The current anthropogenic climate change and its global impact have increased the urgent need for reliable estimates of biomass and carbon pools for terrestrial ecosystems ( Brown, 2002 ). Tropical forests are the most important forest vegetation type with large carbon stocks that have the potential to offset global warming through the removal of atmospheric Carbon dioxide (CO₂) ( Löf et al., 2019 ). Sub-Saharan Africa has 25% (2.1 Gt per year) of the global carbon stock of the forests ( Pearson et al., 2017 ) with the largest from above-ground tree biomass (≥80%), mostly held in stems, wood and branches ( Henry et al., 2010 ). However, this carbon pool is the most vulnerable to anthropogenic activities causing deforestation for agriculture and forest degradation through selective logging ( Hairiah & Rahayu, 2007 ) to directly affect the aboveground biomass of trees ( Gibbs et al., 2007 ). In contrast to deforested areas that are used for agriculture and grazing, most buffering capacity and resilience are high in selectively logged forests that are well managed ( Asner et al., 2006 ) and may recover carbon stocks ( West et al., 2014 ). However, carbon stock from tropical forest regrowth is subject to high uncertainties ( Sierra et al., 2012 ).

Selective logging can diminish timber resources, carbon reserves and habitat quality for local fauna ( Poudyal et al., 2018 ) slowing its recovery ( Hu et al., 2020 ); but these disturbances also create environmental conditions that are favorable for the regeneration and recovery of other tree species and forest biomass ( Goulamoussene et al., 2017 ; Mokake et al., 2024 ) and thus the capture and recovery of carbon ( Herault & Piponiot, 2018 ). This recovery forms part of the reincorporation of carbon into both the soil and the forest biomass ( Houghton, 2005 ). However, the recovery of forest areas as a result of disturbances generated from selective logging has received little attention as little is known about recovery and the dynamics of their carbon stock. Though some studies have been carried out in tropical Africa especially Ghana, ( Duah-Gyamfi et al., 2014 ), most focused on the regeneration of commercial timber species as was the case with Mokake et al. (2022) in Eastern Cameroon, without taking into consideration the recovery of the whole forest biomass and carbon stock especially in the East Region of Cameroon. More importantly, uncertainty remains about the rate of biomass recovery in secondary and primary forests in the landscape ( Sierra et al. 2012 ). Therefore, it is necessary to accurately estimate the carbon stocks of primary and secondary forests in order to determine biomass accumulation at the different stages of succession after disturbances in order to rebuild the ecological functions provided by the forests.

Before using allometric equations, which is a non destructive method of determining the above ground biomass, their validity within a particular area needs to be tested ( Chave et al., 2005 ) in order to be sure of the carbon stored by the forest taking into consideration that the biomass in the tropics is affected by topography, climatic condition and woody species composition along with other factors like natural and anthropogenic disturbances like selective logging. The main source of error in the estimation of biomass and carbon stock is thus the choice of the allometric equation ( Chave et al., 2005 , 2014) which depends greatly on the number of trials carried out and the site in question ( Chave et al., 2014 ). Thus a considerable uncertainty exists in the spatial distribution of biomass. Pantropical models based on large compiled datasets contain more community-level variation in specific forests than local models ( Brown et al., 1989 ; Chave et al., 2005 ; Djomo et al., 2010 ), as site specific equations are usually fitted using a limited number of trees and species ( Kenzo et al., 2009 ), which may not be representative of species pool in other regions ( Fayolle et al., 2013 ). However significant bias in the estimates of Carbon has been reported by Henry et al. (2010) , Alvarez et al. (2012) and Lima et al. (2012) in Ghana, Columbia and Brazil, respectively using pantropical equations. Therefore, there’s a need to investigate if a substantial difference exist between the pantropical, regional and site-specific allometric equations in the estimation of tree/wood plant biomass and carbon of African tropical forests in order to avoid misinformation in the development of specific allometric models for Africa improving on the quality of biomass estimates under the UN-REDD + programme as locally developed allometric equations are thought to provide more accurate estimates of the aboveground biomass (AGB).

For countries without specific allometric equations two options exist based on the IPCC (2006) . The first option is tier 1 which uses existing generalized equations for their national forests; attained in Africa by Brown et al. (1989) and Chave et al. (2005) and Cameroon by Mokake et al. (2022, 2023) . The second option is tier 2 which explores the need and subsequent process of developing country or region-specific allometric equations with few specific allometric equations in tropical African forests in general (in Cameroon by Fayolle et al., 2013 , 2018 ) with consequences in the variation in the attainment of Tier-2 and Tier-3 (tier 3 is the site specific allometric equation attained in Cameroon by Djomo et al. (2010) approaches of the IPCC. Therefore pantropical equations may not be appropriate in reflecting specific growth patterns or conditions. It was against this backdrop that this study sought to determine the carbon stock recovery after selectively logged while assessing the validity of using pantropical multi-species, regional and site specific allometric equations to estimate the biomass of trees in the East Region of Cameroon thereby contributing to the information gap that exists in Africa in general and Cameroon in particular. It was therefore hypothesized that:

The specific objectives were to:

The objective of this study was to determine the carbon stock recovery of a tropical forest after selective logging using different allometric equations in the East Region of Cameroon. A 20% reduction in the number of recensured plants may be due to natural causes (tree fall, mortality) and anthropogenic activities (illegal logging and farming activities) in the logged and unlogged forests decreasing the number of individuals. However, the logged forests had the highest % change in the stand composition due to collateral damages and death of individuals, species and genera after logging. Our findings were consistent with previous studies that have investigated how the removal of large-size trees altered the composition and structure of the forest ( Diway et al., 2023 ; Hayward et al., 2021 ).

There were commercial species of exploitable diameter that were still present in all the forest types. It is unsurprising that there were more trees above harvestable size in the unlogged forests, but the relatively high abundance of exploitable tree species of harvestable sizes in the logged forest especially in the ≥ 50 cm suggests that trees of commercial species in the remnant cohort survived within or were recruited into the logged areas after logging. Commercial trees of ≥ 50 cm DBH size classes in logged forests could also be potential seed/pollen sources left during logging as their absence would generally reduce recruitment and regeneration of species and individuals. Also there was a high density of individuals in the < 10 cm DBH size-class in the forests of 11 YAL. With canopy openness after selective logging, seeds in the soil seed bank and nearby secondary forests became activated due to the presence of light and then eventually germinated to increase the number of species in the < 10 cm DBH size-class. This highlights the potential for regeneration in logged forest and therefore indicates the necessity of an assessment of natural regeneration in selectively logged forests as a useful tool for sustainable forest management. This also calls for silvicultural management of selectively logged sites. This result is consistent with other studies in Sabah, Malaysia ( Berry et al., 2008 ). However, a reduction in the density of individuals in unlogged forests for DBH size classes 10 - 40 cm may be due to illegal logging being carried out or the growth of remnant cohorts to higher DBH size classes. Evidence from the comparisons of PCA axis scores between logged and unlogged forest, indicated that natural regeneration will not be enough even 21YAL to control the effects of selectively logging; thus the need for silvicultural management. This is in line with the studies of Hayward et al. (2021) who indicated that the effect of selective logging can still be observed 23 - 35 YAL.

Though the diversity decreased in the unlogged forest, species evenness was highest which may be due to canopy formation that prevents seed establishment from nearby vegetation. The forests of 21 YAL, though having a canopy, had already been opened to received seeds from the nearby forest vegetation thereby increasing its diversity. This high diversity in the logged forests is different from the results obtained from other selectively logged forests that had lower diversity in Peninsular Malaysia ( Okuda et al., 2003 ) and Central Kalimantan, Indonesia ( Brearley et al., 2004 ). This difference may be due to additional disturbance in the logged forests leading to further degradation and thus decreasing the diversity. Maximum diversity is predicted with repeated local disturbances that are frequent enough to prevent competitive exclusion over an entire area, but not so frequent as to eliminate most species ( Whittaker et al., 2001 ). The unlogged forests and forests of 21 YAL were greatly similar in diversity, indicating their change towards a stable primary forest; consistent with studies carried out in the forest of 13 YAL ( Bischoff et al., 2005 ) and 5 - 20 YAL ( Verburg & van Eijk-Bos, 2003 ). The lack of difference in alpha diversity seen here suggests that the disturbance from logging was not severe enough to prevent the maintenance of a species-rich tree community.

The insignificant difference in the AGB estimate using Pan tropical, regional and site specific allometric equations is an indication firstly that mixed-species regression equations provide good estimates of the total aboveground biomass when the diameter and wood density are input variables. Djomo et al. (2010) in the Campo-Ma’an forest in Cameroon using only diameter as input variable, indicated an average error of only 7.4%. This was confirmed by Fayolle et al. (2018) in Central African forests who indicated that the diameter and wood density had a significant influence on the calculations of AGB. Secondly, this is an indication that the equation of Chave et al. (2005) is a good reference for the calculations of AGB in moist tropical forests as it takes into consideration both the wood density and the DBH ( Gourlet-Fleury et al., 2013 ) which are the variables which have a significant difference in the calculations of the AGB using allometric equations ( Brown, 2002 ). Also the Chave et al. (2005) pantropical allometric equation is attributed to a large input of dataset of 2410 trees ≥ 5 cm diameter from 27 study sites across the tropics used for the development of this equation and one of such sites is in the South eastern part of Cameroon ( Djomo et al., 2010 ; Chave et al., 2005 ; Fayolle et al., 2013 ). Site-specific allometric equations are required only in localized sites when the generalized equations may not adequately represent all forest types ( Gibbs et al., 2007 ). This may be true where forest types have specific growth patterns, and unique species compositions such as young regenerating forests, areas of shifting cultivation mosaic, areas undergoing heavy selective cutting resulting in stunted trees, and abandoned plantations.

The more than 50% of recovery rates of the ACS, may be due to the fact that during regeneration, the recuperation of forest composition, biomass and carbon is initiated, promoting the recovery of the forest areas to a state of maturity or at least to a state prior to that of the disturbance, in a few decades. This is similar to the studies of Rutishauser et al. (2015) who indicated that 7 - 21 years is required to recover their initial ACS, in the Amazon and Butarbutar et al. (2019) who indicated that the recovery time of the aboveground carbon is between 26 - 46 years even after silvicultural treatment in the tropical forest of Indonesia. Thus ours study indicates that at 21 years, the forest had already recovered more than 50% of the initial ACS.

The observed pattern showed an increase in the growth of biomass in logged compared to unlogged forests as reported in GourletFleury et al. 2013. The highest AGB in the forests of 11 YAL may be due to the fast growth of the remnant cohort of species after logging ( Fayolle et al., 2013 ). As observed by some authors ( Gourlet-Fleury et al., 2004 ; Suarez et al., 2019 ), openings increased the growth of remaining trees and the recruitment of new trees, with an increase in the proportion of fast-growing pioneers. Specifically, Suarez et al. (2019) indicated that the biomass accumulation rate of secondary forests (≤20 years) in Africa is 5.8 times that of primary forests. Gourlet-Fleury et al. (2013) in a similar study found that after selective logging, the gain in biomass proved to be high leading to fast recovery rates of the AGB. The gain however, decreased with time since the forests of 21 YAL had a lower AGB in our study; due to competition which increased as stands accumulated biomass.

The observed decreased in AGB and ACS 21YAL may be due to the forests losing large timber species after disturbance mostly which are light demanding timber species (non-pioneer timber species) which have high wood density which takes time to recover (as in the forests of 21 YAL) as opposed to the pioneer species which immediately sprout up after logging (11YAL) with low wood density. With time these pioneer species fade out giving way to non-pioneer species. In literature, it has been well established that it is impossible for a first-time-logged tropical forest to recover its timber stock within the usual duration of a felling cycle (synthesis in Putz et al., 2012 ). Rutishauser et al. (2015) found that the above-ground carbon pool recovered within 75 - 85 years after disturbance as Bourgoin et al. (2024) indicated that tropical moist forests don’t significantly recover from canopy loss, after 30 years following disturbance such as fire and selective logging. Additionally, Poorter et al. (2016) reported the annual ACS recovery for tropical forest is 1 - 3 Mg C/ha/yr in a decade after logging and 3 – 5 Mg C/ha/yr in two decades. These findings are supported by our study with a net annual recovery of 4.3 and 2.1 Mg C/ha/yr after selective logging in the forests of 11 and 21 YAL respectively. Thus, further research is required to determine the required time to get to the initial climax carbon pool in Cameroon.

A higher ACS in the unlogged forests than in the similar forests of 21 YAL may be due to the fact that in the process of forest succession, plant community composition has an important impact on biomass stocks. At the early stages of succession, fast-growing species have a higher contribution, while the contribution of slow-growing is significant at later stages of succession ( Shimamoto et al., 2014 ), due to the change in biomass accumulation patterns over time; plant composition along successional trajectories and the different succession stages that have different biomass allocation patterns under light-limited conditions. This is in line with Vieira et al. (2004) who observed that the 10 - 20 cm DBH size class which is more prominent in the logged forests contains as much as 14% of total ACS and may be highly dynamic in some Amazonian forests.

The high BCS in the forests logged 21YAL may be due to the increased number of individuals that arise after logging. The removal of timber tree species decreases the ACS and subsequently the BCS. This however increases as the forest gets back to a stable primary state. Optimal partitioning theory suggests that plants allocate more resources to the organ that acquires the most limiting resource ( Johnson & Thornley, 1987 ). Accordingly, plants would allocate more carbon to roots if the limiting resources are belowground, i.e. water and nutrients, and would allocate more carbon aboveground when the limiting resource is light or CO₂. In the forests logged 21YAL the limiting factor is water and nutrients due to a rapid increase in the number of individuals just after logging, thus an increase in the BCS. Also the destruction of the canopy encourages the growth of individuals in the <10 cm DBH size class which presumably invest more biomass below ground to take up nutrients and water for fast growth and survival.

The non-significant alteration in soil organic carbon encountered in this study revealed that selective logging doesn’t extensively change the soil entirely but acts locally in the area of logging. However, the amount of most nutrients increased in the logged sites. This could be related to high release of nutrients via the decomposition processes thereby reflecting the greater input from organic matter from the logging process. Additionally, the decrease in soil carbon stock with soil depth might be due to the decomposition of litter at the top soil, which contains more nutrients at the top soil than at the lower soil depths. The effect of selective logging 21 YAL indicates that the impact is felt after a very long time. The duration of nutrients released from decomposed stems, twigs and leaves of trees damaged by logging may thus explain the substantial but long-duration release of nutrients on these nutrient-poor soils.

In a nutshell, this study showed that carbon and biomass recovery after selective logging in a tropical forest like the Eastern forest in Cameroon do not accumulate at the same rate as the unlogged forests due to the change in the species composition after the removal of key commercial species. Thus, further research over a larger surface area and on a longer time frame is required for more conclusive information on the Carbon stock recovery after selective logging; as on a regional scale, alternative management methods of exploitation may coexist to maximize a compromise between timber production and preservation of the environment.

2005 inventory results were provided by CTFS through their Regional Coordinator Professor Chuyong George. We would like to thank Mr Bekolo Bekolo and Mr Njombe Ewusi of the National Forestry Agency (ANAFOR) for introducing us to the timber company through cosupervision, Mr Decolvenaere of SFIL-GVI for allowing us use his forest concessions. We also thank Mr Bounoungou Zibi for field supervision; Mr Blaise Jumbam, Njoh Jean and Mr Nakoe Roger for field assistance and the staff of the forest company SFIL-GVI Ndeng.

Seraphine E. Mokake: In charge of the research design, supervision of field work and data collection, guided analysis and interpretation, and wrote manuscript. Chuyong George was the CTFS representative and project coordinator who suggested the topic and the research design, contributed in perfecting the methodology and analyses and reviewed the manuscript. Egbe E. Andrew contributed in research concept and design and reviewed the manuscript. Lyonga M. Ngoh: analysed data collected, contributed in the research concept and design and methodology. Also contributed in data interpretation and review of manuscript.

Funding for the fieldwork was supported by the International Tropical Timber Organisation (ITTO) grant, the equipment grant was provided by IDEA WILD, the Carnegie Corporation of New York through the Future Africa Research Leadership (FAR-LeaF) grant of the University of Pretoria also supported the fieldwork and data analyses, the mobility grant and research modernization allowance of the Ministry of Higher Education (MINESUP) Cameroon and finally, the Fako America Scholarship.

Data are available within the article and/or its supplementary materials.

Not applicable.

The authors all accept the publication of this work.