Seed yield and quality of soybeans across the US: an overview of management and environmental characterization

Abstract

Soybean (Glycine max L.) is the main oilseed worldwide with the United States (US) producing 29 % of the global production (125 million Tn). The main targets of crop production in soybeans are yield (farmer profit), and seed quality in terms of protein (soybean meal for animal feed), and oil (for human consumption and biofuels). Seed yields are relatively stable due to the capacity of branches and nodes to adapt and maintain yield per area in diverse environmental conditions. However, soybean quality exhibits significant variation both within and between regions due to environmental, genetic, and management factors. Therefore, optimizing growth and understanding how management and environmental factors modulate seed yield and quality is critical for improving productivity and use of soybeans. This dissertation comprises six chapters (Chapter 1, Introduction, and Chapter 6, Conclusions). In Chapter 2, we explored metrics to characterize plant-to-plant spacing and its impact on yield across seeding rate and yield environments. We found that lack of plant-to-plant uniformity negatively impacted yield in both low- and medium-yield environments (<3 Mg ha-1) under low plant densities (<20 plants m-2), but with no effect at high yields (4.3 Mg ha-1). In Chapter 3, we developed an assessment of yield and seed quality allocation to mainstems and branches of an old and modern soybean genotype at varying row spacings. The modern genotype presented 50% greater yields, with a three-fold greater contribution from branches compared to the old genotype at narrow rows (<0.38 m). In addition, the modern genotype maintained protein levels despite high yields. In Chapter 4, we collected 235 on-farm seed and soil samples across 13 states of the United States (US). In addition, we collected management and weather data seeking to relate environment covariates to seed quality variation across regions. We predicted yield and seed oil concentration with better accuracy (R2 0.56 and 0.39, respectively) than protein concentration (R2 0.09). Yield and oil in the south were mainly limited by delayed planting dates that exposed early stages to high temperatures and shortened the season length. In the north, limitations were related to low temperatures during seed formation due to long maturity groups or late planting dates. Finally, in Chapter 5, we built a robust model for predicting oil (R2 0.54, MAE. 0.75%) and protein (R2 0.45, MAE 1.13%) concentration in all soybean fields of the US. We gathered satellite data on weather and soil, calculated the green chlorophyll vegetation index (GCVI) and derived the growing season length at a 1 x 1 km resolution. The outcomes of this study will assist farmers to develop near-real time predictions of soybean quality changes not only for segregation at field-scale but providing essential information for soybean exports.