Potential and limits of exploitation of crop wild relatives for pea, lentil, and chickpea improvement

Legumes represent the second most important family of crop plants after grasses, accounting for approximately 27% of the world's crop production. Past domestication processes resulted in a high degree of relatedness between modern varieties of crops, leading to a narrower genetic base of cultivated germplasm prone to pests and diseases. Crop wild relatives (CWRs) harbor genetic diversity tested by natural selection in a range of environments. To fully understand and exploit local adaptation in CWR, studies in geographical centers of origin combining ecology, physiology, and genetics are needed. With the advent of modern genomics and computation, combined with systematic phenotyping, it is feasible to revisit wild accessions and landraces and prioritize their use for breeding, providing sources of disease resistances; tolerances of drought, heat, frost, and salinity abiotic stresses; nutrient densities across major and minor elements; and food quality traits. Establishment of hybrid populations with CWRs gives breeders a considerable benefit of a prebreeding tool for identifying and harnessing wild alleles and provides extremely valuable long‐term resources. There is a need of further collecting and both ex situ and in situ conservation of CWR diversity of these taxa in the face of habitat loss and degradation and climate change. In this review, we focus on three legume crops domesticated in the Fertile Crescent, pea, chickpea, and lentil, and summarize the current state and potential of their respective CWR taxa for crop improvement.

are still underutilized and considered neglected crops (Foyer et al., 2016). Legumes did not benefit from the Green Revolution phenomenon, which revolved around not only technological advances but also on policy interventions and investment for major staple food crops. This propelled a large-scale planting of major cereals (rice, wheat, and maize) on the best agricultural land coupled with larger amounts of agricultural inputs such as fertilizers. Conversely, legume yield potentials have been limited because of its relegation to marginal lands where various abiotic stresses such as water limitation, short growing seasons, and poor soils commonly occur (de la Peña & Pueyo, 2012). Despite this, legumes represent the second most important family of crop plants after Poaceae (grass family), accounting for approximately 27% of the world's crop production, which is dominated by the oilseeds soybean and groundnut. Collectively, the grain legumes represent about three times of groundnut production and one fourth of soybean production. Dry pea currently ranks second after common bean as the most widely grown grain legume in the world, with primary production in temperate regions and global production of 16M tonnes at 8 Mha, followed by chickpea (14.7M tonnes, 14.5 Mha) and lentil (7.5M tonnes, 6.5 Mha) (FAOSTAT, 2010). Without a rapid increase in yield, the legume production gap is projected to increase to 10 million tons by 2050 (Joshi & Rao, 2017). As a result of these production gaps, there is a rising awareness of the need to increase pulse production to help ensure global food security (Food and Agriculture Organization [FAO], 2010; Godfray et al., 2010). Confounding the goal of increased production is climate change. Climate change is already evident worldwide, with continuing increases in levels of greenhouse gases and an associated rise in temperature, very likely to reach at least 1.5 C and possibly 2 C or more above preindustrial levels by 2050 (Ripple, Wolf, Newsome, Barnard, & Moomaw, 2019). With accelerating climate change, increased abiotic stresses are expected to challenge agriculture and food security (Ripple et al., 2019). High temperature spikes, during crop growth and especially for the most critical reproductive period, are expected to exceed the range encountered during crop domestication, and world temperature rise will be greater over land than sea (Intergovernmental Panel on Climate Change, 2019). The novel genetic variation needed to address this challenge may be available from crop wild relatives (CWRs), among which are the direct progenitor species (Dullo, Fiorini, & Thormann, 2015). These have a much wider genetic diversity, which was only fractionally sampled during domestication and selection of rare genes/mutations for reduced seed dispersal (shattering), reduced seed dormancy, but increased seed size, plant biomass, and harvest index. There is an urgency to breed for climate-resilient crops, particularly for tolerances of heat, drought, and cold (Hatfield & Preuger, 2015). One option that is currently emphasized is a more systematic and targeted use of CWRs in crop improvement programs Vincent et al., 2013). This has been supported by activi- Change Adaptation (https://www.cwrdiversity.org). CWR contain a wealth of genetically important traits due to their adaptation to a diverse range of habitats and the fact that they have not passed through the genetic bottlenecks of domestication. Further, CWR have longer evolutionary history across more diverse environments and today are found on uncultivated and often hostile soils in challenging environments Yadav, Hegde, Habibi, Dia, & Verma, 2019). Dynamic response to climate change with shifts in genetic structure such as increased earliness has been shown in CWRs of wheat and of barley in Israel (Nevo et al., 2012).
Thus, the study of molecular ecology and conservation of these taxa should be of high priority (Castañeda-Alvarez et al., 2016;Heywood & Dulloo, 2006).
At the beginning of the 20th century, leading agronomists and geneticists recognized the need to preserve and characterize the genetic diversity of cultivated plants and their wild relatives. For example, Russian scientist N.I. Vavilov led worldwide systematic collection and classification of agricultural diversity for the Soviet State (Vavilov, 1926 andreviewed in Hummer &Hancock, 2015;Janick, 2015). Similar collections were made across much of the Western world, with collection starting in the colonial period and becoming more systematic around the time of Vavilov (e.g., Griesbach, 2013).
Since Vavilov's era of collecting, crop genetic diversity has eroded, as a result of subsequent breeding efforts and farmers' adoption of more uniform varieties at the expense of locally adapted landraces in conjunction with increased commercialization and market quality standards. The resulting elite cultivated varieties were very productive relative to the unimproved landraces but further reduced the genetic base. Most wild accessions and landraces were abandoned without regard to their genetic value, which was often found in individual locations. Recent genetic and genomic analysis revealed dwindling genetic diversity present in modern agriculture (Diamond, 2002;Gross & Olsen, 2010). Domestication bottlenecks followed by the widespread transition from subsistence to commercial agriculture have caused a high degree of relatedness between crop varieties. This was further pronounced in modern breeding programs, leading to a narrower genetic base of cultivated germplasm prone to pests and diseases (Gur & Zamir, 2004;Harlan, 1976;McCouch, 2004;Zamir, 2001).
After domestication, only favorable haplotypes were retained around selected genes (e.g., for photoperiod adaptation of flowering), which created regions with extremely low genetic diversity. To overcome the narrowing of the genetic base, there is a need to identify beneficial alleles that segregate in wild populations so that we can then use this existing variation to improve elite cultivars.
Plant breeders recognized the potential value of landraces since at least the early 20th century, but their sheer number and the absence of a simple means to determine which landraces might hold valuable genetic variation have severely limited their use. Now, with the advent of modern genomics and computation, combined with systematic phenotyping, it is feasible to revisit wild accessions and landraces and prioritize their use for specific agricultural purposes, for example, disease resistance, drought tolerance, and nutrient density .
In this review, we focus on three Fertile Crescent-originating legume crops-pea, chickpea, and lentil-and summarize the current state and potential of their respective CWRs for crop improvement.  . P. sativum subsp.
abyssinicum A. Braun (Berger, 1928;Maxted & Ambrose, 2001), or classified as P. abyssinicum (Kosterin, 2017;Trněný et al., 2018), is found only in cultivation (Ethiopia and Yemen) and was likely domesticated independently of P. sativum, most likely being derived from a distinct genetic stock of wild P. sativum subsp. elatius . From a taxonomical and phylogenetic perspective, Pisum is paraphyletic and nested in Lathyrus and Vicia (Schaefer et al., 2012).
The primary gene pool for domesticated pea (Harlan & de Wet, 1971) consists of the P. sativum/elatius complex Trněný et al., 2018), although because of the existent nuclear-cytoplasmic conflict (Bogdanova, Galieva, & Kosterin, 2009;Nováková et al., 2019), there are some barriers to gene flow. A secondary gene pool (crosses with less success and lower fertility) extends to the other species in the genus, P. fulvum and P. abyssinicum.
P. abyssinicum has never been found in the wild but has a distinct diversity and karyotype Weeden, 2018). The tertiary gene pool (with strong reproductive barriers between crop and CWR) currently consists of Vavilovia formosa (Stev.) Fed. , which might be reconsidered to be within the secondary pool, as shown by Golubev (1990

| Ecogeographical delimitation and its implications for breeding use
Wild pea (P. sativum subsp. elatius) has a rather broad geographical distribution, with populations scattered over a great area of the Mediterranean basin and central Asia, with the greatest diversity in the Near East (Turkey, Syria, Lebanon, and Israel), whereas the distribution of P. fulvum is mainly restricted to the Middle East (Ladizinsky & Abbo, 2015;Smýkal et al., 2017). Population genetics and spatial genetic modeling approaches were used to disentangle the relative roles of geographic and climatic factors in shaping the population's genetic structure of P. sativum subsp. elatius represented by 187 individuals from 14 populations across the northern part of the Fertile Crescent. Genetic distances between wild pea populations were correlated with geographic but not environmental (climatic) distances and support a mixed mating system with predominant self-pollination. Niche modeling with future climatic projections showed a local decline in habitats suitable for wild pea, making a strong case for further collection and ex situ conservation .
Despite environmental distance not being responsible for wild pea population structure, seed dormancy studies have shown phenotypic variation correlated with environmental conditions, including rainfall patterns (Hradilová et al., 2019). As in other native Mediterranean plants and legume species with physical dormancy barriers, seeds germinate mostly in autumn, after experiencing a hot and dry summer season. As a result, established seedlings benefit from available soil moisture and are ready for early spring growth, avoiding increased temperatures during flowering and terminal drought during seed filling. Thus, the temperature is the most prominent environmental factor regulating seed dormancy and germination (Probert, 2000).  (Weller et al., 2012) whereas less is known about the vernalization requirement of the wild species (Highkin, 1956;Wellensiek, 1973).

| Pea wild relatives as a source of novel variation
Pea diversity held in genebanks has been extensively studied over the past two decades (reviewed in Smýkal et al., 2015), with research focusing mainly on cultivated pea diversity. The Genotyping-bysequencing method was applied (Holdsworth et al., 2017) to a set of 431 P. sativum including 11 P. sativum subsp. elatius, two P. abyssinicum, and 25 P. fulvum accessions and the 13k single nucleotide polymorphism (SNP) assay of mapped genes (Tayeh et al., 2015) on 917 samples, including 50 wild accessions (Siol et al., 2017). The largest samples analyzed so far (3,020 and 4,200 accessions) were dominated by cultivated types and had relatively few (45) markers (retrotransposon-based insertion polymorphisms, Jing et al., 2010, Jing et al., 2012, Smýkal et al., 2011. Genome-wide next-generation sequencing techniques have been used recently to study the diversity of wild peas Trněný et al., 2018). A recent study, which included 143 P. sativum subsp.
elatius and 18 P. fulvum accessions, showed that although diversity is present among cultivated and wild material (Ellis, 2011;Jing et al., 2007;Jing et al., 2010;Jing et al., 2012;Holdsworth et al., 2017;Smýkal et al., 2011), wild material provides distinct genetic diversity (Ellis, 2011;Smýkal et al., 2011). Smýkal et al. (2017) conducted a comprehensive analysis of wild P. sativum subsp. elatius by using 409 P. sativum subsp. elatius and 106 P. fulvum accessions and extracted environmental variables. This study showed that P. fulvum has a distinct and only partially overlapping environmental niche. P. fulvum grows in restricted regions of Middle East, sometimes sympatrically with P. sativum subsp. elatius (Ladizinsky & Abbo, 2015). The spatial diversity of the ecological niche patterns reveals not only the species diversity center of the Near East but also the predicted centers of Northern Africa and on the coast of Turkey and the Southern Aegean islands .
Archeological evidence supports the cultivation of pea spreading from the Fertile Crescent westwards through the Danube valley into ancient Greece, Rome, and Europe. During the same period, pea also moved eastward to Persia (now Iran and Afghanistan), India, and China (Chimwamurombe & Khulbe, 2011;Makasheva, 1979). These separate expansions might explain the novel diversity of Afghan type and Chinese landrace peas (Smýkal et al., 2011;Zong et al., 2009) either through drift or through natural selection in diverse environments (Li, Redden, Zong, Berger, & Bennett, 2013 (Vavilov, 1926), and their pathogens followed this distribution (Turcotte, Araki, Karp, Poveda, & Whitehead, 2014). In addition to abiotic stresses, plant pathogens are a major constraint to agriculture and threaten global food security.
Moreover, ongoing climate change could accelerate temporal and spatial disease spread and severity.
Of these, er1 was identified as an MLO gene (Humphry et al., 2011;Rispail & Rubiales, 2016). A combination of knowledge of pea germplasm diversity with that of the eIF4E gene for virus resistance (Ashby, Stevenson, Jarvis, Lawson, & Maule, 2011) and screening of nearly 3,000 accessions with known geographical origin including of wild Pisum sp. led to the identification of novel alleles of resistance (Konečná et al., 2014). These data highlight the importance of Ethiopian, Central Asia, and Chinese regions as secondary centers of pea diversity, corresponding with the diversity of the pathogen.

| Pea wild relatives as sources of tolerance to abiotic stresses
Besides harboring potential as a source of resistance to biotic stresses, wild pea might provide a source of tolerance to various abiotic factors.

| Other traits explored in wild pea
Positive seed yield and seed yield components were identified in

| Ecogeographical delimitation and its implications for breeding use
The compatible wild relatives of chickpea have a very narrow geographical and ecological range, which has been hypothesized to contribute to the lack of genetic diversity in cultivated chickpea . Both C. reticulatum and C. echinospermum are limited to a few provinces of Southeastern Turkey . It is possible that they also occur in similar habitats in Iran or Iraq, although verification of this is not currently possible. C. reticulatum and C. echinospermum rarely co-occur, except for a few likely hybrid populations in the Euphrates valley north of Cermik (Berger, personal observation) but do have adjacent distributions.
C. echinospermum typically occurs on more basaltic substrates at lower elevations in open pastures and disturbed meadows with lower tree cover than for C. reticulatum, which occurs more frequently on sandstone or granitic substrates in mixed pastures and some disturbed habitats (von . Taxa in the tertiary gene pool have somewhat ecologically and geographically broader distributions.
C. pinnatifidum, in particular, occurs in drier habitats in southeastern Turkey.

| Chickpea wild relatives as a source of novel variation
Of the Middle Eastern founder legumes, the primary and secondary Robertson, . However, given that the global Cicer collection at that time was extremely limited, particularly among the primary and secondary gene pool relatives, it was impossible to adequately define the value of the wild species as a source of adaptive traits . This is important because in many cases, this work showed that the tertiary gene pool species might have more to offer than those that were readily crossable with chickpea. If this were true, then chickpea improvement through wild introgression would be complicated. However, given that the evaluation of C. reticulatum and C. echinospermum was based on very few truly independent accessions, this argument was not very sound, and we were hopeful that wider collection would change this situation. For example, ongoing characterization of our new collection shows that the yield potential of wild chickpea (C. reticulatum) in culture is shown in Figure 2. In contrast to two to four branches as found in natural habitat (Figure 2d), in cultivation, it can be several dozen These contrasting lifecycles subject domestic and wild Cicer to different selection pressures that are likely to have important adaptive ramifications that may be exploited for chickpea improvement. For example, there is no robust reproductive chilling (Berger, 2007;Berger et al., 2012) or vegetative cold tolerance in domestic chickpea relative to wild Cicer Singh et al., 1995), whereas heat tolerance is relatively common (Devasirvatham,

| Chickpea wild relatives as a source of resistance to biotic stresses
The leading biotic stresses for chickpea include Ascochyta blight, Phytophthora root rot, Botrytis blight, and Fusarium wilt, among others.
The annual wild Cicer species have long been recognized as a promising source of resistance or tolerance to a range of important biotic stresses (Fusarium wilt, leaf miner, bruchids, and nematodes) . However, the narrowness of the world's wild Cicer collection at that time made it impossible to evaluate whether this resistance  was prescriptive of the species as a whole or merely a symptom of a limited collection . For example, C. reticulatum was rated as highly susceptible to Ascochyta blight and C. echinospermum as moderately susceptible to susceptible , but these scores were based solely on the evaluation of material derived from 18 and 10 independent accessions, respectively. Making matters worse, five of these 18 independent Although these activities are ongoing and largely unpublished and have not yet contributed to new cultivar release, there is a history of wild Cicer exploitation in chickpea improvement (Singh & Ocampo, 1997). C. echinospermum, in particular, has been used as a source for Ascochyta resistance, particularly in the Australian breeding program (Knights, Southwell, Schwinghamer, & Harden, 2008). Many Australian lines bear a signature of introgression from C. echinospermum as a result of this.

| Chickpea wild relatives as a source of tolerance to abiotic stresses
As outlined previously, the contrasting evolutionary trajectories and life histories of wild and domestic Cicer exposed these species to different climatic stresses at different periods in their lifecycle. Early attempts by ICARDA to convert Mediterranean chickpea from a spring to winter crop demonstrated little tolerance to vegetative cold in domestic chickpea and determined the following ranking: C. bijugum > C. reticulatum = C. echinospermum > C. pinnatifidum > C. yamashitae > C. chorassanicum = C. arietinum > C. judaicum > C. cuneatum Singh et al., 1995). However, this cold tolerance evaluation was based on the same limited collection discussed previously and was extremely unbalanced, comparing 5,515 chickpea accessions to n < 6 for C. echinospermum, C. chorassanicum, C. cuneatum, and C. yamashitae . Subsequent work with a wider range of domestic material confirmed these trends and was equally unbalanced (Singh et al., 1995).
In the relatively mild Australian winters, chickpea has a Mediterranean winter annual lifecycle and is often exposed to chilling temperatures at flowering that can delay podset for >1 month (Berger et al., 2004;Clarke & Siddique, 2004). An evaluation of global chickpea genetic resources from contrasting reproductive phase temperature habitats showed no reproductive chilling tolerance in the cultigen but promising tolerance among wild Cicer (Berger, 2007;Berger et al., 2012). However, this evaluation was subject to the same constraints as the earlier ICARDA work and was equally unbalanced.
orientalis as the most closely related wild progenitor of L. culinaris ssp.

| Ecogeographical delimitation
The cultivated lentils were divided into two subspecies by Barulina (1930) and two races by Cubero (1981), the large-seeded macrosperma and small-seeded microsperma. Alo et al. (2011) detected the divergence, following domestication, of the domesticated gene pool into overlapping large-seeded (macrosperma) and small-seeded (microsperma) groups. Within the cultivated lentils, the extreme specificity of adaptation to ecogeographies limits the scope of the direct introduction of exotic landraces. South Asian landraces are generally early maturing small-seeded red lentils, and the West Asian landraces are late maturing large-seeded mostly yellow lentils.
To widen the genetic base of lentil, ICARDA's breeding program has used parents of diverse origins to combine traits contributing to yield, appropriate phenology, adaptation to major biotic and abiotic stresses, and market preferred traits by manipulating photoperiod and temperature under controlled conditions. Derivatives from crosses between South and West Asian parents have generally shown higher yields mainly because of larger seed size introduced from the West Asian parents in the typically short duration background of South Asian genotypes (Shrestha, Siddique, Turner, Turner, & Berger, 2005).

T A B L E 2
Genetic resources conserved at ICARDA Name of taxon The distribution of all CWR overlaps in Turkey (Figure 1) then differ geographically (Davies, Lulsdorf, & Ahmad, 2007;Singh et al., 2014); L. c. ssp. orientalis extends throughout the Fertile Cres-

| Lentil wild relatives as a source of novel variation
Wild species are valuable sources of novel variation for yield traits and resistance to biotic and abiotic stresses. For example, L. ervoides has been identified as a good source of growth habit, biomass production, and seed traits (Fiala et al., 2009;Tullu et al., 2011;Tullu et al., 2013). Useful genetic variability for crop duration, secondary branches, number of pods, biological yield, grain yield, and seed size has been reported in wild relatives of lentil (Kumar et al., 2014;Kumar, Imtiaz, Aditya, & Gupta, 2011;Singh et al., 2013). Genes for yield traits like seed size and number of seeds and pods have been observed in L. lamottei and L. culinaris ssp. orientalis (Ferguson et al., 1998;Gupta & Sharma, 2006). Variation in root traits, including nodulation and root distribution in soil differences, were discovered

| Lentil wild relatives as a source of resistance to biotic stresses
Screening of CWR of lentil has resulted in identification of resistance/tolerance for key stresses including Ascochyta blight, Stemphylium bight, rust, Fusarium wilt, Sitona weevil, bruchids, Orobanche, powdery mildew, and Anthracnose (Table 1).

| Lentil wild relatives as a source of tolerance to abiotic stresses
Wild relatives of lentil also offer drought tolerance in L. nigricans, L. odemensis, and L ervoides (Gupta & Sharma, 2006; and cold tolerance in L. culinaris ssp. orientalis (Hamdi, Küsmenoglu, & Erskine, 1996).   under hydroponic culture at 120-mM NaCl concentration resulted in the identification of several donors for salinity tolerance . Importantly, flowering and growth responses of wild Lens to light quality have also been studied (Yuan, Saha, Vandenberg, & Bett, 2017).
In extensive studies of lentil CWR in India, Gupta and Sharma (2006) found L. nigricans to show the most drought tolerance.
Many environments for L. orientalis across Syria, Jordan, Tajikistan, Turkmenistan, and Azerbaijan have low rainfall and may provide sources of drought tolerance. A study illustrating the potential of lentil CWRs in relation to root traits showed significant differences for root traits and fine root distribution between and within species, the proportion of root biomass partitioned into each soil layer and number of nodules (Gorim & Vandenberg, 2017b). Omar, Ghoulam, Abdellah, and Sahri (2019) examined drought tolerance in crosses of elite lentil varieties with CWR. Tolerance was associated with pubescent leaves, cell membrane stability, relative leaf water content, increased root:shoot ratio, and reduced wilting, transpiration, and canopy temperature. Tolerant segregants are being advanced for trait fixation. Sanderson, Caron, Shen, Liu, and Bett (2019), with a focus on drought tolerance and disease resistance, studied RILs of crosses of lentil cultivars with L. orientalis, L. odemensis, and L ervoides in the ICARDA lentil prebreeding project. This aims to develop genetic maps and markers for lentil and CWR for the transfer of key drought traits into breeding programs for drought tolerance.
Lentil CWR have yet to be screened for heat tolerance. Heat stress tolerance has been only been reported in cultivated lentils (Sita et al., 2017). Singh et al. (2019) used genome-wide transcription to identify heat responsive genes in the regulatory system of lentil cultivars. However, more analysis of heat tolerance mechanisms is required to elucidate heat tolerance. With the assumption that CWR are adapted to their environment of collection (Baute, Dempewolf, & Reisenberg, 2015) and that the reproductive period occurs in May-June, sources of heat tolerance in L. orientalis may occur in Turkmenistan especially, as well as Tajikistan and northern Syria. An alternative to large-scale field testing is the prioritization of accessions according to the climatic history of their origin. GPS data exists for lentil CWR collected at known locations, opening up the opportunity to download 25 years of historical weather data and analyze vectors across sites for heat, drought, and frost stresses. Sites with extreme distributions for these stresses can be found, providing identification of candidate CWR accessions for stress tolerances with the use of FIGS type prioritization (Street et al., 2008).

| HOW TO EFFECTIVELY USE THE DIVERSITY OF CWRs?
Conventionally, breeders have used CWR in their breeding schemes typically as sources of resistance to various biotic and abiotic stresses (e.g., Hajjar & Hodgkin, 2007). However, this inevitably led to the occurrence of undesired wild type traits, which have been removed or altered through the domestication process (Meyer & Purugganan, 2013). In many cases, these undesired traits are dominant and polygenic and consequently challenging to select against.
Thus, these undesired traits need to be removed via repeated backcrosses of elite crop genotypes accompanied by trait (such as resistance) testing, a process that can be facilitated by the use of molecular markers either for the trait or background selection. This process takes time and resources and needs to be done repeatedly on the case by case basis. To make this process more efficient and applicable, the development of series of introgression lines has been proposed (e.g.,  and initiated in all three Fertile Crescent pulse legumes. The development of introgression lines creates backcrossed lines stabilized by selfing, which are also thoroughly phenotyped and genotyped, providing a "library" of lines with various fragments of CWR parent introgressed into a cultivated background (Prohens et al., 2017). In some cases, the fertility of crosses between a crop and its progenitor or more distant relatives is reduced (Dempewolf, Hodgins, Rummell, Ellstrand, & Rieseberg, 2012;Meyer & Purugganan, 2013). This incompatibility, in some cases, is caused by karyotype differences or genomic rearrangement, which might reduce the ease with which recombinants can be found. Such chromosomal segments are challenging to break up by crosses (Tanksley & Nelson, 1996).
There are now genetic procedures to identify CWR with adaptation to local abiotic stresses. Application of population genomic scans can detect loci with exceptionally high population Fst values, possibly indicating loci with divergent selection for local adaptation (Baute et al., 2015). Newer methods without the biases of Fst have emerged or associations of SNPs with climatic variables and are available with improving power to detect SNP-environment associations (e.g., Baypass, bayscan, bayenv2, Bedassle, and Gradient forests) (Fitzpatrick & Keller, 2015). Identification of outlier markers can be facilitated using high throughput sequencing methods for genetic mapping and identification of candidate genes. Alleles adapted to specific abiotic stresses may be associated with such environments, a means of prioritizing CWR accessions for genetic analysis and introgression into elite crop cultivars (Baute et al., 2015;Sanderson et al., 2019).
The prerequisite to an effective selection of adapted material is the existence of sufficiently precisely georeferenced samples. This information allows not only to extract information on the environment but also to conduct ecological modeling of species occurrence, gap analysis of potential sampling, and conservation of CWRs (Castañeda-A lvarez et al., 2016). Akin to advances in genomics, there is also progress in remote sensing technologies. Geographic information systems (GISs) can provide information on the patterns of terrestrial environmental variation representing topography, ecoclimatological, and soil properties. When coupled with genomics, these data sets offer opportunity to search for adaptive selection, which can also be used in breeding programs.
Transgressive segregants for agronomically important traits have been mined from lentil-wide crosses (Kumar et al., 2014;Singh et al., 2013). A recent development in lentil improvement efforts has been the successful hybridization of the cultivated lentil with L ervoides using embryo rescue (Tullu et al., 2013) and the introgression of resistance to Orobanche crenata (Bucak, Bett, Banniza, & Vandenberg, 2014) and anthracnose (Tullu et al., 2011). Similarly, foreign genes were introgressed for resistance to Ascochyta blight, anthracnose, cold (Fiala et al., 2009), and Stemphylium blight (Podder, Banniza, & Vandenberg, 2013) into cultivated lentil. More recently, crossing of cultivated species with L. tomentosus followed by ovule culture has resulted in the development of several prebreeding lines carrying diversity for flower color, seed coat, and cotyledon color (Suvorova, 2014). The genetic base of cultivated germplasm of lentil, especially improved varieties, is based on repeated use of a handful of germplasms. Pedigree analysis of lentil varieties released in India confirmed the extensive and repetitive use of a few genotypes as one of the parents in hybridization (Kumar et al., 2004). An early flowering exotic line Precoz (ILL 4605) has been utilized extensively to tailor plant architecture having vigorous growth, medium maturity, large seeds, and cold tolerance, particularly for the Indo-Gangetic plains (Kumar et al., 2014). During domestication and directed breeding, many alleles were inadvertently left behind in landraces and wild species; the introgression of these lost alleles using innovative breeding tools could bolster modern improved germplasms. For example, rapid cycling can be used to advance lines quickly as shown in an F 2 population derived from a cross between L. culinaris Medik. and L. ervoides (Lulsdorf & Banniza, 2018). Past research shows marked genetic variability for desired traits among landraces and wild lentils. Use of germplasm in lentil breeding has been restricted mainly because of difficulties in access to exotic germplasm, extreme regional specificity of adaptation, a large number of uncharacterized accessions, linkage drags, and the perception that wide crosses would disturb favorable combinations in cultivated germplasm and result in inferior recombinants.
Collections of CSSLs derived from crosses of cultivated pea (P. sativum) with two wild species (P. fulvum and P. sativum subsp. elatius) were developed (Zablatzká & Smýkal, 2015). Utilization of Cicer wild relative diversity for abiotic stress resistance has lagged behind, a common trend across breeding programs (Hajjar & Hodgkin, 2007

| A NEED FOR FURTHER COLLECTION AND IN SITE ASSESSMENT
CWRs, like other plant species, have evolved in relation to their given environment and habitat. As a result, CWRs have experienced selection resulting in adaptation to a given habitat, reflected by allelic composition across the genome (Piperno, 2017). However, more studies . Crops have about 50% higher yields than respective progenitors, realized by higher biomass and seed size, while reducing pod material. However, there was no difference in the number of seeds per plant (Preece et al., 2017).
Additionally, we believe that the greatest need with these taxa is to harness the diversity present in wild relatives. Although introgression populations have been built from wild-cultivated crosses in all three species, we see considerable power in building the large Nested Association Mapping (NAM)-styled introgression populations that have recently been built for chickpea .
Large introgression populations can be outstanding resources for gene identification of traits that segregate in wild populations and give breeders a considerable benefit on the prebreeding task of harnessing wild alleles Warschefsky et al., 2014).
These hybrid populations, particularly if carefully phenotyped in multiple locations, can be extremely valuable long-term resources. Ensuring that these resources are widely available, in the context of the benefit-sharing mandate of the International Treaty and Plant Genetic Resources and Nagoya Protocol, will be critical to ensuring the widespread value of these plant genetic resources.
Lastly, we also believe that there are benefits to further collecting of CWR diversity of these taxa, particularly in more remote regions of Southeastern Turkey, in the Caucasus mountains, in Central Asia and into the Eastern Fertile Crescent for all three taxa, and in Spain for Given the importance of crops from this region, beyond the three founder legumes, preservation of this natural reservoir of adaptation is among the most important conservation challenges we face.
For effective in situ conservation, both local leadership and international partnerships will be needed. Recent work on in situ conservation has developed a range of principles and some organization. Ideas such as preserving locations with high overlap of CWR taxa, as well as sites with unique characteristics, are important. Setting up preserves to allow migration in response to climate change will also be necessary. However, the social aspects of in situ preservation will likely be more challenging. Funding may be essential and may be one role that international partners can play. However, given declining trends for support for science, particularly for conservation, we may need to be creative to find ways to be optimistic. Political will for preservation must come from local communities and cannot be imposed fairly or effectively by outsiders. In the face of ongoing civil strife, conservation becomes a very low priority. We hope that illustrating the value of CWRs helps build support for CWR conservation, as their value will not diminish over time.

DATA AVAILABILITY STATEMENT
The data sets used during the current study are available from the corresponding author on reasonable request.