Quantifying fragile carbon balance by multiple remote sensing based vegetation optical depth datasets

Tropical forests in the Americas play a pivotal role in the global carbon cycle, yet their net contribution to carbon balance remains uncertain due to competing processes of disturbance and recovery. While widespread deforestation, degradation, and fire drive persistent carbon losses, secondary and old-growth forests are often assumed to function as compensatory sinks. Here, we aim to provide a process-explicit and regionally consistent assessment of aboveground carbon (AGC) dynamics by disentangling carbon losses and gains across tropical America. However, the magnitude and persistence of these carbon losses and gains have not been systematically quantified at wide scale. To address this key issue, we integrate satellite-based AGC estimates derived from L-band vegetation optical depth (L-VOD), high-resolution (0.001°) disturbance mapping from AGC dataset and the Tropical Moist Forest (TMF) dataset, along with simulations from 9 dynamic global vegetation model (DGVMs) to provide a comprehensive assessment of AGC dynamics across tropical America during 2011–2020. L-VOD observations revealed that tropical America stored a mean AGC stock of 87.3 ± 18.8 PgC, with a modest but positive net annual increase of 159.2 ± 35.3 TgC yr⁻ 1 during 2011–2020. This net gain, however, masks substantial internal carbon fluxes driven by opposing disturbance and recovery processes. Disturbances collectively caused 475.6 TgC yr⁻ 1 of AGC loss, 301.7 TgC yr⁻ 1 of which originated from forests, indicating that forest ecosystems remain the primary source of carbon emissions in the region. Among disturbance types, fires dominated carbon losses (252.5 ± 24.1 TgC yr⁻ 1 ), followed by deforestation (115.8 ± 29.4 TgC yr⁻ 1 ), edge degradation (79.1 ± 30.4 TgC yr⁻ 1 ), and other forms of degradation (28.2 ± 7.2 TgC yr⁻ 1 ), highlighting the combined influence of climatic extremes and anthropogenic pressures on forest carbon stability. In contrast, carbon gains were largely restricted to old-growth forests. Secondary regrowth contributed only 20.4 ± 6.2 TgC yr⁻ 1 , suggesting limited post-disturbance recovery capacity, whereas simulations from nine DGVMs under the TRENDY framework estimated that intact old-growth forests sequestered 350.9 ± 66.6 TgC yr⁻ 1 − over 70% of the total regional sink. This dominance of old-growth forests indicates that the current carbon surplus mainly depends on the sustained uptake by a shrinking pool of mature forests rather than on widespread ecosystem recovery. Overall, the proposed remote sensing approach revealed a fragile carbon balance: large-scale disturbance-driven losses are only partially offset by unevenly distributed gains. This study revealed this regional sink was unlikely to persist under intensifying climatic and anthropogenic pressures.