Author(s) :

Cédrick Belmich Nkoua Ondon, Jessy Houston Tchoffo Tsoala, Joseph Otshinga Abedi, Michel Omeno Opelele, Papy-Claude Boliale Bolaluembe, Raphael Muamba Tshimanga, Aimé Patrick Lola Missamba, Roger Vumilia Kizungu, Bonaventure Nyami LeLe, Eric Landui Lutete.

Volume/Issue :

Abstract :

The Mvouti district, located in the Kouilou department (Republic of the Congo), harbors a critical portion of the Mayombe forest massif, which encompasses the Dimonika Biosphere Reserve (136,000 ha, UNESCO-MAB 1988). This reserve was selected as it represents a biodiversity hotspot under severe anthropogenic pressures from artisanal gold mining and logging. Using six Landsat image composites (2000–2025, 30 m spatial resolution) classified into a five-class land-use/land-cover nomenclature via the Random Forest algorithm on Google Earth Engine, this article quantifies the spatio-temporal dynamics of forest cover, analyzes the fragmentation of the massif, and simulates three prospective trajectories through 2050 using CA-Markov modeling. With an overall classification accuracy of 90.6% (κ mean = 0.883), dense forest lost 32,106 ha (9.1% of the initial cover) between 2000 and 2025. The mean deforestation rate is estimated at 0.37%/year, three to four times higher than the Congolese national average. Disaggregation by land tenure status reveals contrasting dynamics: areas outside logging concessions peaked at 2.27%/year between 2015 and 2020, while managed concessions (FMUs) within the reserve buffer zone maintained significantly lower rates. Fragmentation intensified sharply (patch count rose from 442 to 6,189; MESH: −21.5%). Under the integrated governance scenario (SC3, consistent with Law No. 33-2020 and the Nationally Determined Contribution [NDC]), loss by 2050 would be limited to 14,255 ha, versus 45,634 ha under a governance failure scenario (SC2) a divergence of 31,379 ha, equivalent to approximately 9.8% of the 2025 forest cover. These results provide a directly applicable quantitative baseline for revising the Kouilou Land Use Plan and updating the Congolese NDC reports. The main methodological limitations are the 30 m resolution of the Landsat imagery, which constrains detection of sub-pixel degradation, and the stationarity assumption of the Markov transition probabilities, which does not explicitly incorporate future climate-driven shifts in land-use pressure.

No. of Downloads :

0