Physiological adaptive traits are a potential allele reservoir for maize genetic progress under challenging conditions

Yield progress under challenging environmental conditions is crucial to mitigate the negative impacts of climate change and resource limitations. Breeding and improved agronomic practices have significantly increased yields in many crops, including maize. A substantial portion of maize yield gains resulted from optimizing plant cycle duration to match local environmental conditions, leveraging genetic variability in flowering time. However, significant yield increases have also occurred within specific cycle durations, attributed largely to direct selection for yield and, more recently, genomic selection. This study addresses the critical question of whether this rate of genetic progress is sustainable. Evaluating the potential for further improvement requires a comprehensive analysis linking yield progress to underlying traits and their corresponding genomic regions. This approach helps determine which traits can still be improved and assesses the extent to which breeding programs have already fixed alleles for these traits. While some studies have examined genomic changes associated with selection, yield progress over time, or trait changes with variety release, a holistic analysis linking changes in alleles, traits, and yield across diverse environmental conditions has been lacking. This is challenging because many yield quantitative trait loci (QTLs) exhibit positive, negative, or neutral effects depending on the environment. Therefore, a multi-environment analysis encompassing trait effects on yield and generational changes is vital. This study focuses on two categories of traits: adaptive physiological traits (e.g., stomatal control, leaf growth), which vary rapidly in response to environmental conditions and are difficult to phenotype at scale, and constitutive traits (e.g., phenological phase duration, plant architecture), which have longer-term variations and are more amenable to selection. The study's primary objective was to determine the extent to which yield-based selection has influenced both adaptive, environment-dependent traits, and constitutive traits with more stable effects.

The literature reveals significant advancements in maize yield through breeding and agronomic improvements. Studies highlighting the role of flowering time in yield gains are well-documented, showing that fine-tuning cycle duration to environmental conditions significantly impacts yield. However, a significant portion of yield increase has been achieved within specific cycle durations through direct selection for yield. Genomic selection methods have further accelerated this progress. Existing research has investigated genomic changes associated with selection and trait changes over time. However, a comprehensive analysis integrating changes in alleles, traits, and yield across various environmental scenarios is lacking. This is due to the complexity of quantitative trait loci (QTLs), many of which have environment-dependent effects on yield. This research gap underscores the need for this study's multi-environmental approach, examining both adaptive and constitutive traits and their relationship to yield improvement.

This study used a maize panel of 66 European varieties released between 1950 and 2016. The research involved several key steps: (i) Phenotyping: Physiological traits such as stomatal conductance, growth responses to environmental conditions, plant architecture, and reproductive development were measured using novel phenomic methods in ten high-precision experiments across three phenotyping platforms and three equipped fields. Precise definitions of the measured variables are provided in Supplementary data 11. (ii) Yield Measurement: Yield and its components were measured in 30 field experiments across Europe (Supplementary Table 2). (iii) Trait Contribution Analysis: Linear models and Bayesian networks were used to analyze the contributions of different traits to yield improvement. (iv) Genomic Analysis: The study investigated whether genomic regions associated with the traits showed signatures of selection. This was done by comparing genomic regions associated with traits (identified in a diversity panel with a similar genetic background) with signatures of selection in the genetic progress panel (Supplementary Fig. 1). The hypothesis that adaptive traits did not respond to selection due to their conditional effects on yield was also examined. The studied environments and hybrids represent typical conditions in the European maize-growing area, encompassing a wide range of latitudes (43-48°N), covering most of the European maize growing area and capturing the variability of yields and environmental conditions. Environmental scenarios were defined by combinations of air temperature, evaporative demand, and soil water status, which significantly affected yield. The genetic progress panel comprised elite breeding pools of European and American germplasm, showing an appreciable shift in SNP frequency with the year of release and an increasing proportion of lodent material. A diversity panel of 250 hybrids served as a training population for genomic prediction of yield, aiding the comparison of QTLs identified in this panel with signatures of selection in the genetic progress panel. The rate of genetic progress was assessed by analyzing the relationship between yield and year of release, considering various environmental scenarios and plant densities. Trait contributions to yield improvement were investigated using linear models and Bayesian networks. Finally, genomic regions associated with the traits were compared to those showing signatures of selection to determine the extent of exploitation of available genetic diversity by breeders.

The study revealed a consistent annual genetic gain in grain yield of 101 kg ha⁻¹ year⁻¹ across diverse environmental scenarios (favorable and unfavorable), indicating a similar rate of progress in both high- and low-temperature conditions as well as varying soil water status. This rate was also similar at different plant densities, contradicting the hypothesis that breeding for high plant density was a key driver of yield improvement. Yield progress was predominantly driven by an increased grain number, while individual grain weight showed little change. Breeders indirectly improved reproductive development through changes in plant phenology. Though crop cycle duration remained largely constant, the vegetative phase duration increased, and the grain-filling phase duration decreased, resulting in a reduced anthesis-silking interval (ASI). This change increased the number of ovary cohorts and improved the proportion of ovary cohorts becoming grains. Plant architecture also contributed to yield progress. The vertical distribution of leaf area shifted, with a greater proportion of leaf area at lower canopy levels, increasing light interception at the ear level, resulting in improved grain development. Despite considerable genetic variability and heritability, physiological adaptive traits such as leaf growth sensitivity, stomatal conductance, and water-use efficiency showed little change with the year of release. Genomic analysis revealed that regions under selection (RUS) strongly overlapped with QTLs for flowering time, florigens, and architectural traits, but showed little overlap with QTLs for stomatal conductance, leaf growth rate, or water-use efficiency. Further analysis of six published yield QTLs demonstrated that adaptive traits correspond to QTLs with scenario-dependent allelic effects on yield, fluctuating considerably across environments. Conversely, constitutive traits (flowering time and plant architecture) correspond to QTLs with more stable allelic effects. Breeding for yield primarily selected for constitutive traits with stable effects while leaving adaptive traits largely unexploited.

The findings highlight the unexpected observation that adaptive traits, essential for plant survival and reproduction under stress, were not significantly influenced by breeding in elite temperate maize varieties. While adaptive traits might be advantageous only under extreme conditions, the study revealed considerable genetic variability and heritability for these traits, suggesting that available genetic diversity was not fully exploited by breeding. The consistent yield improvement across diverse conditions suggests that selection for yield indirectly enhanced reproductive development and light-use efficiency. The absence of selection on adaptive traits could be due to their context-dependent effects on yield, with oscillations in allelic effects across environments preventing consistent selection for beneficial alleles. This suggests that constitutive traits, with stable effects on yield across environments, were favored by selection. However, these constitutive traits might approach their limits for further yield improvement. The study suggests an untapped potential in physiological adaptive traits, and the use of novel breeding methods, focused on improving yield stability under various environmental conditions and exploiting genotypic responses to environmental variables, should be considered. Future breeding strategies incorporating yield stability, detailed genotype-environment-management interaction analysis, and the genetic variability of yield responses to environmental conditions could lead to further yield gains. The use of big data strategies integrating phenomics, modeling, and genomic prediction will be critical in this effort.

This study demonstrates that while yield in temperate maize has increased significantly over the past 65 years primarily due to selection for constitutive traits affecting plant architecture and phenology, a significant potential remains in the exploitation of adaptive traits that enhance resilience to environmental stress. The absence of selection for these traits likely stems from their context-dependent effects on yield, leading to fluctuating allelic frequencies across years and environments. Future research should focus on incorporating novel phenotyping and breeding strategies to unlock this untapped genetic resource and enhance maize yield stability under increasingly challenging climatic conditions. Further research into high-throughput phenotyping methods and advanced modelling techniques for capturing complex genotype-by-environment interactions is crucial to effectively utilize this genetic reservoir.

The study's focus on European maize varieties limits the generalizability of the findings to other geographic regions. The specific set of traits analyzed might not encompass all the relevant factors influencing yield under various conditions. The availability of historical data could also influence the analysis, as some older hybrids were not readily accessible for phenotyping, introducing a potential bias in the representation of older varieties. While a wide range of environmental conditions was investigated, the study might not fully capture the variability of extreme climate events expected in the future.