Extended Data Fig. 5: Attribution of negative GPP events to specific climate drivers and changes in the frequency of compound events.

a, Percentage of attributed negative GPP events to each climate driver based on the entire period 1982–2016 and global scale (bars; median of each dataset) with corresponding percentage of attributed negative GPP relative to the total negative GPP extremes. The attribution allows for multiple drivers per event thus percentages add up to more than 100% (see Methods for details). In total 68.7% (72.1%) of the events (Cumulative GPP anomaly) can be associated with these climate drivers (median of the three datasets) with potential other drivers such as fire⁷⁶, insects^81,82, wind explaining the remainder⁸³. b, c, Changes in the frequency of compound events of considered droughts (b, SPI and c, SPEI (% of events)) coinciding with high temperatures between the two periods. The black vertical (a) / horizontal (b,c) lines correspond to the defined threshold of 10% where lower attribution rates indicate insignificance of the corresponding driver or occurrence of analysed compound events, respectively (see methods).