Negative relationship between photosynthesis and late‐stage canopy development and senescence over Tibetan Plateau

Abstract

AbstractCanopy greening, which is associated with significant canopy structure changes, is the most notable signal of ecosystem changes in response to anthropogenic climate change. However, our knowledge of the changing pattern of canopy development and senescence, and its endogenous and climatic drivers is still limited. Here, we used the Normalized Difference Vegetation Index (NDVI) to quantify the changes in the speed of canopy development and senescence over the Tibetan Plateau (TP) during 2000–2018, and used a solar‐induced chlorophyll fluorescence dataset as a proxy for photosynthesis, in combination with climate datasets to decipher the endogenous and climatic drivers of the interannual variation in canopy changes. We found that the canopy development during the early green‐up stage (April–May) is accelerating at a rate of 0.45–0.8 × 10−3 month−1 year−1. However, this accelerating canopy development was largely offset by a decelerating canopy development during June and July (−0.61 to −0.51 × 10−3 month−1 year−1), leading to the peak NDVI over the TP increasing at a rate of only one fifth of that in northern temperate regions, and less than one tenth of that in the Arctic and boreal regions. During the green‐down period, we observed a significant accelerating canopy senescence during October. Photosynthesis was found to be the dominant driver for canopy changes over the TP. Increasing photosynthesis stimulates canopy development during the early green‐up stage. However, slower canopy development and accelerated senescence was found with larger photosynthesis in late growth stages. This negative relationship between photosynthesis and canopy development is probably linked to the source–sink balance of plants and shifts in the allocation regime. These results suggest a sink limitation for plant growth over the TP. The impact of canopy greening on the carbon cycle may be more complicated than the source‐oriented paradigm used in current ecosystem models.