Spatiotemporal changes of global landscape ecological risk and its impact on NPP changes at the watershed scale

The rapid increase in the intensity of human activities has continuously changed the global underlying surface, resulting in ecological risks to the regional landscape. In this study, an ecological risk system was constructed for the watershed-scale landscape using a variety of landscape pattern indices. The global major basins and sub-basins were taken as the research objects. The land use raster data was collected for 34 consecutive years from 1982 to 2015. The spatiotemporal ecological risk of the global landscape was analyzed at the watershed scale. The various driving factors of landscape ecological risks were then quantified using geo-detectors. At the same time, the spatiotemporal variation was determined for the global net primary productivity (NPP) of vegetation at the watershed scale, in order to reveal the impact of global landscape ecological risks. The results showed that: 1) The range of spatial variation was 0-0.29 for the multi-year average ecological risk of the global landscape at the watershed scale. Specifically, the high-value areas were distributed mainly in the temperate and cold temperate zones, arid climate zone, and polar zone near the Arctic. Meanwhile, 17.44% and 23.45% of the watershed landscape ecological risks showed a significant increase and decrease trend from 1982 to 2015, respectively (P<0.05). 2) There was a large spatial variation in the coefficient of variation of landscape ecological risk in the arid climate, subtropical, and tropical zones. Furthermore, 36.82% watershed of significant change-points were found in the landscape ecological risk from 1988 to 2006. The spatial distribution of the watersheds with the significant changes in the landscape ecological risk was basically consistent with that of watersheds with significant change-points (P<0.05). The global underlying surface changed greatly in the 1990s, indicating the great variation in the landscape ecological risk. 3) There was the greatest impact of precipitation on the landscape ecological risk of major- and sub-watersheds at the multi-year scale. The influencing factors on the landscape ecological risk of watersheds were ranked the descending order of the precipitation, temperature, NDVI, population density, altitude, slope, and gross domestic product. There were also complex interactions among the various influencing factors on the landscape ecological risk. 4) The annual NPP of 52.32% and 12.60% of the watersheds increased and decreased significantly from 1982 to 2015, respectively. The landscape ecological risk was conducive to the NPP in the 9.11% watershed with a significant positive correlation between annual landscape ecological risk and annual NPP (P<0.05). By contrast, the annual landscape ecological risk was significantly negatively correlated with the annual NPP in the 11.43% watershed (P<0.05), where the reduction of landscape ecological risk was conducive to the NPP. The finding can also provide a solid scientific basis for global carbon neutrality and carbon sinks in the terrestrial ecosystems under climate change.