Leveraging soil mapping and machine learning to improve spatial adjustments in plant breeding trials

Spatial adjustments are used to improve the estimate of plot seed yield across crops and geographies. Moving means (MM) and P-Spline are examples of spatial adjustment methods used in plant breeding trials to deal with field heterogeneity. Within the trial, spatial variability primarily comes from soil feature gradients, such as nutrients, but a study of the importance of various soil factors including nutrients is lacking. We analyzed plant breeding progeny row (PR) and preliminary yield trial (PYT) data of a public soybean breeding program across 3 years consisting of 43,545 plots. We compared several spatial adjustment methods: unadjusted (as a control), MM adjustment, P-spline adjustment, and a machine learning-based method called XGBoost. XGBoost modeled soil features at: (a) the local field scale for each generation and per year, and (b) all inclusive field scale spanning all generations and years. We report the usefulness of spatial adjustments at both PR and PYT stages of field testing and additionally provide ways to utilize interpretability insights of soil features in spatial adjustments. Our work shows that using soil features for spatial adjustments increased the relative efficiency by 81%, reduced the similarity of selection by 30%, and reduced the Moran's I from 0.13 to 0.01 on average across all experiments. These results empower breeders to further refine selection criteria to make more accurate selections and select for macro- and micro-nutrients stress tolerance. Spatial adjustments utilizing soil maps perform better than traditional methods for spatial adjustments of trials. Soil-based spatial adjustments can be used to better understand the spatial variability in breeding trials. Site-specific machine learning models for spatial adjustments perform better than large generalized models. Plant breeding trials are a key component of crop improvement for yield, quality, and stress resistance. Breeding trials typically are grown on small plots of land and are highly affected by the area in the field where they are planted due to field trends. We investigated if using the soil features in a field could explain some of the variability in the early stages of a breeding program and used machine learning techniques to estimate the soil effects on observed yields. We found that by using the soil features for spatial adjustments, we could increase the accuracy of selections and improve the outcomes of decisions made by a breeder. This could have great impacts on increasing the accuracy of selection of early generation breeding trials, resulting in better lines being selected for yield, quality, and stress resistance traits, helping to make agricultural production more resilient and improve genetic gain.