A grand challenge facing humanity is how to produce food for a growing population in the face of a changing climate and environmental degradation. Though empirical evidence remains sparse, management strategies that increase environmental sustainability, like increasing agroecosystem diversity through crop rotations, may also increase resilience to weather extremes without sacrificing yields. We used multilevel regression analyses of long-term crop yield datasets across a continental precipitation gradient to assess how temporal crop diversification affects maize yields in intensively-managed grain systems. More diverse rotations increased maize yields over time and across all growing conditions (28.1% on average), including in favorable conditions (22.6%). Notably, more diverse rotations also showed positive effects on yield under unfavorable conditions, with yield losses reduced by 14.0 to 89.9% in drought years. Systems approaches to environmental sustainability and yield resilience like crop rotation diversification are a central component of risk reduction strategies and should inform enabling policies.

This study involved 11 long-term rainfed experiments on crop rotation that were located in the U.S.A. and Canada (see Table 1 of associated publication). These experiments included all those existing that could be identified, using literature searches (e.g. “crop rotation” AND yield AND [corn OR maize]) and personal communication from experts, that met our criteria in these two countries: at least three full rotation cycles of data; including maize monoculture or two-crop maize rotations as well as more complex rotations; rainfed conditions; no other treatments confounded with rotation (e.g. farming systems comparisons). Beyond maize, other crops in rotation varied depending on suitability for local conditions (Tables 1 and S1). Between 16–58 years of data were available from the experiments and all had numerous complete rotation cycles. In all experiments, each phase of every rotation was present every year, so maize yields were measured from every rotation every year. Maize genotypes were the same across rotations within a given site in a given year. Site specific management (e.g. genotypes, inputs, and field operations), can be found in site references (see Table S1). With one exception (Woodslee, ON), rotations were replicated in an experimental design similar to a randomized complete block design. Historical data on maize yields were requested from researchers at each experiment during spring and summer 2016 and subsequently processed to allow for cross-site analyses.

The dataset provides maize yields from 11 long-term experiments in the U.S.A. and Canada. The 11 long-term experiments correspond to the "site" code, which in turn correspond to the sites described in Table 1 and SI Table 1 of the associated publication. The "system" variable describes the sequence of crops in rotation. See SI Table 1 for abbreviations. The corn yield data are in kg per ha at 15.5% moisture. "block" and "plot" describe within site block and plot IDs. "tillage" describes the three levels of tillage, see SI Table 1 for abbreviations. "fertilization" describes four types of nitrogen fertilization, ZN = zero nitrogen applied; LN = low nitrogen applied; SN = typical nitrogen rates applied; ON = organic nitrogen applied. See the caption for SI Figure 4 for details on fertilization, and Table 1 for actual levels of fertilizer nitrogen applied.