Grassland is one of the most widespread and dominating biomes worldwide, responsible for 30-35% of global terrestrial net primary production and storing approximately one third of the global terrestrial carbon stock. Grassland ecosystems are highly sensitive to changes in climate and human activities. With the continuous atmospheric CO2 concentration increase and recent drying trend widely observed in the North American grassland, it is an unresolved question how the grassland vegetation growth will change. Furthermore, vegetation growth may respond not only to simultaneous climate variability, but also to that of previous months. Yet it is unknown whether and to how long this lagged vegetation response may exist in the North American grassland.
Here, we explored the temporal dynamics of grassland vegetation growth (normalized difference vegetation index, NDVI) in North America and its response to climate variabilities. CRU climate data and UMD global land cover map were resampled in according to the GIMMS NDVI grid. We used piecewise linear regression to detect the potential turning point in NDVI time-series trend, and Akaike Information Criterion for regression model comparison. In addition, we used maximum regression analyses to estimate potential time lag effect of vegetation growth responses to climate.
Results/Conclusions
At continental scale, although a statistically significant positive trend of average growing season NDVI was observed (0.53×10-3 yr-1, P=0.036), we showed two distinct periods with opposite growth trends: growing season NDVI first increased before mid 1990s (2.6×10-3 yr-1, P<0.001), then decreased after mid 1990s (1.1×10-3 yr-1, P<0.001). The timings of this trend shift, however, varied spatially. In addition, we also found significant time lags between vegetation growth and climate factors, particularly for precipitation changes, while the length and strength of the lagged response were spatially variable. Our results highlight that grassland is highly sensitive to temperature-induced water limitation, and it responds to climate changes both at concurrent basis and by imposing time lags.