REVIEWS

Genomics and the future of conservation

Fred W. Allendorf*‡, Paul A. Hohenlohe§|| and Gordon Luikart ¶# Abstract | We will soon have complete sequences from thousands of species, as well as from many individuals within species. This coming explosion of information will transform our understanding of the amount, distribution and functional significance of in natural populations. Now is a crucial time to explore the potential implications of this information revolution for and to recognize limitations in applying genomic tools to conservation issues. We identify and discuss those problems for which will be most valuable for curbing the accelerating worldwide loss of biodiversity. We also provide guidance on which genomics tools and approaches will be most appropriate to use for different aspects of conservation.

Neutral locus The ability to examine thousands of genetic markers The application of to conservation is still A locus that has no effect with relative ease will make it possible to answer many in its early stages, but shows promise. First, functional on adaptation because important questions in conservation that have been metagenomics of microbial communities provides a all genotypes have the intractable until now. Simply increasing the number of novel perspective on ecosystem processes, such as same fitness. neutral loci that we can screen will increase the power nutrient and energy flux. Although some studies have and accuracy of estimating a variety of important compared functions across a broad scale of biomes9,

*Division of Biological parameters in conservation (for example, kin rela- similar comparative approaches may identify aspects Sciences, University of tionships and coefficients (F)). However, of ecosystem function across sites within a habitat. Montana, Missoula, the most exciting contributions of genomics to con- The second potential application of metagenomics to Montana 59812, USA. servation are those that will allow new questions to conservation is in assessment of physiological con- ‡ School of Biological Sciences, be addressed in a wide variety of species (BOX 1). For dition of individual organisms. For instance, Vega Victoria University of 10 Wellington, Wellington, instance, it should be possible to estimate the effect Thurber et al. have found numerous shifts in the New Zealand. size and distribution of loci affecting fitness across the endosymbiont community of corals in response to §Center for Ecology and genome or to ask whether the loci are coincident across multiple stressors, such as reduced pH, increased Evolutionary Biology, populations1,2 (FIG. 1). nutrients and increased temperature. Third, a metage- University of Oregon, Eugene, Oregon 97403, USA. Genomic approaches are currently being used pri- nomic analysis of human faecal samples catalogued ||Department of Zoology, marily with a few species for which genomic informa- 3.3 million microbial and found substantial Oregon State University, tion and tools are available3; for example, wolves, bison differences in the microbial metagenome between Corvallis, Oregon 97331, USA. and bighorn sheep have been studied using genomic healthy individuals and those with inflammatory ¶ Flathead Lake Biological tools developed in related domestic species4. However, bowel disease11. It may be possible in the future to Station, Division of Biological Sciences, University of the range of species is expanding as new approaches are apply metagenomic techniques to faecal samples from Montana, Polson, Montana developed that are not dependent on genomic infor- wildlife species to assess physiological states, such as 59860, USA. mation from closely related species5,6. For example, starvation stress. #Centro de Investigação em van Bers et al.7 obtained over 16 million short sequence Genomics already has provided some interesting Biodiversidade e Recursos contigs Genéticos, Universidade do reads and conducted de novo assembly of 550,000 surprises, such as the discovery of adaptive loci that Porto, 4485‑661 Vairão, covering 2.5% of the genome to discover 20,000 novel show extremely high genetic divergence between popu- Portugal. SNPs in the great tit (Parus major). These markers will lations of marine fish for which there is virtually no e-mails: be used for mapping and whole allele frequency divergence at neutral loci12 (BOX 2). In [email protected]; genome association studies. addition, a multi-faceted genomic approach has pro- hohenlop@science. oregonstate.edu; In addition, multiple taxa can be combined in a sin- vided important insights into the treatment of a facial [email protected] gle sequencing analysis using genomic techniques that tumour disease that threatens the persistence of the doi:10.1038/nrg2844 can assay large amounts of variable DNA sequence8. Tasmanian devil (Sarcophilus laniarius)13.

nature reviews | Genetics Volume 11 | october 2010 | 697 REVIEWS

Box 1 | What is ‘conservation genomics’? proceed through issues that will become more feasible in the future. We consider how genomic approaches will Conservation genomics can be broadly defined as the use of new genomic techniques allow us to understand the genetic basis of inbreeding 72 to solve problems in conservation biology. Frankham recently reviewed the current depression and adaptation. We then apply these insights status of conservation genetics and proposed 13 priorities for development in the field. to important outstanding problems in conservation, Many of these priorities have been intractable through traditional genetic techniques. hybridization Although genomic techniques are not appropriate or necessary in all cases, we believe including understanding the effects of and that genomics will have an important role in addressing several research challenges predicting outbreeding depression, as well as predicting over the next few years. evolutionary responses to climate change. Genomic techniques will be more immediately applicable to some questions than to others (TABLE 1). For example, in estimating neutral population parameters, such as ‘Neutral’ markers effective population size, genomics simply provides a larger number of markers to an The most straightforward contribution of genomics to analytical and conceptual framework that is already widely used in conservation conservation will be to enormously increase the pre- genetics. Genomic identification of functionally important genes is now common in cision and accuracy of estimation of parameters that other fields; conservation genomics can incorporate these approaches to study the require neutral loci (for example, effective population size genetic basis of local adaptation or . By contrast, predicting a (N ) and migration rate (m)) by genotyping hundreds population’s viability or capacity to adapt to climate change based on genomic e information will require not only the identification of relevant loci, but also a to thousands of neutral loci in numerous individuals. quantitative estimate of their connection to fitness and demographic vital rates. The accuracy of parameter estimation will be improved These challenges must be tackled by conservation genomics over the longer term. because examining several loci facilitates the identifica- Understanding genomic approaches is crucial to the success of applying genomics tion and exclusion of loci under selection (outlier loci) to conservation (FIG. 1). A growing list of techniques is available for detecting DNA that cause biased estimates of parameters. For example, a sequence differences across individuals in natural populations, and these vary widely in small proportion (1–5%) of non-neutral loci can change the density of markers across the genome, their ability to target candidate loci, the cost 18–20 estimates of mean FST by 30–50% , and change the per sample, and so on. Genomic techniques can be roughly grouped into three classes: topology and branch lengths of evolutionary trees21,22. marker-based genotyping, including a diversity of array-based SNP genotyping Similarly, the assessment of demographic parameters, platforms; reduced-representation sequencing, which uses next-generation such as population bottlenecks or growth rates, requires sequencing technology to target a subset of orthologous regions across the genome of many individuals; and whole-genome sequencing. A crucial component of all genomic many loci to identify outliers and reliably infer change techniques is . The tools for handling genomic data are changing as fast in population size. Selection can shrink (by bottlenecks) 23 as (and in response to) techniques for gathering the data, and we do not review the or expand genealogies at a locus . Therefore, inferences software and analytical issues here111. Nonetheless, researchers using genomic about population growth should be more robust if out- techniques should plan on a substantial investment of time and resources devoted to lier loci are removed, for example by using a hierarchical data storage and analysis. Bayesian model to assess the parameters of each locus separately24. Increasing the number of markers will also facilitate There have been several excellent reviews on con- estimation of directionality of migration (emigration Inbreeding coefficient servation genomics recently3,14–17. We have attempted to and immigration rates), especially if haplotypes can be The probability that two alleles build on these reviews and to distinguish ours by mak- inferred from linked loci25. Certain questions require in an individual are both descended from a single allele ing specific practical recommendations on how genomic linked loci or can be vastly improved by using haplotype in an ancestor (that is, that approaches can be applied to key problems in conserva- inference; for example, estimating relationships among they are ‘identical-by-descent’). tion (TABLE 1). For example, Ouborg et al.15 present a com- individuals26, population structure27, admixture28, dates of prehensive view of how genomics will provide insights historical bottlenecks and directionality of migration25. Contig An abbreviation for contiguous into the mechanisms behind the interaction between Furthermore, it will become increasingly feasible to sequence; used to indicate a selectively important variation and environmental jointly estimate multiple parameters, which generally contiguous piece of DNA that conditions. Nevertheless, if we are to apply this under- requires more loci than single parameter estimation. For is assembled from shorter standing of fitness to conservation, we need to address example, likelihood, Bayesian and approximate Bayesian overlapping sequence reads. the population-level consequences of genetic variation, estimators combined with coalescent approaches will Metagenomics which include population subdivision, demography and allow the simultaneous estimation of multiple param- 25,29 selection The study of the collective population viability. We have incorporated population eters, such as Ne and m , or Ne and the genomic material contained in structure and demographic effects into FIG. 1, and have coefficient (s)30. This is important because it will improve an environmental sample of distinguished issues that only genomic approaches can parameter estimation, allow parameter estimation in microorganisms, facilitated by metapopulations high-throughput sequencing thoroughly address from issues that can be adequately (not just in isolated populations with technology that allows the tackled with traditional techniques. no gene flow), and facilitate investigations of the relative direct sequencing of We have two primary objectives. The first is to iden- importance and interactions among drift, selection and heterogeneous samples. tify those problems in conservation biology in which migration in populations of conservation concern. genomics will be most valuable in providing new insights By contrast, simulations suggest that as the number Endosymbiont An organism that lives within and understanding. The second is to provide guidelines of loci increases, the accuracy of parameter estima- the cells of a host organism. as to which new genomics approaches will be most tion can decrease owing to non-independence or link- appropriate for the different problems in conservation age among loci31. Failure to account for linkage could Inbreeding depression that can benefit from genetic analysis. limit the utility of SNPs or multi-locus sequencing in The loss of vigour and fitness 32 that is observed when We begin by focusing on issues in conservation studies using genealogical information . Markers are genome-wide homozygosity genomics that are immediately accessible (for example, usually assumed to be independent. Failure to account is increased by inbreeding. increasing the number of neutral markers) and then for non-independence can lead to overestimation of

698 | october 2010 | Volume 11 www.nature.com/reviews/genetics REVIEWS

Population size Population structure directly. Skare et al.26 conducted simulation power analy- ses and showed that relatively distant relationships (for example, cousins) can be inferred using 500,000 SNPs Migration rates and likelihood-based relationship estimators. Nonetheless, pedigrees often will not have sufficient depth or completeness because it is difficult to sample Inbreeding Hybridization most individuals in a population over many years. In such cases, genotyping thousands of loci could poten- tially give more reliable estimates of relationships and Genotype-by-environment Loss of Local adaptation 26,38 interactions individual heterozygosity (inbreeding) than pedigrees or at least greatly improve pedigree reconstruction37. Future research is needed to quantify the trade-off point Loss of adaptive variation Inbreeding depression Outbreeding depression between using pedigree inference versus thousands of genetic markers to estimate individual inbreeding.

Demographic vital rates Individual-based . Individual-based approaches can yield less biased delineation of popula- tions than traditional population-based approaches that Population growth or viability require somewhat subjective grouping of individuals39 (for example, based on morphology or geographic Figure 1 | Schematic diagram of interacting factors in conservation of natural populations. Traditional conservation genetics, using neutral markers,Nature providesReviews | directGenetics origin). For population delineation, an empirical study estimates of some interacting factors (blue). Conservation genomics can address a wider of 377 in humans has shown that using range of factors (red). It also promises more precise estimates of neutral processes (blue) greater numbers of loci can increase statistical power to and understanding of the specific genetic basis of all of these factors. For example, resolve between closely related ethnic groups (FST < 0.05) traditional conservation genetics can estimate overall migration rates or inbreeding and infer the proportion of admixture40,41. coefficients, whereas genomic tools can assess gene flow rates that are specific to Individual-based approaches can give less biased adaptive loci or founder-specific inbreeding coefficients. estimation of contemporary migration rates without assumptions such as mutation–migration–drift equilib- rium42. However, the power to estimate contemporary

precision and overconfidence in subsequent inferences. migration rates is low unless FST is relatively high (for

Fortunately, the problem is likely to be minor unless loci example, FST > 0.10) when using only 10–20 microsatel- 33 43 Adaptation are tightly linked . Failure to consider linkage could also lite loci . Little is known about power when genotyp- 44 Heritable changes in genotype have other effects; for example, human loci in regions ing hundreds of loci, although Rannala and Mountain or phenotype that result in of lower recombination tend to have greater FST, appar- reported that an assignment test method using 50–100 increased fitness. ently because of the greater probability of being associ- loci gave reasonable power to identify individuals with

Hybridization ated with selected loci in chromosomal regions with less grandparents from different countries, although the 34 Interbreeding of individuals recombination . differentiation of allele frequencies among populations from genetically distinct was low. Individual-based approaches are crucial for populations, regardless Description of kin relationships and pedigrees. fine-scale spatial genetic analyses to localize genetic dis- of the taxonomic status of Examining hundreds of loci will vastly increase the continuities (for example, barriers or secondary contact the populations. precision and accuracy of kinship estimates. For exam- zones) on a landscape. Individual-based approaches in 35 45 Outbreeding depression ple, Santure et al. showed that the average pair-wise also allow assessment of the influ-

Reduced fitness of F1 or F2 relatedness estimated from 771 SNPs closely brackets ence of landscape features on dispersal and gene flow individuals after a cross known pedigree relationships for a pedigree population across spatial scales. between two species or populations. It can result from of zebra finch. This suggests that assessments of correla- genetic incompatibility or tions between phenotypes and genetic relatedness and Inbreeding depression reduced adaptation to local thus estimation of heritability will be feasible in natural Genomic approaches can potentially address basic ques- environmental conditions. populations. Nevertheless, the accuracy of estimating tions about the molecular basis and genetic architecture individual levels of inbreeding is somewhat limited, and of inbreeding depression46. For instance, is inbreeding Effective population size The size of the ideal population the variances for relatedness between individuals remain depression caused by a few loci with major effects or by 36 that would experience the substantial even with 771 SNPs . many loci with small effects? How much of inbreeding same amount of genetic drift Pedigree reconstruction will become feasible in some depression results from dominance (or partial domi- as the observed population. wild populations with hundreds of loci33,37. This will nance) versus overdominace (heterozygous advan- epistasis Outlier locus improve estimates of effects of inbreeding and outbreed- tage)? What is the contribution of to inbreeding A genome location (or marker ing on fitness and the detection of paternities or pol- depression? Understanding the number of loci involved or base pair) that shows len flow between populations and over long distances, in inbreeding depression and the mechanism of their behaviour or a pattern of if most individuals can be sampled over many years. effects would allow prediction of the potential efficacy variation that is extremely Santure et al.35 suggested that using marker information of purging. divergent from the rest of the genome (locus-specific effects), to reconstruct the pedigree, and then calculating relat- Recent work indicates that the intensity of inbreed- as revealed by simulations or edness from the pedigree, is likely to give more accurate ing depression can differ greatly depending on which statistical tests. relatedness estimates than using marker-based estimators specific individuals are founders47,48. This suggests that

nature reviews | Genetics Volume 11 | october 2010 | 699 REVIEWS

Box 2 | Detection of cryptic subdivision and local adaptation in marine species There is little genetic drift in many marine fish and invertebrates because of their large population sizes121,122. As a

FST consequence, population genetic studies of many marine species have failed to detect genetic substructure even A measure of population between geographically disjunct subpopulations for which there is evidence of reproductive isolation122. The absence subdivision that indicates the of genetic differentiation at neutral markers, however, should not be taken to mean the absence of adaptive proportion of genetic diversity differences. The amount of genetic divergence among subpopulations at selectively neutral markers is largely a found between populations function of the number of migrants per generation (N m) rather than the migration rate (m). With large population relative to the amount e within populations. sizes, even very low migration or dispersal rates can result in enough migrant individuals to eliminate genetic evidence of population differentiation at neutral loci, but not at locally selected adaptive loci. Population bottleneck We expect this effect to be greatest in marine species because of the large local population sizes, which allow A marked reduction in selection to be more efficient because drift is weaker. The amount of divergence at selected loci is determined by the population size followed by the relative values of migration and selection coefficient (s). Species with larger local populations (Ne) will have much survival and expansion of a lower rates of migration than species with small population size with the same number of migrants and amount of small random sample of the divergence at neutral loci. Therefore, even fairly weak selection may bring about genetic differentiation between original population. It often subpopulations in species with large local population sizes because s is much more likely to be greater than m. results in the loss of genetic This prediction is supported by a recent study123 of Atlantic cod (Gadus morhua) in which almost no genetic variation and more frequent differentiation (F = 0.003) was found at nine loci, but substantial differentiation (F = 0.261) was found matings among closely ST ST 105 124 related individuals. at the PanI locus, which previous studies have shown to be under . Similarly, Haemmer-Hansen et al.

reported an FST of 0.45 at a heat shock protein locus in comparison to a mean FST value of only 0.02 at nine Hierarchical Bayesian model microsatellite loci in the European flounder (Platichthys flesus). This approach of simultaneously comparing many A Bayesian model in which neutral and candidate gene markers has been highly successful in a range of species19. the prior depends on another In addition, the absence of genetic differentiation in marine species should not be interpreted to indicate that parameter that is not in the populations are demographically connected as a single management unit125. Demographic connectivity is the likelihood function and largely a function of the proportional amount of exchange. Therefore, low migration rates (m < 0.001) can result that can vary and have in a substantial number of migrant individuals when local population sizes are in the thousands, resulting in F another prior. ST values near zero. Much greater exchange is necessary for demographic connectivity between populations. 126 Haplotype For example, Waples and Gaggiotti have suggested that m must be greater than 10% for populations to be A set of genetic markers demographically interdependent. that are present on a single chromosome and that show complete or nearly complete the genetic load is unevenly spread among founder affected by selection at a particular locus, but that fine- gametic disequilibrium. genomes and supports the notion that inbreeding mapping of a selected locus will be more difficult. Ideally, They are inherited through depression sometimes results from major effects at a few researchers would study populations with both long and generations without being 49 changed by crossing-over loci . The founder-specific partial F coefficient is the short chromosomal regions of gametic disequilibrium or other recombination identical-by-descent (IBD) probability (for an individual) to allow for initial coarse-mapping and subsequent mechanisms. that is attributed to a particular founder. A study with fine-mapping of loci under selection. Ripollesa domestic sheep found that most of the inbreed- In the future, it could be possible to identify loci that Admixture ing depression resulted from individuals being IBD for contribute to inbreeding depression by sequencing the The production of new 49 genetic combinations in genes from just two of the nine founders . Managing whole genomes of parents and offspring. For example, 53 hybrid populations through founder-specific inbreeding depression using partial Roach et al. analysed the complete genome sequence recombination. inbreeding coefficients could be extremely effective in of two parents and their two children, who suffered cases in which inbreeding depression results primarily from two clinical recessive disorders. They narrowed Coalescent approach A means of investigating the from a few loci with major effects; such partial inbreed- down the candidate genes for both of these Mendelian shared genealogical history ing coefficients could be useful when selecting potential disorders to four using family-based genome analysis. of genes. A genealogy is matings in a captive population. constructed backwards in time Local adaptation starting with the present-day Identifying alleles responsible for inbreeding depression. One of the most promising aspects of applying genomic sample. Lineages coalesce when they have a common Genome scans of large numbers of markers can detect tools to conservation is the simultaneous estimation ancestor. the signature of inbreeding depression. Deleterious reces- of neutral (that is, genome-wide average) processes sive alleles related to inbreeding depression have been along with identification of specific genomic regions Selection coefficient identified in a few species46,50,51. In general, attempts to responding to selection, such as adaptation to local A term that describes the difference in average fitness identify loci responsible for inbreeding depression may conditions that vary across a metapopulation. These between genotypes when be less successful than those aimed at positive selection specific genomic regions appear as outliers from the fitness is measured relative to for a few reasons. First, detecting the multiple genetic patterns observed at the neutral genomic background, the average fitness of one of mechanisms that may underlie inbreeding depression, which is determined primarily by genetic drift and gene the genotypes (known as the including epistasis and genotype-by-environment inter- flow. Researchers have developed multiple approaches reference genotype). action, may prove more difficult52. Second, populations to detect these outliers54,55 (BOX 3). The utility of these Metapopulation of interest are likely to be small, necessitating small approaches depends on the timescale over which selec- A collection of populations sample sizes, which reduce power and accuracy. Third, tion has operated and the study’s taxonomic scale (for of a species found in differing the longer regions of gametic disequilibrium expected example, the study might be investigating divergence geographic locations and with restricted gene flow in small inbred populations (observed in wolves by among species, differentiation among populations 46 (exchange of genes) between Hagenblad et al. ) mean that genotyped anonymous within a species or evolutionary history within a single the populations. markers are more likely to lie within a genomic region population), as well as on the techniques used55.

700 | october 2010 | Volume 11 www.nature.com/reviews/genetics REVIEWS

Table 1 | Primary genetic problems in conservation and how genomics can contribute to their solution* Primary problem Possible genomic solution

Estimation of Ne, m and s Increasing the number of markers, reconstructing pedigrees and using haplotype information will provide greater power to estimate and monitor Ne and m, as well as to identify migrants, estimate the direction of migration and estimate s for individual loci within a population Reducing the amount of admixture Genome scanning of many markers will help to identify individuals with greater in hybrid populations amounts of admixture so that they can be removed from the breeding pool Identification of units of conservation: The incorporation of adaptive genes and gene expression will augment our species, evolutionarily significant understanding of conservation units based on neutral genes. The use of units and management units individual-based landscape genetics will help to identify boundaries between conservation units more precisely Minimizing adaptation to captivity Numerous markers throughout the genome could be monitored to detect whether populations are becoming adapted to captivity Predicting harmful effects of Understanding the genetic basis of inbreeding depression will facilitate the inbreeding depression prediction of the effectiveness of purging. Genotyping of individuals at loci associated with inbreeding depression will allow the selection of individuals as founders or mates in captive populations. Pedigree reconstruction will allow more powerful tests of inbreeding depression Predicting the intensity of Understanding the divergence of populations at adaptive genes will help outbreeding depression to predict effects on fitness when these genes are combined. Detecting chromosomal rearrangements will help to predict outbreeding depression Predicting the viability of local Incorporating genotypes that affect vital rates and the genetic architecture of populations inbreeding depression will improve population viability models Proportion of admixture Predicting the ability of populations Understanding adaptive genetic variation will help to predict the response The proportion of alleles in a to adapt to climate change and other to a rapidly changing environment or to harvesting by humans and allow the hybrid swarm or individual anthropogenic challenges selection of individuals for assisted migration that comes from each of *These problems are listed from top to bottom in sequence of those that can be immediately addressed to those that will become the hybridizing taxa. more feasible to address in the future. m, migration rate; Ne, effective population size; s, selection coefficient. Epistasis The dependency of the For most conservation purposes, only a subset of Genomic databases may even come from related species, effects of alleles at one locus on the genotypes at other these tools will be most appropriate, and application so that rare species of conservation concern are ‘genome- 3 loci in the genome. of the wrong approach could result in type I and type II enabled’ by the resources of better-studied, related taxa . errors. Specifically, detecting genomic regions that are A growing variety of genomic tools can also be used Purging responsible for local adaptation in a species relies on directly to genotype individuals at up to thousands of The selective reduction in (TABLE 2) frequency of deleterious comparisons among related populations that may or candidate loci . recessive alleles in small may not be linked by ongoing gene flow. In this case, The second major approach to detecting local adap- populations because the the most appropriate analyses often will focus on dif- tation searches the genome for signatures of selection increase in homozygosity ferentiation in allele frequencies among populations using anonymous markers66,67. A limitation here is that increases the ability of selection (that is, F 20). Within a single population, the allele fre- markers must be in gametic disequilibrium with selected to act on recessive alleles. ST quency spectrum can indicate regions under selection55. loci to exhibit a signature of selection, and the signa- Identical-by-descent By contrast, techniques for detecting historical selection ture can be quite small depending on the nature of the An allele shared by two based on fixed sequence divergence between species or selection. In particular, local adaptation with ongoing related individuals is said to the relationship between divergence and polymorphism gene flow between populations subject to differential be identical-by-descent if the allele is inherited from the are likely to have only limited applications in conser- selection is expected to produce a soft sweep; such a same common ancestor. vation because of the longer timescale of selection that signature of selection can have a very narrow footprint can be detected (but see Garrigan and Hedrick56). Here along the genome and be difficult to detect, even given Gametic disequilibrium we focus on the first case — local adaptation among strong selection68. Nonetheless, the density of markers A measure of whether alleles at populations within a species. along the genome allowed by high-throughput genomic two loci in a population occur in a non-random fashion. techniques can be sufficient to identify these regions, Methods for assessing local adaptation. There are two especially when replicate populations subject to simi- Type I and type II errors general ways to assess local adaptation in the genome lar selection pressures can be sampled66. The array of Statistical errors in which (BOX 3): the first starts with a list of candidate loci or genomic techniques covers the range of trade-offs a true null hypothesis is rejected (type I) or a false genomic regions and asks whether these lie in the tails of between density of markers and number of individuals null hypothesis is not the genome-wide distribution of population differentia- or populations sampled. Any information on the overall rejected (type II). tion57–60. Genomics can augment these studies indirectly amount of gametic disequilibrium can inform the exper- by providing a list of candidates; for example, expressed imental design of genome scans (see Supplementary Expressed sequence tag sequence tag (EST) databases allow for the bioinformatic information S1 (figure)). A short DNA fragment (several hundred base pairs) produced identification of microsatellites or other traditional There are trade-offs between the two general by reverse transcription of markers closely linked with target genes, and primers or approaches outlined above. The first allows targeting mRNA into DNA. probes can be developed from these EST sequences61–65. of particular loci, which can be valuable if selection is

nature reviews | Genetics Volume 11 | october 2010 | 701 REVIEWS 702 local adaptation; for example, Hohenlohe genomic tools also allow anonymous markers to be assayed across the genome to identify analysis identified 20 as indicated by bootstrap resampling (blue, Wiley and Sons. The image in part to be visualized as continuous distributions along chromosomes. In part populations. Coloured bars above each graph show regions of significantly elevated differentiation between each of the three freshwater populations and the two marine the top panel shows associated DNA (RAD) tags individuals across 5 populations in Alaska. They used sequencing of restriction-site- populations.The image in part indicate genomic regions that have responded to divergent selection in parallel across population differentiation (yellow shading shared among the three populations) consistent peaks of freshwater-versus-marine differentiation. Common patterns of indicates the chromosomes, and yellow shading indicates the nine most significant and heterozygosity Box 3 | Namroud Genome scans for selection can focus on either candidate loci or anonymous loci. genotyped 534 across the genome. This density of markers allows population genetic statistics, such as respectively. Against a background of little population differentiation (F outlier analysis of these data, based on the relationship between b a | t oc FST FST FST FST FST 0.08 0.06 0.0 –0.2 –0.2 –0.2 –0.2 0.0 0. 0. 0. 0.8 0.8 0.8 0.8 0.6 0.6 0.6 0.6 0. 0. 0. 0. 0.2 0.2 0.2 0.2 10 14 12 0 0 0 0 0 4 4 4 4 4 2 Genome scans to detect local adaptation o 0 05 et b er 2010 al. 62 59 SNPs located on 345 candidate genes. Part sampled white spruce (Picea glauca) from 6 populations in Quebec and ; the grey and red lines represent the 95% and 99% confidence levels, 0 F | SNPs (circled dots) in 19 genes above the 95% confidence level. New Volume 11 ST between the two marine populations. The next three panels show 0. 10 100 127 to simultaneously identify and genotype over 45,000 a is reproduced, with permission, from b 150 Expected heter is reproduced from 0.20 Po et sition (Mb) 200 p al. ≤ 10 66 sampled 100 threespine stickleback ozygosit –5 ; red, 0.30 250 REF. 66 y p ≤ 10 a of the figure shows their 300 . Na –7 F ). Vertical grey shading tur ST and expected 0. e Re 40 REF. 62 b 350 of the figure, ST vie

= ws 0.006), this

© |Genetics John 400 0.50 SNPs SNPs F

F F ST ST ST ,

,

methods could be used to effectively plan harvestingplan effectively to used be could methods suchpopulations; independent or distinct towards targeted efficiently be can monitoringprogrammes thatmanagementpopulation so unitsnecessary is and context, a onfocus maintaining adaptive genetic potential. In this byhumans threatenpersistenceofmany the species genetic changes in response to the harvesting of animals ping of associations between adaptive genome regions phenology of adaptive genetic variation related to key traits, such as approachestats.Genomic identificationmay the allow mate change will affect a wide range of species and habi- anticipationofenvironmental change; forexample, cli- understandinghowmaintainto ability evolve the to in importantcomponentAnis genetics conservation of Climatechangeanthropogenicotherand challenges. of probes for candidate and anonymous combinationa printedwith be genotypingcanarrays combined;approaches forbe example, These also can genomiconsimilar regions acting acrosspopulations. is selection testwhether as well as selection, to subject such as estimates of how many regions of the genome are tive pressures and can provide quantitative information, absence of thecontrast, approachBysecondin the useful mostis of individuals or populations for the same overall effort. cies. This approach can also be applied to a larger number tional genetic information is available from related - spe suspected to act on particular phenotypic traits and - func evolutionarilyunits significant importantcontributions conservationtogenetics of most the of one is conservation of units Describing Units of conservation and hybridization flow across landscapes in expected the spatial–temporal landscape change and modelling gene predictingby adaptive alleles of flow genechange on ofenvironmentalforecasting allowcould effects ofthe This time. andenvironmental spaceand in gradients frequencies of those desirable phenotypes reducethat populationscanthe selection wild imposes segments number two priority for conservation genetics Frankham’sis uniformofspecies definitionuponinga identification remainsspecies controversial, andagree- Nevertheless,developed. relativelywellare methods identificationandclassification,principles genetic and diversity.ofbiologicalconservationfor the For species lationunitsfor protection andmanagement essential is identificationThe ofappropriate taxonomic andpopu - before substantial or irreversible changes implement managementeffects the reverseorlimit to potentially harmful changes will maximize our ability to could be extremely important because early detection of ofgenomicsmonitor touse changes Thegenetic these require separate management: these units include units these management: separate require to enough distinct are that species within units ing The harvest of phenotypically desirable animals from A great deal of effort is currently involved in - describ and or drought tolerance, so that management may landscape genomics a

priori

management units management hypotheses about specific loci or - selec www.nature.com/reviews/genetics approach allows the map- (ES . The identificationofThe . U s), distinct populationdistinct future. loci. 70 occur . In addition, 72 71 . . 72 71 69 . .

REVIEWS

Table 2 | Major techniques for detecting DNA sequence variation and considerations for conservation applications Traditional qPCR-based SNP High-density Targeted DNA Anonymous DNA Whole-genome markers chips SNP chips sequencing sequencing resequencing Summary Various techniques Hybridizing array; High-density Fragment High-throughput Sequencing of for small to genotyping by oligonucleotide capture with sequencing whole genome moderate numbers real-time qPCR hybridizing array oligonucleotide of reduced for multiple of markers with fluorescent array; genotyping representation individuals in a probes by next-generation genomic DNA sample sequencing fragments Examples Microsatellites; Fluidigm Affymetrix Exon capture110 RAD sequencing127 Next-generation exon-priming dynamic arrays; GeneChip; and future intron-crossing Illumina Golden Illumina sequencing markers Gate; Applied BeadChip129 technologies Biosystems OpenArray128 General considerations Cost per sample Variable US$10–50 $200–500 $200–1,000 $50–150 $500–5,000 Number of markers 101–102 102–103 104–105 104–105 104–106 Complete genome Applicability to Moderate Low Low Low–moderate High Low new taxa Ability to target Yes Yes Yes Yes No Yes candidate loci (bioinformatically) DNA quality Low Low High High Low–moderate High required Equipment PCR machine; $100,000 platform $150,000 platform $5,000 for PCR machine; Next-generation needed traditional equipment; next-generation sequencer, sequencer next-generation sequencer bioinformatics sequencer resources Utility Pedigree/kin in Limited Limited–moderate Yes Yes Yes Data overkill in wild populations; most cases individual-based population genetics Neutral Yes, but variance Yes, but variance Increased Increased Increased More data than (genome-wide due to few markers due to moderate accuracy; can accuracy; can accuracy; no needed average) number of markers include candidate include candidate previous genomic landscape loci loci resources genetics Detecting loci Useful after Useful after Allows genome Allows genome Dense genome Most appropriate of interest markers at markers at scanning along scanning along coverage for in family-based (inbreeding candidate loci candidate loci with candidate loci with candidate de novo mapping studies depression, have been have been loci and targeting outbreeding identified identified chromosome depression, local regions adaptation) Marker-assisted Efficient screening Efficient screening Allows genomic Overkill after Overkill after Overkill after restoration of few known of few known selection key markers are key markers are key markers are markers markers approaches identified identified identified qPCR, quantitative PCR; RAD, restriction-site-associated DNA.

quotas (to avoid overharvesting, for example) or to modifications by humans. Hybridization has contributed devise ways to translocate and reintroduce individuals to the extinction of many species73,74. Genomics could (to avoid, for example, the mixing of adaptively dif- have an important role in distinguishing between natural ferentiated populations). It is sometimes necessary to and anthropogenic hybridization73. Also, genomics pro- prioritize population units for conservation owing to vides the potential to predict the effects of hybridization limited financial resources. on fitness (heterosis or outbreeding depression). Hybridization is one of the major threats to con- Phenology servation of many plant and animal species73. Rates Units of conservation. The description of conserva- The timing of periodic introgression biological phenomena of hybridization and have increased dra- tion units generally requires two steps: estimating the that are usually correlated matically worldwide because of widespread intentional amount of gene flow among populations and evaluat- with climatic conditions. and incidental translocations of organisms and habitat ing the amount of adaptive divergence. The ability to

nature reviews | Genetics Volume 11 | october 2010 | 703 REVIEWS

genotype many neutral loci will provide much better 12 Atlantic salmon (Salmo salar) populations to exam- estimates of the patterns of reproductive isolation and ine global patterns of gene expression and found they demographic history of populations to address the first were concordant with patterns of divergence at seven step. Genomic approaches for studying functional genes microsatellite loci. These results support the notion that will provide the opportunity to evaluate the amount of patterns of divergence at neutral loci reflect patterns of adaptive divergence among populations required in the adaptive variation in gene expression. second step, and its distribution across the genome. Conservation units have been described on the Detection of hybridization. Molecular detection of basis of divergence at loci that are assumed to be selec- hybridization and estimation of the proportion of admix- tively neutral. It has been suggested that this could be ture between genetically divergent populations can be improved by including genetic divergence at adaptive accomplished accurately with tens of loci73,82. However, markers along with the divergence at neutral loci75–77. accurate description of the dynamics of hybridization Adaptive markers could enhance and help set priori- and introgression can require hundreds of loci83. In addi- ties for the identification and management of units of tion, estimation of the proportion of admixture within conservation. However, a complete understanding individuals will require many more markers. of adaptive divergence is unattainable. Moreover, a For example, Halbert and Derr84 found that 7 of 11 recent comparison of assumed neutral and putatively US federal bison (Bos bison) populations contained selected alleles in over 640,000 autosomal SNPs in introgression from domestic cattle (Bos taurus) based humans concluded that average allele frequency diver- on 14 nuclear loci. The conservation value of admixed gence is highly predictive of adaptive divergence and populations has been controversial73,85,86, and some that neutral processes (population history, migration believe that these herds should not be considered as and effective population size) exert powerful influences bison for conservation purposes87. However, this posi- Landscape genomics 78 87 The study of many markers, over the geographic distribution of selected alleles . tion has not been generally accepted . Regardless, the including markers in genes This result supports the use of neutral loci to provide potential to estimate the proportion of cattle alleles in under selection, in spatially useful descriptions of the patterns of divergence at individual bison will allow the selection of individuals referenced samples collected adaptive loci. to reduce the magnitude of introgression from cattle in across a landscape and often across selection gradients. It There are pitfalls in focusing on individual adap- managed bison herds. uses comparisons of adaptive tive loci rather than neutral patterns or genome-wide Genomics provides exciting opportunities to assess and neutral variation to quantify averages. Genes important for contemporary or past differential rates of introgression across different the effects of landscape adaptations might not be those that will be crucial genomic regions following hybridization88. For exam- features and environmental for adaptation in future environments. In addition, much ple, Fitzpatrick et al.89 found that 3 of 68 markers variables on gene flow and spatial genetic variation. effort has been devoted recently to genome-wide asso- spread rapidly into native California tiger salamanders ciation studies for detecting the genetic basis of com- (Ambystoma californiense), whereas the other 65 markers Evolutionarily significant unit plex traits, particularly disease in humans, using large show little evidence of spread beyond the region where A classification of populations samples of individuals and genetic markers. Although introductions of non-native barred tiger salamanders that have substantial reproductive isolation which has many candidate genes have been identified, often a large (Ambystoma tigrinum mavortium) occurred. Differential 79 led to adaptive differences so proportion of the heritability remains unexplained . A introgression rates of genomic regions raises some dif- that the population represents focus on detectable adaptive genomic regions could ficult issues with regards to treating hybridized popu- a significant evolutionary result in loss of important genetic variation at other lations in conservation89 and brings into question the component of the species. regions. Moreover, even when the same genomic regions efficacy of using a few (that is, ten or so) neutral markers Distinct population segment are implicated in, for example, local adaptation across to detect hybridization. A classification under the populations, the particular alleles involved may be dif- Endangered Species Act ferent and perhaps even result in outbreeding depression Outbreeding depression. Concerns about the possibility of the United States that when combined. of outbreeding depression have restricted, perhaps allows for legal protection of populations that are distinct, Landscape genomics will help to identify manage- unnecessarily, the use of managed gene flow to avoid relatively reproductively ment units by providing sufficient power to localize increased risks of extinction caused by loss of genetic isolated and represent a boundaries on the landscape that separate demograph- variation because of habitat fragmentation and isola- significant evolutionary ically independent groups. Examination of hundreds tion. Frankham72 has identified the development of lineage to the species. to thousands of loci in hundreds of individuals across methods for predicting outbreeding depression as the Management unit landscapes will improve assessments of the interac- top priority in conservation genetics. Outbreeding A local population that is tions of gene flow, genetic drift and natural selection depression can result from either chromosomal or genic managed as a unit owing to its in influencing the evolution and persistence of popula- incompatibilities between hybridizing taxa (intrinsic demographic independence. tions. Landscape genomics will help to identify ESUs outbreeding depression) or reduced adaptation to

Introgression (and spatial locations of boundaries between them) by local environmental conditions (extrinsic outbreeding 90 Gene flow between including both neutral and adaptive variation. depression) . Genomic approaches can potentially populations or species Recent papers have explored the potential of tran- provide valuable empirical information for predicting whose individuals hybridize. scriptomic analysis of gene expression to assess func- the probability of either of these sources of outbreeding tional genetic divergence among populations80; for depression; for example, next-generation sequencing Heterosis 81 When hybrid individuals example, Tymchuk et al. hybridized a microarray using paired-end reads can be used to detect chromosomal 91 have greater fitness than with 16,000 salmonid cDNAs (16K cDNA microarray) rearrangements , such as large inversions or gene copy either of the parental types. to RNA extracted from whole fry raised in captivity in number variation92.

704 | october 2010 | Volume 11 www.nature.com/reviews/genetics REVIEWS

Genomic approaches will also be increasingly Minimizing adaptation to captivity. The emphasis of used to detect outbreeding depression by estimating captive breeding protocols has been to reduce genetic the number of progeny produced by individuals with drift by maximizing effective population size98, which different proportions of admixture. For example, is appropriate for captive breeding programmes of Muhlfeld et al.93 estimated the individual propor- mammals and birds in zoos that have a relatively small tion of admixture between introduced rainbow trout number of individuals that are managed using pedigrees. (Oncorhynchus mykiss) and native westslope cutthroat (FIG. 2) trout (Oncorhynchus clarkii lewisi) . a 16 Captive breeding and assisted migration Genomic tools may assist the management of ex situ 14 emale populations and reintroductions by providing increased 12 precision and accuracy of estimates of neutral popula- tion genetic parameters and by identifying specific loci 10 spring per f of importance, which is essential for selecting founder 8 individuals. First, many neutral loci could be used to construct a more precise pedigree of the captive popula- 6 tion and determine whether the founders from the wild 4 are kin. Second, screening of the founders for known 2

deleterious recessive alleles could substantially reduce Mean number of off any subsequent inbreeding depression in the captive 0 population. In addition, screening of the founders for 0 0.2 0.4 0.6 0.8 1.0 known adaptive alleles could increase the evolutionary Proportion of rainbow trout admixture potential of the captive population. b 5 Managing inbreeding depression. The overarching goal of maintaining genetic diversity in an ex situ population 4 pre-dates genomic techniques. Nonetheless, genome scans may produce better estimates of genome-wide het- 3 erozygosity and genetic diversity than smaller numbers male per spring of traditional markers, such as microsatellites94. Methods are being developed to maximize the sampling of genetic 2 variation for founders of captive breeding colonies 95 based on genomic data . A caveat here is that the rela- 1 tionship between genome-wide average heterozygosity and inbreeding depression is not always strong. As a Mean number of off result, a more powerful application of genomics may be 0 to estimate pedigrees and degrees of relatedness among 0 0.2 0.4 0.6 0.8 1.0 Proportion of rainbow trout admixture captive or founding individuals35,96, allowing captive management plans to minimize inbreeding per se. Figure 2 | Effects of proportion of individual admixture The ability to use genomics to identify specific loci with introduced rainbow trout onNa theture fitness Reviews of | Genetics native related to local adaptation or inbreeding depression and westslope cutthroat trout. Sixteen microsatellite loci Marker-assisted selection the success of marker-assisted selection in livestock were used to estimate the individual proportion of The use of molecular genetic and crops97 raise the possibility of managing specific admixture between introduced rainbow trout and native markers to increase the westslope cutthroat trout93. These same loci were used to loci in some conservation situations. For example, response to selection in a identify the parents of progeny produced in a stream over population by the favouring of individuals with particular adaptive genetic variants a 5-year period. The bubble plots show the mean number reproduction by individuals could be chosen for reintroduction or genetic rescue. of offspring per individual identified plotted against the with a certain allele or In captive breeding programmes, particular genetic proportion of rainbow trout admixture for females (a) and genotype. The marker is variants could be selected against. In one example, the males (b). In a bubble plot, the size of the bubble is closely linked to a quantitative trait locus. small population of the California condor (Gymnogyps proportional to the number of observations with that californianus) has a relatively high frequency of a recessive value. The mean values for first-generation hybrids are Genetic rescue lethal allele causing chondrodystrophy. A condor genom- shown as triangles; these points were not included in the The recovery in the average ics project is seeking a marker to identify carriers of the regression. These results are striking in two ways. First, fitness of individuals through chondrodystrophy allele, and members of this project there was a strong reduction in the number of progeny increased gene flow into small produced as the amount of admixture with introduced have therefore developed several genomic resources, populations, typically following rainbow trout increased in both females and males. a fitness reduction due to including a bacterial artificial chromosome library Second, first-generation hybrids had much greater inbreeding depression. 17 and a fibroblast cell line for transcriptomic analysis , reproductive success than other individuals with 50% with the goal of designing breeding programmes to Chondrodystrophy admixture. This suggests a strong heterotic effect in the A genetically based skeletal select against heterozygotes for the chondrodystrophy first-generation hybrids caused by sheltering of disorder that affects the allele while minimizing loss of genetic diversity deleterious recessive alleles. Figure is reproduced, with development of cartilage. elsewhere in the genome. permission, from REF. 93 © (2009) The Royal Society.

nature reviews | Genetics Volume 11 | october 2010 | 705 REVIEWS

However, adaptation to captivity is a serious problem Choosing genomic approaches associated with captive breeding programmes for many The diverse and growing list of genomic techniques species99,100. This will inevitably reduce the fitness of provides a range of options for experimental design individuals reintroduced to wild or natural conditions. (TABLE 2). Currently, array-based techniques (SNP chips) For example, tameness in response to humans is gener- can efficiently genotype markers across many individuals ally advantageous in captivity but can have serious con- for a range of conservation applications. As the cost of sequences in the wild. In addition, increasing effective sequencing continues to fall, reduced-representation population size for some captive species (for example, sequencing may replace SNP chips as a preferred method fish and plants) may increase the rate of adaptation to in many cases110. Sequence data can provide additional captive conditions. Genetic monitoring101 of many loci information for functional assessment of candidate throughout the genome should become a standard tool genes or detection of haplotype structure or inversion for detecting adaptation to captivity (that is, rapid, locus- polymorphisms, and sequencing is easily applied to taxa specific change in allele frequencies) in species for which without any existing genomic resources. However, at adaptation to captivity is a concern100. least in the near term, array techniques will retain their advantage of having a highly standardized protocol for Restoration. The condor example highlights the com- genotyping a fixed set of markers. This makes them well- plexity of identifying specific loci to allow targeted suited to, for example, long-term genetic monitoring genetic management of populations, even when a single of populations. Mendelian locus is implicated. However, the success of It is becoming feasible to sequence complete genomes marker-assisted selection in livestock is due in part to in a reasonable research timeline and budget111. Whole- the fact that specific alleles and their functional roles genome resequencing of all individuals in a study will need not be determined; rather a correlation between become an option in conservation112. However, while phenotype and genotype at multiple markers is estab- there are potential uses for whole-genome resequenc- lished, and selection on genotype produces a correlated ing, such as detection of Mendelian inherited traits in response in phenotype (for example, growth rate or dis- families53, in most situations it is likely to create more ease resistance). Given the ability to identify genomic challenges than it solves. First, because of linkage dis- regions correlated with local adaptation (BOX 3), con- equilibrium, dense marker genotyping already provides servation genomics could similarly use this informa- a nearly complete view of genomic variation113. Such tion in, for example, selecting source populations for genomic structure is likely to be even more pronounced translocation or reintroduction. A general risk in such in small populations of conservation concern than in efforts is outbreeding depression as a result of differ- traditional model organisms46; whole-genome rese- ent and incompatible genetic bases of adaptation in the quencing is thus data overkill. Moreover, whole-genome two populations. The choice of source population can resequencing introduces many challenges for compu- now be informed by four factors: ecological similar- tational bioinformatics; the resources simply to store, ity, phenotypic similarity, genome-wide similarity as assemble and analyse such large data sets may outweigh indicated by neutral markers, and genetic similarity at their benefits, at least for the near future. adaptive loci. We envision an emerging standard for conservation Genetic rescue has been used as an effective res- genomics in which the starting point will be a reference toration tool to avoid or reverse the consequences of genome sequence. A rapidly growing number of species, inbreeding depression102. However, the identification of particularly vertebrates, have reference sequences avail- individual loci with major adaptive effects (for exam- able already114, or an initial investment can be made to ple, major histocompatibility complex in animals103 and produce one. From this point, genotyping of multiple self-incompatibility loci in plants104) raises the possibility individuals from population samples would be done of allele-specific genetic rescue. Interestingly, other loci with array-based or reduced-representation sequenc- with exceptionally strong fitness effects are being found ing techniques, with the reference sequence providing a in a number of species, such as PanI105 in the cod family valuable resource for sequence alignment and candidate (Gadidae) and Pgi in butterflies and other insects106. It gene identification and annotation. remains to be seen whether such loci are unusual or are present in most species. Perspective Research in community genomics suggests that indi- This is an exciting and challenging time for conserva- vidual alleles can affect community diversity and com- tion genetics. Genomic approaches have the potential position107–109. For example, alleles at tannin loci in to transform the management of populations for con- cottonwood trees increase the palatability and decay servation in various ways, from estimates of pedigrees rate of leaves, which in turn influences the abundance and inbreeding based on large numbers of markers to of soil microbes, fungi and arboreal insects and birds108. identification of loci responsible for local adaptation Loss or restoration of such alleles to populations could and outbreeding depression. Genomics also provides the Community genomics thus influence community diversity and ecosystem potential to understand the genetic basis of interactions The study of the effect of function108. Nevertheless, the complexity of these inter- among species, which could greatly enhance our abil- individual alleles or genotypes on the species composition, actions presents real challenges before it will be pos- ity to manage communities rather than just individual diversity or functioning of a sible to use this information in a practical conservation species. Perhaps the greatest contribution of genomics community or ecosystem. situation. to conservation will be the precise genomic monitoring

706 | october 2010 | Volume 11 www.nature.com/reviews/genetics REVIEWS

Epigenetics of changes in allelic frequency to quantify the effects of the paradox of invasive species that have lost genetic Changes in or gene expression genetic drift, natural selection and hybridization in wild variation during a bottleneck associated with their caused by mechanisms and captive populations. introduction but are nonetheless able to adapt to new other than changes in the Although we have focused on genomic techniques environmental conditions119. underlying DNA sequence, such as DNA methylation that detect variation in DNA sequences, emerging tech- Recognizing the limitations of new techniques is and histone modifications. niques also allow the study of , which may also essential. Improved basic scientific understanding have an important role in conservation genetics in the through genomics will not necessarily lead to improved Vital rates future115,116. There is increasing evidence that epigenetic conservation. For example, genomics will make it pos- Demographic values that processes can be important following hybridization and sible to provide genome-wide estimates of functional affect population growth 115,117 1 (for example, age-specific in outbreeding depression . In addition, epigenetic genetic variation and fitness . Nevertheless, this will survival, fecundity and age effects might be an important source of variation for not be sufficient to improve our estimates of popula- at first reproduction). invasive species. Richards et al.118 have shown that the tion viability unless we are able to make the connec- invasive Japanese knotweed (Fallopia spp.), which has tions between individual fitness and population growth little variation in DNA sequence, maintains substantial rates120 (FIG. 1). To make these connections will require phenotypic variation even under controlled environ- long-term studies of individual fitness and of the effects mental conditions. Epigenetic effects associated with this of fitness differences among individuals on demographic phenotypic variation might enhance knotweed’s ability to vital rates. This is perhaps the most important and invade novel environments. This could partially explain difficult future challenge facing conservation genetics.

1. Ellegren, H. & Sheldon, B. C. Genetic basis of fitness 18. Allendorf, F. W. & Seeb, L. W. Concordance of 35. Santure, A. W. et al. On the use of large marker differences in natural populations. Nature 452, genetic divergence among sockeye salmon panels to estimate inbreeding and relatedness: 169–175 (2008). populations at allozyme, nuclear DNA, and empirical and simulation studies of a pedigreed An important paper that reviews current mitochondrial DNA markers. Evolution 54, zebra finch population typed at 771 SNPs. understanding of the molecular basis of fitness 640–651 (2000). Mol. Ecol. 19, 1439–1451 (2010). differences between individuals in natural 19. Luikart, G. H., England, P., Tallmon, D. A., Jordan, S. & 36. Smouse, P. How many SNPs are enough? Mol. Ecol. populations. Taberlet, P. The power and promise of population 19, 1265–1266 (2010). 2. Slate, J. et al. Gene mapping in the wild with SNPs: genomics: from genotyping to genome-typing. Nature 37. Pemberton, J. M. Wild pedigrees: the way forward. guidelines and future directions. Genetica 136, Rev. Genet. 4, 981–994 (2003). Proc. Biol. Sci. 275, 613–621 (2008). 97–107 (2009). 20. Storz, J. F. Using genome scans of DNA polymorphism A persuasive argument for the value of 3. Kohn, M. H., Murphy, W. J., Ostrander, E. A. & to infer adaptive population divergence. Mol. Ecol. 14, constructing pedigrees in wild populations to Wayne, R. K. Genomics and conservation genetics. 671–688 (2005). investigate major issues in evolutionary biology, Trends Ecol. Evol. 21, 629–637 (2006). 21. Giger, T. et al. Life history shapes gene expression in including the genetic architecture of traits, 4. Pertoldi, C. et al. Genome variability in European and salmonids. Curr. Biol. 16, R281–R282 (2006). inbreeding depression and inbreeding avoidance. American bison detected using the BovineSNP50 22. Landry, P.‑A., Koskinen, M. T. & Primmer, C. R. 38. Sham, P., Cherny, S. & Purcell, S. Application of BeadChip. Conserv. Genet. 11, 627–634 (2010). Deriving evolutionary relationships among populations genome-wide SNP data for uncovering pairwise 5. Thomson, R. C., Wang, I. J. & Johnson, J. R. using microsatellites and (δμ)2: all loci are equal, but relationships and quantitative trait loci. Genetica 136, Genome-enabled development of DNA markers for some are more equal than others. Genetics 161, 237–243 (2009). ecology, evolution and conservation. Mol. Ecol. 19, 1339–1347 (2002). 39. Pritchard, J. K., Stephens, M. & Donnelly, P. 2184–2195 (2010). 23. Nordborg, M. Structured coalescent processes on Inference of population structure using multilocus 6. Kerstens, H. et al. Large scale single nucleotide different time scales. Genetics 146, 1501–1514 genotype data. Genetics 155, 945–959 (2000). polymorphism discovery in unsequenced genomes (1997). 40. Hubisz, M. J., Falush, D., Stephens, M. & using second generation high throughput sequencing 24. Beaumont, M. A. Detecting population expansion and Pritchard, J. K. Inferring weak population structure technology: applied to turkey. BMC Genomics 10, 479 decline using microsatellites. Genetics 153, with the assistance of sample group information. (2009). 2013–2029 (1999). Mol. Ecol. Resour. 9, 1322–1332 (2009). 7. van Bers, N. E. M. et al. Genome-wide SNP detection 25. Beerli, P. & Felsenstein, J. Maximum likelihood 41. Rosenberg, N. A. et al. Genetic structure of human in the great tit Parus major using high throughput estimation of a migration matrix and effective populations. Science 298, 2381–2385 (2002). sequencing. Mol. Ecol. 19, 89–99 (2010). population sizes in n subpopulations by using a 42. Wilson, G. A. & Rannala, B. Bayesian inference of 8. DeLong, E. F. The microbial ocean from genomes to coalescent approach. Proc. Natl Acad. Sci. USA 98, recent migration rates using multilocus genotypes. biomes. Nature 459, 200–206 (2009). 4563–4568 (2001). Genetics 163, 1177–1191 (2003). A review of metagenomics in marine systems, 26. Skare, O., Sheehan, N. & Egeland, T. Identification of 43. Faubet, P., Waples, R. S. & Gaggiotti, O. E. including transcriptomic and functional approaches distant family relationships. Bioinformatics 25, Evaluating the performance of a multilocus Bayesian linking microbial genomes to ecosystem processes. 2376–2382 (2009). method for the estimation of migration rates. 9. Dinsdale, E. A. et al. Functional metagenomic profiling 27. Browning, S. R. & Weir, B. S. Population structure with Mol. Ecol. 16, 1149–1166 (2007). of nine biomes. Nature 452, 629–634 (2008). localized haplotype clusters. Genetics 10 May 2010 44. Rannala, B. & Mountain, J. L. Detecting immigration 10. Vega Thurber, R. L. et al. Metagenomic analysis (doi:10.1534/genetics.110.116681). by using multilocus genotypes. Proc. Natl Acad. Sci. indicates that stressors induce production of herpes- 28. Lecis, R. et al. Bayesian analyses of admixture in wild USA 94, 9197–9201 (1997). like viruses in the coral Porites compressa. Proc. Natl and domestic cats (Felis silvestris) using linked 45. Holderegger, R. & Wagner, H. H. Landscape genetics. Acad. Sci. USA 47, 18413–18418 (2008). microsatellite loci. Mol. Ecol. 15, 119–131 (2006). BioScience 58, 199–207 (2008). 11. Qin, J. et al. A human gut microbial gene catalogue 29. Anderson, C. & Meikle, D. Genetic estimates of An integrated review of the emerging field of established by metagenomic sequencing. Nature 464, immigration and emigration rates in relation to landscape genetics. 59–65 (2010). population density and forest patch area in 46. Hagenblad, J., Olsson, M., Parker, H. G., 12. Nielsen, E. E., Hemmer-Hansen, J., Larsen, P. F. & Peromyscus leucopus. Conserv. Genet. 5 Feb 2010 Ostrander, E. A. & Ellegren, H. Population genomics Bekkevold, D. Population genomics of marine fishes: (doi:10.1007/s10592-009-0033-8). of the inbred Scandinavian wolf. Mol. Ecol. 18, identifying adaptive variation in space and time. 30. Bollback, J. P., York, T. L. & Nielsen, R. Estimation of 1341–1351 (2009).

Mol. Ecol. 18, 3128–3150 (2009). 2Nes from temporal allele frequency data. Genetics 47. Casellas, J., Varona, L., Ibanez-Scriche, N., 13. Murchison, E. P. et al. The Tasmanian devil 179, 497–502 (2008). Quintanilla, R. & Noguera, J. L. Skew distribution reveals Schwann cell origins of a clonally 31. Wang, J. & Santure, A. W. Parentage and sibship of founder-specific inbreeding depression effects transmissible cancer. Science 327, 84–87 (2010). inference from multilocus genotype data under on the longevity of landrace sows. Genet. Res. 90, 14. Avise, J. Perspective: conservation genetics enters the polygamy. Genetics 181, 1579–1594 (2009). 499–508 (2008). genomics era. Conserv. Genet. 11, 665–669 (2010). 32. Glaubitz, J. C., Rhodes, O. E. & DeWoody, J. A. 48. Lacy, R. C., Alaks, G. & Walsh, A. Hierarchical analysis 15. Ouborg, N. J., Pertoldi, C., Loeschcke, V., Bijlsma, R. & Prospects for inferring pairwise relationships with of inbreeding depression in Peromyscus polionotus. Hedrick, P. W. Conservation genetics in transition to single nucleotide polymorphisms. Mol. Ecol. 12, Evolution 50, 2187–2200 (1996). conservation genomics. Trends Genet. 26, 177–187 1039–1047 (2003). 49. Casellas, J., Piedrafita, J., Caja, G. & Varona, L. (2010). 33. Jones, O. R. & Wang, J. Molecular marker-based Analysis of founder-specific inbreeding depression on 16. Primmer, C. R. From conservation genetics to pedigrees for animal conservation biologists. birth weight in Ripollesa lambs. J. Anim. Sci. 87, conservation genomics. Ann. N. Y. Acad. Sci. 1162, Anim. Conserv. 13, 26–34 (2010). 72–79 (2009). 357–368 (2009). 34. Keinan, A. & Reich, D. Human population 50. Charlier, C. et al. Highly effective SNP-based 17. Romanov, M. N. et al. The value of avian genomics to differentiation is strongly correlated with local association mapping and management of recessive the conservation of wildlife. BMC Genomics 10, S10 recombination rate. PLoS Genet. 6, e1000886 defects in livestock. Nature Genet. 40, 449–454 (2010). (2010). (2008).

nature reviews | Genetics Volume 11 | october 2010 | 707 REVIEWS

51. Vermeulen, C. J., Bijlsma, R. & Loeschcke, V. 73. Allendorf, F. W., Leary, R. F., Spruell, P. & 95. Miller, W., Wright, S. J., Zhang, Y., Schuster, S. C. & QTL mapping of inbreeding-related cold sensitivity Wenburg, J. K. The problems with hybrids: Hayes, V. M. Optimization methods for selecting and conditional lethality in Drosophila melanogaster. setting conservation guidelines. Trends Ecol. Evol. founder individuals for captive breeding or J. Evol. Biol. 21, 1236–1244 (2008). 16, 613–622 (2001). reintroduction of endangered species. Pac. Symp. 52. Kristensen, T. N., Pedersen, K. S., Vermeulen, C. J. & 74. Levin, D. A., Franciscoortega, J. & Jansen, R. K. Biocomput. 2010, 43–53 (2010). Loeschcke, V. Research on inbreeding in the ‘omic’ era. Hybridization and the extinction of rare plant species. 96. Blouin, M. S. DNA-based methods for pedigree Trends Ecol. Evol. 25, 44–52 (2010). Conserv. Biol. 10, 10–16 (1996). reconstruction and kinship analysis in natural An important evaluation of the use of new 75. Crandall, K. A., Binindaemonds, O. R. P., Mace, G. M. populations. Trends Ecol. Evol. 18, 503–511(2003). technologies to understand the genetic basis of & Wayne, R. K. Considering evolutionary processes in 97. Goddard, M. E. & Hayes, B. J. Mapping genes for inbreeding depression. conservation biology. Trends Ecol. Evol. 15, 290–295 complex traits in domestic animals and their use in 53. Roach, J. C. et al. Analysis of genetic inheritance in a (2000). breeding programmes. Nature Rev. Genet. 10, family quartet by whole-genome sequencing. Science 76. Hedrick, P. W., Parker, K. M. & Lee, R. N. 381–391 (2009). 328, 636–639 (2010). Using microsatellite and MHC variation to 98. Lacy, R. C. in Conservation Genetics in the Age of 54. Nielsen, R. Molecular signatures of natural selection. identify species, ESUs, and MUs in the endangered Genomics (eds Amato, G., Ryder, O., Rosenbaum, H. & Annu. Rev. Genet. 39, 197–218 (2005). Sonoran topminnow. Mol. Ecol. 10, 1399–1412 DeSalle, R.) 58–81 (Columbia Univ. Press, New York, 55. Oleksyk, T. K., Smith, M. W. & O’Brien, S. J. (2001). 2009). Genome-wide scans for footprints of natural 77. Bonin, A., Nicole, F., Pompanon, F., Miaud, C. & 99. Araki, H., Cooper, B. & Blouin, M. S. Genetic effects of selection. Philos. Trans. R. Soc. Lond. B 365, Taberlet, P. Population adaptive index: a new method captive breeding cause a rapid, cumulative fitness 185–205 (2010). to help measure intraspecific genetic diversity and decline in the wild. Science 318, 100–103 (2007). 56. Garrigan, D. & Hedrick, P. W. Detecting adaptive prioritize populations for conservation. Conserv. Biol. One of the first papers to demonstrate a reduction molecular polymorphism: lessons from the MHC. 21, 697–708 (2007). in fitness in wild populations caused by gene flow Evolution 57, 1707–1722 (2003). 78. Coop, G. et al. The role of geography in human from captive populations. 57. Antao, T., Lopes, A., Lopes, R., Beja-Pereira, A. & adaptation. PLoS Genet. 5, e1000500 (2009). 100. Frankham, R. Genetic adaptation to captivity in Luikart, G. LOSITAN: a workbench to detect 79. Frazer, K. A., Murray, S. S., Schork, N. J. & Topol, E. J. species conservation programs. Mol. Ecol. 17,

molecular adaptation based on a Fst-outlier method. and its contribution to 325–333 (2008). BMC Bioinformatics 9, 323 (2008). complex traits. Nature Rev. Genet. 10, 241–252 101. Schwartz, M. K., Luikart, G. & Waples, R. S. 58. Beaumont, M. A. & Balding, D. J. Identifying adaptive (2009). Genetic monitoring as a promising tool for genetic divergence among populations from genome 80. Hansen, M. M. Expression of interest: transcriptomics conservation and management. Trends Ecol. Evol. scans. Mol. Ecol. 13, 969–980 (2004). and the designation of conservation units. Mol. Ecol. 22, 25–33 (2007). 59. Beaumont, M. A. & Nichols, R. A. Evaluating loci 19, 1757–1759 (2010). A foundation paper that defined and organized the for use in the genetic analysis of population 81. Tymchuk, W. V., O’Reilly, P., Bittman, J., MacDonald, D. emerging field of genetic monitoring. structure. Proc. R. Soc. Lond. B 263, 1619–1626 & Schulte, P. Conservation genomics of Atlantic 102. Tallmon, D. A., Luikart, G. & Waples, R. S. (1996). salmon: variation in gene expression between and The alluring simplicity and complex reality of genetic 60. Foll, M. & Gaggiotti, O. A genome-scan method to within regions of the Bay of Fundy. Mol. Ecol. 19, rescue. Trends Ecol. Evol. 19, 489–496 (2004). identify selected loci appropriate for both dominant 1842–1859 (2010). 103. Piertney, S. B. & Oliver, M. K. The evolutionary and codominant markers: a Bayesian perspective. 82. Cornuet, J.‑M., Piry, S., Luikart, G., Estoup, A. & ecology of the major histocompatibility complex. Genetics 180, 977–993 (2008). Solignac, M. New methods employing multilocus 96, 7–21 (2006). 61. Makinen, H. S., Cano, J. M. & Merilä, J. genotypes for selecting or excluding populations as 104. Holmes, G. D., James, E. A. & Hoffmann, A. A. Identifying footprints of directional and balancing origins of individuals. Genetics 153, 1989–2000 Limitations to reproductive output and genetic rescue selection in marine and freshwater three-spined (1999). in populations of the rare shrub Grevillea repens stickleback (Gasterosteus aculeatus) populations. 83. Witherspoon, D. J. et al. Genetic similarities within (Proteaceae). Ann. Bot. 102, 1031–1041 (2008). Mol. Ecol. 17, 3565–3582 (2008). and between human populations. Genetics 176, 105. Pogson, G. H. & Fevolden, S. E. Natural selection and 62. Namroud, M.‑C., Beaulieu, J., Juge, N., Laroche, J. & 351–359 (2007). the genetic differentiation of coastal and Arctic Bousquet, J. Scanning the genome for gene single 84. Halbert, N. D. & Derr, J. N. A comprehensive populations of the Atlantic cod in northern Norway: a nucleotide polymorphisms involved in adaptive evaluation of cattle introgression into US federal test involving nucleotide sequence variation at the population differentiation in white spruce. Mol. Ecol. bison herds. J. Hered. 98, 1–12 (2007). pantophysin (PanI) locus. Mol. Ecol. 12, 63–74 17, 3599–3613 (2008). 85. Allendorf, F. W. et al. Cutthroat trouth (2003). 63. Nielsen, E. E. et al. Genomic signatures of local hybridization and the U. S. Endangered Species 106. Wheat, C. Phosphoglucose isomerase (Pgi) directional selection in a high gene flow marine Act: one species, two policies. Conserv. Biol. 19, performance and fitness effects among Arthropods organism; the Atlantic cod (Gadus morhua). 1326–1328 (2005). and its potential role as an adaptive marker in BMC Evol. Biol. 9, 276 (2009). 86. Campton, D. E. & Kaeding, L. R. Westslope cutthroat conservation genetics. Conserv. Genet. 11, 387–397 64. Oetjen, K. & Reusch, T. B. H. Genome scans detect trout, hybridization, and the U. S. Endangered (2010). consistent divergent selection among subtidal vs. Species Act. Conserv. Biol. 19, 1323–1325 107. Barbour, R. C., Forster, L. G., Baker, S. C., Steane, D. A. intertidal populations of the marine angiosperm (2005). & Potts, B. M. Biodiversity consequences of genetic Zostera marina. Mol. Ecol. 16, 5156–5167 87. Marris, E. The genome of the American west. Nature variation in bark characteristics within a foundation (2007). 457, 950–952 (2009). tree species. Conserv. Biol. 23, 1146–1155 (2009). 65. Vasemägi, A., Nilsson, J. & Primmer, C. R. 88. Payseur, B. A. Using differential introgression in 108. Whitham, T. G. et al. Extending genomics to natural Expressed sequence tag-linked microsatellites hybrid zones to identify genomic regions involved communities and ecosystems. Science 320, 492–495 as a source of gene-associated polymorphisms for in speciation. Mol. Ecol. Resour. 10, 806–820 (2008). detecting signatures of divergence selection in (2010). 109. Crutsinger, G. M. et al. Plant genotypic diversity Atlantic salmon (Salmo salar L.). Mol. Biol. Evol. 22, 89. Fitzpatrick, B. M. et al. Rapid spread of invasive genes predicts community structure and governs an 1067–1076 (2005). into a threatened native species. Proc. Natl Acad. Sci. ecosystem process. Science 313, 966–968 (2006). 66. Hohenlohe, P. A. et al. Population genomic analysis of USA 107, 3606–3610 (2010). 110. Hodges, E. et al. Genome-wide in situ exon capture for parallel adaptation in threespine stickleback using A fascinating study showing that natural selection selective resequencing. Nature Genet. 39, sequenced RAD tags. PLoS Genet. 6, e1000862 can rapidly accelerate the rate of introgression for 1522–1527 (2007). (2010). certain regions of the genome from introduced into 111. Nowrousian, M. Next-generation sequencing One of the first papers to use genomic scans of native species. techniques for eukaryotic microorganisms: sequencing- thousands of markers to understand the genetic 90. Edmands, S. Between a rock and a hard place: based solutions to biological problems. Eukaryot. Cell basis of adaptation in natural populations. evaluating the relative risks of inbreeding and 2 Jul 2010 (doi:10.1128/EC.00123-10). 67. Wilding, C. S., Butlin, R. K. & Grahame, J. outbreeding for conservation and management. 112. Li, R. et al. SNP detection for massively parallel whole- Differential gene exchange between parapatric Mol. Ecol. 16, 463–475 (2007). genome resequencing. Genome Res. 19, 1124–1132 morphs of Littorina saxatilis detected using AFLP 91. Shendure, J. & Ji, H. Next-generation DNA (2009). markers. J. Evol. Biol. 14, 611–619 (2001). sequencing. Nature Biotech. 26, 1135–1145 113. International HapMap Consortium. A second 68. Pennings, P. S. & Hermisson, J. Soft sweeps II — (2008). generation human haplotype map of over 3.1 million molecular population genetics of adaptation from 92. Hoffmann, A. A. & Rieseberg, L. H. Revisiting the SNPs. Nature 449, 851–861 (2007). recurrent mutation or migration. Mol. Biol. Evol. 23, impact of inversions in evolution: from population 114. Haussler, D. et al. Genome 10K: a proposal to obtain 1076–1084 (2006). genetic markers to drivers of adaptive shifts and whole-genome sequence for 10000 vertebrate 69. Voight, B. F., Kudaravalli, S., Wen, X. & Pritchard, J. K. speciation? Annu. Rev. Ecol. Evol. Syst. 39, 21–42 species. J. Hered. 100, 659–674 (2009). A map of recent positive selection in the human (2008). 115. Bossdorf, O., Richards, C. L. & Pigliucci, M. genome. PLoS Biol. 4, e72 (2006). An important synthesis that evaluates the Epigenetics for ecologists. Ecol. Lett. 11, 106–115 70. Allendorf, F. W. & Hard, J. J. Human-induced evolution importance of chromosomal inversions in (2008). caused by unnatural selection through harvest of wild population genetics and evolution using modern A valuable consideration of the future application animals. Proc. Natl Acad. Sci. USA 106, 9987–9994 molecular approaches. of epigenetics to understanding the ecology of (2009). 93. Muhlfeld, C. C. et al. Hybridization rapidly reduces natural populations. 71. Allendorf, F. W., England, P. R., Luikart, G., fitness of a native trout in the wild. Biol. Lett. 5, 116. Richards, C. L., Bossdorf, O. & Pigliucci, M. What role Ritchie, P. A. & Ryman, N. Genetic effects of harvest 328–331 (2009). does heritable epigenetic variation play in phenotypic on wild animal populations. Trends Ecol. Evol. 23, 94. Ljungqvist, M., Åkesson, M. & Hansson, B. evolution? BioScience 60, 232–237 (2010). 327–337 (2008). Do microsatellites reflect genome-wide genetic 117. Salmon, A., Ainouche, M. L. & Wendel, J. F. 72. Frankham, R. Where are we in conservation genetics diversity in natural populations? A comment Genetic and epigenetic consequences of recent and where do we need to go? Conserv. Genet. 11, on Väli. et al. (2008). Mol. Ecol. 19, 851–855 hybridization and polyploidy in Spartina (Poaceae). 661–663 (2010). (2010). Mol. Ecol. 14, 1163–1175 (2005).

708 | october 2010 | Volume 11 www.nature.com/reviews/genetics REVIEWS

118. Richards, C. L. et al. Plasticity in salt tolerance traits 124. Hemmer-Hansen, J., Nielsen, E. E., Frydenberg, J. & Acknowledgements allows for invasion of novel habitat by Japanese Loeschcke, V. Adaptive divergence in a high gene flow This article is based partially on work supported by the US knotweed s. l. (Fallopia japonica and F. bohemica, environment: Hsc70 variation in the European National Science Foundation grants DEB 074218 to F.W.A. and Polygonaceae). Am. J. Bot. 95, 931–942 (2008). flounder (Platichthys flesus L.). Heredity 99, G.L., and IOS 0843392 to P.A.H. G.L. also received support 119. Allendorf, F. W. & Lundquist, L. L. Introduction: 592–600 (2007). from the Walton Family Foundation and research grants PTDC/ population biology, evolution, and control of invasive 125. Lowe, W. H. & Allendorf, F. W. What can genetics tell BIA-BDE/65625/2006 and PTDC/CVT/69438/2006 from the species. Conserv. Biol. 17, 24–30 (2003). us about population connectivity? Mol. Ecol. 19, Portuguese Science Foundation. We thank D. E. Campton, 120. Coulson, T. et al. Estimating individual contributions 3038–3051 (2010). R. Frankham, O. Gaggiotti, P. Hedrick, L. S. Mills, B. A. Payseur, to population growth: evolutionary fitness in ecological 126. Waples, R. S. & Gaggiotti, O. What is a population? K. M. Ramstad, M. K. Schwartz, P. Sunnucks and D. A. Tallmon time. Proc. Biol. Sci. 273, 547–555 (2006). An empirical evaluation of some genetic methods for useful comments, and W. H. Lowe for endless EndNote 121. Palsbøll, P. J., Berube, M. & Allendorf, F. W. for identifying the number of gene pools and their tutoring to F.W.A. Identification of management units using population degree of connectivity. Mol. Ecol. 15, 1419–1439 genetic data. Trends Ecol. Evol. 22, 11–16 (2007). (2006). Competing interests statement 122. Waples, R. S. Separating the wheat from the chaff: An extremely valuable paper that considers the The authors declare no competing financial interests. patterns of genetic differentiation in high gene flow fundamental problem of defining ‘population’ in species. J. Hered. 89, 438–450 (1998). population genetics. An important paper that considers how to interpret 127. Baird, N. A. et al. Rapid SNP discovery and genetic the low genetic differentiation observed between mapping using sequenced RAD markers. PLoS ONE 3, FURTHER INFORMATION marine populations that are apparently e3376 (2008). Fred W. Allendorf’s homepage: http://www.cas.umt.edu/ demographically isolated. 128. Perkel, J. SNP genotyping: six technologies that keyed casweb/for_faculty/FacultyDetails.cfm?id=873 123. Pampoulie, C. et al. The genetic structure of Atlantic a revolution. Nature Methods 5, 447–453 (2008). cod (Gadus morhua) around Iceland: insight from 129. Decker, J. E. et al. Resolving the evolution of extant SUPPLEMENTARY INFORMATION See online article: S1 (figure) microsatellites, the PanI locus, and tagging and extinct ruminants with high-throughput experiments. Can. J. Fish. Aquat. Sci. 63, 2660–2674 phylogenomics. Proc. Natl Acad. Sci. USA 106, All links are active in the online pdf (2006). 18644–18649 (2009).

nature reviews | Genetics Volume 11 | october 2010 | 709 online only

might be important for understanding the effects of hybridization ToC Blurb and predicting outbreeding depression. • Improved basic scientific understanding through genomics will not 000 Genomics and the future of conservation necessarily lead to improved conservation. For example, under- genetics standing the relationship between genetic variation and fitness Fred W. Allendorf, Paul A. Hohenlohe and Gordon itself will not be sufficient to improve our estimates of population Luikart viability. Understanding the connections between individual fit- This article discusses how genomic techniques are ness and population growth rates is perhaps the most important expected to provide new insights into important and difficult future challenge facing conservation genetics. problems in conservation and to allow questions to be addressed that have previously not been tractable. Author biographies The authors also offer advice on choosing the most Fred W. Allendorf received his Ph.D. from the University of appropriate genomic approaches for studying Washington, Seattle, USA, and was a postdoctoral fellow at Aarhus different aspects of conservation. University, Denmark, and the University of Nottingham, UK. He has been at the University of Montana, Missoula, USA, since 1976 and cur- Online summary rently has a partial appointment at Victoria University of Wellington, New Zealand. He recently has held fellowships at the University of • We will soon have complete genome sequences from thousands of Western Australia in Perth and Australia’s Commonwealth Scientific species. This coming explosion of information will transform our and Industrial Research Organisation (CSIRO) in Hobart. His pri- understanding of the amount, distribution and functional signifi- mary research focus is the application of population genetics to cance of genetic variation in natural populations. conservation biology. • We identify those problems in conservation biology in which genomics will be most valuable in providing new insights and Paul A. Hohenlohe received his Ph.D. from the University of understanding. We also provide guidelines as to which new genom- Washington and worked as a conservation biologist for the US Bureau ics approaches will be most appropriate for the different problems of Land Management and Forest Service under the Northwest Forest in conservation that can benefit from genetic analysis. Plan. He is currently a research associate at the University of Oregon, • The most straightforward contribution of genomics to conserva- Eugene, USA, and Oregon State University, Corvallis, USA. His tion will be to enormously increase the precision and accuracy research is focused on population genomics approaches to questions of estimation of crucial parameters that require neutral loci (for in conservation and evolution. example, effective population size and migration rate). • Genomic approaches can address important questions about the Gordon Luikart received his Ph.D. at the University of Montana, molecular basis and genetic architecture of inbreeding depression. was a Fulbright fellow at LaTrobe University, Victoria, Australia, a Recent work indicates that the intensity of inbreeding depression research scientist at the Centre National de la Recherche Scientifique can differ greatly depending on which specific individuals are (CNRS), Grenoble, France, and a research scientist at the Centro founders. This suggests that the genetic load is unevenly spread de Investigação em Biodiversidade e Recursos Genéticos (CIBIO), among founder genomes and supports the notion that inbreeding Vairão, Portugal. He is a research associate professor at the Flathead depression sometimes results from major effects at a few loci. Lake Biological Station, University of Montana. His research is aimed • Anthropogenic challenges affect a wide range of species and habi- at developing molecular and computational approaches to bridge the tats. Genomic approaches will allow the identification of adaptive gaps between theory, basic science and practical applications in con- genetic variation related to key traits for the response to climate servation biology and evolutionary ecology. change, such as phenology or drought tolerance, so that manage- ment may focus on maintaining adaptive genetic potential. The use of genomics to monitor genetic change caused by the harvesting of animals by humans could be extremely important because early detection of potentially harmful genetic change will maximize our ability to implement management to limit or reverse the effects before substantial or irreversible changes occur. • Genomics provides exciting opportunities to assess differential rates of introgression across different genomic regions following hybridization between native and introduced species. The differ- ential introgression rates of genomic regions raise some difficult issues with regards to treating hybridized populations in conserva- tion and bring into question the efficacy of using a few (that is, ten or so) neutral markers to detect hybridization. • Genomic tools will assist the management of ex situ populations and reintroductions by providing increased precision and accuracy of estimates of neutral population genetic parameters and by iden- tifying specific loci of importance, which are essential for selecting select founder individuals. • There is increasing evidence that epigenetic processes can be impor- tant following hybridization. Therefore, an epigenetics perspective