Dataset from a five-season staggered-sowing maize experiment across contrasting rainfed environments in eastern Kenya
Description
This dataset contains curated experimental, rainfall-exposure, phenological, and statistical-analysis outputs from a five-season maize sowing experiment conducted at Machang’a and Kiamaogo in eastern Kenya between March 2009 and August 2011. The two sites represent contrasting rainfed environments: Machang’a is comparatively moisture-limited, whereas Kiamaogo is wetter and has greater soil-water-storage capacity. The split-plot experiment evaluated three sowing strategies (dry sowing, rainfall-triggered wet sowing, and delayed sowing) with three replications and site-relevant maize varieties. The same physical plots were followed across five growing seasons: LR09, SR09, LR10, SR10, and LR11. The repository includes plot- and treatment-level rainfall-exposure metrics, sowing and phenological dates, grain-failure summaries, covariance-structure comparisons, Type III fixed-effect tests, estimated marginal means, planned sowing-treatment contrasts, pairwise variety comparisons, establishment and cob-production analyses, phenological-model outputs, and covariance-sensitivity results. The archived files support the associated agronomic study on how staggered sowing modifies rainfall capture, grain yield, stover production, harvest index, and reproductive failure across contrasting rainfed environments. All datasets are provided as UTF-8 CSV files with descriptive filenames and supporting documentation, including a README. Treatment terminology, dates, units, and interaction notation were standardised during curation, while all numerical estimates and statistical results were retained at their original precision. The repository is intended to support reproducibility, secondary analysis, agronomic modelling, meta-analysis, teaching, and the development of rainfall-informed maize sowing recommendations for mixed crop–livestock systems.
Files
Steps to reproduce
To reproduce the archived results, download and extract the repository while preserving its folder structure. Import the UTF-8 CSV files from the data folder. Treatment-level rainfall summaries can be reproduced by averaging the plot-level rainfall-exposure data by site, season, and sowing treatment. Covariance selections can be verified by identifying the minimum AIC and BIC for each model. The archived Type III tests, estimated marginal means, planned contrasts, variety comparisons, establishment and phenology outputs, grain-failure summaries, and covariance-sensitivity results can then be checked directly against the corresponding CSV files. Grain-failure proportions are reproduced as failed plots divided by total plots, while contrast t ratios are reproduced as the estimate divided by its standard error.