher than currently documented across various taxa, underscoring the urgency of conservation efforts.
• Further refinement of extinction estimation models is needed to account for factors like detection probability and species abundance to accurately gauge biodiversity loss.