The spatio-temporal dynamics of the peak of the growing season and its responses to climatic driving factors in Africa

Vegetation phenology characterizes the periodic plant life cycle events across a growing season. It is sensitive to climate change and thus important for the understanding of vegetation-climate interactions. Previous studies have revealed the importance of the start and end of the growing season, and investigated their dynamics with climate change. However, the spatio-temporal dynamics of the peak of the growing season (POS), which is also important to understand the response of plants to climate change, has not been well documented. Here, we applied three methods (Savitzky-Golay filtering, SG; Asymmetric Gaussian function, AG; and Double Logistic function, DL) in TIMESAT to estimate the latitudinal and longitudinal gradients and temporal trends of the peak of the growing season (POS) in Africa based on normalized difference vegetation index (NDVI) during 1982-2015. We investigated the relationships of POS with mean temperature, accumulative precipitation, and accumulative insolation over the preseason periods by computing temporal partial correlation coefficients. The results showed similar patterns and minor differences in POS based on the three methods. The averaged differences overall less than 0.4 days (MEANSG-AG = -0.1427 days, MEANSG-DL = 0.307 days, MEANAG-DL = 0.1983 days). For the spatio-temporal dynamics of POS, the latitudinal variations in POS were not very obvious but significant in both northern and southern Africa. In northern Africa, there was no clear longitudinal gradient in POS, while a significant west-east gradient in the temporal trends of POS was exhibited with POS change rate from 0.48 days/year to 0.19 days/year. From the temporal trends, we found an overall delayed trend of POS with 0.28 days/year on average in northern Africa and 0.25 days/year on average in southern Africa. In northern Africa, preseason precipitation was the most important factor controlling POS with a positive correlation, followed by temperature with a positive correlation. The correlations between preseason temperature and POS exhibited spatial heterogeneity, with more areas showing positive relationship than negative relationship. Compared to northern Africa, the response of POS to climatic factors is more complex in southern Africa. The three climate factors all showed significant correlations to POS for most areas. Both decreased accumulative precipitation and radiation over preseason overall delayed POS for most areas. The temperature was also significantly associated with POS changes, while had different effects on POS, i.e., POS can be resulted from lower temperature and higher temperature (depending on the location), implying the complicated interactions between POS changes and temperature in southern Africa. Our study deepens the understanding between phenology and climate change in Africa.