With changing climate and rising population, it will become more challenging to grow more crops to satisfy food demand. Hence, it is highly important to develop strategies for sustainable intensification of agricultural production. Establishing such strategies requires an understanding of historical trends of crop yields to identify potential yield level and its rate of increase. We approach this problem by studying yield patterns across different temporal and spatial scales. Specifically, we use multilayer network structure of countries, where layers of network connections are defined based on synchronism of long-term trends or short-term fluctuations, common climate, etc. We apply community detection algorithms to find clusters of countries. Overall, the clustering results will characterize global patterns of yields for different crops, identify drivers behind the increase/decrease/stagnant yield in groups of countries, and facilitate the reduction of the disparities in access to food resources.