Building strong relationships between conservation genetics and primary industry leads to mutually beneficial genomic advances

Several reviews in the past decade have heralded the benefits of embracing high‐throughput sequencing technologies to inform conservation policy and the management of threatened species, but few have offered practical advice on how to expedite the transition from conservation genetics to conservation genomics. Here, we argue that an effective and efficient way to navigate this transition is to capitalize on emerging synergies between conservation genetics and primary industry (e.g., agriculture, fisheries, forestry and horticulture). Here, we demonstrate how building strong relationships between conservation geneticists and primary industry scientists is leading to mutually‐beneficial outcomes for both disciplines. Based on our collective experience as collaborative New Zealand‐based scientists, we also provide insight for forging these cross‐sector relationships.

One does not need to read beyond the pages of Molecular Ecology to see how emerging technologies are revolutionizing the way we conduct research in ecology and evolutionary biology (i.e. EEB) and conservation biology. This is exemplified by rapid advances in genomics, where in the span of two decades the field of molecular ecology has grown from using Sanger technologies to sequence single target loci to using high-throughput sequencing (HTS) technologies to affordably sequence entire draft genomes (Narum et al. 2013;Payseur & Rieseberg 2016;Tigano & Friesen 2016). When new technologies become available, there is a tendency for reviews to be published heralding their potential to address new and exciting questions. Beyond the value of these reviews, an even more important conversation needs to take place in the peer-reviewed literature: How do we efficiently incorporate new technologies into our research repertoire to make accelerated gains in applied and fundamental science?
The field of conservation genetics is currently in transition given rapid advancements in HTS technologies. Many reviews have highlighted the promise of embracing HTS technologies in conservation (Luikart et al. 2003;Kohn et al. 2006;Primmer 2009;Allendorf et al. 2010;Avise 2010;Frankham 2010a;Ouborg et al. 2010;Angeloni et al. 2011;Ekblom & Galindo 2011;Funk et al. 2012;McCormack et al. 2013;Narum et al. 2013;Steiner et al. 2013;Ellegren 2014;McMahon et al. 2014;Shafer et al. 2015;Andrews et al. 2016;Benestan et al. 2016;Grueber 2016). However, as recently discussed by Shafer et al. (2015Shafer et al. ( , 2016 and Garner et al. (2016), there are a limited (albeit increasing) number of published empirical studies that apply HTS data to conservation. We are aware of empirical genomic studies in EEB that are applicable to questions in conservation (e.g. Defaveri et al. 2013;Hoffman et al. 2014;Knief et al. 2015;B er enos et al. 2016;Hess et al. 2016;Prince et al. 2016) and there are many EEB researchers applying their genomics expertise to improve conservation outcomes for threatened species, including two of our co-authors (MK, AWS). In addition to the EEB sphere, there are conservation geneticists (e.g. our co-authors SJG, TRB, MLH, TES) who are successfully venturing into conservation genomics through collaborations with colleagues in another applied discipline well versed in genomics: primary industry (a collective term referring to scientists in agriculture, fisheries, forestry and horticulture; such as our co-authors RE, JM, RM, PW). Through building these cross-sector relationships, it has become clear that there is immense potential for conservation geneticists and primary industry scientists to collaborate on applied research that addresses aligned questions using similar genomic approaches. In this opinion piece, we use our experience as a collaborative group of New Zealand-based scientists to argue that building strong relationships between conservation genetics and primary industry can lead to improved genomic outcomes for both disciplines and offer advice on how to best build meaningful cross-sector relationships.

Conservation genetics and genomics
Before discussing mutually beneficial genomic synergies between conservation genetics and primary industry, we feel it is important to first address what conservation genetics is, what can be gained using a genomic approach and what obstacles may impede geneticists from adopting genomic technologies. Conservation genetics is a subdiscipline of conservation biology (Soul e 1985) which uses genetic data to inform the management of threatened species in collaboration with conservation practitioners (Frankham 1995(Frankham , 2010bAvise 2008;Haig et al. 2016). While there is overlap between the fields of conservation genetics and EEB, we distinguish conservation genetics as an applied subdiscipline with direct implications for the management of threatened species. Many threatened taxa have experienced significant population declines (i.e. demographic bottlenecks, see Keller et al. 1994), leading to small populations that are susceptible to genetic factors (i.e. loss of genetic diversity, inbreeding and inbreeding depression) associated with extinction risk (Frankham 1995). Conservation geneticists have traditionally used few targeted neutral genetic markers including mitochondrial sequences, microsatellites and amplified fragment length polymorphisms (AFLPs) to measure inbreeding, relatedness and genetic diversity within threatened populations, estimate population genetic structure and gene flow among threatened populations, delineate species boundaries in threatened taxa and detect hybridization and introgression between threatened and nonthreatened species Ouborg et al. 2010).
Advancements in HTS technologies are enabling the development of genomic resources for threatened species including the de novo assembly and annotation of high-quality reference genomes (e.g. Li et al. 2014;Zhang et al. 2014) and characterization of a large number of genome-wide markers such as single nucleotide polymorphisms (SNPs) (e.g. Benestan et al. 2015;Kraus et al. 2015;Lemay & Russello 2015). For conservation geneticists who have traditionally used small panels of neutral genetic markers to estimate population genetic parameters above and below the species level, HTS technologies are appealing as they enable an affordable means to discover and genotype a large quantity of genome-wide SNPs (Avise 2010;McCormack et al. 2013;Shafer et al. 2015) and these large SNP data sets are more representative of genome-wide variation and can result in higher resolution estimates of population genetic parameters (V€ ali et al. 2008;Ljungqvist et al. 2010;Santure et al. 2010;Taylor et al. 2015). In the field of conservation genetics and EEB, a small but rapidly growing number of empirical studies have demonstrated the utility of genomic markers in estimating population genetic structure and gene flow (Bowden et al. 2012;Dierickx et al. 2015;Lew et al. 2015;Oyler-McCance et al. 2015), estimating relatedness (B er enos et al. 2016), measuring genome-wide diversity (Robinson et al. 2016) and detecting hybridization and introgression (Hohenlohe et al. 2011). We anticipate even more conservation geneticists will begin to embrace HTS technologies as empirical evidence demonstrating the superiority of using genomic markers to inform conservation decisions grows and the costs of doing so diminishes (Box 1).
The paradigm underlying many conservation genetic studies is that a genetically diverse population as measured by neutral genetic markers is also likely to be functionally diverse (Bataillon et al. 1996) and therefore better able to adapt to environmental change (Frankham 2005). While many have aspired to move past this paradigm, it remains entrenched in most conservation genetic studies that use neutral markers (Caballero & Garc ıa-Dorado 2013;Vilas et al. 2015). As a result of the lack of empirical data on functional genetic diversity in species of conservation interest, beyond studies that include immunocompetence genes like those in the major histocompatibility complex and Toll-like receptors (reviewed in Grueber 2016), it has been difficult to assess the validity of this conservation genetic paradigm. Further, even if supported by empirical data, neutral genetic data might not be a suitable proxy for functional genetic data for threatened species. For example, the translocation of individuals from a large genetically diverse population to supplement a small genetically depauperate population might introduce new genetic diversity (Weeks et al. 2011;IUCN/SSC 2013), but it might also inadvertently lead to outbreeding depression if source and recipient populations are each locally adapted (Edmands 2007;Frankham et al. 2011;but see Frankham 2015;Whiteley et al. 2015;He et al. 2016).
There is exceptional interest in using a conservation genomics approach to detect regions of the genome that underlie phenotypic variation linked to fitness in threatened populations (i.e. adaptive variation; Luikart et al. 2003;Kohn et al. 2006;Ouborg et al. 2010;Angeloni et al. 2011;Harrisson et al. 2014;Shafer et al. 2015). There are several methods available to study adaptive variation, including gene-mapping approaches (i.e. genome-wide association studies or GWAS, and quantitative trait loci mapping or QTL; Slate et al. 2010;Stapley et al. 2010), outlier locus analysis (Luikart et al. 2003;Haasl & Payseur 2016) and selective sweep mapping (Pardo-Diaz et al. 2015). However, determining the genetic basis of phenotypic traits, especially those linked to fitness, is complex, owing to the fact that most fitness-related traits are likely to be controlled by multiple loci (Savolainen et al. 2013) and many are likely to be under at least some environmental influence (Falconer & Mackay 1996;Lynch & Walsh 1998). In addition, the success of these approaches is often contingent on large sample sizes (e.g. Ball 2005) which will be challenging to generate for most species of conservation concern.
While there are challenges associated with the detection of adaptive variation in threatened populations (reviewed in Shafer et al. 2015), there is potential to answer new questions previously not tractable by employing small sets of targeted genetic markers. In particular, an understanding of the genetic basis of fitness traits will allow more robust predictions of the evolutionary potential of threatened species (Ouborg et al. 2010;Harrisson et al. 2014), including a better understanding of genetic trade-offs between traits that might constrain adaptation (Slate et al. 2010). Further, identifying loci underlying local adaptation is likely to help identify candidate populations for conservation translocations (Seddon 2010;He et al. 2016). Finally, identification of genes responsible for detrimental traits associated with inbreeding depression will have immediate impact on the management of threatened species, especially where matings between individuals are managed (e.g. captive populations; Angeloni et al. 2011;Harrisson et al. 2014;Shafer et al. 2015).
Despite having been available for over a decade (Margulies et al. 2005), a limited number of publications Box 1. The costs of using a conservation genomic approach. Perspectives are those of Tammy Steeves.
Since I arrived in New Zealand from Canada in 2004, I have had the privilege of developing conservation genetic management recommendations in collaboration with several Department of Conservation recovery or specialist groups to assist the recovery of endemic taonga (treasured) bird species. To date, these recommendations have been predominantly based on genetic markers, namely mitochondrial sequences or microsatellite genotypes (e.g. Steeves et al. 2010;Hagen et al. 2011;Overbeek et al. 2016). In collaboration with primary industry colleagues in the MapNet community (see Boxes 2 & 3), I recently assessed the direct and indirect costs associated with shifting from a conservation genetic to a conservation genomic approach and decided to develop genomic markers (SNPs) for the endangered tuturuatu/shore plover (Thinornis novaeseelandiae; Tuturuatu/Shore plover-I was recently invited to be an expert advisor to the Shore Plover Specialist Group. The Specialist Group was interested in sampling captive and wild birds to estimate the extent of population genetic structure and compare levels of genetic diversity, between captive and wild tuturuatu. To achieve this, I knew the cost to develop, screen and genotypẽ 20 polymorphic species-specific microsatellites for 94 individuals (~10K NZD) would be more than using a reduced-representation approach to simultaneously discover and genotype >20 000 SNPs for the same number of individuals (Elshire et al. 2011;~8.5K NZD). I also knew it would be possible to expedite the characterization of SNPs if I was able to use a reference-guided approach. As a member of the Avian Genome Consortium, I was aware bird genomes are small, compact and highly conserved (Zhang et al. 2014), and that one of the newly available high-quality bird genomes (killdeer, Charadrius vociferus) would likely be an appropriate proxy-reference genome for SNP discovery and genotyping in tuturuatu because both species are members of the Family Charadriidae (Card et al. 2014). Thus, the main driver of my decision to embrace a conservation genomic approach for tuturuatu was to ensure that I could develop a comprehensive postgraduate research project that could deliver pertinent results to the Shore Plover Specialist Group in a timely fashion.
Kak ı/Black stilt -As a member of the Kak ı Recovery Group, I have used species-specific genetic markers to inform the conservation genetic management of captive and wild kak ı populations for many years. For example, I routinely use genetic-based measures of relatedness based on microsatellites to inform captive pairing decisions (as per Hagen et al. 2011). However, emerging evidence indicates genetic-based measures are relatively poor indicators of genome-wide diversity, particularly in genetically impoverished species like kak ı, and a better indication of genome-wide diversity should be obtained from genomic-based measures of relatedness based on genome-wide SNPs (Taylor et al. 2015;Willoughby et al. 2015). Thus, the main driver of my decision to generate SNPs for kak ı was to establish the Kak ı Recovery Programme as an exemplar of 'best practice' conservation genomic management.  have applied HTS technologies to conservation (Shafer et al. 2015(Shafer et al. , 2016; but see Garner et al. 2016), with the term 'conservation genomics gap' first being used in 2015 to describe the paucity of conservation geneticists using HTS technologies to inform conservation management (Shafer et al. 2015). While there are a growing number of examples that show how genomic data are being used to inform conservation decisions ; but see Shafer et al. 2016; see Fig. S1, Supporting information) and many conservation geneticists who are currently producing HTS data sets, there has been a substantial time lag between when these techniques have become available and uptake by the conservation research community, especially in comparison with other applied genetic disciplines like primary industry (e.g. agriculture, fisheries, forestry and horticulture; see Fig. 1). In addition, much of the uptake in conservation biology has been restricted to threatened wild fish stocks Shafer et al. 2016). Of the 51 articles in Fig. 1 classified as 'conservation genomics', 30% pertained to the management of declining, overfished or threatened commercially fished species (e.g. Atlantic salmon, Salmo salar; orange roughy, Hoplostethus atlanticus; delta smelt, Hypomesus transpacificus), which provides an excellent example of how conservation genomic research can also be relevant to other applied scientific disciplines including primary industry (e.g., these articles were classified as both 'conservation genomics' and 'primary industry' in Fig. 1). Shafer et al. (2015) predominantly attribute the conservation genomics gap to a persistent disconnect between academia and real-world conservation issues. We agree strong relationships between academics and conservation practitioners are crucial, but argue the conservation genomics gap as defined by Shafer et al. (2015) is more akin to a 'research-implementation gap' (Knight et al. 2008;Hogg et al. 2016). Indeed, if strong relationships between academics and conservation practitioners are absent, the likelihood that any research will be translated into conservation action is exceptionally low (Haig et al. 2016). Here, we predominantly attribute the apparent shortage of conservation geneticists using HTS technologies (i.e. the conservation genomics gap sensu stricto) to several interconnected challenges associated with the generation, analysis and interpretation of genomic data.
Prior to identifying these interconnected challenges, we recognize some questions in conservation are still being readily addressed with genetic data (e.g. Dowling et al. 2015;Li et al. 2015a;Pacioni et al. 2015;Trask et al. 2015;Cubrinovska et al. 2016;Hammerly et al. 2016;Overbeek et al. 2016). We anticipate studies such as these to persist, at least in the short term, because existing panels of genetic markers remain a sufficient low-cost option in some situations (Angeloni et al. 2011;McCormack et al. 2013;McMahon et al. 2014). Although we acknowledge that direct cost can be a factor contributing to the conservation genomics gap, we do not think it underpins it, especially when reduced-representation approaches (e.g. restriction siteassociated DNA sequencing, genotyping-by-sequencing, exome capture and RAD capture; Baird et al. 2008;Elshire et al. 2011;Jones & Good 2016;Ali et al. 2016) make it possible to characterize tens of thousands of SNPs in hundreds of individuals for nonmodel species at a lower cost than developing and screening relatively few novel microsatellite markers (Narum et al. 2013;Andrews et al. 2016; Box 1). Beyond direct cost, the shortage of high-quality reference genomes is an often cited impediment to SNP discovery and genotyping for nonmodel species (e.g. Allendorf et al. 2010;Ouborg et al. 2010;Shafer et al. 2015), particularly when approximate SNP location is of interest (e.g. Kardos et al. 2015). However, an ever increasing number of highquality and high-coverage genomes are becoming available (Ellegren 2014). It has also become apparent that low-coverage draft genomes (sometimes referred to as 'landing-pad' or 'skim' genomes), or even high-quality and high-coverage genomes of closely related taxa, can enable reference-guided mapping assembly and SNP characterization in some taxa (Card et al. 2014;Wang et al. 2014). The lack of bioinformatic expertise and pipelines required to analyse large population genomic data sets has also been frequently cited as a challenge that precludes the use of HTS technologies in conservation (e.g. McCormack et al. 2013;Shafer et al. 2015). Steep analytical learning curves are generally associated with new technologies, particularly for rapidly advancing fields like genomics where bioinformatic expertise is needed to analyse large genomic data sets. However, the analysis of large population genomic data sets is no longer exceptional. For example, in regard to SNP discovery and genotyping alone, several comprehensive bioinformatic pipelines are readily available (e.g. Glaubitz et al. 2014;Puritz et al. 2014;Herten et al. 2015;Sovic et al. 2015;Melo et al. 2016).
Depending on the conservation genetics project at hand, one or a combination of the challenges listed above might impede conservation geneticists from transitioning to HTS technologies. Given the recent developments in HTS technologies and the potential it has for benefitting conservation outcomes, we suggest it is time for researchers to start sharing practical advice on how to expedite the transition from conservation genetics to conservation genomics. Here, we argue that an effective and efficient way to navigate the conservation genomics gap is to capitalize on emerging synergies between conservation genetics and primary industry, and demonstrate how building strong relationships between these two disciplines is leading to mutually beneficial genomic outcomes.
Strong relationships lead to mutually beneficial genomic advances Conservation geneticists are skilled at building strong relationships in an interdisciplinary landscape to improve conservation outcomes (Haig et al. 2016;Hogg et al. 2016). However, by pushing the boundaries of the conservation 'silo', conservation geneticists will be better able to navigate the conservation genomics gap if they forge novel relationships with scientists that have shared genomic goals, albeit in a different discipline such as primary industry (Fig. 2). As a discipline, primary industry represents a diverse group of scientists from universities, private institutions and government organizations that apply scientific data to the benefit of primary production output (e.g. meat, fish, eggs, dairy, fruits, vegetables, fibres and timber). Some of the early draft genomes were published to improve commercial outcomes, including rice (Oryza sativa; Goff et al. 2002), red jungle fowl (Gallus gallus; Hillier et al. 2004), silkworm (Bombyx mori; Xia et al. 2004) and cattle (Bos taurus; Schibler et al. 2004). With these early reference genomes and the accumulation of massive SNP data sets coupled with phenotypic data, many primary industry scientists have years of expertise with the application of genomic data. Approximately 1981 HTS studies using genomic data have been published in primary industry from 2005 to 2015, which outnumbers those produced in conservation biology by more than an order of magnitude ( Fig. 1). Conservation has already benefitted from genomic resources provided by primary industry. For example, genomic resources developed for cattle including the draft genome (Schibler et al. 2004) and the bovine SNP chip (Gunderson et al. 2005;Steemers et al. 2006;Matukumalli et al. 2009) have been used to estimate the extent of introgression from cattle to American bison (Bison bison; Halbert et al. 2005), measure genomic variation in American and European bison (B. bonasus; Pertoldi et al. 2009) and develop genomic resources for scimitar-horned and Arabian oryx (Oryx dammah and O. leucoryx, respectively; Ogden et al. 2012). Similarly, genomic resources developed for domestic sheep (Ovis aries) have been used to describe genome-wide diversity and assess genetic rescue for bighorn sheep (Ovis canadensis; Poissant et al. 2009;Miller et al. 2012). Of course, there are species of mutual interest to both conservation and primary industry, including species in the fishery and forestry sectors (e.g. Monterey pine, Pinus radiata D.Don; New Zealand t otara, Podocarpus spp.; chinook salmon, Oncorhynchus tshawytscha; orange roughy, Hoplostethus atlanticus), and therefore, genomic resources produced by one discipline can be easily used by the other (Dillon et al. 2013;Larson et al. 2014;Marshall et al. 2015;da Silva et al. 2015). We anticipate conservation geneticists may opt to use closely related commercial or model species to inform adaptive variation studies in threatened species, given that genemapping approaches are contingent on large sample size (Ball 2005; see discussion above) and the small census size of threatened populations may be inadequate.
Collaborations between conservation geneticists and primary industry scientists are logical because researchers in these two disciplines are beginning to address similar questions in an applied genetic discipline (see Table 1). For example, primary industry scientists have been using neutral genome-wide SNPs to calculate inbreeding coefficients in domestic sheep (Li et al. 2011), reconstruct parentage assignments in cattle (Hayes 2011) and calculate diversity measures for genetic improvement in poultry (red jungle fowl, Muir et al. 2008; domestic turkey, Meleagris gallopavo, Aslam et al. 2012). Pipelines that have been used or developed to address these questions in commercial species are likely to be of interest to conservation geneticists, but are sometimes published in discipline-specific peer-reviewed journals such as the Journal of Dairy Science or Plant Biotechnology Journal (e.g. Allen et al. 2012;Li et al. 2015b). Similarly, there are some conservation genomic articles from nonacademic sources that are not represented in peer-reviewed literature . These examples highlight how relationships between conservation genetics and primary industry scientists can enable the dissemination of discipline-specific publications and will allow scientists from both disciplines to learn about recently developed pipelines.

Conservation
Understanding the genetic basis of desired commercial traits is also a main focus in primary industry (Womack 2005;Tuberosa & Salvi 2006;Sellner et al. 2007;Collard & Mackill 2008;Neale & Kremer 2011;Sonah et al. 2011;Hu et al. 2013). Primary industry has benefitted from collaboration with researchers in human health to determine the genetic basis of phenotypic traits in complex pedigrees and structured populations using QTL mapping and GWAS (George et al. 2000;Aulchenko et al. 2007;Price et al. 2010). In turn, these gene-mapping approaches have been successfully applied to understanding the genetic basis of ecologically relevant traits in wild populations (Schielzeth & Husby 2014). While there are numerous research groups outside of primary industry exploring adaptive variation (e.g. Rietveld et al. 2013;Brachi et al. 2015;Chaves et al. 2016), we anticipate that conservation geneticists in particular will benefit from forging relationships with primary industry scientists given that both groups work in an applied discipline with species characterized by small effective population sizes. Additionally, there is potential for conservation geneticists to adopt a genomic selection approach (e.g. Hayes et al. 2009;Heffner et al. 2009) to generate breeding values to inform the selection of individuals for captive breeding. Lastly, we recognize that both conservation geneticists and primary industry researchers routinely work with species with complex genomes (Clevenger et al. 2015), and therefore researchers from these two disciplines have an opportunity to work together and think of creative bioinformatic solutions for species that present bioinformatic challenges (Box 3). Given these commonalities, synergies between both conservation genetics and primary industry can lead to the development of improved HTS techniques and pipelines to address mutual problems in species of both conservation and commercial interest (Boxes 2 and 3; Table 1).
Relationships between conservation geneticists and primary industry scientists can result in improved commercial outcome for primary species as well. Conservation geneticists strive to preserve genetic diversity and the ecological and evolutionary processes that generate it (Groom et al. 2006;Haig et al. 2016). There is growing discussion among primary industry scientists regarding the need for commercial breeding programs to maximize genetic diversity and minimize inbreeding (Medugorac et al. 2009 (Eteqadi et al. 2015) and Iberian pigs (Sus scrofa; Saura et al. 2015). There is also an increasing awareness of the risks associated with deploying very few genotypes, particularly in the presence of novel crop pathogens (Kim et al. 2015) and an increasing concern among rare breeds regarding the loss of genetic variation associated with traits that might be useful in future markets (e.g. Catalonian donkey Equus africanus, Gutierr ez  Trait mapping for economically important traits using GWAS and QTL mapping 6,7 in rice 8 , dairy cattle 9 , pig 10 and soy bean 11 Gene copy number variation Quantifying genome-wide copy number variation and estimating its contribution to phenotypic variation Quantifying genome-wide copy number variation and estimating its contribution to economically important traits in apple 12 , pig 13 , wheat 14 Inbreeding and relatedness Measuring inbreeding (f), detecting inbreeding depression and estimating relatedness (r) for small populations to maintain evolutionary potential Measuring inbreeding (f), detecting inbreeding depression and estimating relatedness (r) in sheep 15 , pigs 16,17 and salmon 18 to enhance traits for commercial selection Box 2. Retrospective and prospective of genotyping by sequencing (GBS). Perspectives are those of Rob Elshire.
In 2007, I joined the Buckler Lab at Cornell University and the next-generation sequencing revolution simultaneously. My first task was to develop a new library preparation method for the nascent Illumina sequencing platform. The technology was not nearly as robust as it is today and the reads were very short (i.e. 32 bp in length). Our challenge was to sequence the nonrepetitive fraction of the maize (Zea mays) genome. To do that, we used a combination of digestion by restriction enzymes and gel-based size selection to exclude the repetitive fraction. The data generated formed the basis for the first Maize Hapmap paper (Gore et al. 2009). When that project neared completion, I was tasked with building a low-cost, high-throughput genotyping method as an extension of my previous work. The overall goal was to develop a genotyping system that would allow simultaneous marker discovery and genotyping and also address the issue of marker discovery bias. Other researcher groups at the time were developing similar methods, as there was a high demand for an affordable and reproducible method of genotyping and it was the next logical thing to try. One aim was to provide enough genetic markers at the right price point to enable plant breeding by genomic selection. To maximize the benefit of our work and encourage others to take what we did and create new methods appropriate for new questions, we made our work openly available. The resultant genotyping-by-sequencing (GBS) method was published in PLoS One in 2011 (Elshire et al. 2011).
We achieved our goal of developing a new genotyping method that was inexpensive, both in terms of cost per sample and cost per data point (i.e. fractions of a cent per marker). The low-cost and high-throughput nature of GBS allows plant breeders to genotype thousands of plants per cycle in genomic selection driven breeding programs (He et al. 2014). Primary industry programs in animal breeding have also taken up GBS. Unlike microsatellites or SNP chips, no previously generated genomic resources are necessary to deploy GBS. This allows researchers working in nonmodel species, such as orphan crops (i.e. crops of regional commercial importance, but not global), to take advantage of powerful genomic tools (Varshney et al. 2012). The situation for researchers in ecology and conservation biology is not dissimilar to that of those working with orphan crops. The budgets are small, resources meagre and the questions are of local importance with small (if any) obvious economic returns. It is no wonder that ecologists were among the earliest adopters of GBS.
During the development of the GBS, we tested it on species other than maize. Confident that it worked in a variety of kingdoms, we welcomed interested early adopters to the laboratory for assistance. Two of those early adopters worked in the ecology space. Dr. Thomas White worked with the invasive bank vole (Myodes glareolus, White et al. 2013) in Ireland which had small sample sizes and no reference genome. Dr. Nancy Chen studied the Florida scrub jay (Aphelocoma coerulescens) and developed a method using GBS data and Mendelian inheritance to improve SNP discovery (Chen et al. 2014). It became clear that we had developed a generally useful genomics research tool and it could be used by researchers across disciplines. We had already published the method in an open-access journal and provided analysis software under a free software licence. To allow researchers to more easily use this technology, we set up a GBS service at Cornell. By early 2016 the Cornell service had performed GBS analysis on over 1500 species.
After our initial GBS publication, a plethora of method modifications and additional software tools have emerged. The recently published epiGBS method (van Gurp et al. 2016) allows the interrogation of the methylome and does not require a reference genome, thereby extending the utility of the base method greatly. The GBSX toolkit (Herten et al. 2015) is a set of software designed to assist in the design of GBS based experiments. Many software packages have been developed to analyse GBS data (e.g. TASSEL-UNEAK, STACKS, GBS-SNP-CROPS, GLBPSs; Lu et al. 2013;Catchen et al. 2013;Hapke & Thiele 2016;Melo et al. 2016) that are appropriate for species without reference genomes. Extensions to the molecular method and new software tools make these types of genomics approaches more broadly accessible; however, barriers to using this technology still exist in many disciplines, including the cost of laboratory and informatics setup and reservations in transitioning to new analytical tools.
Marker technology adoption has a long tail distribution. In 2013, I gave a talk on GBS at the Molecular Markers in Horticulture Symposium. Perusing the poster session, I found that researchers were using every type of marker technology that I knew about: from isozymes to GBS. Why were some researchers using cutting edge technologies? Why were others using antiquated, expensive and low information content technologies? Researchers in conservation genomics are in a similar situation. Across disciplines, the biological sciences are encountering rapidly changing technologies and increasingly larger data sets. Industry service providers with expert knowledge and experience, like my small New Zealand-based company (Elshire Group, Ltd.) and many others, can help bridge the gap. By developing relationships spanning human health, primary industry and conservation, as well as actively participating in research communities like MapNet (Box 3), we can work together to expedite the adoption of genomic technologies applicable to the questions at hand, effectively, efficiently and with confidence. provide this biodiversity expertise to commercial species for improved primary production (Fig. 2).
Conservation biologists and primary industry scientists also share similar goals regarding how best to mitigate the impact of climate change (Kristensen et al. 2015). For example, plant and animal breeders are prioritizing the selection of heat-tolerant plants (Ye et al. 2015) and low-emission animals (Hayes et al. 2013) and conservation scientists are debating a role for intentional introgression of desired phenotypic traits (e.g. heat tolerance) among locally adapted species or populations (Hamilton & Miller 2015;Kovach et al. 2016;Miller & Hamilton 2016). Given these shared goals, there is merit for scientists in primary industry and conservation to work together to maintain the evolutionary potential of commercial and threatened species in a changing climate.
A compelling rationale for building strong relationships between primary industry and conservation biology is that scientists in both disciplines conduct applied genetic research. Whereas primary industry scientists respond to the needs of primary industry practitioners (i.e. plant and animal breeders, farmers, fishermen and loggers), conservation scientists respond to the needs of conservation practitioners (i.e. wildlife managers and policy makers; Gordon et al. 2014;Haig et al. 2016). Considering the research-implementation gap that has been discussed in conservation genetic and genomic literature (Knight et al. 2008;Laikre et al. 2010;Shafer et al. 2015;Taylor & Soanes 2016), researchers from conservation genetics and primary industry can collaborate on how to best communicate research needs and results between scientists and practitioners. In the policy arena, both conservation geneticists and primary industry scientists work to develop improved policy regarding the utilization and dissemination of genetic and genomic information (e.g. the Nagoya Protocol on Access to Genetic Resources and the Fair and Equitable Sharing of Benefits Arising from their Utilization, https://www.cbd.int/abs; the International Treaty on Plant Genetic Resources for Plants for Food and Agriculture, http://www.planttreaty.org) and we anticipate that relationships between the two disciplines will allow for discussion on how to best form policy regarding the application of genomic information to threatened and commercial species.
Cross-sector collaborations will provide exciting opportunities to strategize how best to engage with stakeholders (e.g. private landowners, local governments and researchfunding bodies; Jacobson & Duff 1998;Dubbeling & Merzthal 2006); but where we see an even greater opportunity for considerable gains is for conservation geneticists Box 3. Building strong interdisciplinary relationships: MapNet and VISG. Perspectives are those of Phil Wilcox.
MapNet is a genomics collaboration that was formed in 2005 by a collective of New Zealand-based researchers from agriculture, horticulture, forestry and human medical genetics that quickly identified analytical gaps in international statistical genetics research. In response, MapNet members formed the Virtual Institute of Statistical Genetics (i.e. VISG) in 2007 and successfully obtained research funding to address these gaps. Through these synergies, methods developed for large human data sets (e.g. CNVrd, CNVrd2, selectionTools; Nguyen et al. 2013Nguyen et al. , 2014Cadzow et al. 2014) have been successfully applied to apple (Malus pumila) data to identify genes of interest in commercial species (e.g. Boocock et al. 2015). Other workflows, such as the selectionTools pipeline developed and applied to human data sets such as the 1000Genomes human data (Cadzow et al. 2014), are applicable to other outcrossed species where genetic maps are available. Recently, these relationships have also expanded to include cross-sector projects with scientists from the EEB and conservation genetics sector, who are able to provide insight into how these pipelines can be more broadly applicable to other applied genetic disciplines.
Critical for these cross-sector collaborations is effective and ethical behaviours among researchers, distributed leadership, commitment to an explicitly articulated vision, and effective resourcing for method development and testing. Ongoing cost reductions in both high-throughput sequencing and genotyping will constantly challenge data analyses. Thus, collaborations among researchers in primary industry, human medical genetics, EEB and conservation genetics are an effective option to develop and apply genomic methods in a financially limited environment. The benefits of the above-mentioned collaborations would ensure (a) relevant data analysis tools could be produced by adding relevance and utility to primary-sector researchers proposing to develop such tools, and (b) providing a platform for more efficient utilization of resources such as laboratory spaces and analytical capabilities, further reducing costs and therefore increasing data generation capacity. Collaborating with primary-sector researchers working on closely related species would also benefit conservation genetics by improving efficiency. In some cases, the same species may be endangered within its natural range, but be of commercial value in other regionssuch as Pinus radiata D. Don, which is widely planted as an exotic in the southern hemisphere but endangered in its natural range in Baja and northern California. An additional benefit of such collaborations is valuable experience and learnings from primary-sector colleagues regarding experiment design, data analyses and interpretation of results. The MapNet collective was formed and run at essentially no cost, by utilizing the resources of collaborating institutions and labour of those who were committed to this initiative, thus such cross-sector networks are easy to establish and operateand often professionally rewarding for all involved. and primary industry scientists to learn from one another about the importance of building meaningful partnerships with local and indigenous communities. Partnering with these communities enriches conservation and primary industry science because it creates research projects that are informed by the traditional knowledge and needs of these communities from the initial research proposal to the final report. In New Zealand, scientists and practitioners have clear directives to engage with M aori (indigenous peoples of Aotearoa/New Zealand) regarding the management of taonga (treasured) species (i.e. Ko Aotearoa T enei/This is New Zealand, conventionally known as WAI 262, http:// www.waitangitribunal.govt.nz/) and various approaches have been developed to facilitate such engagement (Tipene-Matua & Henaghan 2007;Wilcox et al. 2008;Hudson et al. 2010). In addition, researchers are required to consult with relevant M aori tribes (iwi or hap u) when applying to receive permits for scientific research on taonga species from the Department of Conservation. New Zealand endemic species of cultural importance include threatened species (e.g. tuturuatu/shore plover and kak ı/black silt; Box 1) and commercial species (e.g. k utai/green-lipped mussel/Greenshell TM mussel, Perna canaliculus), and therefore, we urge conservation genetic and primary industry scientists to collaborate on how to build productive partnerships with relevant M aori communities to develop research that is responsive to the needs and expectations of those communities. Beyond New Zealand, researchers based in any of the 92 countries around the world that are signatories to the Nagoya Protocol on Access to Genetic Resources and the Fair and Equitable Sharing of Benefits Arising from their Utilization (https:// www.cbd.int/abs/) have an opportunity to do the same. However, we argue that as global citizens, all scientists should be acting as if their country was a signatory, because as we get closer to generating population genomic data sets that include whole genomes for species of cultural importance we need to be more aware of how these genomic resources can affect and benefit local and indigenous communities.

Moving forward
While multitasking empirical research, relationships with practitioners, stakeholders and interdisciplinary partnerships can be cumbersome, we are confident that the biggest gains in both conservation genetics and primary industry will be made under this approach. Given the mutual problems that can be solved when conservation geneticists and primary industry scientists work together, we encourage scientists in both disciplines to be leaders in interdisciplinary research and we offer the following advice on how to best forge these relationships: Get out of your silo The first step to building successful interdisciplinary relationships is for researchers to get out of their silos and meet people with aligned research goals across disciplines.
To accomplish this task for conservation genetics and primary industry, we advocate for small (<100 people) and diverse cross-sector meetings that allow participants from academia, government agencies and private institutions to actively engage with every presentation, especially those outside of their silos. In a New Zealand context, annual meetings such as MapNet (see Box 3), the Canterbury 'Omics Symposium and the Queenstown Research Week exemplify small, diverse, cross-sector meetings that allow scientists from both conservation and primary industry to meet and expand their research networks. For larger countries, these diverse and small meetings might be more effective on a regional vs. a national level. In addition to meetings, we encourage conservation geneticists and primary industry scientists to attend genomic and networking workshops to meet people with aligned vision for genomic research, albeit in another discipline.

Practice leadership in interdisciplinary research
The second step to forging mutually beneficial partnerships between conservation and primary industry is to actively communicate with and collaborate with researchers outside of one's silo. Doing so invariably requires leadership, respect and motivation to tackle shared problems (see Table 1), generally by expanding your own research program to incorporate collaborative interdisciplinary projects between conservation and primary industry (e.g. Banks 2004;Knowler & Bradshaw 2007;Hobbs et al. 2008;Blank 2013;Sardinas & Kremen 2015;Box 3). Upon launching these collaborations, it is essential that leaders from both parties open an honest dialog concerning expectations, limitations and potential hindrances to interdisciplinary work such as intellectual property issues. If collaborative groups choose to develop new methods or bioinformatic pipelines, we encourage these groups to test these tools on different species representing a wide range of genomic complexities (i.e. ploidy levels, genome size and number of repetitive elements, see Table 1) so these tools are robust and widely applicable to any research study (see also Boxes 2 and 3). We also advocate for these collaborative groups to develop methods and pipelines that are open source (see Box 2), which inspires others to use and improve upon cross-disciplinary tools. Pursuing co-funding opportunities between conservation and primary industry can be an excellent means of building mutually beneficial research collaborations, especially given that most grant providers favour collaborative proposals that tackle complex problems with broad research impact (Ledford 2015; but see also Bromham et al. 2016). Worldwide, there are groups that are forming to tackle complex problems through an interdisciplinary approach, including the Virtual Institute of Statistical Genetics (see Box 3) and Te P unaha Matatini (translated to 'the meeting place of many faces', http://www.tepunahamatatini.ac.nz/). As leaders from conservation and primary industry initialize interdisciplinary research, we encourage the formation and utilization of these groups to facilitate the scientific process and encourage the involvement of new partners.

Promote a community of interdisciplinary research
Leaders in both the conservation and primary industry sphere can go beyond collaborating with interdisciplinary scientists to promote a culture of interdisciplinary research. To accomplish this, we encourage editorial teams at conservation and EEB journals with a broad readership like Molecular Ecology to periodically invite perspective articles from colleagues in primary industry. We equate this approach to the recent decision made by the editorial team at Animal Conservation to invite submissions from conservation practitioners so conservation academics can better understand the needs and challenges of real-world conservation (Gordon et al. 2014). Leaders who are organizing meetings and conferences in primary industry, conservation and genomics can strive to incorporate cross-sector talks and break down organizational silos by minimizing field-specific sessions, as proposed by Taylor & Soanes (2016) and practised by cross-sector meetings like MapNet (see Box 3). We also challenge scientists in both primary industry and conservation to become good interdisciplinary mentors to promote a culture of interdisciplinary research. This can involve mentors in conservation and primary industry promoting genomic seasonal internships or research positions to students in different silos. Not only will this encourage an interdisciplinary field, but it will also produce wellrounded and informed early-career researchers with excellent interpersonal skills and a network of colleagues to help solve shared problems.
After relationships between conservation genetics and primary industry are forged, we do not anticipate relationships will end once genomic gains are made in both disciplines. Instead, we envision these relationships will continue to grow and enable both disciplines to problemsolve and incorporate new technologies for the improvement in threatened and commercial species. With other emerging techniques being discussed and used in both conservation and primary industry, including other -omic techniques (e.g. transcriptomics, proteomics, metabolomics; Diz & Calvete 2016;Todd et al. 2016), epigenetic studies ) and genome editing , we expect conservation genetics and primary industry to continue to collaborate and solve mutual problems while incorporating new technologies in applied disciplines.
We are confident that building strong interdisciplinary relationships will enable genomic advances in both conservation genetics and primary industry. However, we appreciate our colleagues in the global conservation community may be pursuing different strategies to successfully navigate the transition from genetics to genomics and we look forward to hearing about them in due course. In the meantime, our hope is that new technologies including genomics will be effectively incorporated into applied genetic disciplines like conservation and primary industry, because there is much to gain using HTS technologies to improve outcomes for the world's threatened and commercial species.

Data accessibility
Text files including all literature search results presented in Fig. 1 and Fig. S1 (Supporting information) are available on Dryad: doi: 10.5061/dryad.32j55.

Supporting information
Additional supporting information may be found in the online version of this article.
Data S1 An ISI Web of Science literature search was conducted to compare the number of publications from 2005 to 2015 that have been produced in the fields of conservation biology and primary industry. It should be noted that the number of publications may be underestimated in both disciplines, given that conservation and primary industry researchers can publish their results in the grey literature .