Search

Frost exposure and poor pollination are great challenges for the cultivation of perennial orchard systems such as apples, whose adaptive capacity to climate impacts is limited. Unfortunately, such climatic risks remained largely unexplored due to the complexity in future climate predictions, uncertainties in apple phenological simulations, and difficulties in quantifications of risks of late-spring frost and poor pollination. Here, focusing on the main apple production Sub-regions I–IV (i.e., the Loess Plateau, the Bohai Bay, the Southwest Cool Highland, and Xinjiang) in China, we used an ensemble of seven phenological models to estimate changes in apple first flowering and fruit-setting. Based on phenological sensitivity window proposed, we evaluated the late-spring frost risk and climatic risk of poor pollination based on two individual indices developed for these two kinds of meteorological risks. The phenology model ensemble was driven by climatic data derived from an ensemble of 27 global climate models (GCMs) from CMIP6 to simulate apple phenology in two periods of 2021 − 2060 (Near Future) and 2061 − 2100 (Far Future) under two climate scenarios of SSP245 and SSP585. Model ensemble could largely improve phenology prediction accuracy. Our results also confirmed the general earlier occurrence of apple phenological stages and the shortening of phenological sensitive windows. The first flowering advanced about 0.19 − 0.32 d∙y-1 and fruit-setting advanced about 0.24 − 0.47 d∙y-1. Although the shortening of phenological sensitivity window might decrease frost frequency in some regions and reduce the general risk of late-spring frost (e.g. in Sub-region II), the increase of frost intensity could offset this kind of alleviation and even exacerbate frost risks in Sub-regions I and IV. Finally, except for Sub-region III and a few sites in Sub-regions I and IV, shifts in climatic risks of poor pollination in spatial contrasts were lower than late-spring frost risk. This highlighted late-spring frosts would be more possibly the reason for apple yield or quality losses than the unfavorable climatic pollination conditions under climate change. Model ensemble provides a realistic assessment for quantifying future risks of late spring frost and poor pollination in apple production systems, and helps to identify urgent meteorological risks and provide options for apple production systems in response to climate change. [Display omitted] • Model ensemble approach improved the accuracies in apple phenophase prediction. • Apple phenophase is advanced while phenological sensitive window is shortened. • Warming may not reduce frost due to discrepancies in frost intensity & frequency. • Change of pollination risk is small in most regions of China under climate change. [ABSTRACT FROM AUTHOR]