Based on the county level census yield data as well as the corresponding weather observations during 1961-2014, yield limiting factors for oilseed rape in Hubei province were analyzed by employing two flexible nonparametric models such as Classification and Regression Trees (CART) and Random Forests model. The results showed that temperature, sunshine hours, as well as drought/waterlogging had significant effects on yield, and such impacts varied among different phenological stages. Oilseed rape was sensitive to low temperature before flowering while sensitive to high temperature from then on, and sensitive to drought during seedling and budding stages, while sensitive to waterlogging during flowering and podding stages, effect of sunshine hours was highly dependent on water condition, with a negative effect under water deficits, while a positive effect when there was excess rainfall. Among all the limiting factors, waterlogging had the highest occurrence frequency, followed by drought and cold damage. Waterlogging reduced yield directly by limiting root aerobic respiration, as well as decreased in photosynthesis and increased in disease incidence due to absence of sunshine and high humidity, and the latter was more important.
