Gaps between farmer and attainable yields across rainfed sunflower growing regions of Argentina

► We report gaps between farmer (reporting district) and attainable (trial) yields. ► Gaps are significant (p85%) by environment (year, site) effects. ► Magnitude of gaps justifies research into their causes and mitigation thereof. We computed three estimators of attainable yield for each of between 5 and 8 rainfed sunflower-growing regions of Argentina using between 5 and 9 years of data over the 2000–2007 interval. The estimators were based on comparative yield trial (CYT) data for commercial hybrids, on individual commercial field (ICF) data, and on reporting district (RD) yield information. Contrasts between these estimators led us to prefer the attainable (CYT) yield estimator over the other two. Attainable (CYT) yields ranged from 2.21 to 2.83tha−1 across regions. Yield gaps between mean farmer (RD data) and attainable (CYT) yields were computed using best linear unbiased estimator (BLUE) values for both variables obtained using mixed linear models. These gaps were statistically significant (p≤0.05) for all 8 regions and ranged from 0.37 to 1.18tha−1 across regions, for a country average of 0.75tha−1, equivalent to 41% of the mean country yield of 1.85tha−1. We also used CYT data to examine the issue of recurrent, albeit infrequent, reports of unusually high yields. Mean yields for the top decile of comparative yield trial data ranged from 3.2 to 4.2tha−1 across regions, and the highest yields for this decile in any of the years of record ranged from 3.9 to 4.8tha−1 across regions. Individual commercial field yields were available for 5 regions. Gaps between BLUEs for this variable and attainable (CYT) yields were smaller than those between reporting district and attainable (CYT) yields, but were nevertheless significant in all 5 regions. A notable feature of reporting district, individual field, and yield trial data was their variability. At reporting district level within regions, contributions of spatial and temporal variability were roughly similar. The mean relative contribution of the trial effect to non-error variance of the CYT data exceeded 85% across regions, dominating the contributions of genotype and of genotype by trial effects. We conclude that the magnitude of mean farmer/attainable (CYT) yield gaps for this crop in Argentina justifies further research aimed at