Crop rotations and canola yields: Evidence from field-level data in Western Canada

Abstract

Canola (Brassica napus) acreage increased in Western Canada in recent years, leading to rotations with fewer break years between canola plantings. Field trials suggest that frequent plantings of canola reduce canola yields. However, there is considerable disagreement about the magnitude and persistence of these effects. We analyze the effect of rotational practices on canola yields in Saskatchewan using over 20 years of observational data, representing 61% of canola hectares in the province. We examine how the impact of rotations varies across time, soil zone, soil moisture conditions and the distribution of yields. We regress canola yields in Saskatchewan on the share of land that was planted with particular crops in previous years, using a battery of covariates and fixed effects to address potential bias in the model. After including these fixed effects, we cannot reject the hypothesis that there is no sample selection bias. We use an unconditional quantile estimator to investigate how rotations affect different deciles of the yield distribution. Our analysis confirms that crop rotations significantly influence canola yields, albeit more modest than field trials suggest. We find a 7.5% yield reduction when canola follows canola, compared to cereals, with this penalty persisting for 4 years but diminishing in magnitude with each additional year. The adverse effects of consecutive canola plantings are more pronounced in wetter regions and at lower yield deciles. Conversely, canola yields are higher when planted after pulse crops (as opposed to after cereal crops).