The Carolina parakeet (Conuropsis carolinesis) was the only endemic North American parrot species north of the Rio Grande. The species likely went extinct in the first half of the 20th century (Snyder and Russell 2002, Elphick et al. 2010), though the proximate and ultimate causes of their extinction remain unknown. Despite being a charismatic and colorful bird, the Carolina parakeet has not received as much attention as other recently extinct North American birds, such as the passenger pigeon (Ectopistes migratorius, e.g. Stanton 2014), the ivory-billed woodpecker (Campephilus principalis, e.g. Gotelli et al. 2011), or Bachman's warbler (Vermivora bachmanii, e.g. Carlson et al. 2018). Aside from the efforts of two individual researchers, Daniel McKinley (e.g. McKinley 1965, McKinley and Hardy 1985) and Noel Snyder (Snyder and Russell 2002, Snyder 2004), there has been relatively little research on the Carolina parakeet since the last captive individual died in the Cincinnati Zoo in 1918. However, with the progression of increasingly sophisticated analytical approaches, such as genetic sequencing and species distribution modeling, researchers are starting to provide new insights into Carolina parakeets' evolutionary relationships within the parrot phylogeny (Kirchman et al. 2012) and the range size and migratory behavior (Burgio et al. 2017). To facilitate more research on this understudied, extinct species, we present the most detailed dataset yet compiled of georeferenced observations and specimens of the Carolina parakeet, ranging from 1564 to 1944.

It was only through Daniel McKinley's efforts were we able to compile this dataset. A vast majority of observations were detailed, with primary sources, in McKinley's numerous publications from 1959 to 1985 (e.g. McKinley 1965, McKinley and Hardy 1985). Within our dataset, we include observations gleaned from a wide variety of sources, including hunters' logs, the correspondence of well-known historical figures including Thomas Jefferson (Jefferson 1894) and the explorers Lewis and Clark (Lewis and Clark 1904), and even letters to the editor of local newspapers. However, some sightings may be fraudulent, as collectors from the early 20th century would often pay high prices for the eggs of a nearly extinct bird. As an example, included in the dataset is the date and location of one such potentially fradulant clutch of eggs, collected by a prolific egg collector and dealer, Charles Doe, who was notorious for passing off eggs of one species for those of other, less rare species (Snyder 2004, Snyder and Russell 2002). Fraud surrounding the Carolina parakeet was not uncommon, as evidenced by a video of living parrots painted to look like Carolina parakeets mixed in with taxidermied specimens that began circulating in the 1950s (Snyder et al. 2010). While we include these "sightings" in our dataset, they are flagged as likely invalid.

Adding to the lore surrounding Carolina parakeet observations, as late as the 1940s, ornithologist Oscar Bayard claimed that he knew of a remnant population of Carolina parakeets in rural Florida, but he refused to tell anyone where the birds were for fear that people would trap or kill the birds (Snyder 2004). In an effort to try to find information of this long lost population of Carolina parakeets, Snyder even visited his family to see if any journals were left in his affects after he died, yet he found nothing (Snyder 2004).

While data about Bayard's "lost population" has disappeared, some data may still be buried in obscure texts, waiting to be found. For instance, McKinley, who spent over 25 years writing about Carolina parakeets focusing primarily on the sightings recorded in each state, referenced two manuscripts in his writing that were never published, including a detailed account of sightings of Louisiana. We contacted his former university department to see if any of his writing or materials remained after his passing but his former collleagues could not find anything. We included as many observations as we could find from Louisiana, but it remains one of the most poorly represented states in our dataset. If other researchers find new citations or localities for the Carolina parakeet not found here, we encourage them to share them with us so we can update this dataset.

These data can be used in many avenues in research for this poorly understood species. In conjuction with modeling approaches, they may be able to help shed new light on the likely cause of the species' extinction, which remains a mystery. Accurate and precise georeferenced specimen data can be used in combination with genetic and stable isotope sampling, for example, to learn more about the relationship between the two subspecies and their dietary composition or foraging behaviour. Of the known extinct parrots, the Carolina parakeet had the largest range, by far (Olah et al. 2016). Learning how and why they went extinct may yield lessons that can inform modern conservation efforts for other parrot species, which is of pressing need, since parrots are the most-threatened major order of birds (Olah et al. 2016) with many obstacles related to conservation prioritization, especially when considering the future effects of climate change.

The authors would like to thank K. Block and G. Nuttall for help with data collection, M. MacFerrin for help with data formatting, the University of Connecticut Inter-Library Loan department for handling hundreds of interlibrary loan requests, and The Natural History Museum for funding. We'd also like to thank W.P. Corgan for making this all possible. Thanks to J. Wieczorek, R. Mounce, J. Stanton, and C. Hermes for their helpful feedback that greatly improved this manuscript. KRB was funded by NSF grant # DGE-0753455.

KRB designed data collection protocol & collected data. All authors verified data and wrote the manuscript.