Climate drives variability and joint variability of global crop yields

In this study, long-term national-based yields of maize, rice, sorghum and soybean (MRSS) from 1961 to 2013 are decomposed using Robust Principal Component Analysis (RPCA). After removing outliers, the first three principal components (PC) of the persistent yield anomalies are scrutinized to assess their association with climate and to identify co-varying countries and crops. Sea surface temperature anomalies (SSTa), atmospheric and oceanic indices, air temperature anomalies (ATa) and Palmer Drought Severity Index (PDSI) are used to study the association between the PCs and climate. Results show that large-scale climate, especially El Nino-Southern Oscillation (ENSO) and North Atlantic Oscillation (NAO) are strongly correlated with crop yield variability. Extensive maize harvesting regions in Europe and North America, rice in South America, Oceania and east of Asia, sorghum in west and southeast of Asia, North America and Caribbean and soybean in North and South America, Oceania and south of Asia experienced the influence of local climate variability in this period. Sorghum yield variability across the globe exhibits significant correlations with many atmospheric and oceanic indices. Results indicate that not only do the same crops in many countries co-vary significantly, but different crops, in particular maize, in different PCs also co-vary with other crops. Identifying the association between climate and crop yield variability and recognizing similar and dissimilar countries in terms of yield fluctuations can be informative for the identified nations with regard to the periodic and predictable nature of many large-scale climatic patterns. (C) 2019 Elsevier B.V. All rights reserved.