Soybean Phenology Simulation in the North-Central United States

Forecast simulation of soybean [Glycine max (L.) Merr.] vegetative stem node (Vn) and reproductive (Rn) phases can be used to improve timing of agronomic practices to optimize input use efficiency. A soybean model (SoySim) was used to predict soybean phenological stages in researcher and producer fields and to evaluate sources of variance. Evaluation of SoySim phenology simulation involved 21 and 22 Nebraska soybean producers in 2009 and 2010, respectively, with cultivars of various maturity groups (MGs 2.5–3.8) and planting dates (21 April–11 June). SoySim evaluation in research fields was conducted in Nebraska, Iowa, and Indiana with cultivar MGs 0.8 to 4.2 planted 27 April to 17 June. The combined root mean square error (RMSE) in research fields for V1 (first-node), R3.5 (mid-pod), R5 (beginning seed), and R7 (beginning maturity) was 3.6 d. In producer fields, Vn and Rn stages RMSE was 5.5 and 5.7 d, respectively. With few exceptions, the simulated and observed Vn or Rn stage data clusters in the research and producer fields did not differ (α = 0.05). Aside from subjectivity in most Rn calls that required quantifying pod and seed sizes, and pod maturity, variance in simulated and observed Vn or Rn in producer fields was related to the reliability of VE (emergence) calls, or the use of planting dates instead of VE dates in some producer fields. For no-till to minimum-till fields, the accuracy and precision of SoySim forecast simulation requires a reliable VE date call for scheduling soybean growth-stage dependent critical farming decisions.