Constraints to gene flow increase the risk of genome erosion in the Ngorongoro Crater lion population

Abstract

Small, isolated populations are at greater risk of genome erosion than larger populations. Successful conservation efforts may lead to demographic recovery and mitigate the negative genetic effects of bottlenecks. However, constrained gene flow can hamper genomic recovery. Here, we use population genomic analyses and forward simulations to assess the genomic impacts of near extinction in the isolated Ngorongoro Crater lion (Panthera leo) sub-population. We show that 200 years of quasi-isolation and the recent epizootic in 1962 resulted in a two-fold increase in inbreeding and an excess in the frequency of highly deleterious mutations relative to other populations of the Greater Serengeti. There was little evidence for purging of genetic load. Furthermore, forward simulations indicate that higher gene flow from outside of the Crater is needed to prevent future genomic erosion in the population, with a minimum of one to five effective male migrants per decade required to reduce the risk of long-term inbreeding depression and reduction in genetic diversity. Our results suggest that in spite of a rapid post-epizootic demographic recovery since the 1970s, continued isolation of the population driven by habitat fragmentation and potentially male territoriality, exacerbate the effects of genome erosion.

Fig. 1. Sampling, population structure and gene flow.

a Sampling locations for African lions. Inset shows the 15 newly-sequenced genomes for this study including 10 Ngorongoro Crater, 3 Greater Serengeti and 2 Selous* lions. b Principal Component Analysis. (c) Admixture plot for K = 2-6 and for the 15 newly-sequenced and five lion genomes from de Manuel et al. . Cross Validation error values are given. *Note that Selous is nowadays known as Nyerere National Park.

Fig. 2. Past demography.

a Long-term past demography using the PSMC with a substitution rate of 4.5 × 10⁻⁹ substitutions/site/generation and a generation time of 5 years. b Recent past demography for Crater lions over the past ~200 generations and assuming a generation time of 5 years. The black line depicts the mean and shaded area the 95% CI for 10 replicates (Supplementary Data 2) each calculated as the geometric mean over 40 independent estimates from the observed spectrum of linkage disequilibrium using the Panthera sp. recombination rate. c Population census size from 1962 to 2022. Note that no data was collected between 1972-75 and 1978 and 1980 (Supplementary Data 3).

Fig. 3. Heterozygosity and inbreeding.

a Heterozygosity estimates showing theta (N. Het. sites/1000 bp) for 16 lion genomes with sequencing depth $\ge$14X. Horizontal lines depict the mean and whiskers the standard deviation. Colours refer to the four different populations. b Inbreeding coefficients (F_ROH) based on the identification of Runs of Homozygosity (ROH) using a 100kbp sliding window. Empty bars show the proportion of genomes in ROH$\ge$100 kb (i.e., background relatedness) and grey bars the proportions in ROH$\ge$ 2Mb (i.e., recent inbreeding events). Bars extending from the mean values represent the standard deviation. Statistical significance was assessed with Wilcoxon signed-rank tests (n = 4). Only significant differences are shown.

Fig. 4. Genetic load.

a R_xy of derived alleles for high- and moderate-impact variants for Crater lions relative to Serengeti and Selous lions. R_xy < 1 or >1 corresponds to a deficit or an excess in allele frequency, respectively, in population x (i.e., Crater, n = 10) relative to population y (Serengeti and Selous lions, n = 5). Whiskers represent ±1 SD. b Total load estimated as the ratio of High and Moderate impact to Synonymous variants for 16 lions (with sequencing depth $\ge$14X). c Realised load. Horizontal lines depict the mean and whiskers the standard deviation. Statistical significance was assessed with Wilcoxon signed-rank tests (n = 4). Only significant differences are shown. Colours refer to the four different populations.

Fig. 5. Genome-informed simulations recapitulating the recent population history of Crater Lion from 1820 to 2020.

a Temporal changes in census size (N), mean heterozygosity, mean inbreeding (F_ROH), mean realised load and masked load (rescaled for 20,000 genes; See Methods). Dashed and full lines depict the mean across all runs for the Greater Serengeti Ecosystem (GSE) and Crater (C.) populations, respectively. b Average number of deleterious mutations of each category (for 5000 genes) for the GSE (2020) and the Crater population (1950, 1964, 1972 and 2020). Note that 1964 corresponds to 2-years and 1972 corresponds to 10 years (i.e. 2 generations) after the epizootic, respectively. Horizontal lines within boxplots depict the median, bounds of boxes represent the first and third quartiles and whiskers extend to 1.5 times the interquartile range.

Fig. 6. Prediction of future changes in genome-wide variation over 100 years for various migration rates.

a Heterozygosity; (b) Inbreeding F_ROH; (c) Realised load; (d) Masked load. We simulated 0, 1, 5 and 10 effective migrants per decade and K_Crater values of 50 to 200. Values > 0 and <0 indicate increase and reduction, respectively. Points represent mean and whiskers represent the 95% CI. Dotted lines depict a 5% change and dashed line no change.