Rice yield improvements through plant breeding are offset by inherent yield declines over time

Meeting the challenge of feeding a growing population with limited resources will require increasing the yield potential of staple crops, such as rice. Yet many high-yielding, intensive production systems have experienced slow rates of yield improvement in recent years despite a demonstrated increase in the yield potential of new crop cultivars. We analyzed experimental data from one such cropping system, i.e., California (CA) rice, in order to quantify improvements made in the genetic yield potential obtained through plant breeding. California rice systems are among the highest in the world and close to maximum yield potential. Specifically, the hypothesis was tested that if rice cultivar yields decline over time then apparent yield increases in side-by-side yield comparison tests will not reflect increases in yield potential. This hypothesis was tested using 33 years of experimental yield data from the California Cooperative Rice Research Foundation Rice Experiment Station. Based on side-by-side comparisons of old and new rice cultivars which do not consider yield decline over time, there was an apparent increase in yield. However, the yields of older cultivars were found to decline at an estimated rate of 29.3 kg ha(-1) year(-1) (90% credible interval -4.4 to -53.3) after initial selection. Once this effect was considered, the yield advantage of newer cultivars over old was uncertain (-3.3 kg ha(-1) year(-1), 90% credible interval -36.1 to 31.5). These results highlight (1) the importance of continuous crop improvement and deployment of new cultivars simply to maintain existing yields, and (2) to increase the genetic yield potential, higher yield targets are needed. Importantly, when breeding near the yield potential, despite the limited yield gains, significant advances in improving quality and reducing crop duration have been made.