Focused Identification of Germplasm Strategy (FIGS): polishing a rough diamond

Focused Identification of Germplasm Strategy (FIGS) has been advocated as an efficient approach to predict and harness variation in adaptive traits in genebanks or wild populations of plants. However, a weakness of the current FIGS approach is that it only utilizes a priori knowledge of one evolutionary factor: natural selection. Further optimization is needed to capture elusive traits, and this review shows that nonadaptive evolutionary processes (gene flow and genetic drift) should be incorporated to increase precision. Focusing on plant resistance to insect herbivores, we also note that historic selection pressures can be difficult to disentangle, and provide suggestions for successful mining based on eco-evolutionary theory. We conclude that with such refinement FIGS has high potential for enhancing breeding efforts and hence sustainable plant production.


Introduction
Pests and pathogens pose major threats to domesticated crops [1] and optimal strategies are needed to minimize damage. Intrinsic plant resistance should play a fundamental role in these strategies, especially since Integrated Pest Management (IPM) has been globally embraced as the future paradigm for plant protection [2]. Unfortunately, the strong contemporary demand for resistant cultivars cannot be easily met, as important resistance traits have been lost during domestication [3,4]. Thus, landraces and wild relatives of crops that still harbor important resistance traits must often be revisited to meet needs for resistance [5 ,6]. This is because continuous natural selection (including by herbivores) prevents the loss of adaptive traits in their wild or cultivated environments. Thus, restoring lost resistance traits (known as 'rewilding') in modern cultivars using these wild genetic resources is a major goal for IPM [7 ].
Unfortunately, mining for rare genetic resources in wild populations and genebanks is often extremely difficult and time consuming, due to the enormous amounts of material available for screening ex situ and in situ. Searching for material with specific rare traits is thus frequently likened to looking for a needle in a haystack [8]. Core or representative subsets of germplasm collections [9] can be established to reduce search efforts. However, their creation and classification are not typically based on specific resistance traits, so this approach (in current formats) does not help breeders to focus on the most promising material. However, Focused Identification of Germplasm Strategy (FIGS) provides a new way to maximize the likelihood of capturing plant genotypes with specific adaptive traits from genebanks and wild populations [10]. The rationale is that certain types of environments mediate natural selection, for or against, the adaptive evolution of focal plant traits in wild populations and cultivated landraces. Hence, the spatial distribution of adaptive plant traits should be predictable from ecogeographic profiles of plant trait distributions, allowing trait miners to focus on limited numbers of georeferenced germplasm samples, originating solely from evolutionary 'hotspots'. Thus, the 'focus' in FIGS may refer to both the specific adaptive trait(s) of interest and specific environmental profiles corresponding to evolutionary hotspots.
In the last decade, the FIGS approach has been used to harness various traits, including inter alia drought tolerance [11], plant size and flower number [12 ], as well as resistance to pathogens [13][14][15], and herbivores [16][17][18]. However, as noted by various authors [10,12 ,15], FIGS is still a nascent approach, with scope for significant improvement. The main weakness of current FIGS is that it only utilizes a priori knowledge about one evolutionary factor: natural selection. This limits its potential utility because focal traits may not have been strongly influenced by natural selection and/or historic selection pressures (exerted by herbivores for example) may be difficult to determine. Thus, associations between focal traits and ecogeographic data may be weaker than expected [19]. Further challenges are associated with insect resistance, as the underlying traits must be identified. Here we discuss these challenges in the light of recent literature and suggest possible ways to improve FIGS, particularly in relation to insect resistance.

Challenges and opportunities related to resistance traits
Plants can be resistant to insects in myriad ways, involving inter alia trichomes, leaf toughness, toxic compounds, kairomones, or extrafloral nectar [20,21]. While highly specialized plant-insect interactions may depend on single plant traits, most herbivores -especially chewing species -are sensitive to several types of resistance traits [22], which may be shaped by multiple selection pressures. Thus, individual plants within their species' range may allocate different amounts of resources to specific resistance traits, just as the overall resistance level typically varies [18,23 ].
As FIGS focuses on individual traits, the first step in a FIGS process is to identify all known or likely types of resistance traits in a plant species, then select the most optimal trait according to several criteria. First, the selected resistance trait should clearly reduce herbivore damage as efficiently as possible [24]. Second, it should be compatible with IPM, and thus not hamper biocontrol agents or other beneficial organisms [2,25]. Third, it should be acceptable to human consumers, as consumer rejection is a major reason for the selection against, and loss of, some resistance traits during domestication [26 ]. Such undesirable traits include, among others, bitterness, toughness, and some characteristics related to unacceptable metabolic costs that may incur yield penalties. 'Rewilding' will not necessarily reintroduce these 'problematic' traits, as wild relatives of crops and landraces often have a rich palette of resistance traits from which breeders can select. Finally, to be compatible with FIGS the selected trait should, of course, be prone to adaptation under directional selection.
Identifying specific resistance traits that are particularly compatible with FIGS is beyond the scope of this review, but generally we believe that modern crossbreeding should focus more on indirect resistance, that is, traits that recruit natural enemies of herbivores [27]. Traits underlying indirect resistance are typically induced, rarely require substantial resources from the plant's metabolic budget [28] and are available in landraces and wild relatives of most crops [29,30]. Moreover, they are adaptive [31,32] and the underlying genes are increasingly being identified, for example, through genome-wide association studies [32].

Predicting historic selection
Generally, biotic selection is a key driver of plants' evolution [33] and can rapidly shape and reshape their resistance traits [34,35]. Hence, plant traits related to insect resistance are among those with the clearest potential for FIGS. However, although some or most resistance traits are under selection by herbivores, the same traits may also respond to other, conflicting selective pressures, potentially leading to low correlation between historic selection by herbivores and resistance [36]. For example, resistance compounds in floral nectar may deter pollinators, leading to conflicting selection by herbivores and pollinators, and thus 'diffuse' plant evolution [37 ,38]. The outcome of such diffuse selection may depend on the relative importance of the players involved, as well as the extent of their spatial overlap, which may lead to spatially divergent plant populations. In other cases, some traits may have multiple positive roles. For example, a trait may confer resistance to both herbivory and abiotic stress, like frost or drought [39,40 ]. In such cases, abiotic selection pressures on resistance traits may lead to strong insect resistance even in areas where herbivory is low [41]. Thus, knowledge of the multifunctionality of resistance traits is crucial for robust predictions of historic selection pressures for use in FIGS.
Nevertheless, although other biotic and abiotic factors are important, in most cases the historic distribution of keystone herbivores is the most important factor for predicting distributions of historic selection pressures on resistance traits [42]. Generally, climatic clines (e.g. latitude and altitude) and resource availability have received the most attention for their roles in shaping herbivore distributions, and evolution of plant resistance [23 ,43,44]. Latitude and altitude correlate negatively with temperature and length of the growing season, and thus are relatively good proxies for herbivore diversity and herbivore pressure, although no such relationships (or even opposite patterns) have been observed in analyses of several plant species' environmental interactions [45].
Early reporters of significant associations between host plant resistance to insects and ecoclimatic profiles of the collection sites of accessions in genebanks included Flanders et al. [46]. They found that accessions of Solanum from hot and arid zones showed resistance to Colorado potato beetle, potato flea beetle and potato leafhopper, while species from cool or moist areas showed resistance to potato aphid. In further research involving use of data pertaining to 92 Solanum species obtained from two decades of field trials, the same group [47] demonstrated that insect resistance of the genus was not evenly distributed in the American continent, but varied with the altitude and in some cases latitude of their origins.
Although many plant species show higher overall resistance in areas at low latitudes or altitudes, this does not necessarily mean that these areas are always hotspots for valuable resistance traits in FIGS. As single resistance traits rather than overall resistance levels are ideally used in FIGS, focal individual traits may respond more strongly in low-diversity communities, where specific herbivores contribute more strongly to the direction of net selection. There is, in any case, no doubt that climatic and resource clines should be utilized in FIGS, but the direction of the putative correlations should ideally be roughly confirmed by pre-studies. Clines shaping plant traits can be very narrow, so small-scale patterns warrant attention as well as larger scale patterns [12 ]. Other natural factors affecting herbivore distributions, such as precipitation, winter temperature and habitat have also been successfully integrated into FIGS [10, Stenberg, unpublished data], but are not further considered here due to space limitation.
In addition to identifying hotspots based on natural environmental factors, as discussed above, awareness is needed that human activities repeatedly expanded distributions of crop wild relatives. Thus, they escaped from some of their old herbivores, leading to loss or remolding of resistance in their new areas [48]. For these reasons, identifying native areas where both the plants and herbivores have interacted historically, and filtering out more novel distributions, has been a key step in practical application of FIGS [10].
In practical terms, proxies of historic natural selection have been implemented as ecoclimatic layers to form ecogeographic profiles for focal traits in FIGS (Figure 1). Although these search profiles have helped efforts to harness important traits for several plants, many authors have also noted lower than expected correlations between ecoclimatic data and focal traits, highlighting the need to develop FIGS [10,12 ,15]. However, as shown below, evolutionary hotspots can be more precisely identified if biogeographic layers of nonadaptive proxies are added to the ecogeographic profiles.

Incorporating nonadaptive evolutionary processes into FIGS
The best way in our view to improve FIGS' precision is to consider nonadaptive evolutionary factors in parallel with directional natural selection. The importance of gene flow and genetic drift in the evolution of resistance and other traits of wild plants has been demonstrated in several recent studies [49 ,50 ]. Our understanding of how selection, gene flow and genetic drift jointly shape the geographic distribution of traits has increased tremendously in recent years following developments in metapopulation ecology and the geographic mosaic of coevolution theory [51,52]. A key assumption in these frameworks is that species' distributions are patchy, consisting of local plant populations that are connected to varying degrees within larger metapopulations [51,52]. Some local populations may be colonized by certain keystone herbivores, which impose selection pressure for specific traits. Other populations may host other herbivores selecting for other traits, and another set of local populations may inhabit an enemy-free environment that promotes loss of resistance   Conceptual framework for Focused Identification of Germplasm Strategy (FIGS). The first step in the FIGS process is to identify the type of trait that would be most ideal to confer a desired outcome (e.g. reduction in damage pests cause to a particular plant). Then information about the adaptive and nonadaptive evolutionary processes shaping this trait and its geographic distribution should be obtained and applied in the form of ecoclimatic and biogeographic layers in an ecogeographic profile that highlights the most intense evolutionary hotspots in the plant's range. Focusing trait mining efforts on germplasm from these hotspots should be considerably more fruitful than focusing on other areas. [42]. Local herbivore populations in turn can go extinct or recolonize populations, molding and remolding natural plants over time and space, creating geographic mosaics with predictable evolutionary hotspots. The more isolated a local population is, the more it will be affected by local selection and small populations may also be characterized by strong genetic drift. By contrast, more connected populations will be affected not only by local selection, but also by higher gene flow that dilutes local selection through inflows of phenotypes that are evolutionary products of neighboring populations with different selection regimes [52]. Asymmetric gene flow (e.g. higher immigration than emigration, leading to dilution of local selection) is expected for small populations neighboring big ones and for low-altitude populations exposed to genetic 'rainfall' from higher-altitude populations [50 ].
Obtaining knowledge about gene flow and drift may necessitate zooming in and utilizing biogeographic data at more local geographic scales. In practical terms, data on landscape (habitat) isolation and geographic distances between populations can be used as proxies of gene flow [11]. Genetic drift is partly dependent on gene flow, but relatively independent proxies include population size and recent colonization ('founder effects'). Together, data on these nonadaptive proxies can be described as biogeographic layers, which can be applied in ecogeographic profiles for focal traits in FIGS (Figure 1).

Concluding remarks
All plant traits and their distributions are products of evolution, but the relatively few published attempts to apply FIGS in practice, to enhance crop plants' insect resistance for instance, suggest that partial knowledge of natural selection does not always provide sufficient precision to capture elusive traits. To improve FIGS it is important to catch up with corresponding progress in basic eco-evolutionary theory. First, to improve prediction of adaptive processes it is necessary to identify conflicting or concurrent selection pressures imposed by other agents and integrate them into the ecogeographic profile (Figure 1). Second, following previous suggestions [11], biogeographic proxies of nonadaptive processes (gene flow and genetic drift) should ideally be routinely considered in future FIGS dealing with elusive traits (Figure 1). These adjustments to FIGS should help breeders to identify the most promising evolutionary hotspots for trait mining.
Furthermore, resistant genebank accessions identified via FIGS can be used for genomic prediction based on characterization with genotyping by sequencing as training populations. This may lead to identification of useful diversity hotspots hosting genotypes that can be advantageously used introgressively in breeding programs. This approach was proposed to improve grain yield in wheat [53], and subsequently shown to have promising potential to raise biomass yields of sorghum [54]. Such 'turbocharging' of genebanks offers a cost-effective strategy to tap their valuable plant genetic resources by shifting from 'gene mining' to estimating accessions' breeding values through genomics.

Conflict of interest statement
Nothing declared.

5.
Amanuel T, Paliwal R, Manthi SJ, Odeny DA, Midega CAO, Khan ZR, Pickett JA, Bruce TJA: Genome wide association analysis of a stemborer egg induced "call-for-help" defence trait in maize. Sci Rep 2020, 10:11205. This research group has previously shown that landraces of maize retain important indirect resistance traits, incl. herbivore-induced plant volatiles that attract natural enemies of herbivores. Here, the authors carried out GWAS to identify genomic regions associated with indirect defence. Together with other recent papers from this group, these results show that indirect resistance traits are of key interest for rewilding of staple crops.

12.
Egan PA, Muola A, Stenberg JA: Capturing genetic variation in crop wild relatives: an evolutionary approach. Evol Appl 2018, 11:1293-1304. This paper represents the first, and so far only, attempt to integrate nonadaptive processes into FIGS. The results show that gene flow is an important evolutionary factor shaping wild strawberry traits, suggesting that proxies of nonadaptive processes routinely should be considered in future FIGS attempts.