Puma genomes from North and South America provide insights into the genomic consequences of inbreeding

Abstract

Pumas are the most widely distributed felid in the Western Hemisphere. Increasingly, however, human persecution and habitat loss are isolating puma populations. To explore the genomic consequences of this isolation, we assemble a draft puma genome and a geographically broad panel of resequenced individuals. We estimate that the lineage leading to present-day North American pumas diverged from South American lineages 300-100 thousand years ago. We find signatures of close inbreeding in geographically isolated North American populations, but also that tracts of homozygosity are rarely shared among these populations, suggesting that assisted gene flow would restore local genetic diversity. The genome of a Florida panther descended from translocated Central American individuals has long tracts of homozygosity despite recent outbreeding. This suggests that while translocations may introduce diversity, sustaining diversity in small and isolated populations will require either repeated translocations or restoration of landscape connectivity. Our approach provides a framework for genome-wide analyses that can be applied to the management of similarly small and isolated populations.

Fig. 1

Puma range past and present. The current range of pumas (hashed) compared to their historic range (blue). Circles denote the geographic coordinates of the puma populations sampled in this study. Panels show zoom-ins of puma habitat distribution (dark gray) within the known range of the species in the contiguous United States as predicted by the USGS. Current range data are from the IUCN Red List of Threatened Species. Historic range data are approximated based on prior reports. Base map generated with Natural Earth.

Fig. 2

Demographic history of pumas. a Mitochondrial maximum likelihood phylogeny of the ten pumas in this study plus an additional puma from Big Cypress (KP202261.1) and the African cheetah (KP202271.1) as the outgroup. We calculated divergence times by determining the number of pairwise divergences between sequences and used a mitochondrial divergence rate of 1.15% bp per Myr^,. We estimate a common maternal ancestor of these pumas 278,000 ± 5,639 years ago (star; 100% bootstrap support), divergence between North American and South American mitochondrial lineages 201,000 ± 1952 years ago (pentagon; 63% bootstrap support), and a common maternal ancestor of North American pumas 21,000 ± 10,412 years ago (circle; 100% bootstrap support). b Inferred changes in effective population size (N_e) over time using the pairwise sequentially Markovian coalescent (PSMC) model for the ten pumas. We assume a generation time of 5 years and a per generation mutation rate of 0.5e-8 per bp per generation. The PSMC model for EVG21 shows a sharp increase in inferred N_e that is probably attributable to its hybrid ancestry.

Fig. 3

Stratification of pumas based on the geographic population. a Principal component analysis of 166,037 sites separates the sampled pumas based on population. The first component primarily separates South and North American pumas, while the second component distinguishes the variation within North America. All California pumas (Santa Cruz and Santa Monica) cluster closely. b TreeMix analysis, using the African cheetah as the outgroup, indicates the best tree separates pumas based on population and includes one migration event (weight = 0.453911) from the branch of South American diversity into the admixed Everglades puma (EVG21). c The mean of ten permuted matrices of STRUCTURE analysis for each of K = 2 through 4, performed using CLUMPP. Both delta K and L(K) values indicated that K = 3 was the best K (Supplementary Fig. 10).

Fig. 4

Heterozygosity and runs of homozygosity. a Sliding window heterozygosity (black dots) and called ROH (colored boxes) across a single scaffold for three pumas from three different populations (Big Cypress, Yellowstone, and Brazil). Plots for all pumas are provided as Supplementary Fig. 12. b Average genome-wide heterozygosity versus the proportion of the genome in ROH for the ten pumas sequenced. c Distribution of lengths of ROH. The length in Mb is indicated, as is the associated expected number of generations since the individual’s maternal and paternal lineages shared a common ancestor. d Heat map showing the percent of the genomes that are in ROH that are shared IBD between pairs of pumas (Supplementary Table 6).