Objectives Under the rotation of winter wheat and spring maize (3 years for winter wheat and 1 year for spring maize), the effects of tillage and fertilization on crop yield in rain-fed agricultural area of Northwest China under different precipitation years were studied to provide theoretical basis for high and stable crop yield in the Northwest upland.

Methods This study is based on a long-term positioning experiment of tillage and fertilization in dryland soil set up in Zhenyuan County, Qingyang City, Gansu Province in 2005. This study is a two-factor experiment, with tillage as the main treatment and fertilization as the secondary treatment, to analyze the yield and water effects of long-term cultivation and fertilization in a 17-year rotation system of wheat and maize, as well as the mechanisms of soil fertilization and sustainable yield increase. The tillage treatment included traditional tillage and no tillage, while the fertilization treatment included no fertilization (CK), single nitrogen fertilizer (N), single phosphorus fertilizer (P), single organic fertilizer (M), nitrogen fertilizer combined with phosphorus fertilizer (NP), and NP combined with organic fertilizer (NPM). Combining analysis of wheat and maize yields, this study explored the mechanisms of crop yield differences across rainfall years under tillage and fertilization treatments.

Results Different rainfall years, tillage, and fertilization methods significantly affected the yields of wheat and maize. Under different precipitation methods and tillage methods, the NPM treatment yields (winter wheat 3441 kg/hm², spring maize 8991 kg/hm²) were significantly higher than those of other fertilization treatments, with a more obvious effect of NPM treatment on crop yields under wet years. Compared to conventional tillage, no-tillage significantly reduced crop yields during drought years and wheat yields during normal rainfall years. In wet years, no significant difference was observed in crop yields between traditional tillage and no-tillage treatments. However, in drought and normal years, the average yield of winter wheat decreased by 15.6% and 25.7%, respectively, and maize yields decreased by an average of 17.9% and 4.6%. Compared to conventional tillage, no-tillage decreased wheat and maize seasonal water use efficiencies by 6.7% and 17.0% in drought years, and wheat seasonal water use efficiencies decreased by 24.0% in normal rainfall years, while there was no significant difference in wet years. In contrast, soil organic matter, total nitrogen, total potassium, alkaline dissolved nitrogen, Olsen-phosphorus and readily available potassium contents were significantly higher under no-tillage than under conventional tillage treatments.

Conclusions  Crop yield reduction caused by no-tillage is due to differences in crop water use efficiency rather than soil nutrient differences. In summary, tillage managements for rain-fed agriculture on the dryland Loess Plateau with dark loessial in eastern Gansu should be selected based on rainfall conditions, so as to improve the yield stability and sustainability of crops in this region. This knowledge provides a theoretical basis for the efficient use of precipitation and stable high yield in dryland farming in eastern Gansu.