Dear Sagang and co-authors,

In my opinion, your manuscript tackles an interesting topic about how different environmental factors, including soils and forest fires, mediate forest expansion in the humid savannas of Central Africa. I think the manuscript is ready for publication upon minor revisions.

My only comment was related with Figure 2. The non-linear adjustment in Figure 2d doesn’t seem to fit the raw data well, as there are no visible points above 12 Mg ha⁻¹. Please revise these adjustments by using less flexible splines, such as quadratic adjustments or polynomials with maximum three degrees of freedom, or even linear regressions, as the trends point to opposite patterns compared to those plotted, i.e., higher grass biomass with higher fire frequency. Additionally, it is unclear why not all the panels in Figure 2 display these linear or non-linear adjustments that could justify the results stated in lines 207 to 210.

For the results shown in panels d and e in Figure 2, additional statistical analysis accounting for the large number of zeros (e.g., Hurdle models) is recommended to provide more robust results.

Furthermore, in Figure 2, why is FFI included in panel f along with the regression line between FFI and the gradient of AGBGRASS? Consider including that adjustment in panel d, and in panel f, include the regression line between AGBwood and AGBgrass.

Other minor comments:
Abstract:
“Northern humid hemisphere” please use another word instead of hemisphere at it can be misleading.
Include some punctuation in the 5th line, like separating by commas “including fire frequency derived from remote sensing”

L 212. Relationships, instead of interactions.

L 138. “the allometric equation A specific to savanna” & L 140. “allometric equation 7”, is this a typo? otherwise include the equation here as the reader is not going to check the reference to figure out equations A and 7, please rephrase as “the individual using the allometric equation specific to savanna found in reference [80]” & “the pantropical allometric equation found in reference [81].

L255. Given in parenthesis not in brackets