Table 1 presents a summary of nine papers that provide extreme examples of population persistence with low genetic diversity. Many of the examples of long-term persistence with low genetic diversity provided have several points in common with each other. These commonalities provide lessons for conservation biologists, which I discuss below.
Environmental quality and density-dependent growth
Ellegren et al. (1993) describe Scandinavian beavers as reintroduced into an intact habitat, without competitors and protected from hunting. Broders et al. (1999) point out that the bottlenecked Canadian moose were introduced into ‘ideal’ sites with abundant food, no predators and no competitors. The Mauritius kestrel was driven to the brink of extinction by the use of DDT and the presence of invasive predators. This species was recovered by banning the use of DDT, rearing hatchlings in incubators, supplemental feeding of wild populations in order to double the number of clutches, predator control measures and the construction of artificial nests in order to increase carrying capacity. The Chillingham cattle live in a heavily managed environment without predators.
These examples demonstrate the power of large population growth rates to rapidly extricate populations from the extinction vortex and, with sufficient carrying capacity, to become robust populations with little short-term risk of extinction. The beavers, moose and kestrels were all reduced to very low population sizes relative to carrying capacity, providing the opportunity for very rapid density-dependent population growth. They also had moderate to very large environmental carrying capacities to sustain that growth and buffer them against future demographic and environmental stochasticity. In the case of the Mauritius kestrel and the Chillingham cattle, potential population growth rates are increased (and variance decreased) through food supplementation and other forms of population management.
Sometimes inbreeding leads to such low fitness that habitat restoration to increase growth rates cannot remedy the decline and only the infusion of new genetic material can do so (e.g. Westemeier et al., 1998; Hogg et al., 2006; Pimm et al., 2006; Willi et al., 2007; Trinkel et al., 2008; Hedrick & Fredrickson, 2010). For example, even though Chillingham cattle have survived for 300 years with very little genetic variation, another bovine population managed under semi-natural conditions for the past 130 years seems destined to extinction from inbreeding depression without the inducement of gene flow from another herd (Halbert, Grant & Den, 2005). This illustrates the stochastic nature of the fixation of the deleterious load of mutations under inbreeding. By chance, some populations will become fixed for such a load of deleterious mutations that they cannot rebound without genetic rescue.
However, most inbred populations will not be in such dire straits. Modeling has been explicit in suggesting under what conditions inbreeding depression is likely to be important (Reed et al., 2003; Robert, Couvet & Sarrazin, 2005; O'Grady et al., 2006; Theodorou & Couvet, 2006; Reed, 2008). Further reductions in fitness due to an increase in the inbreeding level or from environmental deterioration have little effect on population viability until their impacts on the stochastic population growth rate are such that it approaches zero, at which time even small changes can have huge impacts on the probability of extinction (see fig. 2.1 and accompanying text in Reed, 2008). If a population is already adapted to the environment, the environment is very benign, and/or the species is colonizing a habitat with available and suitable niche space with few generalist pathogens and predators to impede its population growth – lack of genetic diversity will likely seldom be a limiting factor in their success.
In the cases presented above, populations were expected to have strong positive growth rates due to environmental conditions. Thus, these examples do not run contrary to current ecological-evolutionary theory concerning population viability. However, the Mauritius kestrel example especially, provides us with hope that some populations with even extreme inbreeding will rebound on their own, or with human intervention, as long as suitable habitat remains.
Neutral genetic variation
Eight of the studies inferring long-term persistence despite little genetic variation rely on estimates of neutral genetic variation from molecular markers, three assayed MHC diversity in addition to neutral genetic variation, and one relied solely on a region of mitochondrial DNA to infer genome-wide levels of genetic diversity. There are a number of problems associated with these methods.
Estimates of genome-wide heterozygosity from molecular markers usually correlate poorly with estimates of genetic variation for quantitative traits (Reed & Frankham, 2001; Badri et al., 2007; Knopp et al., 2007) and microsatellite diversity may correlate only moderately with estimates of genetic diversity from single nucleotide polymorphisms (SNPs) that can be found within coding regions of genes and directly selected upon (Väli et al., 2008). Fundamental differences in the maintenance and production of genetic variation at neutral loci versus loci under selection suggest that the evolutionary potential of a population for ecologically important quantitative traits are unlikely to be accurately predicted from assays of neutral molecular markers. This problem is further exacerbated if the number of loci used is small.
The three studies that did not rely solely on neutral genetic markers used variation at MHC loci. The use of genes known to be under selection and to be important to a trait (disease resistance) that impacts population viability is certainly a step in the right direction. However, while many studies have shown a correlation between MHC diversity and disease resistance, many others have not. As pointed out by Acevedo-Whitehouse & Cunningham (2006), variation at MHC loci may explain only half of the variation in disease resistance and molecular studies usually focus only on a small portion of all MHC loci. Combined with the fact that populations face many other stressors in addition to pathogens, it is not surprising that populations with little or no genetic variation at a particular MHC locus have been often been found to thrive.
Several of the studies on long-term persistence with little genetic diversity were initiated because the populations were known to have passed through a population ‘bottleneck’. Unfortunately, the somewhat vague definition of a population bottleneck (i.e. a short-term restriction in the total size of a population) can apply to cases where little genetic change is expected to occur (i.e. the term bottleneck has been applied to short-term restrictions of ≥100 individuals, where the loss of genetic diversity is expected to be negligible). That aside, the ability of even populations that have been severely bottlenecked to undergo rapid evolutionary change has been verified repeatedly both experimentally and during natural founder events (e.g. England et al., 2003; Sax et al., 2007), so that such occurrences should no longer be surprising. Bottlenecks may reduce genetic variation in accordance with their severity and the number of generations which they persist. However, due to the ability of selection to maintain genetic variation and purge the genetic load in inbred populations (Kristensen et al., 2005; Bensch et al., 2006; Kaeuffer et al., 2007; Fox, Scheibly & Reed, 2008; Demontis et al., 2009) as well as the potential conversion of dominance and epistatic variation to additive genetic variation during bottlenecks (e.g. Bryant, McCommas & Combs, 1986; van Heerwaarden et al., 2008), many populations will suffer little or no loss of quantitative genetic variation even with severe inbreeding.
The following four papers deserve special attention because of the very long persistence times involved:
Hadly et al. (2003) provide an interesting and impressive case of a social rodent that seems to have existed for around a 1000 years with very little mitochondrial haplotype variation in the cytochrome b gene. This finding is no doubt noteworthy and raises the potential that this species may have persisted at a small population size for a remarkably long time. However, the claim that genome-wide genetic variation and genetic variation for quantitative traits was ‘negligible’ requires quite a leap of faith. The rank correlation coefficient between mitochondrial and allozyme diversity among 97 species of animals is only a moderate 0.36 (Piganeau & Eyre-Walker, 2009). Perhaps further work will elucidate the demography and genetics of this species over the past millennia, using a more comprehensive sampling of the genome, but it would be rash to declare this population as any kind of exception based merely on mitochondrial DNA.
Milot et al. (2007) sample a large number of markers and a huge number of individual wandering albatrosses as well. There is no doubt that the two albatross species have little molecular genetic variation at the markers assayed and that they almost certainly inherited the low genetic variation from a common ancestor. Thus, wandering albatrosses seem to have existed for hundreds of thousands of years with low molecular genetic variation. This is very informative and I found it impressive. However, no minimum heterozygosity level for neutral genetic loci has ever been suggested to be compatible with viability. The wandering albatross exists at a population size that should provide it with long-term viability and likely an effective population size large enough that mutation can replace genetic variation faster than drift can remove it (Reed et al., 2003). The Amsterdam albatross is a different story, but the results of its recent bottleneck and its loss of 60% of the heterozygosity found in the wandering albatross is yet to be determined. Milot et al. (2007) do an excellent job of pointing out potential reasons for the ecological success of wandering albatrosses despite low genetic variation. I concur with them wholeheartedly that some species will have naturally occurring low levels of neutral genetic variation based on their life history and other factors and that this should not be cause for alarm or spark genetic management.
The results of Babik et al. (2009), while very interesting for a number of reasons, are not surprising from an ecological-evolutionary perspective on population viability. Crested newts from a refuge population subsequently spread across Europe after the retreat of the glaciers, despite reduced levels of MHC and overall genomic genetic diversity. It is well established that species can expand their geographic ranges drastically from only a small number of founders (e.g. Pascual et al., 2007; Zayed, Constantin & Packer, 2007; Eales, Thorpe & Malhotra, 2008), even when not repopulating depleted ecological communities after the retreat of glaciers. Thus, while always informative, such examples do not constitute a challenge to conventional thinking on population viability as the potential for expansive openings in niche space and reduced competition can allow for rapid population growth and huge carrying capacities for establishing species.
Johnson et al. (2009) represents probably the biggest challenge to current ecological-evolutionary thinking concerning population viability. However, the challenge is not on the grounds of the low molecular diversity seen in the fish eagles. The heterozygosity levels in this species are much higher than in albatrosses (Milot et al., 2007), the Parnassius apollo butterfly (Habel et al., 2009), and many other assayed species. However, it would be a wondrous feat if this population has managed to exist at 100–120 breeding pairs for 2800 years despite environmental and genetic stochasticity. Unfortunately, the confidence intervals around the timing of the start of population decline range from 600–9000 years. To test how unusual persistence over these time intervals might be, I used the stochastic population viability models for 102 species summarized by Reed et al. (2003). For an eagle of this size, 600 years is c. 45 generations. The probability of persistence over 45 generations averaged across models scales as:
where P is the probability of persistence and K is the ceiling carrying capacity. Thus, if we take 120 pairs (240 adult animals) as the carrying capacity, the probability of population survival over 600 years is expected to be 36%. These models include catastrophes, whereas the genetic data for the Madagascar fish eagle suggests a stable population size during this time. Catastrophes greatly decrease persistence times of populations and also increase the importance of genetic diversity if those catastrophes are repeated bouts of drought, similar pathogens or directional environmental change (Reed, 2008). However, the probability of persistence at this carrying capacity for 2800 years (>200 generations) is negligible and, if true, would make this an astounding feat of population persistence.