Evolutionarily Significant Units

Biodiversity conservation efforts traditionally focus on maintaining one or more minimum viable populations (MVP) of a focal taxon that, in theory, should persist on ecological time scales (Shaffer 1981). However, the past decade has seen an increased emphasis on conserving population-level genetic diversity and the processes that promote adaptive evolution within species (Crandall et al. 2000, Moritz 2002). Given the intensifying nature of ongoing threats to biological diversity, the latter strategy is aimed at enhancing the capacity of species to adapt to future environmental conditions and thus to increase the probability of persistence over evolutionary time scales. Conserving adaptive potential may be especially important for insular populations that have limited ability to move in response to environmental changes (e.g., shifts in climate) and instead must adapt in situ.

The term “evolutionarily significant unit” (ESU; sometimes also termed “evolutionary significant unit”) was coined by Ryder (1986) to describe intraspecific taxonomic units worthy of conservation. The ESU concept is central to the development of population management strategies and the application of conservation legislation, particularly in determining “distinct population segments” (DPSs) as units of conservation for vertebrate species under the U.S. Endangered Species Act (Fay and Nammack 1996, Groom et al. 2005). The conservation genetics of Pacific salmon (Oncorhynchus spp.) provided important empirical examples for application of an ESU to natural populations under the Endangered Species Act (Waples 1991, 1995, CDFG 2002). However, the criteria for delineating ESUs have varied considerably over time (Crandall et al. 2000), and the interpretation and application of ESUs and DPSs has historically spurred scientific and public debate (Waples 1998). Though Ryder's (1986) original definition focused on adaptive differentiation, an increase in the availability of genetic data facilitated a movement toward ESU definitions that focus solely on the presence of genetic differentiation or reciprocal monophyly at neutral loci (Moritz 2002, Zink 2004). Moritz (2002) asserts that the use of reciprocal monophyly provides an unambiguous definition of an ESU and preserves the genetic diversity of irreplaceable, isolated lineages. In contrast, Crandall et al. (2000) argue that the original ESU definition put forward by Ryder (1986) is more conservation relevant and that adaptive variation and “ecological exchangeability” (i.e., the degree to which populations are adapted to the same ecological niche and are thus exchangeable) must be considered, not just neutral genetic divergence and reciprocal monophyly (Shimizu 2008). Both viewpoints focus on the conservation of genetic diversity, as does the federal requirement of genetic or morphological distinction for delineating distinct population segments (Fay and Nammack 1996), but each differ in their emphasis on adaptive versus neutral regions of the genome and the type of data required to demonstrate that a population qualifies as an conservation unit. Fraser and Bernatchez (2001) argue that instead of debating the relative merits of each definition, we should recognize that both definitions have strengths and weaknesses and we should apply the appropriate definition(s) on a case-by-case basis. Island flora and fauna represent a special case for determining which ESU concept to apply.

We argue that the most useful ESU definition for island populations is one that incorporates both adaptive variation and measures of neutral genetic divergence. Compared to mainland and other insular populations, island populations often show pronounced divergence at neutral loci because of restricted gene flow across oceanic barriers and the strong effects of genetic drift in small isolated populations (Patirana et al. 2002). Exceptions to this phenomenon include highly mobile taxa that do not perceive oceans as a hard barrier to dispersal (Postma and van Noordwijk 2005). Insular terrestrial populations of most taxonomic groups tend to be sedentary and thus likely to show reciprocal monophyly at neutral loci (e.g., mtDNA), even if the time since divergence was relatively recent (Neigel and Avise 1986, Walker et al. 2006). This necessitates separate conservation units on each island or between island and mainland sites under the Moritz (1994) ESU definition.

The ESU definition is especially important for the conservation of insular populations because these populations are frequently subject to intensive, population-specific management actions (e.g., Coonan et al. 2010). We suggest that using neutral genetic divergence alone for delineating ESUs on islands is insufficient because the method may overestimate the evolutionary significance of any island population relative to other island and mainland populations. The method may also impede consideration of translocations as a management strategy aimed at facilitating demographic or genetic rescue for small insular populations. Conservation efforts on islands should focus on identifying units of insular endemism that are adaptively differentiated from other mainland or insular populations and that show marked genetic divergence.

Delineating ESUs on Islands

Effective management of insular populations requires an objective framework for identifying and prioritizing intraspecific conservation units (Pullin and Stewart 2006). To assess how others have defined ESUs on islands, we conducted an ISI Web of Knowledge literature search in March 2012 using the following terms: [islands AND (evolutionarily significant unit OR evolutionary significant unit OR distinct population segment)]. The search returned 71 articles, 39 of which were deemed relevant (Appendix). We included only empirical studies of island taxa at the population or subspecies level and excluded review articles, perspective pieces, empirical studies of marine or human populations, and studies conducted on island taxa that had been described as insular endemic species (rather than populations or subspecies relative to the mainland). We considered studies only when an island ESU was based on a comparison to a mainland population or other surrounding island populations. Of the 39 studies, 45% used genetic distance as the sole criterion for delineating a conservation unit, and 55% based their designation on evidence of genetic and phenotypic differentiation (Fig. 1). Bottin et al. (2007) was the only study to demonstrate that the phenotypic traits unique to an insular population represented genetically based, adaptive differentiation, as opposed to phenotypically plastic differences caused by environmental effects. In the 39 studies, an ESU definition focusing on neutral genetic divergence was applied much more frequently for insular populations than was ESU frameworks that integrate data on neutral genetic and adaptive phenotypic differentiation. Moreover, most studies documenting genetic divergence at neutral loci showed genetic differentiation as significant differences in allele or haplotype frequencies, rather than demonstrating reciprocal monophyly (but see Kanthaswamy et al. 2006, Hoglund et al. 2011; Appendix).

Fig. 1.

Most studies of island taxa use genetic data alone for ESU designation. While some studies include data on morphological differences, there remains a paucity of studies that test whether these phenotypic differences are genetically based and adaptive prior to making conservation recommendations.

We suggest that management strategies on islands should transition to putting a greater emphasis on adaptive differentiation. We are particularly concerned with cases where populations are distributed on multiple adjacent islands or where populations are found on islands and nearby mainland locations. In such cases, insular populations are likely to exhibit neutral genetic divergence relatively quickly, even in the presence of some gene flow, because of initial founder effects and subsequent genetic drift.

We propose a modified version of the framework described by Crandall et al. (2000) for delineating ESUs on continental islands and on island archipelagos. Our framework is aligned with Crandall et al. (2000) in that popu lations must be genetically differentiated (rejecting genetic exchangeability). However, we put greater emphasis on the need to demonstrate that observed phenotypic differences are both genetically based and adaptive (i.e., increase fitness). We deemphasize the requirement that a population must have a unique ecological role that is nonexchangeable with sister populations on adjacent islands or the mainland. We also argue that even if island populations do not qualify as ESUs using our framework, they may still warrant some level of conservation prioritization, depending on the degree of neutral, phenotypic, and adaptive divergence from mainland or other island populations (see “Summary and Recommendations” and Table 1). The 3 components of this modified framework are (1) neutral genetic divergence, (2) phenotypic divergence, and (3) genetically based local adaptation; and we present the data required to test these criteria.

Neutral Genetic Divergence

Despite our a priori expectation that most insular populations should show some degree of genetic differentiation, we recognize the importance of neutral genetic data to confirm isolation of island populations. Therefore, as a first step in identifying insular endemic populations, we recommend comparing genetic divergence at neutral loci across island and mainland populations or between island populations. We also agree with the recommendation of Crandall et al. (2000) for measuring genetic differentiation on both historical and recent time scales. The magnitude of recent genetic differentiation is expected to vary among taxa due to differences in life history and dispersal abilities and thus should be considered with reference to the level of genetic differentiation between species, subspecies, and populations in other parts of the taxon's range (Oliva-Tejera et al. 2006).

Table 1.

Criteria for delineating evolutionarily significant units (ESUs) and conservation priority for insular populations based on neutral genetic, phenotypic, and adaptive divergence. Supporting evidence for divergence or designation as an ESU (yes), failure to find evidence for divergence or ESU (no), missing data (no data), and unresolved (?).

Traditionally, phylogeographic studies use mitochondrial, chloroplast, or nuclear DNA sequence data to infer deeper historical relationships among populations, whereas landscape genetic studies use microsatellite loci to quantify genetic structure on more recent time scales. Both approaches have been applied to delineate island ESUs (Appendix). Gene flow can also be estimated from measurements of genomic variation across populations based on single-nucleotide polymorphisms (SNPs), thereby allowing characterization of adaptive differentiation (Luikart et al. 2003, Hohenlohe et al. 2010, De Wit et al. 2012).

Phenotypic Divergence

The second component of our framework addresses variation in morphological, behavioral, or demographic traits among islands and between continental island and mainland populations. The first step in documenting adoptive differentiation is quantifying divergence in phenotypic traits that are easily measured, and this practice has been applied to several island studies to date (Appendix). Insular populations often exhibit the insular syndrome: a suite of morphological, life history, or be havioral traits that appear to be adaptive for island environments (e.g., Postma and van Noordwijk 2005). Divergence in traits used for courtship and mating might also reveal island populations that are in the early stages of speciation (West-Eberhard 1983, Coyne and Orr 2004). However, the mere detection of unique phenotypic traits is not sufficient for establishing whether adaptive differentiation has occurred. We recommend that these data be used to frame hypotheses and develop appropriate tests for local adaptation.

Local Adaptation

The third component of our framework requires testing whether island populations are adaptively differentiated from mainland or other island populations. Many studies include phenotypic divergence as supporting evidence for subspecific or ESU designation of island populations (Fig. 1). However, few test whether these differences are the products of adaptive evolution (Appendix), which testing requires that traits have a genetic basis and confer a fitness benefit in the local environment (Funk and Fa 2006). Experimental, genomic, and field studies can elucidate patterns of local adaptation. Below we describe several different approaches that can be applied to determine whether observed pheno typic differences of island populations are genetically based local adaptations with fitness consequences.

Identifying Adaptive Differentiation

Experimental studies.—Reciprocal transplant (RT) and common garden (CG) experiments are the most common approaches for investigating local adaptation (reviewed by Kawecki and Ebert 2004, Rader et al. 2005). These experimental designs are well suited for sessile or small organisms (e.g., plants, some insects, and small vertebrates; Losos et al. 2000) and are powerful for distinguishing evolved responses from environmentally mediated ones. A review of RT and CG experiments detected high levels of adaptive variation in marine taxa, but few of those studies were evaluated in an ESU framework (Conover et al. 2006), emphasizing the disconnect between studies of adaptive variation and conservation. RT and CG experiments can be challenging to conduct on vertebrates, especially species of conservation concern (Mittelbach et al. 1999, Ballentine and Greenberg 2010, Herczeg and Valimaki 2011, Svanback and Eklov 2011). In light of this limitation, we propose that in situations where RT and CG experiments are not feasible, population genomic and field studies be used to determine the genetic basis of phenotypic differences observed between island and mainland populations.

Population Genomic Studies.—Population genomics is a powerful approach for characterizing adaptive differentiation among island populations (Hudson 2008, Morozova and Marra 2008, Stapley et al. 2010). Sequencing SNPs can be used to identify adaptively divergent populations and to study reproductive isolation and incipient speciation processes (Nosil and Feder 2012), and this method has been validated for known divergent ecotypes (Lumley and Cusson 2013). One approach is to com pare genetic distances across thousands of loci to identify outlier loci, presumably under divergent selection, to characterize adaptive differentiation among populations (Luikart et al. 2003, Funk et al. 2012). Ideally, many loci of adaptive significance should be examined to quantify overall adaptive differentiation in response to the multiple dimensions of environmental variation. We advise against using few genes of known function because this does not characterize overall adaptive differentiation (Funk et al. 2012). Genomic data are rapidly becoming more affordable and easier to obtain, which will increase the importance of population genomics in conservation (Allendorf et al. 2010, Lumley and Cusson 2013). For example, one can conduct a population genomic study using RAD tag technology for $50 to $100 per individual (Allendorf et al. 2010), a cost that is less than many long-term monitoring programs.

Field studies.—Studies that measure selection or test whether divergent traits are adaptive in the field can help distinguish between local adaptation and phenotypic plasticity, especially in cases where funding is limited and RT and CG experiments are not feasible. Field research, particularly behavioral studies, need not be invasive or time consuming, contrary to assertions by some evolutionary biologists (e.g., Zink 2007). For example, Peluc et al. (2008) used a simple experimental design to demonstrate in one field season that the unique nest-site selection behavior exhibited by Orange-crowned Warblers (Oreothlypis celata) on Santa Catalina Island reflected the birds' ability to respond plastically to variation in the nest-predator community, rather than a genetically based phenotypic differentiation. Quantifying patterns of morphological and life history variation among populations typically requires more intensive, long-term study (e.g., Radar et al. 2005). For example, selection gradient analysis of marked individuals can help identify whether divergent traits are under selection (Lande and Arnold 1983). Closely related species could act as proxies for an insular population that is not amenable to experimental studies. (Friesen et al. 2006, Bottin et al. 2007).

Applying the Framework

Combining data on genetic and phenotypic divergence with considerations of local adaptation will make it possible to set appropriate priorities for conservation management on islands. Such an approach may require more effort in determining ESUs but could improve efficiency in delineating ESUs, particularly as population genomics become more feasible and cost effective. We will also gain a deeper understanding of the processes that drive adaptive genetic differentiation by explicitly considering and quantifying the continuum of genetic divergence that is expected for island populations connected by varying levels of gene flow and exposed to varying levels of divergent selection.

We make specific recommendations in Table 1 for ranking populations based on the degree of neutral, phenotypic, and adaptive differentiation compared to other island and mainland populations. We define an ESU as a population that shows both genetic divergence at neutral loci and adaptive, genetically based phenotypic differences. Assuming that the demographic characteristics among populations are equal, we place the highest conservation priority on populations where evidence exists for adaptive genetic differentiation, even when tests for the genetic basis of phenotypic divergence are pending. Mid-priority is assigned to populations that are genetically isolated and that exhibit phenotypically plastic differentiation from neighboring populations. Populations that are divergent at neutral loci but have no phenotypic or adaptive differences are lower priority. Populations that show no detectable differentiation at neutral genetic loci (even with sufficiently variable loci) but exhibit phenotypic differences (e.g., Ballentine and Greenberg 2010) should have some conservation value. Populations that are indistinguishable both genetically and phenotypically are not considered a distinct ESU and are the lowest conservation priority. We suggest that this ranking system be used to identify island populations that are most deserving of limited economic resources for bio diversity conservation efforts. Nonetheless, other factors must also be considered when making decisions about conservation prioritization, including population status, our ability to influence population viability through management actions, management objectives of different land owners, and socioeconomic factors.

Beyond prioritization, data on neutral versus adaptive genetic differentiation can also inform the management of populations that have been identified as a conservation priority. For instance, insular populations are particularly susceptible to the demographic and genetic consequences of small population sizes, especially when anthropogenic stressors reduce numbers below historic levels. Translocation of individuals from neighboring populations is one management strategy to consider in those situations. This strategy has the potential to ameliorate demographic stochasticity or inbreeding depression but could also lead to outbreeding depression if the translocated individuals originate from an adaptively divergent population. Outbreeding depression is most likely when populations are adaptively divergent and least likely when they only differ at neutral loci (Frankham et al. 2011). The benefit of demographic or genetic rescue may outweigh the potential cost of outbreeding depression for a population undergoing severe declines, but information on the degree of adaptive population divergence could still be valuable for management.

We recognize that identifying conservation units at the intraspecific level can be challenging with real-world data and pressing management concerns. The degree of population-level divergence can fall along a spectrum that ranges from slight divergence at neutral loci, to adaptive phenotypic divergence, to incipient speciation. These gradations in divergence should be translated into priority management categories. However, insular populations require a particularly strong emphasis on identifying and understanding patterns of adaptive divergence, not just neutral genetic divergence.

Loggerhead Shrike

San Clemente Island is home to an endemic population of Loggerhead Shrike (Lanius ludovicianus mearnsi) that was driven nearly extinct by habitat destruction due to livestock grazing and other anthropogenic land-use disturbances. This population was listed under the U.S. Endangered Species Act (ESA) in 1977 (USFWS 1977), and the population size dropped to an estimated low of 14 individuals in 1998 (USFWS 2009). Since then, an intensive and costly management program ($25 million from 1993 to 2008; DOD 2010) has involved invasive species removal, manage ment of shrike breeding habitat, captive breeding, and the removal or eradication of 5 species of native shrike predators (Elliot and Popper 1999). The population increased to nearly 200 individuals by 2009 (including juveniles and nonbreeding adults) but remains listed under the ESA due to small population size and other ongoing threats (USFWS 2009).

The management actions used to conserve shrikes on San Clemente Island have been controversial, in part because of their cost and their impact on other native vertebrate species, including the endemic island fox (Urocyon littoralis, Roemer and Wayne 2003), and the debate over the population's genetic distinctiveness. Lanius l. mearnsi was described by Ridgeway (1903) and later Miller (1931) based on plumage and morphological characters. Subsequent genetic studies concluded that the San Clemente shrikes are genetically distinct from populations on the mainland and on the northern Channel Islands (Mundy et al. 1997, Eggert et al. 2004, Caballero and Ashley 2011). However, genetic analyses of museum specimens collected in the late 1800s and early 1900s have detected a decline in genetic diversity over time, and those specimens have a different genetic composition than the postmanagement birds sampled after 1990. Moreover, the postmanagement population of L. l. mearnsi is more genetically distinct from the mainland and other islands than was the historical population. Thus, the apparent genetic distinctiveness of L. l. mearnsi may reflect, to a certain extent, the effects of genetic drift during the extreme population bottleneck of the past century, as well as the effects of population management since 1990—a situation that may warrant very different management strategies than would be the case for a population with strong historical isolation.

Two questions arise regarding future conservation of shrikes on San Clemente Island: (1) whether continued investment in intensive management is still justified and (2) whether the management strategy for L. l. mearnsi should involve genetic rescue via shrikes from other populations. These 2 questions invoke both the logistical and philosophical complications surrounding conservation management. Our framework provides an objective template to help with these decisions. San Clemente Island shrikes would be recognized as “higher” conservation priority (one level below “highest”; Table 1) but not designated as an ESU (Table 1) based on currently available data (evidence of both genetic and phenotypic divergence). For this population to be designated formally as an ESU, further studies of adaptive differentiation would be needed to show that shrikes on San Clemente Island have evolved genetically based local adaptations and have diverged from the mainland population and other islands in the archipelago. The decision about whether to pursue genetic rescue also relies on understanding whether or not the source and receiving populations share genetically based adaptive variation. The introduction of alleles from another island via captive breeding or translocation could result in maladapted offspring if the source population was locally adapted to divergent environmental conditions. Studies of adaptive variation in Channel Island shrikes would provide these answers.

Island Fox

Consistent with the insular syndrome (Lomolino 2005), the island fox has evolved a much smaller body size compared to its mainland ancestor, the gray fox (Urocyon cinereoargenteus; Collins 1993). Island foxes are found on 6 of the 8 Channel Islands. Subspecies status was granted to foxes on each of those islands based on interisland differences in morphology and neutral genetic structure (Gilbert et al. 1990, Wayne et al. 1991, Collins 1993). However, we do not know whether the morphological differences observed between islands (Collins 1993) are a product of local adaptation (e.g., to different climate regimes or different community composition), genetic drift in small founder populations, or phenotypic plasticity.

In the early 2000s, these subspecific designations were central to management decisions after catastrophic population declines of foxes on 4 of the Channel Islands. Captive breeding programs were established for each of the affected subspecies on their respective islands, and each was listed as endangered under the ESA in 2004 (USFWS 2004). Additional intensive management actions included a vaccination program for canine distemper, the removal of Golden Eagles (Aquila chrysaetos) from the northern Channel Islands, and the reintroduction of Bald Eagles (Haliaeetus leucocephalus), which through agonistic interactions may deter A. chrysaetos (Morrison 2008). These management programs ultimately proved successful. Fox populations have rebounded on all 4 of the affected islands, although population monitoring and vaccination efforts continue (Coonan et al. 2010).

The island fox example provides the opportunity to evaluate whether, given the same data, the previous designations of island populations (based on Gilbert et al. 1990) and our framework reach the same conclusions regarding the conservation priority of island foxes. As with the San Clemente Island shrikes, our framework would not have identified each island fox population as an ESU due to lack of evidence of adaptive differentiation but would have classified the islands as “higher” priority, owing to evidence of genetic and phenotypic divergence. Resource managers followed the precautionary principle, given the limited data on local adaptation in the island fox populations. Going forward, however, limited resources for conservation management and new knowledge may call for a different approach.

The island fox is likely a “conservationreliant” species (Scott et al. 2005) that will require continued monitoring and active management for long-term persistence (Coonan et al. 2010). All 6 U. littoralis populations have low levels of genetic variation (Wayne et al. 1991) and thus may lack the adaptive variation to survive future environmental changes (Allendorf and Luikart 2007). Although we currently have no evidence of inbreeding depression on any of the islands, managers have recognized that a future conservation option may include translocations between islands, with the goal of genetically rescuing fox populations (Coonan et al. 2010). If this is deemed a potentially necessary strategy, then knowledge of the degree of adaptive population divergence between islands will be critical for developing an effective translocation strategy and for understanding the potential risks of outbreeding depression.