Harnessing Crop Wild Diversity for Climate Change Adaptation

Abstract

Warming and drought are reducing global crop production with a potential to substantially worsen global malnutrition. As with the green revolution in the last century, plant genetics may offer concrete opportunities to increase yield and crop adaptability. However, the rate at which the threat is happening requires powering new strategies in order to meet the global food demand. In this review, we highlight major recent 'big data' developments from both empirical and theoretical genomics that may speed up the identification, conservation, and breeding of exotic and elite crop varieties with the potential to feed humans. We first emphasize the major bottlenecks to capture and utilize novel sources of variation in abiotic stress (i.e., heat and drought) tolerance. We argue that adaptation of crop wild relatives to dry environments could be informative on how plant phenotypes may react to a drier climate because natural selection has already tested more options than humans ever will. Because isolated pockets of cryptic diversity may still persist in remote semi-arid regions, we encourage new habitat-based population-guided collections for genebanks. We continue discussing how to systematically study abiotic stress tolerance in these crop collections of wild and landraces using geo-referencing and extensive environmental data. By uncovering the genes that underlie the tolerance adaptive trait, natural variation has the potential to be introgressed into elite cultivars. However, unlocking adaptive genetic variation hidden in related wild species and early landraces remains a major challenge for complex traits that, as abiotic stress tolerance, are polygenic (i.e., regulated by many low-effect genes). Therefore, we finish prospecting modern analytical approaches that will serve to overcome this issue. Concretely, genomic prediction, machine learning, and multi-trait gene editing, all offer innovative alternatives to speed up more accurate pre- and breeding efforts toward the increase in crop adaptability and yield, while matching future global food demands in the face of increased heat and drought. In order for these 'big data' approaches to succeed, we advocate for a trans-disciplinary approach with open-source data and long-term funding. The recent developments and perspectives discussed throughout this review ultimately aim to contribute to increased crop adaptability and yield in the face of heat waves and drought events.

Keywords: abiotic stress tolerance; ex situ conservation; genebanks; genetic adaptation; genome-wide selection scans (GWSS); genome–environment associations (GEA); genomic prediction (GP); germplasm collections; machine learning (ML).

Figure 2

A pipeline for machine learning (ML) applications capable of predicting abiotic stress tolerant and susceptible germplasm accessions. First, a subset of the germplasm collection is (a) characterizing genomically, phenotypically (whenever possible), and environmentally (i.e., abiotic stress adaptation indices based on geo-referencing). This subset is later on partitioned between (b) training and (c) testing populations. The training population is used to calibrate (d) ML models that aim using genomic information to predict genomic estimated adaptive values (GEAVs, an analogous rank to the polygenic risk score (PGS) and genomic estimated breeding value (GEBV) from the quantitative genomics literature, e.g., [102,136]). The computer screen depicts a hypothetical hidden neural network (HNN) algorithm, which is one among many potential ML tools; the repertoire includes several regressions, classification, and deep learning models, thoughtfully reviewed this year by Sebestyén et al. [137] and Tong and Nikoloski [138]. Meanwhile, the testing population is used to compute the (e) unbiased predictive ability of the model by comparing the GEAVs with the recorded environmental (or phenotypic) abiotic stress tolerant/susceptible indices. Broadly speaking, calibrated and validated ML models can serve two main purposes when applied on germplasm collections. First, (f) they could enhance our knowledge on the genomic architecture (i.e., genetic basis) of abiotic stress tolerance via ML-based genome-wide association studies (GWAS), and on the genomic landscape of adaptation via ML-based genome-wide selection scans (GWSS) and genome–environment associations (GEA). Second, (g) calibrated and validated ML models can be applied on a (h) query population such as extended germplasm samples for which environmental-based indices or phenotyping are not viable, informing GEAVs and (i) abiotic stress tolerance on a wider genepool. Clusters of abiotic stress tolerance and susceptibility based on phenotypic information and/or environmental-based indices can be built using traditional classification tools such as the ones listed in Table 1, or may also leverage ML prediction approaches (Table 2).

Figure 1

A roadmap of trans-disciplinary approaches aiming at harnessing genebank utilization for climate change research in the face of heat, and water scarcity. Compiling (a) previous characterizations and (b) geo-referencing-derived climate data/indices of available genetic resources in genebanks is a starting point to (c) assess the extent of abiotic stress tolerance among existing accessions, and the need of (d) new habitat-based population-guided collections targeting isolated pockets of cryptic diversity in dry and semi-arid regions. Planning question-oriented collecting trips of crop wild relatives and hidden landraces across contrasting environments/agro-ecologies is needed now more than ever, despite a century of gathering and preserving diversity in plants throughout genebanks. Coupling ex situ agro-ecological screenings together with (e) ongoing in situ genebanks characterizations for morphological and genetic variation is essential to define (c) putative tolerant reference collections, while understanding the (f) heritability (h²) of adaptive traits and their genetic architecture (i.e., underlying genes) via genome-wide selection scans (GWSS), genome–environment associations (GEA), and genome-wide association studies (GWAS). Since identifying these novel sources of heat and drought tolerance demands merging heterogeneous datasets, (g) machine learning (ML, in red letters) promises speeding up genebank characterization. The distinction that clustering (Table 1) and ML (Figure 2 and Table 2) strategies can provide between abiotic stress tolerant and susceptible accessions is essential to (h) transfer useful genetic variation from wild crop donors and early landraces into elite cultivated lines, either by designing (i) genomic-assisted breeding programs such as genomic prediction (GP) and inter-specific marker- and genomic-assisted backcrossing (MAB and GABC) schemes, or by envisioning (j) multi-trait gene editing strategies (e.g., CRISPR-Cas9). Once (k) abiotic stress tolerant varieties are validated across different environments, (l) legal inscription, seed multiplication, seed delivery system to farmers’ associations, and (m) follow-up given the regional needs, market demands, and adoption potential, are necessary downstream validation steps. These heterogeneous datasets are also likely to be inputted into ML, and in turn feedback new needs beyond heat and drought tolerance such as other types of resistances and nutritional quality. For ML to succeed speeding up the breeding of heat and drought-tolerant crops, there must be long-term funding to generate and maintain an assortment of datasets at each step, which in turn need to be publicly available through open access repositories from various geographic locations. Red boxes highlight different reservoirs of wild and cultivated diversity within the Cartesian space, gray boxes are mixed datasets built around these collections, and connectors are methodological approaches.