Field sampling and laboratory analyses

The plant sample collection schedule is outlined in detail in Supplementary Table 2, but is described here in brief. Every season, the center 13.5 m² area in each plot in Embu and Harare and the center 7.5 m² area in each plot in Kiboko were harvested for grain yield estimation (differences in the harvested area reflected differences in the land availability at each site and, thus, the plot sizes). At harvest of the ninth season in Embu (2015SR), the fifth and sixth seasons in Kiboko (2013LR and 2014SR), and the fifth season in Harare (2014/15), 6 to 10 plants were sampled from the center of each plot at harvest and partitioned into grain and stover biomass. A Retsch SS MM200 Ball Mill Plant was used to grind these partitioned plant tissue samples to < 100 µm. A flash 2000 CHN Analyzer (ThermoFisher Scientific Inc.) was used to analyze for total C and N concentrations in the plant samples via the combustion method (Etheridge et al. 1998).

Following the 2015 harvest in all sites, composite soil samples comprised of 5 or 15 soil cores were taken near the center of each plot (both in row and between row) to a depth of 0.9 m in 5 depth increments (0–0.15, 0.15–0.3, 0.3–0.45, 0.45–0.6, and 0.6–0.9 m). Fewer cores (5) were taken in Embu due to the excessively compacted nature of the soil. Soil samples were ground, sieved through a 2 mm screen, and sub-sampled before shipping to Purdue University for analyses. Bulk density for each depth increment was measured using intact cores (Blake 1965).

Chemical and physical analyses of soil were performed to measure non-N essential plant nutrients and other factors (texture, CEC, pH etc.) that have been found to enhance or limit nutrient/water availability in maize systems (see Supplementary Table 1; Pasley et al. 2019 for more details). Soil samples were extracted with a 1 M KCl solution (10:1) and analyzed for NO₃-N and NH₄-N using an AQ2 Discrete Analyzer (SEAL Analytical) (Keeney and Nelson 1982). A sub-sample was ground to 100 µm diameter using a Retsch SS MM200 Ball Mill and analyzed for total C and N using the combustion method (Bremner and Mulvaney 1982; Nelson and Sommers 1982). Carbonates in soils were negligible, thus soil organic matter (OM) was estimated using the total C concentration and a single factor estimation of OM to C conversion factor of 2:1 (Pribyl 2010).