Modelling maize silk extension using segmented exponential and linear functions

Abstract

Silk extension determines the timing of silking, subsequent pollination and kernel setting, which has not been accounted in functional-structural plant (FSP) modelling for maize. In this study, a two-year field experiment with two local maize hybrids i.e. AnNong 591 (referred to as AN591) and ZhongDan 909 (referred to as ZD909) was carried out in Hefei, Anhui Province, China in both 2019 and 2020. The silks at basal, middle and top sections of a maize ear were destructively sampled daily in capturing silk extension. Silk extension prior to its senescence is described with a segmented model i.e. an exponential function driven by cell division (Phase I), and two linear functions corresponding to rapid silk extension (Phase II) and slow extension (Phase III) driven by cell expansion. Data in 2019 were used to fit silk extension on basal, middle and top ear positions, and the associated parameter values for the exponential and linear functions were extracted while the highest R² was achieved. The model was then compared with the Logistic model by evaluating the normalized root mean square error (nRMSE) between simulations and observations in 2019 and 2020. It was shown that the ability of a segmented model in 2019 fitting was sound though a bit poorer than a Logistic model, however, the validation for the former model performed better than the latter when independent data in 2020 were applied. This indicated that the simple segmented model is competent in predicting silk extension. In addition, silk extension at basal, middle and top sections of a maize ear for both maize hybrids was visually demonstrated. In conclusion, modelling silk extension was successfully realised by the segmented model composed of exponential and linear functions, which complements FSP modelling of maize.