A three-season field study found that hybrid barley was less likely than pure-line barley to produce severely reduced yields under weather stress.
Researchers analyzed yield distributions from a five-factor winter barley field experiment, varying cultivar, sowing density, nitrogen rate, growth regulation and disease control across three contrasting seasons. Instead of comparing only average yields, they estimated the probability that yields would fall below selected thresholds.
Across the experiment, the probability of yields at or below the overall annual mean of 9.62 megagrams per hectare was 0.40 for the hybrid cultivar and 0.62 for the pure-line cultivar. The hybrid's advantage was most apparent among the lowest yields, particularly in seasons with water deficit or excessive precipitation.
Lower risk at the bottom
The hybrid cultivar's yield distribution generally indicated less exposure to unfavorable outcomes than the pure-line cultivar's distribution. At the overall annual mean threshold of 9.62 Mg per hectare, yields were at or below that level in 40% of hybrid observations and 62% of pure-line observations, a difference reported as statistically significant (P = 0.0002).
The gap was largest in the lower end of the yield range during both drought and excess-rain seasons. The researchers also identified management combinations associated with lower conditional risk; the reported risk values under optimized combinations were 0.09 for the pure-line cultivar and 0.24 for the hybrid cultivar.
Why poor harvests matter
Average yield alone does not show how often a crop produces a particularly poor harvest. By examining the full spread of yields, this approach gives cultivar and management comparisons a direct interpretation as the probability of falling below a chosen yield level.
The results suggest that hybrid winter barley may offer more protection against severely low yields under the two weather stresses studied. The framework could support evaluations of cultivars and management choices where avoiding poor outcomes is important, while the findings remain specific to the production system tested.
Evidence and caveats
The study used 108 plots in a five-factor factorial field experiment at one experimental location in north-eastern Poland, across three growing seasons: wet, dry and near-normal. Researchers constructed empirical cumulative distribution functions, which estimate the share of observations below each yield threshold, and compared the cultivar distributions using stochastic-dominance analysis and Kolmogorov–Smirnov tests.
The observations represent variability created within a structured set of management treatments at one site, not randomly sampled environmental conditions across many farms or regions. The authors note that limited observations in some treatment combinations can reduce the precision of the estimated distributions. The method describes the pattern of risk but does not by itself identify its biological cause or separate the effects of individual management factors. Broader generalization would require experiments at multiple locations.
// Source
Scientific Reports · 2026 · DOI: 10.1038/s41598-026-72925-w
Authors: Dariusz Załuski, Krzysztof J. Jankowski, Bogdan Dubis
Institutions: University of Warmia and Mazury in Olsztyn
