Reconstructing the extinction of the Carolina Parakeet: historical data reveal that two distinct human activities drove two separate subspecies' declines

Background/Question/Methods

Species distribution modeling has been used to model the ranges of species from both contemporary and paleontological data, yet little attention has been paid to the middle ground of recently-extinct species. Using species distribution models (SDMs) to reconstruct the pattern of a species’ decline may reveal important details about spatial patterns that may shed light on the cause of the extinction. Here, we model the decline of the Carolina Parakeet (Conuropsis carolinensis) using a database of georeferenced sightings spanning from the early 1500s to the mid 1930s. We apply maximum entropy, or MaxEnt, to model to the distribution of the species using data subsetted for regularly-spaced time intervals, in order to reconstruct the decline of its habitat occupancy. Moreover, we present both an application of SDMs to determine the most likely extinction date of a species and a method for testing the likelihood of questionable late sightings, using a sampling-based method that reconstructs likely extinction dates from the rate of decline in sightings. We use this method to address the credibility of the controversial late sightings of the species between 1911 (the presumed extinction date) and 1934 (the last credible sighting).

Results/Conclusions

Results suggest that the two subspecies of Carolina Parakeet underwent independent declines with different causes. The Eastern subspecies appears to have shifted its range southward in association with the decline of bottomland hardwood forest, while the range of the Western subspecies declined uniformly, possibly due to hunting pressure (the parakeets were widely viewed as agricultural pests). Moreover, we found that the extinction dates of the subspecies differed: while the Western subspecies was entirely extinct by 1903, the Eastern went extinct much later. Although the most likely scenario was a 1915 extinction, accepting at least one of the post-1911 sightings as credible moved the extinction up to 1920, almost a decade later than previously supposed. We conclude that SDMs have the potential not only to reconstruct the decline of extinct species, but can also reveal valuable information about their biology. Given that more than half of all parrot species are currently declining globally and 42% are IUCN listed, determining the cause of recent parrot extinctions may help focus conservation efforts, making them more effective. We conclude by presenting a set of potential applications of these methods to other “unsolved” recent extinctions that could shed light on species currently near extinction.