Typical variability in abundance, failure to detect initial declines and errors in classification of extinction risk

Population Abundance and the Classification of Extinction Risk

HOWARD B. WILSON, BRUCE E. KENDALL and HUGH P. POSSINGHAM

Conservation Biology, Volume 25, Issue 4, pages 747–757, August 2011

In conservation practice classifying species according to their risk of extinction is common. The authors of this paper say the reliability of such classifications rests on the accuracy of threat categorizations, but very little is known about the magnitude and types of errors that might be expected.

The process of risk classification depends on information gleaned from multiple sources. The quality of information from each source is critical to evaluating the overall status of the species.

Abundance is a direct indicator of effectiveness of measure of conservation. So counts of individuals are generally the preferred method of evaluating whether there  is a decline in abundance

Using the thresholds from criterion A of the International Union for Conservation of Nature (IUCN) Red List (critically endangered, decline in abundance of >80% over 10 years or 3 generations; endangered, decline in abundance of 50–80%; vulnerable, decline in abundance of 30–50%; least concern or near threatened, decline in abundance of 0–30%), the researchers  assessed 3 methods used to detect declines solely from estimates of abundance: use of just 2 estimates of abundance; use of linear regression on a time series of abundance; and use of state-space models on a time series of abundance.

The researchers generated simulation data from empirical estimates of the typical variability in abundance and assessed the 3 methods for classification errors.

The researchers say the estimates of the proportion of falsely detected declines for linear regression and the state-space models were low (maximum 3–14%), but 33–75% of small declines (30–50% over 15 years) were not detected. Ignoring uncertainty in estimates of abundance (with just 2 estimates of abundance) allowed more power to detect small declines (95%), but there was a high percentage (50%) of false detections. For all 3 methods, the proportion of declines estimated to be >80% was higher than the true proportion.

The researchers contend that use of abundance data to detect species at risk of extinction may either fail to detect initial declines in abundance or have a high error rate.