Population-level genetic variation and climate change in a biodiversity hotspot

Abstract

Introduction Estimated future climate scenarios can be used to predict where hotspots of endemism may occur over the next century, but life history, ecological and genetic traits will be important in informing the varying responses within myriad taxa. Essential to predicting the consequences of climate change to individual species will be an understanding of the factors that drive genetic structure within and among populations. Here, I review the factors that influence the genetic structure of plant species in California, but are applicable elsewhere; existing levels of genetic variation, life history and ecological characteristics will affect the ability of an individual taxon to persist in the presence of anthropogenic change.

Factors influencing the distribution of genetic variation Persistence in the face of climate change is likely determined by life history characteristics: dispersal ability, generation time, reproductive ability, degree of habitat specialization, plant–insect interactions, existing genetic diversity and availability of habitat or migration corridors. Existing levels of genetic diversity in plant populations vary based on a number of evolutionary scenarios that include endemism, expansion since the last glacial maximum, breeding system and current range sizes.

Regional priorities and examples A number of well-documented examples are provided from the California Floristic Province. Some predictions can be made for the responses of plant taxa to rapid environmental changes based on geographic position, evolutionary history, existing genetic variation, and ecological amplitude.

Conclusions, Solutions and Recommendations The prediction of how species will respond to climate change will require a synthesis drawing from population genetics, geography, palaeontology and ecology. The important integration of the historical factors that have shaped the distribution and existing genetic structure of California’s plant taxa will enable us to predict and prioritize the conservation of species and areas most likely to be impacted by rapid climate change, human disturbance and invasive species.

INTRODUCTION

The extent to which high levels of genetic variation within and among populations are associated with the ability of a species to persist in the face of both short- and long-term disturbance regimes has long been deliberated (Lowe and Allendorf, 2010; Davies et al., 2016). Understanding the distribution of genetic variation on the landscape and the life history or ecological parameters that influence this variation will be important in identifying the taxa most vulnerable to anthropogenic change (Peter and Slatkin, 2013; Brandvain et al., 2014). High levels of heterozygosity have been repeatedly shown to confer resistance to environmental change (Hanski et al., 2006; Willi et al., 2006; Bonin et al., 2007). Model simulations suggest that species with low heterozygosity but large ranges will suffer a spatial redistribution of heterozygosity, depending on the level and length of the disturbance, whereas taxa with limited distributions with severe and frequent disturbance will suffer dramatic declines in heterozygosity regardless of initial levels of heterozygosity (Davies et al., 2016).

Here I focus primarily on molecularly-based population genetic structure and the life history characteristics, historical landscapes and climatic events that have shaped, and will shape, the distribution of genetic variation within and among California plant populations. California, specifically the California Floristic Province (CFP), is a recognized biodiversity hotspot (Myers et al., 2000) with high levels of endemism (Loarie et al., 2008; Kraft et al., 2010; Burge et al., 2016) that has been, and will continue to be, impacted by anthropogenic change. Various models have predicted different outcomes in climatic shifts and at this juncture, because it is difficult to predict how whole plant communities may respond, it will be important to focus on the adaptive ability of individual species and conserving regions that contain high levels of diversity. Genetic diversity as measured by neutral markers does not necessarily provide a measure of adaptive ability, but the study of neutral genetic markers can be useful in the interpretation of past landscapes, refugia and gene flow (Holderegger et al., 2006). An examination of fitness and heterozygosity via a meta-analysis of multiple datasets substantiates that a loss of heterozygosity as measured by neutral markers does indeed reflect a negative impact on fitness (Reed and Frankham, 2003). However, genes under selection are preferred for interpreting how populations may respond to climate changes as neutral genetic markers can lead to incorrect interpretations, particularly in highly fragmented populations (Hoffman and Willi, 2008). High levels of genetic diversity within a taxon dispersed across a diverse array of climate zones and habitat types should confer resilience to varying temperatures. Taxa with low or fragmented patterns of genetic diversity, limited geographic ranges or high habitat specificity are those most in danger of population extirpation or extinction loss due to anthropogenic change. Based on a review of the distribution of genetic variation of California plant taxa, examples of existing and potential refugia under climate change are also identified and discussed. Although the examples here are drawn primarily from the California Floristic Province, the conclusions and recommendations are broadly applicable, particularly in areas with high topographic diversity.

Molecularly-based population genetic data have been very useful in determining past historical events and will be useful in estimating the ability of taxa to respond to contemporary climate change. Extant species that have survived via glacial refugia are likely to show genetic signatures of within-refugium genetic drift and selection or, conversely, harbour historical genetic diversity not found throughout the range of a taxon; however, combinations of these scenarios may also occur. Even where phenotypic variation is present, measuring genetic variability due to recent divergence can be a problem, particularly in places like California, where speciation and evolutionary diversification have occurred recently (Boykin et al., 2005; Burge et al., 2011). Detection of patterns of recent genetic divergence now can be achieved, however, through the use of markers more suitable for measuring population-level variation, such as microsatellites, restriction site-associated DNA markers (RADseq) (Davey and Blaxter 2010) and whole-genome sequences (Brandvain et al., 2014).

Varying levels of diversity among refugia can help infer historical ecological patterns (Gugger et al., 2010) and may be used to identify those areas most valuable for conservation. Depending on the temporal scale, dispersal events, bottlenecks followed by population growth and gene flow among populations will all leave signatures of genetic variation (Slatkin and Hudson, 1991). The efficacy of contemporary refugia in conserving taxa will depend on the capacity of refugia to support enough genetic diversity for population viability and to maintain evolutionary processes (Ovaskainen, 2002). The topographical and latitudinal variation present in California will require the regional identification of areas that are spatially appropriate in size and proximity to the current ranges of the maximum number of species. Those areas that have been identified as containing high levels of species diversity and high genetic diversity within and among populations should be prioritized for conservation.

Connectivity associated with gene flow is further influenced by population demographics that contribute to patterns of genetic variation on physical landscapes; however, evolutionary genetic models more strongly support dynamics among populations than demographic models in Clarkia xantiana spp. xantiana (Moeller et al., 2011; Davies et al., 2016). Population differentiation is easier to resolve in species with low rates of dispersal due to high genetic structuring (high F_ST values), whereas high levels of gene flow (high N_m) can increase the difficulty of interpretation and will be reflected by low F_ST values. Divergent taxa residing within the same region with similar ecological requirements, life histories and behaviours often share similar patterns of genetic diversity (Bermingham and Moritz, 1998). However, loci under selection can alter expected patterns based on demographic patterns or the historical distribution of a population. Thus, precise identification of refugia and vicariant events requires data from a number of species, as differences in life history characteristics and ecological requirements affect lineages differently in time and space. Ultimately, understanding the genetic structures of populations and how they have been influenced by vicariant and dispersal events is important for understanding the potential of fragmented populations to adapt to anthropogenic change. The recognition of regional patterns of genetic diversity offers the opportunity not only to conserve regions with the highest biological diversity but also to conserve evolutionary processes in the face of climate change.

Range shifts of species in response to climate change are a serious concern (Warren et al., 2001; Zacherl et al., 2003). Mean annual temperatures in California increased by 1 °C between 1950 and 2000 (LaDochy et al., 2007) and are projected to rise 3·8–5·8 °C by 2100 under high emission scenarios by 2050 (Cayan et al., 2008). Vulnerability to climate change cannot be predicted by geographic range and habitat specificity, based on a lack of correlation of Climate Change Vulnerability Index scores and rarity type (Anacker et al., 2013); however, geographic range, connectivity and habitat specificity all influence genetic variation. The velocity of temperature change is estimated globally to be 0·42 km year⁻¹ under some IPCC scenarios, a rate that most plants cannot meet, with the highest velocities in flooded grasslands and deserts and the lowest in temperate montane regions (Loarie et al., 2008). An argument has been made to prioritize conservation in regions with high levels of evolutionary diversity to capitalize on genetic and functional diversity (Cadotte and Davies, 2010). Measures of evolutionary diversity need to be developed to maximize ecosystem health and minimize losses to biological diversity. Unfortunately, genetic diversity is often ignored in climate change decision-making (Cadotte and Davies, 2010). Montane areas will likely provide refuge for many species and the creation of large desert reserves may reduce losses of xeric taxa (Loarie et al., 2008). Small reserves are problematic in that they provide only temporary protection. The question remains as to how to mitigate for the loss of habitat in California for 4844 native plant species (Baldwin et al., 2012) in over 435 vegetation alliances (Sawyer et al., 2009), in a rapidly changing climate on a landscape constantly disturbed by humans. Whether we can effectively estimate the ability of a species to migrate based on past migrational events further complicated by human landscape disturbance remains to be seen. Regardless, a combination of ecological and genetic data will provide the best foundation for conserving the ability of the California biota to respond to rapid environmental change (Eckert, 2011; Gugger et al., 2011). In the following sections I review the factors that have shaped the extant genetic structure of plant species in California and argue that existing levels of genetic variation, combined with life history and ecological characteristics, affect the ability of an individual taxon to persist in the presence of anthropogenic change. Of particular importance are dispersal ability, generation time, reproductive ability, degree of habitat specialization, plant–insect interactions and the availability of habitat or migration corridors.

FACTORS INFLUENCING THE DISTRIBUTION OF GENETIC VARIATION

Habitat specificity

Little is known about the interaction of range shifts, habitat specificity and genetic diversity. Dispersal and habitat limitation studies in Mimulus leptaleus and Mimuluslaciniatus in the California Sierra Nevada found that habitat limitation was more important than dispersal ability in limiting range expansion (Sexton and Dickman, 2016). However, in some cases there may be a lack of overlap between a species’ future latitudinal or elevational range and the habitat needed to support it (Bellard et al., 2012). Genetic structuring among populations of serpentine endemics can be significant and, as with many edaphic endemics, the level of genetic variation contained within each outcrop or island can have important conservation implications (Wolf et al., 2000). If there is little variation contained in a small population and the population becomes isolated due to the extirpation of adjacent populations, it may succumb to processes such as genetic drift and inbreeding depression. In a resource-limited world, populations that are genetically depauperate and spatially isolated may not be compelling for other than ex situ conservation. In contrast, populations with high levels of variation in both small and large populations are more viable for conservation (Lesica and Allendorf, 1995).

The California Coast Ranges were formed from a variety of geological substrates, which further complicates the inference of the evolutionary history of many California plant taxa. The late Cenozoic formation of much of the Coast Ranges has resulted in low levels of genetic differentiation among plant species, as found in Ceanothus (Burge et al., 2011) and Collinsia (Baldwin et al., 2011). Despite this recent evolutionary divergence, among-population genetic variation can be high within these species if they are endemic to specific habitats. Increasingly dry summer periods in California beginning about 15 million years ago (MYA) (Baldwin and Sanderson, 1998) may have resulted in intense selection via the restriction of some species populations to the fine-grained, high moisture capacity serpentine soils which often contain seeps (Raven and Axelrod, 1978). Without experimentation for adaptive variation, it will be difficult to prioritize the conservation of populations. For example, 32 Coast Range populations of Calystegia collina, a serpentine endemic, were found to have low gene flow among populations in Napa, Lake and Sonoma counties. Gene flow is thought to be primarily mediated by pollinators, which, should they decline, will result in even lower gene flow (Preston and Dempster, 2012). The maintenance of gene flow and genetic diversity will be important in maximizing the ability of populations to respond to evolutionary processes.

Plant–insect interactions

Insects are important conduits of gene flow in many plant species, maintaining within- and among-population levels of diversity. Insects are particularly susceptible to temperature and the availability of food resources, primarily the plants with which they evolved (Pelini et al., 2009). However, because insects and plants disperse and reproduce at varying rates their respective responses to climate change could become a problem for either group (Schweiger et al., 2008). Butterfly species have demonstrated migration to higher elevations or latitudes of 240 km/30 years (Parmesan et al., 1999), whereas tree migration responds to climate change at 20–40 km/100 years (Davis and Shaw, 2001). The context of the plant–animal interaction also needs to be taken into consideration, specifically the strength of the relationship or the availability of alternative host plants or pollinators. Differential phenological shifts among plants and their pollinators decrease the diet breadth of pollinators via the loss of floral resources and are predicted to result in increased inbreeding and eventual population declines in the plant hosts (Memmott et al., 2007). For example, the butterfly species Erynnis propertius, an important generalist pollinator, occurs on the Pacific coast from Baja California to British Columbia, where the host plants, species of Quercus, occur (Guppy and Shepard, 2001; Pelini et al., 2009). Southern and central California populations of E. propertius feed on Quercusagrifolia and higher-elevation populations feed on Quercuskelloggii. Host plant experiments have found that E. propertius have zero survival on other Quercus species. Southwestern Oregon populations are dependent on Quercusgarryana; however, overlaps between Q. agrifolia and Q. garryana are very limited. If E. propertius has reduced fitness on alternative host plants and the host plants migrate at much slower rates than E. propertius, its survival could be prevented if it is unable to switch hosts (Pelini et al., 2009). The loss of generalist pollinators such as E. propertius at a large scale has the potential to be devastating to plant taxa dependent on outcrossing. Specialist pollinators are expected to be the most vulnerable to extinction when plant–pollinator phenologies are decoupled, but further experimentation is necessary to make more accurate predictions (Forrest, 2015).

Comparative phylogeographic studies among Yucca brevifolia, two obligate pollinators (Tegeticula antithetica and Tegeticulasynthetica) and two species of parasitizing yucca moths (Prodoxus weethumpi and Prodoxussordidus) provides evidence for the effects of climate change on species distributions and genetic structure (Smith et al., 2011). Coalescent-based DNA sequence analyses and distribution modelling of Y. brevifolia trees since the Last Glacial Maximum (LGM) indicate that demographic changes in all species were present prior to climatic change during the Holocene (Smith et al., 2011). For the insect species some divergence is apparent, with average F_ST values of 0·12; however, Y. brevifolia variation is quite different, with high global F_ST values of 0·87, which strongly correlate with geographic distance, which is structured by pollinator association. The differences in F_ST values between Y. brevifolia and their associated insects likely reflect differences in dispersal ability. Demographic and genetic analyses indicate that all of the species had concerted population growth in the late Pleistocene beginning about 0·2 MYA, were little affected by the LGM, and reached their current distribution about 50 thousand years ago (KYA). These populations are projected to contract with increased CO₂ levels and increased temperatures; however, some new habitats may become available (Dole et al., 2003). Decoupling plant–insect interactions via phenological changes or changes in distribution has the potential to become severe enough to result in extinction events of insect and plant taxa or both (Gilman et al., 2012).

Connectivity and disturbance

Populations residing in refugia from the Pleistocene glaciation cycles harbour much genetic diversity and are often found in cooler and more northern microclimates. However, low latitudinal or elevational portions and disjunct populations of a species’ range can also be important reservoirs of adaptive genetic variation (Hampe and Petit, 2005). Although these latter populations may not be currently connected, they likely persisted under both cold and warm conditions and are important sources of genetic distinctiveness, i.e. they may contain genetic diversity not found in other populations.

Increased human settlement can result in less connectivity and increases in regional genetic structure via increased population differentiation. Due to rapid environmental change, this increase in population differentiation is likely to be based on random effects such as genetic drift rather than selection due to sudden declines in population size. The resulting non-adaptive variation will lead to greater susceptibility of a population to further environmental change. The disruption of previously contiguous habitats can result in inbreeding depression and the artificial introduction of genotypes from other locations may be maladaptive (Ellstrand and Elam, 1993; Elam et al., 1998). Alternatively, artificially introduced gene flow can have negative consequences due to differential local adaptations and lead to outbreeding depression (Templeton, 1986; Sexton et al., 2011; Halbur et al., 2014). Generally, if gene flow levels have been historically high, the introduction from similar populations is unlikely to be harmful (Elam et al., 1998). Additionally, gene flow can be beneficial where range limits among populations are similar, and for populations at risk gene flow may be recommended as a conservation option (Sexton et al., 2011).

The extreme loss of mesic and wetland habitats throughout California, but particularly in the Central Valley, has isolated populations and removed opportunities for adaptive radiation and gene flow. For example, formerly widespread but habitat-specific taxa such as those that occur in vernal pools would have historically experienced occasional recolonization events due to periodic widespread flooding that would have resulted in gene flow among existing populations. Chloropyron palmatum, a federal endangered species and vernal pool endemic, is found only in five genetically distinct populations (F_ST = 0·23) in the Central Valley. Seed dispersal historically occurred via flooding, which, due to development, now rarely occurs and is thought to have reduced gene flow among populations (Ayres et al., 2015). Similarly, Allium munzii, which has suffered 80–90 % loss of its coastal sage scrub habitat, primarily due to urbanization, has very strong genetic structure and low gene flow among the remaining 18 small, isolated populations (Mashayekhi and Columbus, 2015). Species such as C. palmatum and A. munzii are unlikely to survive climate change without ex situ conservation. Unfortunately, assisted migration as a remedy for climate change has both risks and benefits, and translocating genotypes requires caution (Frankham, 2010; Vitt et al., 2010).

Climate alone does not predict the distribution of plant species, and even at large scales there may be other factors that contribute to the range of a species (Serra-Diaz et al., 2013). In addition to biotic interactions and traits such as dispersal ability, disturbance can be a major factor, changing light and soil conditions and influencing the demographics of a population (Serra-Diaz et al., 2015). Interacting with climate change, disturbance regimes and land management can hinder or benefit recruitment (Boulangeat et al., 2014). Diverse, intact landscapes, particularly in mountainous areas, may provide suitable habitat for many taxa, due to varying microrefugial climatic conditions, such as found on north-facing slopes (Vanderwel and Purves, 2014). In general, human-caused disturbances such as increased fire frequency will likely influence species distributions more than climate change, but understanding the interaction between disturbance and climate change will be important for the persistence of not just individual species but also of ecosystems (Penman et al., 2014; Syphard and Keeley, 2015).

Human activities, such as introducing non-native species and disturbing, fragmenting or otherwise degrading natural ecosystems, promote hybridization of previously allopatric congeners (Vilà et al., 2000). When introgression accompanies hybridization, dilution of the native gene pool can occur (Abbott, 1992; Johnson JR et al., 2010; Johnson MG et al., 2016), and can result in extinction of the native species (Anttila et al., 1998). Often underrated, human-mediated hybridization is one of the leading causes of biodiversity loss via homogenization of formerly distinct populations or taxa (Muhlfeld et al., 2009). Through introgressive hybridization, populations become less differentiated, which can limit evolutionary flexibility as locally adapted genes become lost along with the populations that once contained them (Rieseberg, 1991; Ellstrand and Elam, 1993). Conversely, hybridization between previously isolated populations may, in some cases, confer genetic resilience via new gene combinations that may be necessary to survive climate change. For example, populations isolated due to human disturbance may benefit from hybridization from populations in more southern latitudes.

Dispersal ability

Life history traits strongly influence dispersal distance (Stevens et al., 2012; Tamme et al., 2014) and consequently influence the distribution of genetic variation on the landscape. Seed and fruit mass and morphology will greatly influence dispersal distance depending on the vector (Thomson et al., 2010), but plant height has generally been found to be more important than seed mass in predicting dispersal distance (Moles et al., 2004; Tamme et al., 2014). Wind- and vertebrate-dispersed species, not surprisingly, disperse further than insect-dispersed species and generally have higher levels of gene flow and genetic similarity among populations (Hamrick et al., 1979), conferring some resilience to climate change depending on habitat availability. It is important to consider, however, how heterogeneous environments also influence dispersal ability. The highest levels of genetic diversity of Notholithocarpus densiflorus (four cpDNA haplotypes) are in southern Oregon, the Klamath Ranges and the northern Sierra Nevada, with one haplotype in the Coast Ranges from Humboldt to Santa Barbara counties (Nettel et al., 2009). The distribution of these haplotypes is likely a reflection of a complex terrain combined with heavy fruits that limit seed dispersal. In contrast to the cpDNA data, the nuclear genome reflects coastal groupings south of San Francisco and in an interior northern Sierra Nevada, Klamath and Southern Oregon group, which indicates that genetic mixing is more effective through pollen movement than through seed (Nettel et al., 2009). The occurrence of this species in complex topography combined with genetic mixing via wind pollination may facilitate its survival under climate change.

Species lifespan and reproductive rates

Species that have long lives have a different vulnerability to extinction than species that have short lives. Regardless of the type of lifespan, critical to survival and recolonization is the availability of migration corridors and source populations at a number of spatial scales. Tree species are long-lived and often have high effective population sizes and high rates of reproduction; along with delayed seed production, they often show genetic signatures of expansion and contraction and lag behind climate change. Other considerations affecting response to climate include a number of factors that influence reproduction rate, such as demographics within a population, variation of seed production within the range, sensitivity of germination to climate and biotic factors, and the availability of suitable microsites across the range of the species (Kroiss and Hillerislambers, 2015). Testing the hypothesis that climate change affects reproductive outcomes, seedling to mature tree size classes were used as a surrogate for age distribution for 46 tree species from northern California through the Pacific Northwest. Seedlings of 33 species occurred in ranges with a mean annual temperature of 0·120  °C lower than the mature trees, a difference consistent with climate change (Monleon and Lintz, 2015). Interestingly, Pinus lambertiana, Pinus jeffreyi, Calocedrus decurrens and Abies concolor, common components of montane regions of California, showed statistically significant shifts towards warmer areas.

Short-lived taxa are expected to respond more rapidly to climate change than the long-lived tree species. Encelia farinosa is a short-lived perennial that occurs in the Sonoran, Mojave and Peninsular Deserts and shows genetic signatures that reflect migration from refugial populations (Fehlberg and Ranker, 2009). A sample of cpDNA sequences reveals that 80·64 % of the variation is explained by differences within locations, consistent with past fragmentation and range expansion. Fossil records from middens for at least 13·8 KYA suggest that Encelia expanded west and north following the LGM. In general, short-lived species with effective regional dispersal, such as E. farinosa, will likely survive in microclimatic areas in adjoining mountainous regions in the face of rapid environmental change.

Two short-lived desert species, Cryptantha flava and Carrichtera annua, were found to be able to buffer stochastic changes under climate change due to elasticity in seed bank emergence (Salguero-Gómez et al., 2012), and this is consistent with model predictions (Koons et al., 2009). Long-lived species may have lower annual reproductive rates but their capacity to withstand incremental changes to their environment may be buffered by wider demographic reaction norms, i.e. greater phenotypic plasticity within different demographic stages (Morris et al., 2008; Koons et al., 2009; Chevin et al., 2013). A number of taxa indicate a non-linear, species-specific response to climate change that will require experimental assessment on an individual basis (Morris et al., 2008; Doak and Morris, 2010).

Existing genetic diversity

There are extreme examples of small isolated populations with low levels of diversity within and among populations (Syring et al., 2007). The long-lived Pinus torreyana, known from two populations with about 9000 individuals, has very low levels of genetic diversity. An analysis of 59 isozyme loci identified all as homozygous, suggested to be a result of reduction to less than 50 individuals during the xerothermic period 8·5–3·5 KYA (Ledig and Conkle, 1983). Pinus balfouriana occurs at 1830–3400 m in the southern Sierra Nevada and the Klamath Ranges, separated by 500 km. There is some morphological divergence but little allozyme divergence among the populations (Mastrogiuseppe and Mastrogiuseppe, 1980; Oline et al., 2000). An analysis of cpDNA, mtDNA and nucDNA genes or regions indicates divergence during the Sherwin glacial maximum event during the Pleistocene, extirpating populations in the central and northern Sierra Nevada, but the genetic data are not consistent with the Holocene xerotherm (Eckert and Carstens, 2008). Pinus balfouriana shows greater genetic diversity than other range-restricted California conifers and, although at great risk in the White Mountains due to climate change, it should survive in some populations in the Rocky Mountains. Another long-lived species, Pinus longaeva, once widespread throughout the Great Basin during the LGM, is now limited to alpine habitat, and the White Mountains represents the western extent of the species distribution. Despite the fact that they grow in isolated alpine habitats, they retain high levels of genetic diversity because of the life history characteristics of longevity, large population size and high seed set, and no evidence of genetic bottlenecks (Lee et al., 2002). However, due to their limited number of populations in a habitat that is predicted not to exist in the next century, their survival in situ is dubious (Dirnböck et al., 2010; Chen et al., 2011).

Hesperocyparis macrocarpa consists of small, isolated populations but has fairly high levels of genetic diversity, indicating a former historic, widespread distribution (Kafton, 1976). Although widely planted as an ornamental, the remaining native populations, although protected, are in urban areas along the coast, which are prone to human disturbance. Further placing this species at risk in situ will be the reduction of coastal fog projected to occur as a result of climate change (Farjon, 2013). Hesperocyparis forbesii, an endemic in the southern California Peninsular Ranges and northwest Baja California, has widely separated native populations; however, genetic diversity as measured by allozymes was not associated with geographic distance (Truesdale and McClenagan, 1998). A close relative of H. forbesii, Hesperocyparisguadalupensis, is found only on Guadalupe Island off the Pacific coast of Baja California. Paternally inherited cpDNA markers compared genetic diversity among populations of H. forbesii and H. guadalupensis and identified higher genetic diversity in H. guadalupensis than in its mainland congener. Contrary to expectations, H. guadalupensis shows no evidence of a genetic bottleneck and may reflect the environmental stability of the island habitat over a long evolutionary history. In contrast, other species support a pattern indicating that the mainland and more northern Peninsular Ranges experienced glacial and interglacial periods during the Pleistocene, which may have led to less predictable interactions within and among populations (Escobar García et al., 2012).

REGIONAL PRIORITIES AND EXAMPLES

It would be ideal to conserve as much ecological, species and genetic diversity as possible; however, climate change and the addition of 14 million people to California’s population by 2060 (State of California, 2014) will result in significant losses. Conservation priorities may have to be focused on areas that contain the greatest number of endemic species or high biodiversity, e.g. the central Coast Ranges, the Klamath–Siskiyou region, the eastern Transverse Ranges, the Tehachapi Mountains, the Santa Lucia Mountains and the eastern Mojave Desert. Areas with high levels of endemism that include species with the greatest range restrictions are on the central coast, in the Sierra Nevada and in the San Bernardino Mountains (Loarie et al., 2008; Kraft et al., 2010). The Mojave Desert and Great Basin have the highest level of neoendemics, or species that are range-restricted, in part because of their recent evolutionary emergence. Further complicating the diversity of California, however, is the presence of a number of palaeoendemic taxa, or those that were once widespread and are now range-restricted. Issues to consider in conservation include local, regional and taxonomic rarity, ecological function and threats to biodiversity at multiple levels.

Klamath–Siskiyou region

The International Union for Conservation of Nature (IUCN) has identified the Klamath–Siskiyou ecoregion as an ‘area of global botanical significance’ (DellaSala et al., 1999; Villa-Lobos, 2003). The biodiversity of this region is partially a result of species from the north reaching their southern limit and species from the south reaching their northern limit of distribution. Geological complexity, especially in those areas that provide edaphic soils, further contributes to the botanical diversity of the area. This area escaped oceanic submergence, major glaciation effects and volcanic activity throughout its history. A fairly stable climate and geographic complexity likely buffered plant community change and population loss for over 100 million years (Whittaker, 1961; Villa-Lobos, 2003). The characteristics that will continue to make the Klamath–Siskiyou ecoregion an important refugial area are topographic diversity, an array of soil types and high levels of biological diversity.

Studies of genetic variation within and among populations of plant species in northern California are primarily focused on conifer taxa due to their economic importance. The Klamath Ranges contain forests that are considered roughly equivalent to the Paleogene circumboreal forests (Sawyer and Thornburgh, 1977). Currently, the Klamath–Siskiyou region contains more than 281 endemic plant species and 3500 plant species (Sawyer, 2006). Pleistocene glaciations throughout California were significant forces affecting intraspecific, interspecific and intergeneric evolutionary patterns in plants and animals, particularly those with low dispersal ability. In contrast, the Klamath–Siskiyou region experienced minimal glaciation and served as a refugium for many plant taxa (Smith and Sawyer, 1988; Soltis et al., 1997; Sawyer, 2006).

Pleistocene glaciations are increasingly recognized as important in shaping the genetic structure of California plant species, as has been demonstrated in Pinus albicaulis, P. lambertiana (Liston et al., 2007) and Pinusmonticola (Steinhoff et al., 1990). Pinus albicaulis is likely a post-diversification migrant from Europe across Beringia, which was unglaciated during the Pleistocene and LGM, and is a known refugium (Brubaker et al., 2005). Pinus albicaulis shows little differentiation between populations, even though it has a discontinuous distribution in alpine habitat (Jorgensen and Hamrick, 1997). Both P. albicaulis and P. lambertiana demonstrate divergence into northern and southern clades, with secondary contact zones in the Klamath–Siskiyou Mountains (Eckert and Hall, 2006). Pinus lambertiana occurs from Oregon south to northern Baja California and individuals from Oregon, and the Klamath and the north coast ranges of California share a common cpDNA haplotype (‘N’), with plants from the Sierra Nevada and the Transverse and Peninsular Ranges having a different haplotype (‘S’) (Gernandt et al., 2005). There is a contact zone between the two haplotypes in northeastern California in a region about 150 km from Mount Lassen. Recent secondary contact suggests connections from northern and southern refugia following Holocene glaciations. The southern Cascades/Klamath–Siskiyou region is proposed to be a northern glacial refugium for P. lambertiana (Eckert and Hall, 2006; Syring et al., 2007). As climates warm, those conifer species with high genetic diversity, ranges that span the largest latitudes, and the broadest ecological amplitude may fare well, such as P.lambertiana, but the outlook is more tenuous for high-elevation species such as P. monticola, P. longaeva, and P. albicaulis. Pinus albicaulis populations were recently found to have a range of seedlings 90·0 m lower in altitude but 91·8 km north in latitude compared with the range of mature trees; mean annual temperature of the populations is similar and consistent with a gradual shift north due to climate change (Monleon and Lintz, 2015).

Patterns of genetic diversity in California conifers indicate a complex array of palaeoendemics, neoendemics that are variously uncommon or widespread. Generally, California conifers generally maintain high levels of genetic diversity, and many of these species, such as Abies concolor, Pinus ponderosa and Pseudotsuga menziesii, are likely to persist, but there are many notable exceptions. Neotoma spp. midden, palynological and genetic data support two refugia for P.ponderosa during the Pleistocene, the southern Sierra Nevada and the southern Rocky Mountains (Betancourt et al., 1990; Latta and Mitton, 1999). Due to high levels of genetic diversity and a large range, P. ponderosa populations are expected to contract but likely survive climate change, whereas even with high levels of diversity those species with more limited ranges are more precarious (Jump and Peñuelas, 2005).

Picea breweriana is currently found only in the Klamath Ranges but was formerly widespread in Idaho, Nevada, southern California and central Oregon in the Pliocene and Miocene. It now occurs in roughly the north–south extent of the Klamath region at elevations of 560–2300 m in small, disjunct populations. Isozyme loci show a lack of correspondence between genetic and geographic distance that is likely a reflection of some genetic drift (Ledig et al., 2005). Levels of heterozygosity and migration between populations are lower than for other conifers (Hamrick et al., 1992) but also decrease with latitudinal increases, which are estimated to be due to postglacial dispersal (Hamrick and Godt, 1996; Ledig, 2000; Ledig et al., 2005). The response of species such as P. breweriana is equivocal; because it is long-lived it may not be able to disperse rapidly enough to respond to warming climates, but due to the microclimates available within the complex topography of the Klamath Ranges it may survive.

It will be difficult to predict, even with large-scale experimentation or measures of adaptive genetic variation, how species with large ranges but low genetic diversity will fare in the face of climate change. For example, Cornus nuttallii is endemic to North America with a distribution from the southern California mountains to southwestern British Columbia. Extensive sampling from throughout the range of the species indicates that there is no isolation by distance and there is generally low diversity range-wide for both the nuclear and cpDNA genomes. A decline in diversity with latitude suggests there was long-distance dispersal from a southern refugium following glaciation. The presence of two chloroplast haplotypes in a north–south division in southern Oregon, low genetic distance and a lack of private alleles also support a long bottleneck during the LGM that resulted in a north–south division in southern Oregon (Keir et al., 2011). Cornus nuttallii has been significantly affected by the introduced Discula destructiva (dogwood anthracnose), particularly in moist sites, which occur in the northern range of the species and result in more mature trees at lower elevations and latitudes and indicates that factors other than climate change are more important in affecting distribution (Monleon and Lintz, 2015). The scenario for C. nuttallii is consistent with the prediction that taxa isolated by the environment will result in more locally adapted genotypes that may be able to migrate long distances but ultimately are poorly adapted to different environments. In contrast, species that are genetically isolated by distance may be able to respond more rapidly than those that require specific habitat types, but gene flow may be limited by dispersal ability (Sexton et al., 2014).

Alpine habitats

Alpine habitats in California are generally considered to be those at 3500 m or higher, with a growing season that is limited to the period between the melt of winter snowpack and first snowfall, depending on the latitude, and with intense summer drought (Sharsmith, 1940). A virtually contiguous alpine habitat exists from Tuolumne to Inyo counties in the Sierra Nevada, in addition to habitats on Mount Shasta in northern California, on peaks in southwestern California (e.g. San Gorgonio Mountain) and in the Great Basin on Mount Patterson and the White/Inyo Mountains (Rundel, 2011). Additional alpine habitats at slightly lower elevations are found on Mount Eddy and Thompson Peak in the Klamath–Siskiyou Mountains, and in the Panamint Mountains east of the White/Inyo Mountains in southern California. Species that occur in these habitats can be divided into those that are endemic to subalpine and alpine habitats and those that have broader distributions. California has 31 endemic alpine species (Rundel, 2011).

Significant changes are projected to occur in the subalpine and alpine flora and fauna of California as climate change progresses (Hayhoe et al., 2004; Moritz et al., 2008). Of the plants found in the high Sierra Nevada and Transverse Ranges, those with the smallest ranges are projected to be the most heavily impacted by climate change (Loarie et al., 2008). Alpine organisms often experience population contraction during glacial periods and dramatic population expansion during interglacial periods (DeChaine and Martin, 2004). Depending on dispersal ability, generation time and habitat specificity, genetic diversity is variable but often alpine organisms leave strong genetic signatures. Neoendemics in general are expected to exhibit a variety of patterns due to isolation, colonization of extreme environments, and strong selection that can result in rapid speciation (Schoville et al., 2010).

Colonization of the high Sierra Nevada is suggested to have occurred from north to south, based on a greater number of endemic species in the southern Sierra Nevada (Raven and Axelrod, 1978). North-to-south colonization of high montane taxa more generally may explain why Mount Lassen shares greater floristic affinity with Mount Shasta than the closer northern Sierra Nevada (Gillett et al., 1995). Similarly, the Klamath Mountains are the southern limit of a number of boreal alpine species (Howell, 1944). Generally, these patterns support a northern origin for the alpine floras of the Cascade Range and Sierra Nevada. Species with their southern limit of distribution in the northern or central high Sierra Nevada are Carex whitneyi, Podistera nevadensis, Claytonia megarhiza, Thalictrum alpinum, Galium grayanum and Salix nivalis (Rundel, 2011). As a result of this colonization pattern and generally decreasing levels of genetic diversity from north to south, southern endemic species and populations will be most vulnerable. Nine of the Sierra Nevada alpine species are restricted to elevations of 3500 m and 36 species are endemic to the Sierra Nevada at ≥2700 m (Rundel, 2011).

Geographically, the alpine flora has diverse origins. The largest group of species (34·3 %) are those that are widespread in the mountainous western USA, 20·5 % occur in the Intermountain Great Basin, 15·8 % in the Sierra Nevada and into the Cascade Range, and 13·6 % are widely distributed in boreal or arctic–alpine regions (Rundel, 2011). A majority of both the Sweetwater (94 %) and White Mountain (90 %) alpine floras are shared with the Sierra Nevada (Scott, 1995). These floras should be considered at great risk. However, many species present there may survive in the northern Sierra Nevada. Models to predict how populations will respond to climate change will need to include changes in population size, demographics and connectivity during changing warm and cold phases, with the intensity of gene flow dependent on individual species characteristics.

Models examining the effects of climate change on European alpine species for temperature increases of 2·5–4  °C found significant changes in population-level genetic structure that became more dramatic as temperatures increased (Jay et al., 2012). Although these models are based on neutral genetic variation, adaptive alleles are expected to follow the same patterns (Thibert-Plante and Hendry, 2010; Bierne et al., 2011). Additionally, invasions by lower-elevation or more southern populations has the potential to increase genetic variation in these populations, provided there are opportunities for gene flow (Verhoeven et al., 2011).

Although there are molecularly-based population genetic studies of alpine plant species in the European Alps and the Spanish Sierra Nevada, beyond Abronia alpina there are no range-wide population genetic studies of California alpine taxa. Abronia alpina is a rare alpine species endemic to the southern Sierra Nevada that is obligately outcrossing, making it dependent on pollinators to maintain genetic diversity. Although genetic diversity is currently high, a loss or upward elevational migration of pollinators will minimally result in a loss of genetic diversity and possibly extinction (Jabis et al., 2011). It is critical to the persistence of California’s alpine plant taxa that assessments of population genetic diversity and connectivity and projections to habitat change be conducted.

Central Valley wetlands

The vernal pools of California have suffered devastating losses in the past 150 years, mostly due to habitat conversion. Formerly located throughout the Central Valley, less than 10 % of vernal pools are estimated to remain (Holland, 1976). Almost 200 plant species are either endemic or associated with vernal pools, 55 % of which are endemic to California (Holland, 1976). The presence of vernal pools in flat, easily accessible land has made them prone to development pressure and alterations to surrounding areas can negatively affect their hydrological properties (Cheatham, 1976). In addition to their direct loss, reduced connectivity among vernal pools has resulted in the extirpation and loss of gene flow in a number of taxa, both plant and animal.

Conservation of vernal pool taxa requires the assessment of gene flow within and among populations to determine their degree of genetic isolation (Sloop et al., 2011) and their ability to persist under altered hydrological and climate change. An assessment of genetic variation in Neostapfia colusana, a federal threatened species endemic to vernal pools, found high levels of heterozygosity (H_O = 0·68, H_E = 0·71), but an F_ST value of 0·268 (P < 0·0001) between northern and southern populations indicates that gene flow is very limited among the regions and likely due to the loss of geographically intermediate populations (Sloop et al., 2011). Some populations had lower than expected heterozygosity that was likely due to geographic isolation and increased inbreeding. Similarly, Tuctoria greenei, a federal endangered vernal pool endemic species evaluated for population-level genetic variation, found high heterozygosity (H_O = 0·77, H_E = 0·79), also with regional subdivision between northern and southern populations, although gene flow was not as restricted as in N. colusana (F_ST = 0·11, P = 0·0001) (Gordon et al., 2012).

The Limnanthaceae is a small clade endemic to western North America, with the exception of Floerkea proserpinacoides, which occurs in pockets throughout North America (Kishore et al., 2004). The genus Limnanthes consists of 18 taxa that occur primarily in wetland habitats in the Central Valley, 12 of which are listed as threatened, rare or endangered. Limnanthes floccosa ssp. californica is a federal and state-listed endangered species endemic to vernal pools along the eastern edge of the Sacramento Valley and currently known from only 14 geographically isolated populations. The species has fairly low genetic diversity, with H_O = 0·10 and H_E = 0·19, and high among-population genetic structure, as measured by pairwise F_ST values of 0·12 – 0·79 (Sloop et al., 2011). Of particular concern for this species is its high rate of inbreeding and barriers to gene flow. Pollinator declines and reduced aquatic dispersal due to habitat loss will likely further erode genetic diversity in this species.

Quercus lobata, endemic to California and the western USA, occurs in remnant populations in the Central Valley and the surrounding valleys of the Coast Ranges, Sierra Nevada and Transverse Ranges of California. The Mediterranean climate that became established by the mid-Pliocene in west central California and continued through the Pleistocene glaciations affected distributions of Quercus and associated species (Griffin, 1988). Genetic data support strong geographic structuring of populations within the species, determined to be about 200 km (Grivet et al., 2006); however, the genetic connectivity of current populations supports long-distance colonization (Sork et al., 2010). Combined with palynological data, these results suggest the Neogene was an important period for Quercus woodlands as they expanded and contracted with glaciations. It will be important to maintain the remaining connectivity in Q. lobata to protect its ability to adapt, as southern populations are more likely to be lost.

Taxa such as Carex, a worldwide genus with its centres of diversity in Northern Hemisphere temperate regions, are likely to fare better in the face of warming climates; however, more isolated populations or species are likely to be in peril. Only 30 known plant species have a bipolar distribution, six of which are found in the genus Carex (Moore and Chater, 1971). Diversification via long-distance dispersal in the late Miocene through the Pliocene is hypothesized for bipolar Carex species (Escudero et al., 2010). Molecular studies using chloroplast and nuclear DNA of the subgenera Vignea and Carex, both of which occur at high latitudes in the Northern and Southern Hemispheres, suggest that periodic glaciations events may have played a role in speciation (Martín-Bravo and Escudero, 2012). This pattern is consistent with Carex as whole (2000 species) as it has an inverse latitudinal gradient of species richness (Hillebrand, 2004), which will likely become more pronounced as climates warm.

Coast, Transverse and Peninsular Ranges

Taxa with populations that have close proximity to the confluence of two or more divergent floras often have high levels of genetic diversity, reflecting a complex evolutionary history. It will be essential to maintain landscape connectivity in these regions to maintain migrational corridors and gene flow in the context of anthropogenic change. The Tehachapi Mountains are at the confluence of major regional biotas and historically have provided migrational corridors between the central coast, Transverse Ranges, the Sierra Nevada and the Mojave Desert. The Tehachapi Mountains and surrounding areas and have long been recognized as containing high biological diversity and evolutionary divergence (Stebbins and Major, 1965; Patton and Smith, 1990). Examples of non-plant taxa with clade breaks across the Tehachapi Mountains are numerous (Schierenbeck, 2014); however, surveys of the population genetics or high-resolution phylogenetic analyses of plants in this area are few. High genetic diversity with evidence of northern expansion into the Sierra Nevada or western expansion into the Coast Range is found in the Tehachapi Mountains for some taxa. The area has long been recognized as an evolutionary hotspot, with the proposal of the area for genetic connectivity between the coast and Transverse Range (Peabody and Savage, 1958) and later as an area for high hybridization between xeric and mesic lineages (Remington, 1968). High levels of relictual plant species, or palaeoendemics with very restricted distributions, provide evidence of range expansion of cooler-adapted species into southern latitudes, with concentrations in the San Jacinto and Santa Rosa Mountains (Stebbins and Major,1965; Raven and Axelrod, 1978; Baldwin, 2014). The San Gorgonio Pass in the San Bernardino Mountains has a very strong cline of habitat types and is the location of many divergent clades (Patton et al., 2008). Conservation efforts have improved in recent years but much of the area remains under threat. The Tejon Ranch, recently placed under conservation easement, is 109 265 ha at the juncture of the Tehachapi Mountains, the San Joaquin Valley, the Transverse Ranges and the Mojave Desert and is home to 20 state-listed and federally listed species and 60 other rare taxa. This area will continue to serve as an important reserve for genetic diversity and cooler microclimates will provide refuge for many species unable to migrate with climate change.

The Transverse Ranges and Los Angeles Basin Pliocene embayment break is an important phylogeographic break and often separates central and southern California for a number of clades (Calsbeek et al., 2003; Chatzimanolis and Caterino, 2007; Polihronakis and Caterino, 2010). Species distributions were affected by expansion and contraction of mesic habitats during the Pleistocene. Expansion in southern clades happened from southern refugia and relationships among genetic divergence and geographic distance indicate a gradual movement north. Unfortunately, the Transverse Ranges have only about 30 % of their original habitat intact, including the Ventana Wilderness Area in the Santa Lucia Range and the Ventura region. There are very high levels of fragmentation and isolation of contiguous habitat. Although much of the land falls within the Los Padres National Forest, the area is heavily impacted by logging, air pollution, disturbance to aquatic habitat, overgrazing, heavy recreational use and fire suppression (David Olson, TNC, pers. comm.).

Genetic distances among taxa in the Transverse and Peninsular Ranges are generally low and reflect late Pleistocene divergences (Calsbeek et al., 2003). Populations of species with northern populations may persist but not those with limited southern California distributions, limited dispersal or strong sensitivity to human disturbance. There are few plant studies of this region, but amphibian studies indicate that it will be important to identify those areas that will be suitable for reproduction and ongoing gene flow in this evolutionary hotspot (Devitt et al., 2013).

Taxa with limited distributions in heavily populated areas are particularly at risk. For example, Dirca occidentalis, a California endemic found only in the San Francisco Bay Area, is known from four genetically distinct populations hypothesized to have resulted from climatic events within the last 20 000 years (Graves and Schrader, 2008). Species such as D. occidentalis, dependent on mesic habitat in areas with dense human populations, are in jeopardy.

CONCLUSIONS

The prediction of how species will respond to climate change will require a synthesis drawing from population genetics, geography, palaeontology and multiple ecological parameters (Table 1). The important integration of these fields will enable us to predict and prioritize conservation areas during a time of rapid climate change, human disturbance and invasive species. Phylogeographic methods, ecological niche modelling, fossils and palaeoclimatic reconstructions all contribute to the inference of refugia during the LGM, which are now reflected as areas with high genetic diversity or as contact zones. These refugia will continue to be significant sources of genetic variation, but because climate is now warming, not cooling, populations at lower elevations and latitudes will also provide important reservoirs of genetic diversity. Although earlier interpretations of the Pleistocene fossil record indicated that responses to climate change were idiosyncratic (Graham et al., 1996), recent evidence shows a number of patterns among groups with similar life history characteristics. Vagility, isolation of the habitat, reproductive output and generation time all affect the ability of species to respond to climate change.

The risk of extinction can be affected by genetic diversity, range size, demography, body size and life history characteristics, especially fecundity. The most vulnerable species are endemic species because of their habitat specificity and rarity, and the most vulnerable ecosystems are those with high levels of spatial isolation, e.g. alpine habitat and serpentine islands. Spatial isolation can mean different evolutionary dynamics in different populations of the same species in terms of gene flow, genetic drift, and strong selection from climate change and invasive species. Ecosystem function is dependent on biological diversity as many ecosystem processes are sensitive to declines in biodiversity. Local declines in biodiversity can be more devastating than global declines as the benefits that organisms provide for local ecosystems can be lost before a global extinction.

POTENTIAL SOLUTIONS

Issues to consider in mitigating climate change include local, regional and taxonomic rarity, ecological function and current threats to biodiversity. The specific nature of species–environment interactions sometimes appears idiosyncratic because of our lack of specific understanding of an ecosystem or the ecological requirements of particular species. Experimental examination of the biotic factors that control genetic diversity and their interaction with climate change should focus on areas that contain the greatest number of endemic species or large amounts of biodiversity, e.g. the central Coast Ranges, the Klamath–Siskiyou region, the eastern Transverse Ranges, the Tehachapi Mountains, the Santa Lucia Mountains and the eastern Mojave Desert. Areas with high levels of endemism that include species with the greatest range restrictions are on the central coast, in the Sierra Nevada and in the San Bernardino Mountains (Kraft et al., 2010). The Mojave Desert and Great Basin have the highest level of neoendemics. High levels of endemism, however, do not necessarily coincide with high levels of genetic diversity, which will likely affect the ability of a species to persist in a rapidly changing climate. Due to its extant biodiversity, latitudinal position, topographical complex and minimal development, and the presence of many taxa with splits between northern and southern clades, the Klamath–Siskiyou region will remain an important refugial region.

Far northern California represents a confluence of the volcanic Cascades, the geologically ancient and complex Klamath–Siskiyou region, the northern Sierra Nevada and the Warner Mountains, separated from the Sierra Nevada by the Modoc Plateau. The species that have hybridization zones or reach the northern or southern limit of their distribution provide varying degrees of divergence. Population genetic studies of plant taxa in the Warner Mountains are virtually nil outside of conifers, but there have been a number of vertebrate and invertebrate studies that establish this region as a contact zone. Because divergence patterns of plant and animal follow similar patterns (Calsbeek et al., 2003), this region should reveal interesting population structures within and among plant taxa. Although not currently considered a priority area, this region has been poorly explored and merits consideration. For example, this region has been documented for the evolutionary divergence of northern and southern subspecies of the bird species Picoides albolarvatus (Alexander and Burns, 2006) and Pacific Northwest and California lineages of Strix occidentalis (Barrowclough et al., 2005). Habitat and climatic differences correspond to a split between the subspecies bird taxa Aphelocoma californica woodhouseii and Aphelocoma californicacalifornica over a distance of about 500 km, and similar scenarios exist between bird taxa Baeolophus inornatus and Baeolophusridgwayi and mammal taxa Pica nuttalli and Picahudsonia (Delaney et al., 2008).

RECOMMENDATIONS

The general long-term goal of conservation genetics is to conserve historical genetic variation that may be critical to the long-term evolutionary survival of a species. Usually this is accomplished through the conservation of genetically distinct populations. In addition, quantitative trait locus analysis can identify traits of functional diversity or those that are important in ecological interactions. In addition, molecular clock data, which inform divergence times combined with the knowledge of vicariant events, can be critical to the survival of a species (Shaffer et al., 2004). For example, highly restricted palaeoendemic taxa with little genetic variation, such as Hesperocyparis forbesii, will be difficult for large-scale conservation outside of its current range. Conservation strategies for recently emerged taxa will vary depending on the time since emergence, genetic variation, existence of sister taxa, and range. As populations become demographically at risk, some reintroductions may be necessary and an understanding of the genetic basis of adaptive variation will need to be accomplished with genetically and ecologically similar populations.

Genetically divergent populations occurring in fragmented habitats need designation as evolutionarily significant management units. Recommendations to ensure the survival of species include the use of genetically appropriate choices for plant materials in ecosystem restoration to maintain biological diversity, and include provision of the genetic diversity necessary for evolution to continue to occur. Appropriate genetic linkages among fragmented populations must be established, evolutionary interactions conserved, surrounding biological diversity maintained, and changes to ecosystem function minimized (Rogers and Montalvo, 2004).

Genotype specificity can be very important in site restoration. Species without quantitative or molecular differentiation suggest that propagule transfer should have a lower risk of maladaptation. Every effort should be made to assess projected climatic shifts and whether the reproductive and phenotypic traits are suitable to future scenarios. The use of native genotypes in restoration is important in the avoidance of outbreeding depression, which can result in the decline of a native population (Erickson, 2008; Aitken and Whitlock, 2013). Local ecotypes are adapted to site conditions, and even though the use of natives is often more expensive there is greater restoration success with their use (Lesica and Allendorf, 1999). However, there have been recent recommendations that local genotypes could be supplemented with non-local genotypes as bet-hedging against unforeseen environmental change (Doak and Morris, 2010).

Phylogeographic studies can be very useful in defining evolutionarily significant units for conservation purposes (Roderick, 1996). Concordance among divergent major clades provides the basis needed for major conservation efforts in particular regions (Avise et al., 1987). Many regions of high diversity are in need of further support in terms of conservation efforts that will strengthen legal protection, provide and strengthen migratory corridors, and increase the size of parks and preserves to protect biological diversity. Population genetic data can be informative for management and should be utilized to determine whether genetic diversity is being lost (Funk et al., 2010).

LITERATURE CITED