Genetic load and viability of a future restored northern white rhino population

Abstract

As biodiversity loss outpaces recovery, conservationists are increasingly turning to novel tools for preventing extinction, including cloning and in vitro gametogenesis of biobanked cells. However, restoration of populations can be hindered by low genetic diversity and deleterious genetic load. The persistence of the northern white rhino (Ceratotherium simum cottoni) now depends on the cryopreserved cells of 12 individuals. These banked genomes have higher genetic diversity than southern white rhinos (C. s. simum), a sister subspecies that successfully recovered from a severe bottleneck, but the potential impact of genetic load is unknown. We estimated how demographic history has shaped genome-wide genetic load in nine northern and 13 southern white rhinos. The bottleneck left southern white rhinos with more fixed and homozygous deleterious alleles and longer runs of homozygosity, whereas northern white rhinos retained more deleterious alleles masked in heterozygosity. To gauge the impact of genetic load on the fitness of a northern white rhino population restored from biobanked cells, we simulated recovery using fitness of southern white rhinos as a benchmark for a viable population. Unlike traditional restoration, cell-derived founders can be reintroduced in subsequent generations to boost lost genetic diversity and relieve inbreeding. In simulations with repeated reintroduction of founders into a restored population, the fitness cost of genetic load remained lower than that borne by southern white rhinos. Without reintroductions, rapid growth of the restored population (>20-30% per generation) would be needed to maintain comparable fitness. Our results suggest that inbreeding depression from genetic load is not necessarily a barrier to recovery of the northern white rhino and demonstrate how restoration from biobanked cells relieves some constraints of conventional restoration from a limited founder pool. Established conservation methods that protect healthy populations will remain paramount, but emerging technologies hold promise to bolster these tools to combat the extinction crisis.

Keywords: fitness; genetic load; genetic restoration; in vitro conservation; runs of homozygosity; simulation.

FIGURE 1

(a) Census population sizes of NWR and SWR over the past two centuries. When our study individuals were born, the SWR population was recovering from <200 individuals and the NWR population was still declining. (b) The demographic model that best fit the two‐dimensional site frequency spectrum (2dSFS), showing population decline until 37 generations ago, after which both populations underwent more rapid size changes, with some evidence of very low levels of gene flow from SWR into NWR. (c) Cumulative fraction of genomes represented by ROH lengths (thin lines are individuals, thick lines are population means; inset shows raw counts across all 22 genomes). The relative abundance of ROH lengths is shown in the heatmap above, with more orange indicating higher abundance in NWR genomes and more blue indicating higher abundance in SWR. NWR genomes had smaller ROH likely stemming from older inbreeding (>500 generations ago), whereas SWR had larger ROH from recent inbreeding (as recently as ~15 generations ago).

FIGURE 2

(a) Deleterious alleles tend to segregate at lower frequencies in both populations, but proportionally more conserved (RS > 4) and non‐synonymous (missense and loss of function) deleterious alleles are fixed in SWR (p = 0.025) than NWR. (b) Compared to NWR, SWR genomes have more homozygous and fewer heterozygous deleterious alleles at conserved and missense conserved sites, but fewer homozygous and heterozygous loss of function (LOF) alleles by count. (c) SWR genomes have proportionally more homozygous deleterious alleles at conserved and missense conserved sites than SWR, but proportionally fewer homozygous and heterozygous LOF alleles than NWR.

FIGURE 3

Longer ROH are enriched for homozygous deleterious mutations. The proportion of homozygous mutations at conserved sites (a) and the ratio of homozygous moderate‐impact conserved (RS > 4) to homozygous low‐impact sites (b) significantly increased with ROH length (bins: 0.1 MB ≤ ROH < 0.5 MB, 0.5 MB ≤ ROH < 1 MB, 1 MB ≤ ROH < 3 MB, and ROH ≥ 3 MB) in both white rhino subspecies. There were likely too few ROH ≥ 3 MB (n = 7 across all individuals) to accurately estimate load in large ROH for NWR. Dashed lines show the proportions of heterozygous sites with deleterious alleles and homozygous sites with deleterious alleles in non‐autozygous regions of the genome.

FIGURE 4

Genetic load and fitness simulated over 10 generations of growth in northern white rhino populations restored from cryopreserved cells. Simulated populations were either founded once in generation 0 (unsupplemented) or founded with subsequent reintroductions of one randomly selected cryopreserved genome each generation (supplemented). Simulation replicates (thin lines; 10 of the 50 replicates are shown) and medians across replicates (thick lines) are colored by growth rate. Dashed black horizontal lines show the median values for SWR genomes. (a, b) Median number of homozygous (upper plots) and heterozygous (lower plots) alleles per genome with severe, moderate, and mild deleterious effects in unsupplemented (a) and supplemented (b) populations. (c, d) Median per‐individual fitness effects of deleterious genetic load each generation relative to the median fitness of SWR genomes. (e, f) Median realized load each generation. (g) Number of individuals per generation in populations with different growth rates.