Agrometeorological models to forecast acai (Euterpe oleracea Mart.) yield in the Eastern Amazon

Abstract

BACKGROUND The increasing demand in Brazil and the world for products derived from the acai palm (Euterpe oleracea Mart) has generated changes in its production process, principally due to the necessity of maintaining yield in situations of seasonality and climate fluctuation. The objective of this study was to estimate acai fruit yield in irrigated system (IRRS) and rainfed system or unirrigated (RAINF) using agrometeorological models in response to climate conditions in the eastern Amazon. Modeling was done using multiple linear regression using the 'stepwise forward' method of variable selection. Monthly air temperature (T) values, solar radiation (SR), vapor pressure deficit (VPD), precipitation + irrigation (P + I), and potential evapotranspiration (PET) in six phenological phases were correlated with yield. The thermal necessity value was calculated through the sum of accumulated degree days (ADD) up to the formation of fruit bunch, as well as the time necessary for initial leaf development, using a base temperature of 10 degrees C. RESULTS The most important meteorological variables were T, SR, and VPD for IRRS, and for RAINF water stress had the greatest effect. The accuracy of the agrometeorological models, using maximum values for mean absolute percent error (MAPE), was 0.01 in the IRRS and 1.12 in the RAINF. CONCLUSION Using these models yield was predicted approximately 6 to 9 months before the harvest, in April, May, November, and December in the IRRS, and January, May, June, August, September, and November for the RAINF. (c) 2019 Society of Chemical Industry