Conservation of genetic uniqueness of populations may increase extinction likelihood of endangered species: the case of Australian mammals

Defining significant species and populations for the purpose of biological conservation can be fraught with problems. There are over 26 separate definitions of species hindering conservation efforts [1], with some definitions leading to a 50 % increase in currently recognised species [2], yet none able to solve the apparent species ambiguity problem [3]. Within species, importance is often given to “unique” populations, labelled as subspecies, chromosomal races, morphospecies, ecotypes and so on (e.g., [4–8]). The biological importance of these populations is not always clearly understood, and can be based on geographical/political boundaries (e.g., the subspecies status of the American puma, Puma concolor), morphology, ecology, genetics and/or a mix of the above [5, 7, 9–11].

In species of conservation concern, the Evolutionarily Significant Unit (ESU) was proposed to identify unique population(s) that have evolved independently for a long period of time, are genetically differentiated and uniquely adapted to their environment [11, 12]. By identifying ESUs, it was argued that managers could prioritise conservation efforts so that at the very least, unique populations would be preserved, thereby maintaining the processes that lead to adaptive differentiation [13]. The advent of molecular techniques, however, led to ESUs being largely defined by neutral genetic markers (e.g., monophyly at mitochondrial markers [10]), with further delineation of populations into Management Units (MUs) depending on the degree of differentiation at nuclear loci [10, 13]. This provided a routine way of characterising unique populations, and has gained widespread use in conservation despite the focus being on relatively few neutral genetic markers.

While there is often acknowledgement that adaptation and ecological diversity should also be considered when defining ESUs (and MUs), as originally defined by Ryder [11], frequently ESUs and MUs are defined entirely through such marker systems (e.g., [14–19]). For threatened species programs, this creates a challenge because conservation efforts are targeting populations rather than the species more generally, and thereby potentially promoting fragmentation. This strategy might make sense if past unique evolutionary trajectories in populations are being conserved; but what if these populations aren’t really “unique” at all, and therefore being managed in such a way that may increase their risk of extinction due to reductions in genetic diversity and loss of population fitness?

Random genetic drift effects will affect small populations greater than larger populations, and we therefore hypothesise that in many threatened species there will be a relationship between the level of genetic diversity and population “uniqueness”. We show using neutral marker genetic datasets from five Australian mammals that random genetic drift (and not mutation) is likely to be responsible for the uniqueness of populations of many threatened taxa; by managing such populations as subspecies, ESUs and/or MUs, the extinction risk of the entire species is likely to be increased based on genetic grounds. Introducing new genetic variants into populations through translocation is likely to benefit populations that have undergone declines in genetic diversity through strictly drift processes. While ideally ecological diversity and adaptive genetic potential should also be considered when conserving populations, conservation decisions in most threatened species programs currently remain based on relatively few neutral markers, and we suggest a novel method for identifying unique populations based on such data.

Using the approach in Coleman et al. [30] and consistent with expectations [31], we found a highly significant negative relationship (linear or quadratic) between both genetic diversity measures (H _e and A _r) and genetic uniqueness (population-specific F _ST) for each dataset (Fig. 1; Additional file 1). Therefore, the more unique a population, the lower the level of neutral genetic variation present within that population. We hypothesised that this relationship is likely to be driven by random genetic drift processes and small effective population size (leading to loss of genetic diversity and changes in allele frequencies). On the other hand, mutation-driven divergence without much loss of genetic diversity could reflect long term adaptive evolutionary change, if it is assumed that mutation-driven divergence in microsatellites also reflects mutation-driven changes in functional genes. We explored this using simulations combining observed microsatellite data from populations for each species and found a consistent and highly significant negative relationship between population-specific F _ST and genetic diversity (see Additional file 1), as expected by theory [31]. Therefore, the strongly negative relationship found in observed data for each species and shape of this relationship is likely to be largely explained by drift processes.

The tight relationship found between A _r and population-specific F _ST under a drift model provides a way of defining unique populations, even in the presence of strong drift effects. This negative relationship is expected by theory [31] and the simulations (Additional file 1) with real datasets confirm this expectation. However, populations that sit above the 95 % prediction intervals (and therefore more likely to be genetically differentiated than expected purely by drift) are likely to have unique mutations that have accumulated within those populations through time. We tested this by undertaking the same simulation for a large population (N = 5000) for 5000 generations, where we would expect mutation to introduce more novel alleles. For each species, the simulated population sits well above the 95 % prediction intervals for A _r (Additional file 1), indicating genetic uniqueness (differentiation) higher than that expected by drift effects alone. H _e generally sits within the 95 % prediction intervals, but this is not surprising, as novel alleles will not necessarily increase heterozygosity. As well as the microsatellite loci having a higher number of novel mutations, there is also an increased likelihood of novel mutations at functional loci, raising the issue of whether these more differentiated populations are adaptively diverging and now heading down different evolutionary trajectories.

Using real datasets from five threatened Australian mammal species, we have shown that there is a strong negative relationship between population neutral genetic diversity and genetic divergence, and that random genetic drift is likely to be driving this relationship. Importantly, this relationship appears to be widespread amongst threatened species, as our examination of the Australian EPBC Act 1999 listed terrestrial mammal species showed that for species where genetic data exists, over 73 % had the same significant negative relationship. This is not surprising as populations of threatened species are typically small and isolated, and therefore prone to the erosion of genetic diversity through random genetic drift [32]. A similar relationship is also observed in endemic Australian freshwater fish species that are predisposed to population bottlenecks and founder events through droughts and floods [30, 33]. While this negative relationship is expected by theory [31], to our knowledge, this is the first time empirical data from a broad range of threatened species has highlighted the profound effects of drift in natural populations.

These findings have implications for how we manage threatened species more generally. By not considering overall levels of genetic diversity across a species, and instead focussing on the conservation of unique populations defined as subspecies, ESUs or MUs, managers may inadvertently increase the likelihood of species extinction. Relatively small reductions in genetic diversity measured through neutral markers increase the threat of population extinction, particularly in changing environments [34]. Heterozygosity is intrinsically linked to additive genetic variance, or the ability to respond to environmental change [35] and theory predicts for small populations (e.g., effective population size of 100) that have low levels of genetic variation, the time to extinction in an environment with modest change may be as few as 100 generations (Additional file 1). These effects, of course, do not take into consideration factors that are already likely operating in populations such as reduced fitness through inbreeding depression (which are more severe in stressful conditions, such as suboptimal habitat) [36, 37]. Some mammal populations and species are therefore likely on a trajectory where recovery is improbable without genetic intervention.

Our analyses highlight that managers should give priority to preserving populations that account for the greatest genetic proportion of the total gene pool for a species, thereby increasing the potential for adaptation to environmental change [37–39]. Populations that have previously been considered genetically unique (and therefore managed independently) are more likely to be in need of genetic rescue and/or restoration (e.g., Perameles gunnii unnamed subspecies from Victoria, Australia; the southern ESU Burramys parvus). Guidelines for genetic rescue/restoration have been developed elsewhere [38, 40, 41], but in many cases translocation between populations in situ represents a powerful strategy for increasing genetic diversity within and between populations of threatened species where drift processes predominate [38]. Despite some risks [38, 42], translocations for genetic reasons should be considered an option in all threatened species programs at a time of rapid environmental change where ongoing adaptation is crucial for species persistence.

The perceived presence of local adaptations in geographically isolated populations in response to selective pressures associated with different environmental conditions has been a barrier to undertaking in situ translocations in threatened species [42–45]. Local adaptation is generally strongest in large populations [46], and, at least in plants, generally absent or relatively weak in small populations [44, 46]. Similarly, random genetic drift decreases local adaptation (due to a decrease in additive genetic variance), particularly in very small populations where migration is absent [37, 47]. Our results highlight that neutral genetic variation for populations of many threatened species has been decreased because of random genetic drift. By conserving populations with low levels of neutral genetic variation and in the absence of gene flow, there is a likelihood that these small populations will have reduced potential for local adaptation if it is assumed that levels of neutral variation reflect adaptive genetic variation (Additional file 1). In some cases, drift effects will be so extreme that populations could be maladapted to their environment through the random fixation of deleterious alleles [46]. Therefore, genetic uniqueness should not necessarily be considered a barrier to undertaking genetic translocations in many threatened species.

Neutral genetic markers such as microsatellites and mitochondrial sequencing have been used prolifically for determining conservation significant populations (e.g., [14–19]). Here we have highlighted how this can place emphasis on populations that are genetically depauperate and potentially maladapted. However, our approach, which focuses on neutral variation and the effects of genetic drift and gene flow, could be used to evaluate conservation significant populations with as few as six microsatellite loci. For instance, for the observed data from all five species, only three populations found in the B. parvus dataset sit clearly above the 95 % prediction intervals for the regression line for A _r (Additional file 1), suggesting that these populations are more differentiated than that expected under a drift model and therefore contain unique genetic diversity (novel alleles that are likely to have arisen through mutation and increased in frequency by chance through time). These three populations are from the northern ESU and have been separated from populations in the central and southern ESUs for at least 20,000–152,000 years [22]. The southern ESU has previously been regarded as the most unique B. parvus population [22], but this population has the least genetic variation of all populations (Additional file 1), does not sit above the regression line and therefore is not genetically unique. Similarly, the three mainland populations of P. gunnii are considered a subspecies, yet have the least genetic variation and do not differ from the regression line (Additional file 1), again highlighting their lack of uniqueness. It is important to note that the entire distribution of a species would need to be sampled and genotyped to be confident that a population or several populations are unique using this methodology, and this is the case here for B. parvus, P. gunnii, D. viverrinus and D. hallucatus, but not D. maculatus (where no samples were genotyped from populations at the northern and southern end of their distribution).

Other methods exist which can also provide information on the uniqueness of populations (e.g., [48, 49]). Similarly, with the advent of genomic approaches in conservation [50, 51], genome-wide estimates of differentiation can now be obtained for natural populations. These incorporate both neutral and adaptive loci, providing a more accurate estimate of population uniqueness, and allowing patterns of genetic diversity across populations to be contrasted for different marker systems. Ultimately, characterising variation in genes involved in local adaptation is likely to provide a clearer understanding of the link between genetic diversity, population uniqueness and adaptive potential.

We have shown that random genetic drift effects in threatened animals are widespread and lead to the erosion of neutral genetic diversity within species. This places many populations and species at a greater risk of extinction in a changing environment even if other threats can be obviated. Conservation strategies need to focus on the preservation of genetic diversity at the species level, rather than that of the population, subspecies or evolutionary significant unit, and augmented gene flow from genetically diverse populations needs to be considered as a way of increasing fitness and the adaptive potential of populations.

A _r , allelic richness; EPBC, Environment Protection and Biodiversity Conservation; ESU, evolutionary significant unit; H _e , expected heterozygosity; IUCN, International Union for Conservation of Nature; MU, management unit