Invasions and extinctions through the looking glass of evolutionary ecology

(a) Genetic variation and response to selection

Empirical studies show that moderate to severe population bottlenecks can be relatively common for both endangered and invasive species [13–15]. However, post-bottleneck population dynamics differ in several important ways that translate to large differences in genetic diversity maintained within populations (figure 1). One key difference is the length of the bottleneck period [16], as this directly affects the rate at which genetic diversity is lost [17]. While paired demographic and genetic data are rare for early stage invasions, the magnitude of genetic bottlenecks estimated from neutral markers shows that, on average, invasive populations suffer a detectable but relatively minor loss of genetic diversity [13]. This result suggests that most introduced populations remain small for only a relatively short period before increasing again, because most introductions begin with a strong demographic bottleneck involving a small fraction () of all individuals present in the native range. Endangered species far more commonly experience extended periods of small population size, greatly increasing the loss of genetic diversity. Even if endangered populations do recover in census size, the genetic effective size of the population (N_e) will recover slowly as new mutations introduce variability back to the population.

Another factor mitigating loss of genetic diversity from demographic bottlenecks in invasive species is the high rate of gene flow common among introduced populations, which can promote ‘genetic rescue’ (alleviating inbreeding depression; [18]) of inbred populations and ‘evolutionary rescue’ (introduction of an adaptive variant; [19]) of maladapted populations [20–22]. Such increased gene flow is particularly important within a metapopulation context, because the negative demographic and genetic consequences of small population size and local extinction can be mitigated by connectivity with other populations in the region [15,23,24]. Under this model, gene flow shares diversity among populations so the metapopulation as a whole maintains higher N_e and sustains the invasion even if individual invasive populations are small. By contrast, declining populations of endangered species tend to be few in number and highly isolated, with little connectivity and opportunity for genetic or evolutionary rescue. Efforts to increase population connectivity for endangered species may be particularly beneficial when isolation is prolonged and population sizes are especially small; however, the benefits of increasing gene flow must be evaluated relative to the risk of compromising locally adapted genotypes.

Because the efficacy of selection is directly proportional to N_e, genetic drift has a much stronger effect on evolutionary change in species with limited genetic variability and among small, isolated populations, compared with larger well-connected populations. The reduced ability to respond to selection makes endangered populations particularly susceptible to environmental changes that lead to loss of local adaptation. Here too, the metapopulation context distinguishes biological invasions and provides additional opportunities for these species to respond to selection despite founder effects and genetic drift during establishment. Response to selection is more likely for invasive populations when (i) population sizes are large or increasing, (ii) gene flow from neighbouring populations acts as a genetic or evolutionary rescue, and (iii) local extinction of maladapted populations are followed by recolonization from populations with pre-adapted genotypes. This latter example of inter-demic selection within a meta-population context occurs as a result of non-random extinction and recolonization; this may be an under-appreciated mechanism for rapid evolution of invaders in novel environments [23,25]. In this way, invasions might be useful experimental systems for understanding how population connectivity contributes to the survival of individual subpopulations in variable environments, applying this knowledge in turn to restoration strategies and expectations for native species. Population genetic models that assign source populations to colonists and test for genetic associations between source population characteristics (e.g. density, inbreeding rate and phenotypes) and colonization probability may be a particularly useful approach for investigating the effect of metapopulation context on population persistence [24,26].

It is important to recognize that the genetic consequences of small population size affect variation at neutral marker loci differently than quantitative genetic variation in ecologically important traits, and indeed these two types of variance are often poorly correlated [27]. This is likely a consequence of the broadly polygenic basis of most quantitative traits, as individual loci each have only a small effect on the overall trait variance. Previous studies have shown that substantial quantitative genetic variation (V_G) can persist for quantitative traits, even when populations show evidence of bottlenecks at individual loci [28,29]. As a result, rapid population growth of invasive species following even a severe demographic bottleneck will have little effect on quantitative genetic variance [30]. Moreover, given that most introductions fail, repeated founding events create opportunities for natural selection to filter the genetic variants that do establish at the scale of genotypes or even entire populations [31], though direct evidence of this process is lacking to date. By contrast, persistent small population sizes over multiple generations in endangered species will erode quantitative genetic variance through genetic drift, limiting adaptive evolutionary responses to changing selection pressures. Few studies have directly addressed the consequences of invasion or colonization on V_G, although one such study showed a loss of V_G in younger colonizing populations during (non-invasive) range expansion [32]. A few studies have compared V_G between native and introduced populations of invasive species [33,34], but in these cases it is difficult to rule out differences among populations in the strength of stabilizing selection. Increasingly, population genomics is enabling the discovery of genes (or closely linked loci) involved in local adaptation, which may allow a more direct integration of molecular data into conservation strategies. More generally, genomic sequencing can now be combined with ecological surveys and field manipulations to better understand the relationship between genome-wide genetic variation and adaptive trait variation in natural populations, and how both are affected by demographic bottlenecks, gene flow and natural selection.

(b) Genetic load and inbreeding depression

One intriguing difference between endangered and invasive species is the effect of small population size on genetic load and inbreeding depression. Genetic load is a well-established conservation concern for many endangered taxa, as elevated inbreeding in small populations exposes the negative fitness effects of recessive deleterious alleles [17]. These effects directly contribute to increased extinction probability, especially under stressful environments [35]. The genome-wide load of deleterious mutations is sensitive to N_e, which determines how effectively purifying selection can purge genetic load. In endangered populations that are small, declining, or have undergone a bottleneck, avoiding or mitigating inbreeding is critical to maintain genotypes that carry few deleterious alleles (e.g. avoiding Muller's Ratchet). Invaders may also suffer from inbreeding depression if a colonization bottleneck is severe, as exemplified by a multi-species analysis of birds introduced to New Zealand. In this case, species that experienced strong bottlenecks (less than 150 individuals) showed persistent increases in hatching failure compared with less ‘bottlenecked’ species [36]. These effects may be transient in growing populations as some empirical studies have found increases in heterozygosity over time, despite bottlenecks as severe as a single pair of breeding individuals [37,38]. This is consistent with an increase in the efficiency of natural selection to ‘weed out’ homozygous individuals as population sizes increase. However, further studies are needed to determine whether this is a general phenomenon in successful invaders. Invasive species may also accumulate genetic load as a by-product of the range expansion process. Recent theoretical work has shown that small populations at the wavefront of an expanding range face increased probabilities of deleterious mutations drifting to high frequency, termed ‘expansion load’ [39]. The accumulation of expansion load is predicted to lead to legacies of reduced fitness following expansion [40,41]. Long-range dispersal between multiple introductions from distinct genetic sources could partially mitigate this effect by helping to shelter the genetic load in recently expanded invasive populations [42], but to our knowledge this has not been explored in colonization models.

Why do invasive populations not suffer from inbreeding depression more often? First, as described above, if invaders recover from demographic bottlenecks more quickly, they maintain genetic variation and experience less severe genetic drift and inbreeding. Second, as invasive populations grow in N_e, selection should become more effective at reducing the frequency of weakly deleterious alleles, thereby reducing the genetic load [43]. Third, the fitness effects of recessive alleles may be conditional on the environment. Phenotypic effects of deleterious mutations may be conditionally neutral when the environment is benign but amplified by environmental stress, resulting in inbreeding × environment (I × E) interactions [44]. Invasive species tend to experience more benign environments than endangered species, for example, by invading resource-rich or enemy-free environments, and this could mask the expression of deleterious alleles [45]. Consequently, invasive species may avoid the negative fitness effects of inbreeding more often than rare and declining species, even when subject to the same evolutionary dynamics of small population size. An important question is thus how much of the among-population variance in fitness (and underlying allele frequency differentiation) is caused by different degrees of purifying selection acting on weakly deleterious or conditionally neutral (I × E) genetic load? Answering this question could improve insights into the population-level consequences of different amounts of inbreeding in both invasive and endangered species. Spatial comparisons of population demography could be coupled with experimental and/or genomic assessments of genetic load to assess the conditions associated with effective purging. Here, an experimental strength is the high replication potential afforded by many invading populations with different demographic histories.

(c) The pros and cons of genetic admixture

Admixture is a well-established outcome for invasions and arises when multiple genotypes from genetically divergent populations in the native range come into secondary contact during invasion [46–48]. Hybridization and admixture are also a major issue for species at risk, where declining populations may be intentionally admixed during management efforts or may unintentionally hybridize with more abundant species [49,50]. The immediate fitness consequences of admixture can be complex, depending on the degree of divergence of the parental lineages, and can vary among F₁ and more advanced recombinant hybrid generations [51,52]. Fitness effects can range from highly beneficial outbreeding to severely deleterious hybrid incompatibilities, decreasing in benefit as parental populations become more locally adapted [53].

A distinct benefit of admixture for both invasive and endangered species is increased heterozygosity at loci containing recessive deleterious mutations. Positive effects of admixture have been observed in invasive populations that show heterozygosity-fitness correlations in the introduced range [46,54]. The fitness benefits of a heterozygous genome are likely to be especially strong for small, declining and inbred populations of endangered species that are unable to purge deleterious mutations [55]. This is frequently seen in zoo or other extremely bottlenecked vertebrate populations. For example, the endangered Florida panther population shrank in size to as few as 20 individuals, with reduced heterozygosity at neutral markers and phenotypic evidence of inbreeding depression, including sperm deformities, kinked tails and reduced survival [56]. Intentional release of eight female Texas pumas into Florida created opportunities for admixture, doubling heterozygosity in the population and alleviating inbreeding depression in many traits.

In many cases, admixture also contributes to increases in standing V_G and can broaden the genotypic space available to selection. When adaptive variants from a genetically distinct population introgress, natural selection can act on this enhanced standing variation without waiting for new mutations to arise de novo [57]. For these reasons, and because invasive species frequently experience novel selective environments in their introduced range, admixture has been implicated as a potential factor fuelling rapid evolution and the generation of novel invasive genotypes [33,48,58]. Nevertheless, direct evidence of a link between introgressed variation and the evolution of invasiveness is largely lacking to date. Population genomic studies of admixture/hybrid zones during invasions could test the adaptive introgression hypothesis using methods that identify the signal of differential introgression of positively selected loci against the null expectation based on the degree of admixture across the genomic background (e.g. genomic clines analysis) [59]. The benefit of adaptive introgression should be most pronounced when rates of gene flow and introgression are low relative to N_e and recombination rate, such that selection can efficiently incorporate beneficial variants while eliminating detrimental alleles. Therefore, adaptive introgression of positively selected genomic regions is likely to be much less common in endangered species, where low population sizes and low effective recombination limit the ability of selection to decouple maladaptive from adaptive introgressed alleles.

Populations of endangered species can also benefit from expanded genotypic variation resulting from admixture. Indeed, this is one reason that wildlife corridors are promoted for conservation [60]. The benefit of admixture is reduced when populations are locally adapted and therefore may be less beneficial for native relative to non-native species. Such an influx of genetic variation would be even more detrimental in small populations where natural selection is less efficient at eliminating locally maladapted alleles [61,62]. This genomic swamping of endangered species has become a serious conservation concern when small native populations meet abundant populations of reproductively compatible invasive species. Many native species show relatively low reproductive isolation from introduced species with which they have had no historical contact, and high numbers of introduced genotypes increase the likelihood of hybridization [63,64]. In addition to swamping locally adapted alleles, these native-invasive hybrids can put legal protections of native species in jeopardy as species definitions become questionable [65].

(a) Plasticity and tolerance of novel environments

Adaptive plasticity allows individual organisms to survive and reproduce in a variety of environmental conditions. Such plasticity can buffer populations or species from extinction when conditions change rapidly [70–72]. For example, some birds and mammals can advance life-history schedules through plastic responses to seasonal cues, allowing them to keep pace with altered timing of food availability due to rapid climate change (e.g. [73]); the many taxa whose populations lack such plasticity may face an enhanced risk of extinction [74,75]. Adaptive plasticity can include effects of parental environments on offspring phenotypes (transgenerational plasticity). For instance, in the sheepshead minnow, offspring growth rates, body mass and expected fecundity were highest at temperatures previously experienced by parent fish, regardless of whether that temperature was high or low [76]. As a result, warmer water temperatures did not cause negative effects on development and fitness. Adaptive transgenerational effects such as this may be most beneficial to species encountering gradual changes in the environment, including the rise in sea temperatures predicted under current models of global climate change.

Along with promoting species persistence, adaptive plasticity can facilitate the rapid spread of invasive species across diverse new habitats [77–79]. In both animals and plants, the ability to maintain net reproductive rates in contrasting environments promotes invasive spread [80,81]. Additionally, a plastic response to sharply increase fecundity in resource-rich environments may be an important attribute of invasive taxa [82,83] because it increases the ‘propagule pressure’ that fuels colonization [84,85]. Theoretical work indicates that greater adaptive plasticity should be associated with higher rates of colonization of new, and more diverse, environments [86]. However, recent meta-analyses disagree as to whether invasive species consistently show higher plasticity in general than native congeners [82,87]. This inconsistency in part reflects the different plasticity metrics and choice of traits used in various studies. Results of native versus invasive comparisons also depend on environmental variability in the home range that favours high plasticity, and on norm of reaction evolution following a species' introduction (for instance, canalization of a new adaptive phenotype following initially high plasticity) [88].

In addition to maintaining fitness across environments, adaptive plasticity can prevent a decrease in ecological breadth when a genetic bottleneck occurs, as each genotype can accommodate diverse conditions. Such plasticity can mitigate the ecological consequences of a prolonged bottleneck in an endangered species [89] or a short-term bottleneck due to an introduction event [90,91]. In these cases, geographically isolated populations may share similar broad patterns of individual plasticity instead of divergent, locally adapted norms of reaction [92,93]. However, as adaptive plasticity itself has a genetic basis, it can be compromised by the negative consequences of inbreeding and sustained bottlenecks. Moreover, two critical factors will determine the effectiveness of plasticity relative to selective evolution in maintaining the viability of populations exposed to novel environments. First, norms of reaction that evolved under past selection pressures may not encompass the range of phenotypes required to maintain fitness in the new circumstances, and can even produce a disrupted, maladaptive phenotype in response to a novel stress [94]. Second, even with sufficient existing plasticity, adaptive phenotypes will not be produced if changed cues fail to provide accurate information to elicit appropriate and timely responses [69,95]. In such cases, plasticity can promote extinction rather than persistence [29]. This potential adaptive limit may be particularly important in human-altered environments or following introduction to a new continent where abiotic and biotic factors that organisms have evolved to use as plasticity cues may be absent or disrupted.

(b) Plasticity and evolutionary potential

Phenotypic plasticity can allow a population to persist following a change in local conditions or introduction to a new range. If pre-existing reaction norms do not produce sufficiently adaptive phenotypes to maintain a population's viability (e.g. [96]), then further adaptive evolution of plasticity is an essential step to either avoid extinction or permit establishment (modelled by Chevin & Lande [97]; see [98–100] on selective evolution of reaction norms). In invasive species, rapid plasticity evolution can promote subsequent spread into new habitats. For instance, cane toads (Rhinella marina) that have spread to colder regions in Australia have evolved higher metabolic plasticity [101]; invasive freshwater populations of the marine-native copepod Eurytemora affinis have evolved increased ion-transport plasticity [96]; and a shrubby South African Senecio introduced into Spain has evolved greater reproductive output in wet conditions without any loss of fitness in its ancestral dry habitat [102].

Expanded repertoires of adaptive plasticity can result from the environmental heterogeneity encountered within and among sites in the new range, rather than from a changed directional selection pressure compared to the native range [103]. This finding is consistent with theoretical predictions that increased adaptive plasticity is selectively favoured in populations and metapopulations that encounter variable environments ([68,86,104–108] and references therein). This particular selective property may lead to a positive evolutionary feedback for greater invasiveness in non-native taxa that have sufficiently broad norms of reaction to survive their initial introduction. Because successful non-natives tend to (i) colonize disturbed, variable habitats and (ii) have high dispersal capabilities that would cause them to encounter diverse sites, they may be especially likely to evolve greater adaptive plasticity post-introduction [109]. Evolution of increasingly generalist norms of reaction will in turn promote an even broader ecological distribution across habitats ([67] and references therein), expanding the invasion front and possibly creating more contiguous populations that will accelerate colonization of new sites [85,110]. Conversely, if endangered taxa initially have less-plastic norms of reaction that confine them to a narrow range of conditions, they may not encounter the environmental variability that promotes evolution of broader plasticity.

Although a cost of plasticity could in theory limit this type of evolution, evidence for such a cost is weak [107,111], and recent models assume that any plasticity costs are outweighed by benefits (e.g. [97]). Indeed, the previous examples show that increasingly broad adaptive norms of reaction can evolve, at least in certain taxa, without fitness trade-offs that would entail a loss of adaptive responses to ancestral environments. In any given case, the evolution of greater plasticity—like that of any adaptive trait—depends on whether or not genetic or developmental constraints are present that limit the potential for selective change. How widespread among invasive taxa is the potential to evolve ‘jack of all trades’ genotypes? Is the lag phase between introduction and invasion commonly characterized by the evolution of greater adaptive plasticity? These questions can be explored through resurrection experiments designed to compare environmental response patterns of genotypes sampled from introduced taxa across time (e.g. [103]).

As with any phenotypic trait, evolution of plasticity requires genetic variation—in this case, genetic differences in reaction norms measured as genotype × environment (G × E) variance in a statistical model ([67,68,112,113] and references therein). Like other aspects of genetic diversity, populations and taxa will differ in G × E variation due to previous mutation, drift and selection. A defining feature of G × E variation is that a given set of genotypes may express different phenotypes in certain environments, but converge on similar phenotypes in others: in other words, genetic variance itself differs from one environment to another [114,115]. A novel environment such as a new range or altered habitat can lead to rapid adaptive evolution if genotypes in a population express different phenotypes [116]. By contrast, evolutionary response to natural selection is buffered in a new environment where similar phenotypes occur [69,95,117]. Consequently, differences in patterns of G × E variation will influence the ability of populations to evolve new adaptive responses following introduction or in situ environmental challenges. In addition, a better understanding of the genetic architectures underlying reaction norms, such as the effects of modularity versus pleiotropy of regulatory pathways [118], will better inform models of constraint on plasticity evolution in natural populations [67,119].

In general, it is not known whether phenotypes in natural populations are more likely to differ or converge in predicted future environments such as high CO₂ and higher temperatures [120,121]. The question of adaptive evolutionary potential is of particular concern with respect to species that may face extinction in the absence of altered plasticity patterns. A case in point is reptiles with temperature-dependent sex determination, which are predicted to produce female-biased sex ratios in warmer future climates [122]. In a study of the leopard gecko Eublepharis macularius [123], populations were found to contain G × E variation for the proportion of males produced at different likely incubation temperatures. Such variation could fuel the selective evolution of new temperature thresholds for sex determination in future populations, promoting the species' persistence. Similarly, certain European populations of great tits (Parus major) contain genetic variation for temperature-based plasticity of reproductive timing, providing the potential for adaptive evolution of life-history plasticity in response to altered seasonal timing of food availability [124]. Examining norm of reaction diversity in populations of various organisms, including its expression under predicted future conditions, is a crucial step to assess (i) the potential for plasticity evolution to prevent extinction and (ii) the critical differences in evolutionary potential between endangered and invasive species.