Genetics-inspired data-driven approaches explain and predict crop performance fluctuations attributed to changing climatic conditions

Abstract

Genetics-focused approaches have been widely used to uncover major genetic variants associated with performance variation. Selecting, manipulating, and editing genetic variants significantly improve crop performance. Meanwhile, the genetic component explains a portion of performance variation, and the environmental component contributes to the remaining, often large, portion (Laidig et al., 2017Laidig F. Piepho H.P. Rentel D. Drobek T. Meyer U. Huesken A. Breeding progress, environmental variation and correlation of winter wheat yield and quality traits in German official variety trials and on-farm during 1983–2014.Theor. Appl. Genet. 2017; 130: 223-245https://doi.org/10.1007/s00122-016-2810-3Crossref PubMed Scopus (118) Google Scholar; Bonecke et al., 2020Bonecke E. Breitsameter L. Bruggemann N. Chen T.W. Feike T. Kage H. Kersebaum K.C. Piepho H.P. Stutzel H. Decoupling of impact factors reveals the response of German winter wheat yields to climatic changes.Glob. Chang Biol. 2020; 26: 3601-3626https://doi.org/10.1111/gcb.15073Crossref PubMed Scopus (33) Google Scholar; Li et al., 2021Li X. Guo T. Wang J. Bekele W.A. Sukumaran S. Vanous A.E. McNellie J.P. Cortes L.T. Lopes M.S. Lamkey K.R. et al.An integrated framework reinstating the environmental dimension for GWAS and genomic selection in crops.Mol. Plant. 2021; 14: 874-887https://doi.org/10.1016/j.molp.2021.03.010Abstract Full Text Full Text PDF PubMed Scopus (32) Google Scholar). To ensure superior and robust performance, elite varieties are extensively tested across multiple years and locations. These extensive performance records, coupled with climatic profiles, could be leveraged to understand climate’s impact on agriculture through approaches parallel to quantitative genetics approaches (Figure 1A). We have conducted a soft white spring wheat trial spanning 16 years. On average, 15 varieties were tested in five locations each year (Figure 1B; Supplemental Table 1). Grain yield, two grain-quality traits (test weight and protein content), and two agronomic traits (heading date and plant height) have been measured consecutively from 2005 to 2020, while falling numbers, a quality trait measuring starch degradation by endogenous α-amylases, has been evaluated since 2013 (Supplemental Table 2). In total, 56 unique varieties (half are marketed) were tested in 80 environments (year-location combinations). Among four commonly tested (>50 times) varieties, two were grown in all 80 environments. Of 30 varieties that were tested less than 20 times, 10 were evaluated before 2013, and the other non-overlapping 20 were tested after 2013 (Supplemental Figure 1A). The two-dimension (row for variety and column for environment) performance records from a multi-environment trial could be aggregated row- or column-wise (Figure 1A). We assessed the impact of data sparsity on row- or column-wise marginal mean aggregations (Supplemental Figures 1B, 1C, 2, and 3). The column-wise aggregation (environmental means) was affected less, which corroborated the large contribution (from 61.3% to 81.9%; Supplemental Figure 4A) from the environmental component. The large number of environments enabled the characterization of performance variation at the environment level. The performance fluctuations, represented by the environmental mean estimated from all tested varieties, epitomized the profound contributions from changing climate conditions to crop performance (Figure 1C). Grain yield was the most volatile trait, with a coefficient of variation (CV) of 33%. The most favorable condition (Aberdeen in 2014, 161.5 bu/A) yielded 12 times more than the worst condition (Soda Springs in 2007, 13.7 bu/A). On the other hand, test weight was the most stable trait (60.7 ± 2.1 lb/bu, CV = 3.4%). Location was the major contributor for variation for grain yield and agronomic traits, while year was the main contributor for grain-quality variation (Supplemental Figure 4B). In the row-wise space, searching through genome-wide genetic variants could uncover major variants associated with performance variation at the variety level (Figure 1A). With a whole-season climatic variant matrix, the analogous concept could be applied in the column-wise space to identify influential ones to explain performance variation at the environment level. We compiled such a matrix (80 × 52 200) with five parameters derived from temperature and precipitation (Supplemental Methods). For example, one climatic variant, GDD10–20, the average daily temperature from 10 to 20 days after planting (DAP), ranged from 5.6°C (10.1°F) (Rupert in 2011) to 17.1°C (30.7°F) (Soda Springs in 2008). The 80 environments could be grouped into two major clusters with this whole-season climatic variant matrix. Environments from Soda Springs and Ashton (both near mountains) were in one cluster, while environments from the other three locations (in the Snake River Plain) were clustered together (Supplemental Figure 5), which suggested that the terrain feature has a significant impact on weather conditions among these testing sites. We used Critical Environmental Regressor through Informed Search (CERIS) (Li et al., 2018Li X. Guo T. Mu Q. Li X. Yu J. Genomic and environmental determinants and their interplay underlying phenotypic plasticity.Proc. Natl. Acad. Sci. U S A. 2018; 115: 6679-6684https://doi.org/10.1073/pnas.1718326115Crossref PubMed Scopus (97) Google Scholar, Li et al., 2021Li X. Guo T. Wang J. Bekele W.A. Sukumaran S. Vanous A.E. McNellie J.P. Cortes L.T. Lopes M.S. Lamkey K.R. et al.An integrated framework reinstating the environmental dimension for GWAS and genomic selection in crops.Mol. Plant. 2021; 14: 874-887https://doi.org/10.1016/j.molp.2021.03.010Abstract Full Text Full Text PDF PubMed Scopus (32) Google Scholar; Guo et al., 2020Guo T. Mu Q. Wang J. Vanous A.E. Onogi A. Iwata H. Li X. Yu J. Dynamic effects of interacting genes underlying rice flowering-time phenotypic plasticity and global adaptation.Genome Res. 2020; 30: 673-683https://doi.org/10.1101/gr.255703.119Crossref PubMed Scopus (35) Google Scholar; Mu et al., 2021Mu Q. Guo T. Li X. Yu J. Phenotypic plasticity in plant height shaped by interaction between genetic loci and diurnal temperature range.New Phytol. 2022; 233: 1768-1779Crossref PubMed Scopus (11) Google Scholar) to search through this climatic variant matrix (Supplemental Figures 1D, 1E, and 6). Uncovered major climatic variants for all six traits were from growth periods earlier than the time of the corresponding traits being fully developed (Figure 1F). A single variant explained 34.2%–67.1% of performance variation (Figure 1G and Supplemental Figure 7). Increasing 0.6°C (1°F) of GDD in early season (represented by GDD1–41) promoted heading 1.7 days earlier, which quantified the impact from climatic conditions. The temperature fluctuations during the grain-filling period (captured by dGDD69–136) had a significant contribution to protein content (Supplemental Figures 6 and 7). A smooth change in GDD (lower dGDD69–136) generally promoted protein accumulation. A composite variant from precipitation and temperature, PRDTR111–137, explained 65.6% of variation for falling numbers (Figure 1G and Supplemental Figures 6 and 7). The combination of heavy rainfall and low diurnal temperature range (DTR) near the end of the 2014 season broke the grain dormancy and induced pre-harvest sprouting, which contributed to low falling numbers and low test weight. Grain yield and plant height, two highly correlated traits (Supplemental Figure 8), were associated with GDD from growth periods around heading stage, GDD33–74 and GDD50–74, respectively (Supplemental Figure 6). High early-season temperature (GDD33–74) inhibited wheat tillering and decreased yield. The late-season temperature (GDD124–136) also significantly contributed to grain yield but in the opposite direction as high temperature promotes grain filling. Considering GDD33–74 and GDD124–136 together increased the proportion of explained variation from 0.43 to 0.54 (Figure 1G), suggesting an additivity of effects between two independent variants. Meanwhile, the late-season temperature had a minimal contribution to plant height (Supplemental Figure 6), which corroborated that plant height was generally determined near the heading stage. With the strategy paralleling genetic dissection, biologically meaningful environmental variants that significantly contribute to crop performance could be systematically identified. In the row-wise space, genomic selection leverages genome-wide genetic variants to predict performance for new varieties (Meuwissen et al., 2001Meuwissen T.H. Hayes B.J. Goddard M.E. Prediction of total genetic value using genome-wide dense marker maps.Genetics. 2001; 157: 1819-1829Crossref PubMed Google Scholar; Millet et al., 2019Millet E.J. Kruijer W. Coupel-Ledru A. Alvarez Prado S. Cabrera-Bosquet L. Lacube S. Charcosset A. Welcker C. van Eeuwijk F. Tardieu F. Genomic prediction of maize yield across European environmental conditions.Nat. Genet. 2019; 51: 952-956https://doi.org/10.1038/s41588-019-0414-yCrossref PubMed Scopus (109) Google Scholar). For the column-wise space, we tested the analogous concept, enviromic prediction, of using whole-season climatic variants to predict the performance of new environments. Two cross-validation schemes had consistent prediction accuracies (Figure 1H). Based on contribution modes (Supplemental Figure 4), we conducted two forecasting schemes. The first scheme, using the chronicle order to separate training and forecasting environments (i.e., environments from 2005 to 2015 were used to build the model to predict the performance for the 2016 environments), predicted grain yield and agronomical traits well (Figure 1I). The second scheme, using the location as the separator (i.e., performance from Ashton was predicted by environments from other four locations), captured grain-quality traits better (Figure 1J). The high prediction accuracies suggested that environomic prediction is a promising approach to predict expected performance under the constantly changing climate conditions. Variation at the variety or environment level summarizes information embedded in the same performance records. Because of the parallelism, principles from the row-wise space (i.e., genetic component) could be applied to the column-wise space. For example, significant genetic variants for the same trait may vary among surveyed populations comprising different individuals. Thus, major environmental variants may vary among surveyed multi-environment trials. Extrapolating either genomic selection or enviromic prediction models should be done cautiously because models are context dependent. With the advancement of high-throughput envirotyping technologies (Xu, 2016Xu Y. Envirotyping for deciphering environmental impacts on crop plants.Theor. Appl. Genet. 2016; 129: 653-673https://doi.org/10.1007/s00122-016-2691-5Crossref PubMed Scopus (162) Google Scholar), a whole-season environmental matrix could consist of other climatic parameters, management, soil profile, and microbiome dynamics. Identifying major environmental conditions will provide a new approach for characterizing and understanding environmental conditions to improve crops (Resende et al., 2021Resende R.T. Piepho H.P. Rosa G.J.M. Silva-Junior O.B. FF E.S. de Resende M.D.V. Grattapaglia D. Enviromics in breeding: applications and perspectives on envirotypic-assisted selection.Theor. Appl. Genet. 2021; 134: 95-112https://doi.org/10.1007/s00122-020-03684-zCrossref PubMed Scopus (54) Google Scholar). Like genome-editing technology that can be applied to correct adverse mutations occurring in key known genes to modify traits, in future, some cost-effective technologies may also be applied to intervene in adverse environmental conditions, which could be quantified by explicit climatic variants. This work was supported by the Agriculture and Food Research Initiative competitive grant (2021-67013-33833) and the Federal Hatch Funds (IDA01312) from the USDA National Institute of Food and Agriculture, by the USDA-ARS In-House Project 2090-21000-033-00D, by the Idaho Wheat Commission, and by the Iowa State University Plant Sciences Institute.

Topics & Concepts