Understanding the weather signal in national crop-yield variability

31 Year-to-year variations in crop yields can have major impacts on the livelihoods of subsistence farmers 32 and may trigger significant global price fluctuations with severe consequences for people in developing 33 countries. Fluctuations can be induced by weather conditions, management decisions, weeds, diseases, 34 and pests. Although an explicit quantification and deeper understanding of weather-induced variability 1 of crop yields is essential for adaptation strategies it has so far been left to empirical models. Here, we 2 provide conservative estimates of the fraction of reported national yield variabilities that can be 3 attributed to weather by state-of-the-art process-based crop model simulations. We find that observed 4 weather variations can explain more than 50% of the variability in wheat yields in Australia, Canada, 5 Spain, Hungary, Romania and Morocco. For maize weather sensitivities exceed 50% in six countries 6 including the USA. For rice the explained variance exceeds 50% in Japan and South Korea and for soy 7 in Argentina. Inhibiting water stress by simulating yields assuming full irrigation shows that water 8 limitation is a major driver of the observed variations in most of these countries. 9 Identifying the mechanisms causing crop yield fluctuations is fundamental not only when aiming at 10 damping it but also in context of the attribution of loss and damage debate in context of climate change. 11 As process-based crop models not only account for weather influences on crop yields but also provide a 12 representation for human management measures they could become essential tools to separate the 13 effects of the individual forces and explore options to reduce future yield fluctuations. 14 15