Effectiveness of grazing exclusion on the restoration of degraded alpine grasslands on the Northern Tibetan Plateau from 2010 to 2017

Allowancing and awarding local herdsmen families for degraded grassland restoration, via grazing exclusion, is an important eco-compensation policy in China. A scientific assessment on how grazing exclusion and climate change affects ecological restoration is essential for improving the layout of enclosures at a broad spatial scale in the future. In this study, we focused on 2 413 enclosures that are distributed in Nagchu and Nagri, Tibetan Autonomous Region, China, and calculated the differences in the means and trends of the Normalized Difference Vegetation Index (DMean_NDVI and DTrend_NDVI) in these exclosures between two subperiods, prior to (2000–2009) and after (2010–2017) the current policy of payment for grassland restoration by fencing. The differences in the means and trends of the growing season temperature (GST) and precipitation (GSP) between corresponding subperiods were also calculated, for better understanding of how climatic variables affect NDVI change in fenced grasslands over time and across space. Our results show that exclosures with a ΔMean_NDVI > 0 and ΔTrend_NDVI > 0, account for 67.3% and 40.5% of all exclosures on the Northern Tibetan Plateau, respectively, indicating that grazing exclusion has effectively restored these degraded grasslands. The proportion of exclosures with a ΔMean_NDVI < 0 and ΔTrend_NDVI < 0 are 22.5% and 31.0%, respectively, indicating that grazing exclusion has negative influences on those fenced grasslands. A ΔMean_NDVI ≈ 0 and a ΔTrend_NDVI ≈ 0 can be found in 10.2% and 28.4% of exclosures, respectively, indicating that grazing exclusion by fencing has no evident influence there. In addition, we found that the NDVI of fenced grasslands were correlated with the GST (r = 0.27) and GSP (r = 0.37), and the correlation coefficients of NDVI with GST did not change between the two subperiods. In alpine and desert steppe zones, the correlation coefficients of NDVI with GSP increased from 0.236 in 2000–2009 to 0.370 in 2010–2017. The results of the linear regressions and analyses of variance further indicate that the GST and GSP have significant influences on the NDVI but can explain only a small fraction of its variance (approximately 2%). In summary, the effectiveness of grazing exclusion on the restoration of degraded grasslands showed a heterogeneous pattern over space across the Northern Tibetan Plateau. Therefore, we suggest opening the exclosures where fencing has negative influences, clarifying the optimal time for grazing-exclusion where fencing has positive influences, and exploring the potential effects of other anthropogenic factors where fencing has no evident influences on grassland restoration.