CRONOSOJA : a daily time-step hierarchical model predicting soybean development across maturity groups in the Southern Cone

Abstract

Accurate prediction of phenology is the most critical aspect for the development of models aimed at estimating seed yield, particularly in species that exhibit variable sensitivity to environmental factors throughout the cycle and among genotypes. With this purpose, we evaluated the phenology of 34 soybean varieties in field experiments located in Argentina, Uruguay and Paraguay. Experiments covered a broad range of maturity group (MG)s (2.2–6.8), sowing dates (SDs) (from spring to summer) and latitude range (24.9–35.6 °S), thus ensuring a wide range of thermo-photoperiodic conditions during the growing season. Based on the observed data, daily time-step models were developed and tested, first for each genotype, and then across MGs. We identified base temperatures specific for different developmental phases and an extra parameter for calculating the photoperiod effect after the R1 stage (flowering). Also, an optimum photoperiod length for each MG was found. Model selection showed that the determinants of phenology across MGs were mainly affecting the duration of vegetative and early reproductive phases. Even so, early phases of development were better predicted than later ones, particularly in locations with cool growing seasons, where the model tended to overestimate their duration. In summary, we have constructed a soybean phenology model that simulates phenology accurately across various geographic locations and sowing dates. The model’s process-based approach has resulted in root mean square errors ranging from 5.8 to 9.5 days for different developmental stages.