The role of phenology in the climate change impacts and adaptation strategies for tree crops: a case study on almond orchards in Southern Europe

Abstract The assessment of the impact of climate change on Mediterranean crop systems is affected by a large number of uncertainties. To overcome this issue in an impact and adaptation assessment for tree crops, an experimental dataset containing 556 phenological observations was used to calibrate and validate a modelling framework based on the Dynamic Model and ASYMCUR approach, for assessing the flowering date of common Mediterranean almond cultivars. Data were collected over 12 years for 15 almond cultivars in 5 locations in Andalusia (Southern Spain), covering a wide range of weather conditions. The model performance was good: for late-flowering cultivars Root Mean Square Error (RMSE) was 3.6 days and Nash-Sutcliffe model efficiency (NSE) was 0.89, while for early-flowering cultivars RMSE was 3.4 days and NSE was 0.83. Weather projections from an ensemble of 12 climate model outputs including 3 Representative Concentration Pathways (RCPs) were used with the modelling framework for each almond cultivar to quantify the changes in flowering date and the associated weather conditions during this stage, under the future weather conditions of the Iberian Peninsula (IP). Thus, in future scenarios, depending on location, start of full bloom was delayed (in mild-winter areas) or advanced (in cold-winter areas), with the change in weather conditions during critical phenological stages potentially affecting the yield in different ways, depicting a high spatial variability in the projections within the IP. For this reason, a spatial analysis was applied to demarcate those areas with adverse weather conditions related to flowering stage. In light of our results it is concluded that the identification of impacts and adaptation strategies for Mediterranean agriculture requires a careful prior evaluation of the systems at local scale, due to the marked spatial heterogeneity and the remaining high uncertainties associated with the crop modelling.