China’s deserts greening and response to climate variability and human activities

Desert, and in the Kubuqi Desert. The spatial distribution of the NDVI trends in summer was the most similar to the spatial distribution of the annual NDVI trends; this may have occurred because the annual NDVI was calculated by the MVC method, while the maximum values of NDVI in most regions occurred in summer. Correlations between the NDVI and climatic factors The Pearson correlation coefficients between monthly, seasonal, and annual NDVI and cli- matic factors were calculated to measure the effect of climate variability on NDVI in deserts and to compare the impact of different time scales on the correlations, as presented in Table 3. Comparing the differences in the effects of climatic factors on NDVI at different scales, it was found that the correlations between climatic factors and NDVI were significant at seasonal and monthly scales, while only precipitation was significantly correlated with NDVI at annual scales. This suggested that climate variability was the main cause of monthly and seasonal vari- ation in NDVI, while only precipitation had some influence on NDVI variation at the interan- nual scale. Lagged effect of climatic factors on NDVI. Considering the possible lag in the effect of temperature and precipitation on the NDVI, the Pearson correlation coefficients for the corre- lations between the current NDVI of each month and the climatic factors of the current month and previous months were calculated (Table 4). The results showed that temperature and precipitation had no lagged effect on the monthly

NDVI because the largest correlation coefficients for the correlations between the NDVI and temperature and precipitation in the same month were 0.796 and 0.869, respectively, and the correlation decreased as the number of lagged months increased. When the lagged months increased to 4 months, NDVI started to be negatively correlated with temperature and precipitation. Table 3. The Pearson correlation coefficients of the climatic factors and NDVI at different time scales. Scale Temperature Precipitation Annual -0.087 0.544� Seasonal 0.865�� 0.866�� Monthly 0.869�� 0.796�� � Significant at level 0.10; �� significant at level 0.05. https://doi.org/10.1371/journal.pone.0256462.t003 Table 4. The Pearson correlation coefficient for the relationship between the monthly NDVI and temperature and precipitation for different time lags. Time lag Temperature Precipitation 0 0.796�� 0.869�� 1 0.764�� 0.864�� 2 0.532�� 0.622�� 3 0.115 0.226�� 4 -0.286�� -0.227�� 0 indicates the correlation coefficients are calculated by the NDVI and climatic factors in the same month, 1 indicates the correlation coefficients are calculated by the NDVI in the current month and climatic factors in the previous month, etc. �� Significant at level 0.05. https://doi.org/10.1371/journal.pone.0256462.t004 PLOS ONE China’s deserts greening and response to climate variability and human activities PLOS ONE | https://doi.org/10.1371/journal.pone.0256462 August 30, 2021 10 / 20