Climate change may outpace current wheat breeding yield improvements in North America

Wheat is a key global staple, with the North American Great Plains producing about 10% of global wheat and 30% of high-quality wheat exports. Rising temperatures and associated extremes threaten regional wheat yields, with prior estimates indicating 1–10% yield losses per degree of warming without adaptation and reduced yield stability due to extremes. Modeling studies often assume adaptation via shifting to later-maturing varieties to maintain growing period length under warming, but such assumptions may not reflect real-world breeding outcomes. Long-term, controlled field observations comparing constant check varieties with contemporaneous advanced breeding lines have been scarce, limiting understanding of how genetic advances have altered climate sensitivity of yields. This study asks whether current wheat breeding has improved climate resilience and whether breeding gains can keep pace with projected warming, using multi-decade nursery data to benchmark real-world breeding effectiveness for winter and spring wheat in North America.

Prior empirical and modeling work has documented negative impacts of warming and heat extremes on wheat yields, particularly around flowering and grain filling, and suggested adaptation via variety selection to maintain season length. Meta-analyses and process-based modeling have projected yield declines under warming but often rely on stylized adaptation (e.g., late-maturing cultivars) that may be unrealistic. Studies have highlighted the role of extreme heat (EDD) and freezing injury (FDD) on wheat, with GDD effects often weaker. Some regional analyses reported mixed evidence on the temperature tolerance of newer winter wheat cultivars. Recent global studies suggest climate change may slow genetic gains, given long breeding and adoption cycles. This study situates itself by providing empirical, multi-site, multi-year evidence comparing advanced lines and constant checks grown side-by-side, thereby addressing limitations in previous assessments.

Data: Annual regional reports from Northern Regional Performance Nursery (NRPN) and Southern Regional Performance Nursery (SRPN) for winter wheat and the Hard Red Spring Wheat Uniform Regional Nursery (HRSWURN) for spring wheat across 92 sites in the North American Great Plains (1960–2018), totaling 85,770 observations (58,472 winter; 27,298 spring). Each site-year trial included a constant check variety (Kharkof for winter; Marquis for spring) and a rotating set of advanced breeding lines. For each year, advanced lines were summarized into percentiles: high-yielding genotype (HYG, 97.5th), median-yielding genotype (MYG, 50th), and low-yielding genotype (LYG, 2.5th). Analysis considered rainfed conditions. Climate variables: Daily Tmin, Tmax, and precipitation obtained from GHCN-D and MSC. Hourly temperature was reconstructed by fitting a cosine curve to daily Tmin and Tmax. Over each growing season, cumulative indices were computed: freezing degree-days (FDD; temperatures below freezing injury threshold), growing degree-days (GDD; between base and optimum thresholds), and extreme growing degree-days (EDD; above optimum threshold). Precipitation totals were also used. Temperature thresholds followed established literature; sensitivity tests varied windows and thresholds. Statistical model: Panel fixed-effect regressions of yield on FDD, GDD, EDD, and precipitation were estimated separately for CK, LYG, MYG, and HYG, with site and time controls as specified. Uncertainty was quantified by bootstrapping site-years (n=1000) to derive 95% confidence intervals. To estimate the net impact of uniform warming, observed daily temperatures were perturbed by +1 °C across the season, and resulting changes in FDD, GDD, and EDD were used with regression coefficients to compute yield impacts. Future projections: Using the historical regression relations (representing current breeding progress), future yield changes were projected under CMIP6 climate models (ACCESS-ESM1-5, BCC-CSM2-MR, CNRM-CM6-1, CNRM-ESM2-1, GFDL-ESM4, IPSL-CM6A-LR) for SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5. Projections included changes in GDD, EDD, FDD (generally increasing GDD/EDD and decreasing FDD), with mixed precipitation signals. Results were summarized as changes relative to baseline CK yields, with temporal evolution and 95% ranges across climate models.

• Sensitivity to +1 °C warming (historical baseline):
  • Winter wheat: Advanced lines (LYG, MYG, HYG) show 3.4–3.6% yield reductions per +1 °C, less sensitive than the check Kharkof (−5.5% per +1 °C). Reduced sensitivity reflects smaller EDD-related losses and larger benefits from reduced FDD in advanced lines; e.g., MYG EDD impact −4.5% (95%: −7.0 to −1.8) vs CK −6.1% (−9.0 to −2.8), and MYG FDD benefit +1.58% (0.95–2.34) vs CK +0.88% (0.09–1.75).
  • Spring wheat: Advanced lines are slightly more sensitive than the check Marquis: MYG −7.5% per +1 °C (−10.0 to −5.55) vs CK −7.1% (−9.90 to −4.18). This arises from larger EDD penalties (MYG −9.2% vs CK −8.4%) and smaller FDD benefits (MYG +0.97% vs CK +1.33%).
• Regression significance across genotypes:
  • EDD: significantly harmful for both winter and spring wheat (multiple t-tests with p < 0.01 to p < 1e−11 across CK, MYG, HYG, LYG).
  • FDD: significantly harmful for both types (p-values < 0.03 to < 1e−6 across genotypes), highlighting winter-kill and freezing effects.
  • GDD: generally weak and statistically insignificant effects in both winter and spring wheat.
• Future projections (relative to baseline CK yields), using current breeding-climate relationships:
  • Check varieties decline markedly: Kharkof −12.7% by end-century in SSP1-2.6 and −47.2% in SSP5-8.5; Marquis −24.8% (SSP1-2.6) and −62.7% (SSP5-8.5).
  • Winter wheat MYG: benefits over CK persist but are eroded with warming; MYG yields drop to historical CK levels around +6 °C mean growing season warming in SSP3-7.0 and SSP5-8.5.
  • Spring wheat MYG: more vulnerable; around +3.6 °C warming reduces MYG yield to historical Marquis levels in SSP2-4.5, SSP3-7.0, and SSP5-8.5.
  • Yield gains over CK (percent): Winter MYG gains around 46% (2030s) and 50% (2050s), and by 2090s range from 47.2% (SSP1-2.6) to 64.8% (SSP5-8.5); LYG 3.9–7.3%; HYG 85.5–115.4%. Spring MYG gains ~44.9% (2030s–2050s) but decline by 2090s to 43.8% (SSP1-2.6) and 32.6% (SSP5-8.5); LYG ~9.6–9.9%; HYG ~67.4–76.0%.
• Overall, current breeding has improved winter wheat climate resilience but not spring wheat; projected warming can outpace yield gains from current breeding, especially for spring wheat.

By directly comparing advanced breeding lines against constant check varieties grown side-by-side across decades and sites, the study isolates the breeding contribution to climate sensitivity. It shows that winter wheat breeding has improved tolerance to high-temperature extremes and reduced sensitivity to freezing, thereby lessening yield losses per degree warming relative to the check. In contrast, spring wheat advanced lines show greater sensitivity to heat extremes than the check, likely reflecting the alignment of spring wheat flowering with hotter periods and differences in developmental genetics (e.g., vernalization requirements). Consequently, breeding-derived yield advantages diminish under projected warming, with spring wheat particularly vulnerable as modest warming (~3.6 °C) negates MYG gains. These results challenge modeling assumptions that simple shifts to late-maturing varieties can fully offset warming, and highlight that real-world breeding progress alone is unlikely to maintain yields under high-emission scenarios. The findings underscore the need for both mitigation (limiting warming toward SSP1-2.6 to preserve breeding gains) and targeted adaptation, including geographic and phenological adjustments (northward shifts for winter wheat, earlier spring planting) and deployment of advanced breeding technologies to develop heat- and cold-resilient cultivars. Robustness checks against prior studies, county-level data, alternative climate windows, and temperature formulations support the reliability of the conclusions.

Using multi-decade, multi-site nursery data under rainfed conditions, the study benchmarks how current wheat breeding affects climate sensitivity. It finds improved climate resilience for winter wheat but not for spring wheat, and projects that future warming will outpace yield gains from current breeding—particularly under higher-emission scenarios—thereby eroding advantages of advanced lines. To sustain productivity, both slowing climate change and accelerating adaptation are essential. Future work should (i) identify and introgress major genetic determinants of heat and cold stress tolerance, (ii) leverage speed breeding, doubled haploids, and genomic selection to shorten breeding cycles, (iii) evaluate advanced lines across diverse, extreme-prone environments, and (iv) refine integrative assessments that couple genomics with climate-crop modeling to guide cultivar development under evolving climates.

• CO2 fertilization effects are not included due to high uncertainty in estimating year-to-year CO2 impacts from the available data.
• The modeling reflects historical breeding progress and does not capture potential acceleration from new breeding technologies or future changes in breeding speed and adoption.
• Projections rely on climate model outputs; precipitation changes are mixed and add uncertainty to yield projections.
• Analysis is based on rainfed nursery trials and constant check comparisons; results may not generalize to irrigated systems or management changes not represented in the nurseries.
• Temperature index thresholds and growing season definitions were tested for sensitivity, but residual uncertainties remain in translating thermal exposures to yield responses.