Spatiotemporal patterns of phenological metrics and their relationships with environmental drivers in grasslands

Climate-induced changes in plant phenology and physiology are crucial in regulating terrestrial productivity and ecosystem functions. However, the spatiotemporal patterns of grassland phenology and its relationships with environmental factors remain unclear. We extracted phenological metrics from grasslands using the FLUXNET dataset (34 sites and 169 site-year). We then explored the spatiotemporal variations in phenological metrics, their relationships with gross primary productivity (GPP), and the driving mechanisms behind them using regression analysis and structural equation modeling methods. The start of the growing season (SOS) significantly advanced, whereas the end of the growing season (EOS) was slightly delayed (non-significant), leading to an extension of the growing season (LOS) (marginally significant) with increasing latitude northward. The multi-year averaged GPP in grassland sites was exponentially correlated with LOS and linearly correlated with maximum GPP (GPP_max). Phenological metrics exhibited linear relationships with mean annual temperature and quadratic relationships with mean annual precipitation (MAP). EOS, LOS, and GPP_max increased (SOS decreased) with MAP initially, then leveled off or decreased (SOS increased) when MAP reached a threshold of 1000 mm. Spatiotemporally, preseason soil water content (SWC) and air temperature significantly affected SOS, and wind speed was the dominant environmental driver for EOS. Structural equation modeling further suggested that decreasing wind speed might delay the EOS by reducing the atmospheric and soil dryness. In conclusion, our findings suggested that an improved grassland phenological model could project an advancing SOS, a delaying EOS, and an extension of LOS in response to decreasing wind speed and increased moisture in the future.