Modelling photoperiod and temperature responses of flowering in quinoa (Chenopodium quinoa Willd.)

Abstract

Two modelling approaches were used to quantify photoperiod and temperature responses of time from emergence to visible flower buds in nine quinoa (Chenopodium quinoa Willd.) cultivars. The first, non-interactive model, considers temperature and photoperiod responses as independent, and the threshold photoperiod, critical photoperiod, and base temperatures as constants. The second, interactive model, considers these attributes as variable, and allows for interaction between photoperiod and temperature responses. Controlled-environment experiments with a factorial combination of temperature and photoperiod provided information on responses, and data from field experiments were utilized in tests of the predictive capacity of the models. The two models were very similar in their goodness of fit and predictive capacity, but testing revealed that some assumptions about the interactive model were not fulfilled, whereas the non-interactive model is more consistent with the data. Both the models failed to predict dates of visible flower buds when average temperatures during the phase were >20°C; it is proposed that interaction between irradiance receipt and high temperature in controlled environments result in lower optimum temperatures there than in the field. Differences between field data and predicted values were eliminated when predictions were recalculated assuming no optimum for the temperature response. All nine cultivars examined are short-day plants. A juvenile sub-phase was observed in the six cultivars for which it was tested; and its duration was negatively associated with the latitude of origin of the lines (R2 = 0.9, p < 0.05). Photoperiod sensitivity was negatively associated with the latitude of origin of the lines (R2 = 0.55, p < 0.05) and positively associated with duration of the basic vegetative phase (minimal time between emergence and visible flower buds) (R2 = 0.55, p < 0.05) using the non-interactive model. Photoperiod and temperature response parameters were not significantly associated with the latitude of origin for the interactive model (p > 0.05).