Sowing-time strategies alter maize grain yield, stover production and biomass allocation across contrasting rainfed environments in eastern Kenya

Uncertain rainfall onset complicates maize establishment in eastern Kenya, where both grain and stover contribute to farm livelihoods. We evaluated pre-rain dry-soil sowing, rainfall-triggered wet sowing and delayed sowing over five seasons at water-limited Machang’a and wetter Kiamaogo. Sowing treatment was assigned to split-plot main plots and locally predominant maize varieties to subplots, with three replications and repeated observation of the same physical plots. Site-specific mixed models quantified grain yield, stover yield, harvest index, establishment, productive-cob formation and grain-yield failure, while sowing-specific rainfall exposure was reconstructed for all ten site–season environments. At Machang’a, dry sowing increased grain yield by 51–86% relative to wet or delayed sowing in the 2009 short rains and 2010 long rains. In three other seasons, 94.4–100% of plots failed to produce grain despite established plants and measurable stover, showing that vegetative biomass could persist under conditions that prevented grain production. At Kiamaogo, sowing responses varied among variety–season combinations. Delayed sowing increased stover by 1.96–3.69 Mg ha⁻¹ in six of 15 combinations and reduced harvest index by 0.086–0.125 in four of 15 combinations, indicating shifts in biomass allocation rather than a consistent grain-yield advantage. Across sites, treatment responses reflected strong interactions among environment, season and variety. The study shows that sowing-time strategies can substantially alter both grain production and residue supply in rainfed maize systems and provide a field basis for developing climate-informed sowing strategies under variable rainfall.