The randomized complete block design had 9 treatment combinations (3 soil cropping patterns × 3 rain simulation durations), each replicated in 3 blocks (altogether 27 plots). The three maize cropping systems were FP, RF, and FMRF, respectively, and the three rain simulation durations were 2, 4, and 6 min, respectively, at a sprinkling intensity of 10 mm min−1. On each of the RF and FMRF plots, narrow (15 cm high × 40 cm wide) and wide (10 cm high × 70 cm wide) ridges were alternated (Fig. 1). We assumed that maize plants were grown in all three treatments, with the same spacing (30 cm) in crop lines (or furrows) and the same density (about 57,000 plants ha−1) in the fields, as those in the local farms. Because the narrow and wide ridges alternated in both RF and FMRF treatments, the spacing between furrows (where maize was seeded in a straight line) interchanged (Fig. 1). Therefore, we assumed that in the treatment of FP, the spacing between maize lines was also interchanged.