Climate extremes and persistent deforestation pose significant threats to Africa's vegetation carbon stocks. However, the patterns of aboveground carbon (AGC) loss, recovery, and their driving factors in Africa remain poorly understood. Here, we utilize low-frequency microwave satellite data to analyze AGC dynamics across Africa during 2010-2020. Results indicate a small net AGC increase of +0.16 +/- 0.03 PgC yr-1 during the study period, composed of gross losses of -1.56 +/- 0.26 PgC yr- 1 offset by gross gains of +1.72 +/- 0.29 PgC yr- 1. The total loss in forested areas amount to -0.50 +/- 0.07 PgC yr-1, of which degradation accounting for twice as much loss as deforestation. In non-forested areas, the total AGC loss was -1. 06 +/- 0.21 PgC yr-1, primarily driven by wildfires (-0.78 PgC yr-1). Following the 2015-2016 El Nino event, 66 % of affected regions exhibited AGC recovery ratios exceeding 100 % during 2015-2020, predominantly in non-forest vegetation, suggesting a higher recover ratio for non-forest vegetation. In contrast, the remaining 34 % of regions did not fully recover, with an average recovery rate of 58 %, predominantly concentrated in forested areas. A machine learning analysis based on random forest suggests that recovery ratios are primarily influenced by vapor pressure deficit (VPD), followed by precipitation and human footprint. Our study provides a comprehensive understanding of the dynamics of African AGC by distinguishing the loss into forest and non-forest vegetation, and also highlights the key drivers of AGC recovery after disturbances. These findings offer valuable insights for ecological conservation, climate adaptation, and global carbon budget assessments.