More than one third of the bird species found in the Caribbean are endemic to a set of neighboring islands or a single island. However, we have little knowledge of the evolutionary history of the Caribbean avifauna and the lack of phylogenetic studies limits our understanding of the extent of endemism in the region. The Sharp-shinned Hawk (Accipiter striatus) occurs widely across the Americas and includes three endemic Caribbean taxa: venator on Puerto Rico, striatus on Hispaniola, and fringilloides on Cuba. These island populations have undergone extreme declines presumably due to ecosystem changes caused by anthropogenic factors, as well as due to severe hurricanes. Sharp-shinned Hawks in general, and Caribbean Sharp-shinned Hawks in particular, have not been placed in a modern phylogenetic context. However, the island taxa have historically been presumed to have some ongoing gene flow with mainland populations. Here we sequenced ultraconserved elements (UCEs) from 38 samples, focusing on Caribbean taxa. Using a combination of UCEs and thier flanking regions, mitochondrial genome sequences, and single-nucleotide polymorphisms (SNPs), we investigated the phylogenetic relationships among Caribbean lineages and their relationships to mainland taxa. We found that Caribbean Sharp-shinned Hawks are reciprocally monophyletic in all data sets with regard to mainland populations and among island taxa (with no shared mtDNA haplotypes) and that divergence in the NADH dehydrogenase 2 gene (ND2) between these mainland and island groups averaged 1.83%. Furthermore, sNMF analysis indicated that Hispaniola, Puerto Rico, and mainland samples each form separate populations with very limited admixture. We argue that our findings are consistent with the recognition of the three resident Caribbean populations as species-level taxa, because both nuclear and mitochondrial genetic data indicate reciprocal monophyly and have species-level divergences there is no sharing of mitochondrial haplotypes among or between island taxa and those on the mainland; and they are diagnosable by plumage.

For full methods see the associated paper, but briefly- libraries were enriched for UCEs using either the 2.5k tetrapod probe kit (2,386 UCEs) supplemented with approximately 100 avian exons, or the standard 5k tetrapod probe kit (5,060 UCEs; Faircloth et al. 2012). Libraries were sequenced with either 100 bp (2.5k kit) or 150 bp (5k kit) paired end reads on an Illumina HiSeq 3000/4000.  After duplicate removal and quality trimming we assembled UCEs using aTRAM v. 2.3.0 (Allen et al. 2018) and the longest contig for each sample was aligned in MAFFT (Katoh 2013).  Each alignment was verified by eye using Geneious 8.0.4 (https://www.geneious.com). To address sequencing error associated with degraded DNA extracted from toepads we first removed unalignable portions of a contig.  Next, we then looked for individual mismatches within each alignment. In the case of tissue or blood samples, if mismatches occurred towards the center of a sequence (outside of the 15 bases on either end), these mismatches were kept. However, for toepad samples, we re-coded mismatches as “N” regardless of where in the sequence the mismatch occurred because coverage was lower across the entire sequence, whereas tissue and blood samples had low coverage typically only at the ends. When a mismatch was identical to any other sequence, regardless of whether the sequence it matched was extracted from frozen tissue, blood, or a toepad, we retained the base call.