Assessing Forest Degradation in the Congo Basin: The Need to Broaden the Focus from Logging to Small-Scale Agriculture (A Systematic Review)

Abstract

While the methods for monitoring deforestation are relatively well established, there is still no compromise on those for forest degradation. We propose here a systematic review on studies about forest degradation in the Congo Basin. Our analysis focused on seven key anthropogenic causes of forest degradation. Shifting agriculture emerged as the most significant driver, accounting for 61% ± 28.58% (mean ± SD) of canopy opening, 73.16% ± 16.88% aboveground carbon loss, and 30.37% ± 30.67% of tree species diversity loss over a 5–60-year period. Our analysis reveals a significant disconnect. Only 29% of the reviewed studies address this driver, while over 64% focus primarily on the consequences of industrial timber harvesting. Despite its comparatively minor contribution to degradation, with effects range from only 8.98% ± 13.63% of canopy opening, 14.79% ± 22.21 aboveground carbon loss, and 4.27 ± 21.07 tree species diversity loss over 1–20 years. Indeed, most of the methods focus on detecting changes in canopy structure associated with forest logging over a short period (0–5 years). These illustrate the need for a shift in focus in scientific research towards innovative methods, which can be developed over time, to monitor the various impacts of all causes of forest degradation.

1. Introduction

African tropical forests are one of the world’s largest forest carbon sink and reservoir of biodiversity [1,2,3] and are currently threatened by climate change and anthropogenic pressures, both contributing to deforestation and forest degradation. Deforestation and forest degradation are distinct processes with different causes and impacts on forest ecosystems [4,5,6]. However, degradation is a relatively recent subject of study, introduced with the REDD+ (Reducing Emissions from Deforestation and Forest Degradation) initiative, primarily to address emissions from selective logging [7,8,9]. Between 2005 and 2010, emissions from tropical forest degradation accounted for around 25% of total deforestation and degradation emissions [10]. In the Brazilian Amazon, it is estimated that forest degradation contributed three times more to gross aboveground biomass loss than deforestation, accounting for 73% of 670 million tons of net carbon lost between 2010 and 2019 [11]. In the Congo Basin, the annual rate of deforestation is increasing, rising from 3.3% between 1990 and 2000 to 3.9% between 2010 and 2020 [12]. Between 2000 and 2005, the rate of forest degradation in the Congo Basin was estimated at 0.09% [13].

The governments of the Congo Basin countries have developed national strategies for the implementation of the REDD+ mechanism, encouraging the promotion of conservation, sustainable forest management, and increased forest carbon stocks [14]. However, there is still a lack of consensus on the definition and approaches to monitoring forest degradation [5,8,15], which leads to a lack of coordinated and effective action.

Recently, European Union member states agreed on a common definition of forest degradation as a structural change in forest cover taking the form of conversion of primary forest to plantation forest or other wooded land [16]. On the other hand, the governments of the Congo Basin countries propose a variety of definitions, which complicates the assessment and monitoring of forest degradation on a regional scale. The Republic of Congo and Gabon define forest degradation as a process that leads to a reduction in biomass without a reduction in forest cover [17,18], whereas, for the Democratic Republic of Congo (DRC) forest degradation refers to the conversion of primary forests into secondary forests [19].

Moreover, the current methods for monitoring forest degradation do not cover all causes or indicators for assessing the forest ecosystems attributes [5,6,20]. Indicators of forest degradation, such as changes in structure, biodiversity, storage, and forest production, can be assessed through three types of approaches. One can rely solely on methods based on field data analysis, primarily relying on forest inventories, or solely use remote sensing tools based on satellite imagery and aerial photographs, or combine both approaches [20,21]. Moreover, assessing these indicators requires considering the temporal dimension to determine the time needed for forest characteristics to recover after a disturbance, evaluate its condition, and measure its post-disturbance resilience [5,22].

To this end, by focusing on studies referring to forest degradation, this synthesis aims to evaluate the effectiveness of current methods for characterizing degraded forests in the Congo Basin. To achieve this, we propose to discuss the following three hypothesis: (1) Forest degradation encompasses various types of anthropogenic causes that are crucial to distinguish in order to effectively characterize and monitor the degradation of tropical forest ecosystems in the Congo Basin. (2) Monitoring methods for forest degradation in the Congo Basin are not sufficiently applied to track the most important causes of forest degradation by assessing various forest properties over time. (3) In studies that take the time factor into account, the most widely measured degradation indicators are assessed over relatively short periods, which makes it impossible to determine the time required for recovery of the respective indicators measured as a result of specific anthropogenic causes of forest degradation.

2. Materials and Methods

2.1. Bibliographical Research and Articles Selection

We performed a literature search according to the PRISMA methodology [23]. We selected articles studying tropical moist forest degradation for the period 2000 to 2024 using the following inclusion criteria: (1) Forest degradation is caused by human disturbances that impact the state of the main forest attributes [5,6,20]. For example, studies focusing on soil degradation and analyzing the impact of agriculture on soil fertility to maintain field yields and studies analyzing the impact of hunting or logging on the depletion of animal populations without comparative analysis with an undisturbed forest or with a relationship to the loss of potential key functions such as seed dispersal, pollination, or pest reduction were not selected. (2) The study is not a literature review, it is localized in the dense tropical moist forests of the Congo Basin and in one of the following six Congo Basin countries: Cameroon, Gabon, Equatorial Guinea, the Central African Republic, the Republic of Congo, and the Democratic Republic of Congo (Figure 1). (3) The sources and type of dataset analyzed in the study are clearly described.

The first phase of the bibliographic research involved extracting studies focused on the Congo Basin region, cited in previous literature reviews at both the global and regional scales [5,6,21,24,25,26,27,28,29]. The analysis of these studies was important to identify keywords associated with the main anthropogenic drivers of forest degradation (

Table 1. Keywords associated with each of the anthropogenic causes of forest degradation identified in previews literature reviews.

Reference	Anthropogenic Drivers	Keywords
Gao et al. [6]	Forest logging	“logging” OR “forest logging” “timber logging” OR “selective logging”
Murdiyarso et al. [24] and Gao et al. [6]	Agriculture	agricultural OR cultivation OR clearing
Gao et al. [6]	Wood energy	“wood energy” OR “fuelwood” OR “fuelwood charcoal” OR “charcoal”
Gao et al. [6]	Anthropogenic fires	fire OR fires OR burn
Gao et al. [6]	Livestock grazing in forest	livestock OR grazing OR “forest gazing”
Murdiyarso et al. [24]	Non-timber forest products	“non-timber forests products” OR NTFPs OR “non-wood forest products”
Abernethy et al. [25]	Hunting	hunting OR overhunting OR bushmeat

).

Next, in December 2024, for each anthropogenic driver, we used the Scopus and Google Scholar^® online databases to expand our search. The query for bibliographic search based on keywords in the article title, keywords, and abstract was structured as follows:

(keywords related to specific anthropogenic activities) AND (keywords related to forest degradation processes: “forest degradation” OR degraded OR degradation OR disturbed OR disturbance OR REDD OR sustainability OR “canopy gap”) AND (keywords related to dense rainforests and the Congo Basin region: rainforest OR “rain forest” OR “tropical moist forest” OR “tropical humid forest” OR “central Africa” OR “Congo Basin” OR Congo OR Cameroon OR Gabon OR “Equatorial Guinea”)

In Google Scholar^®, the most relevant articles were selected from the first 10–15 pages of results.

2.2. Articles Analysis and Data Extraction

Each selected publication was reviewed to extract data on the anthropogenic causes of forest degradation, the indicators linked to specific forest attributes (FA) used to assess them, the different approaches used to estimate these indicators, and, when available, the monitoring time scale and the evaluation of relative importance of different causes of forest degradation due to recovered rates of specific FA.

Anthropogenic causes were initially classified according to defined categories and sub-categories. Subsequently, we undertook a detailed description of the forest degradation process for the studied causes (Table 1).

The indicators were first classified according to criteria or types of forest attributes related to structure, biodiversity, carbon storage, production, and protective functions. Then, based on their prevalence in the literature—estimated using the citation index (If %) and the relative citation index (Ifr %), respectively, derived from Equations (1) and (2) adapted from [30]—we classified the indicators used to assess the impact of forest degradation according to the three measurement approaches employed to quantify them: field data, remote sensing, or a combination of both (mixed approach). The If (%) index ranges from 0 (when the indicator is never cited) to 100 (when it is the only one estimated using a given approach), while the Ifri (%) index ranges from 0 (when the indicator is never cited) to 100 (when it is the most frequently measured by different approaches).

$$I{f}_{iz}=100\times {\displaystyle \frac{{\mathrm{f}}_{\mathrm{i}\mathrm{z}}}{{\mathrm{n}}_{\mathrm{c}}}}$$

(1)

$$I{f}_{ri}={\displaystyle \frac{{{\sum }_{z}f}_{\mathrm{i}\mathrm{z}}}{{{({\sum }_{z}f}_{\mathrm{i}\mathrm{z}})}_{max}}}$$

(2)

where i is the indicator for which the index is calculated, z is the type of approach used to measure indicator i, fiz is the citation frequency of indicator i measured by the z approach, and nc is the sum of the citation frequencies of all indicators measured by the z approach.

We present the temporal scale of monitoring of frequently measured indicators, with a minimum of 10% Ifr, evaluated based on the maximum number of years of monitoring following experimental methods described in the literature. These methods use diachronic monitoring or chronosequences to track forest succession or the number of years since logging.

We identified the most significant anthropogenic factors contributing to forest degradation by analyzing the variation in three most measured indicators related to aboveground carbon (AGC), forest canopy gaps, or opening area tree species diversity. This indicator was expressed as canopy gaps area fraction (%), which represents the ratio between gaps area and the total forest area of the specific case study site. FA loss ($∆FA$) and recovery rates of forest attributes (RFAs) for AGC and tree species richness were evaluated following Equation (3) and Equation (4), respectively:

$$∆FA=-\left({\displaystyle \frac{{FA}_{0{-FA}_{i,n}}}{{FA}_0}}\right)\ast 100$$

(3)

$$\mathrm{RFA}=100+∆FA$$

(4)

where $∆FA$ is the proportion of forest attribute loss (−) or gained (+) post disturbance, FA₀ is the value of interested FA before disturbance or equivalent to nearly undisturbed primary forest (when it is not precise, an approximative value of 200 Mg/ha of AGC was used as the mean regional reference for central Africa primary forest [31]), and FA_i,n is the value of interested FA after disturbance of type i in time n since disturbance. The biomass was considered to contain 50% of carbon [32].

3. Results and Discussion

3.1. Selected Articles

The research expression encoded in Scopus in December 2024 for each selected anthropogenic forest degradation drivers (Table 1) matched with 2.990 scientific articles (Figure 2). Those articles were completed with 57 relevant articles found in Google Scholar. In total, 442 duplicates articles were removed by human or manually abstract and title screening lead to the exclusion of 2.340 articles. Finally, 192 articles were fully read, and 116 of them were included in the synthesis.

Although the period selected for the articles is 2000–2024, the articles included in this synthesis have been published since 2003 (Figure 3). This shows that it is only more recently that researchers have begun to address the issue of forest degradation [7,8,9]. Apart from studies carried out on a regional scale, i.e., those involving at least two countries, 36 articles, or 30% of the articles analyzed, come from Cameroon (Figure 3). Only 15 to 16 articles, or about 14%, come from Gabon and the Democratic Republic of Congo, respectively (Figure 3), even though the latter accounts for at least 60% of the forest area in the Congo Basin region [33]. Finally, we note that, since 2016, there have been almost no studies from the Central African Republic. Generally, the Congo Basin region attracts little interest from researchers compared with the tropical regions of America and Asia [34].

3.2. Towards a Typology of Anthropogenic Causes of Forest Degradation Relevant to Monitoring Forest Degradation in the Congo Basin

Our analysis focused on seven anthropogenic causes of forest degradation identified in the preview literature reviews (Table 1), in which we identified nine specific anthropogenic causes of degradation in the dense tropical forests of the Congo Basin, which we classified into six different categories: selective wood extraction, agriculture, livestock grazing in the forest, anthropogenic fires, hunting, and harvesting of non-timber forest products (Table 2).

Examining data from REDD+ programs in developing countries, refs. [4,35] identified four main causes of forest degradation: logging for timber, logging for wood energy, uncontrolled bush fires, and cattle grazing in forests. In a more recent review of the literature on the contribution of remote sensing to monitoring the causes of forest degradation, ref. [6] also considered other causes of forest degradation, including shifting cultivation and hunting. Even more recently, ref. [36] distinguished the causes of degradation based on the industrial or artisanal nature of the human activities carried out on the forest, in particular: industrial and artisanal agriculture, industrial and artisanal logging, and industrial and artisanal mining. Regarding selective timber extraction, ref. [37] identified three main types: industrial logging, artisanal logging, and the use of wood as a source of cooking energy. However, as described below, apart from industrial logging, there are still few studies on the process of forest degradation by other anthropogenic causes most relevant to the Congo Basin.

3.2.1. Selective Industrial Timber Harvesting

The expansion of industrial logging concessions has led to the emergence of the concept of forest degradation or degraded forests in international climate change policies [8]. Although our analysis reveals that this activity is the most studied cause of forest degradation internationally and in the Congo Basin (Table 2), it is not necessarily the main driver in the region, as several studies indicate [10,36,38]. In the context of sustainable forest management, over the past two decades, initiatives aimed at promoting reduced impact logging practices have seen significant development in Central Africa. The forest legislation of the Congo Basin countries has made the development of management plans for industrial forest concessions mandatory [39]. In addition, in response to consumer concerns about sustainable forestry practices, certification schemes such as the Forest Stewardship Council (FSC) have also been adopted by some forestry companies [40]. The compliance with these standards by forestry companies explains the generally low rate of environmental damage caused by industrial logging on forest structure and carbon storage [38,41,42]. Between 2018 and 2020, in the Congo Basin region, the annual rate of degradation by industrial logging appears to be stabilizing at around 5% of the total number of clearings detected [41]. In the DRC, disturbances caused by industrial logging represent less than 1% of the total number of clearings detected [36]. Furthermore, intensive logging can significantly reduce biodiversity. In Asian tropical forests, the species richness of mammals and amphibians is halved when the logging intensity reaches removal rates of 38 m³/ha and 63 m³/ha of timber, respectively [43]. In Central Africa, high-intensity logging alters the regeneration of liana sapling layers, forbs, and grasses species [44]. However, the logging intensity in the Congo Basin is generally considered low, ranging from 1 to 4 trees/ha [45] compared to 13 to 18 trees/ha in Southeast Asia and South America [46]. Despite this lower intensity, logging concessions in Central African forests can play a complementary role in conservation by providing habitat for many of the region’s threatened animal species and maintaining key ecosystem services [42,47].

Table 2. Typology of studied anthropogenic causes of forest degradation studied in the Congo Basin in the 116 total reviewed articles.

Categories	Sub-Categories	Other Sub-Categories	Articles (Number)	Articles (%)	Ref.
Selective wood extraction	Timber	Industrial forest logging	74	63.8%	[10,26,36,38,41,42,44,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114]
		Artisanal forest logging	10	8.7%	[41,84,111,114,115,116,117,118,119,120]
		No specified	5	4.3%	[121,122,123,124,125]
	Wood energy	Charcoal production and fuel wood production	6	5.2%	[10,115,119,126,127,128]
Agriculture	Shifting cultivation with slash-and-burn or small-scale agriculture		33	28.7%	[36,41,48,52,80,84,88,90,92,99,115,117,118,119,120,123,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144]
	cash crops or agroforests		9	7.8%	[36,41,129,130,133,134,142,145,146]
	No specified		1	0.9%	[147]
Anthropogenic fires			14	12.2%	[10,86,90,122,124,125,128,134,140,147,148,149,150,151]
Hunting			9	7.8%	[58,65,71,94,99,100,122,152,153]
Collection of Non-Timber Forest Products (NTFPs)			8	7.0%	[92,100,119,122,154,155,156,157]
Livestock grazing in forest			5	4.3%	[100,122,125,128,147]

Table 2. Typology of studied anthropogenic causes of forest degradation studied in the Congo Basin in the 116 total reviewed articles.

Categories	Sub-Categories	Other Sub-Categories	Articles (Number)	Articles (%)	Ref.
Selective wood extraction	Timber	Industrial forest logging	74	63.8%	[10,26,36,38,41,42,44,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114]
		Artisanal forest logging	10	8.7%	[41,84,111,114,115,116,117,118,119,120]
		No specified	5	4.3%	[121,122,123,124,125]
	Wood energy	Charcoal production and fuel wood production	6	5.2%	[10,115,119,126,127,128]
Agriculture	Shifting cultivation with slash-and-burn or small-scale agriculture		33	28.7%	[36,41,48,52,80,84,88,90,92,99,115,117,118,119,120,123,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144]
	cash crops or agroforests		9	7.8%	[36,41,129,130,133,134,142,145,146]
	No specified		1	0.9%	[147]
Anthropogenic fires			14	12.2%	[10,86,90,122,124,125,128,134,140,147,148,149,150,151]
Hunting			9	7.8%	[58,65,71,94,99,100,122,152,153]
Collection of Non-Timber Forest Products (NTFPs)			8	7.0%	[92,100,119,122,154,155,156,157]
Livestock grazing in forest			5	4.3%	[100,122,125,128,147]

3.2.2. Selective Artisanal Forest Logging

The artisanal forest logging sector in the Congo Basin region is often described as an informal and illegal activity [158,159]. However, artisanal logging remains one of the least documented sectors regarding its ecological impacts in the Congo Basin region [158]. The available data show that studies on forest degradation caused by this activity represent approximately 9% of research conducted in the Congo Basin region (Table 2). However, in countries of the Congo Basin region, artisanal timber production far exceeds that of industrial logging. In the Democratic Republic of Congo (DRC), for example, this activity is a significant supply sector for both local and international markets, with an estimated annual production of 3.4 million cubic meters of wood, which is ten times more than industrial forest production [160]. Between 2018 and 2020, the annual rate of forest degradation caused by artisanal logging in the Congo Basin region increased, rising from approximately 5–10% of the total number of detected clearings. Moreover, 18% of these clearings were observed in intact forests [41]. The future study of emissions related to deforestation and forest degradation should focus on artisanal logging.

3.2.3. Wood Charcoal Production or Fuel Wood Production

Wood energy remains the main source of cooking energy for households in sub-Saharan African countries, filling the energy gap [161]. In the DRC, wood energy production accounts for 95% of total wood production [162]. This wood often comes from natural forests or agricultural land, and its demand continues to grow, particularly in urban areas where charcoal is widely used [163]. Although wood energy is considered a sustainable energy source within sustainable wood resource management systems [163], less efficient carbonization and cooking equipment [164], quality of governance characterized by corruption, low institutional capacity, and high rural population density [126,127] contribute to the increase in wood energy consumption. However, for the Central African region, there is still a mismatch between the high quantity reported on wood energy consumption and its less significant impact on forest degradation, most notably in the DRC [69,126]. The current methods therefore need to be adapted to establish this link through local-scale studies in the main wood energy supply basins for large urban agglomerations. However, research into forest degradation caused by this activity represents around 5% of the research carried out in the Congo Basin region (Table 2).

3.2.4. Shifting Cultivation with Slash-And-Burn

Shifting cultivation with slash-and-burn, widespread in sub-Saharan Africa, is a most important historical agriculture model for the food subsistence of rural households. This practice involves partially clearing sections of forest for short-term crops, followed by regeneration and succession of mature secondary forests [165]. It accounts for approximately 28% of studies on forest degradation in the Congo Basin (Table 2). In areas with high population density, this type of agriculture is undergoing significant changes, notably the reduction of fallow periods. This compromises the long-term productivity of forest soils and the capacity for forest regeneration, thereby increasing pressure on intact forests [166]. It is often cited as both a cause of forest degradation and deforestation. However, deforestation only occurs if the land does not regenerate into forest during the fallow period, meaning that the use of agricultural land is not necessarily a temporary use of the forest [6,124]. For example, fallows that reach at least 5 m in height after a reference year and are subsequently cleared are not considered a form of deforestation [86]. Visually, in the Congo Basin region, shifting cultivation takes the form of rural complexes, where small active fields, fallows, and secondary forests are intermingled with dwellings [41,132]. In the DRC, it accounts for 76% of the total area of rural complexes [132].

To date, in the Congo Basin region, slash-and-burn agriculture is the main cause of both deforestation and forest degradation [41,86]. It leads to the clearing of areas with a median size estimated at 1.8 ha [86]. Between 2016 and 2020, the annual rate of forest degradation corresponding to small-scale agriculture in the Congo Basin remains high, at around 70% of the total number of clearings detected each year, of which 8% are reported in intact forests [132].

3.2.5. Anthropogenic Fires

The use of fire in agriculture and grazing management has impacts on increasing carbon emissions from forest edges [90]. Carbon emissions are due to both the direct impact of fire intrusion into forests and the indirect impact of changes in local atmospheric circulations that increase forest canopy dryness [151]. Anthropogenic fires can penetrate up to 2.4 km from forest edges [167]. In 2010, fires occurred in 52% of forest edges in sub-Saharan Africa and increased the carbon deficit by 5.5 Mg C/ha [151]. Assessing the impact of fire on forest degradation accounts for around 12% of studies carried out in the Congo Basin (Table 2).

3.2.6. The Collection of Non-Timber Forest Products (NTFPs)

Non-wood plant resources produced by tropical forests can be grouped into various categories: fruits and seeds; plant exudates (latex, gums, and resins); and vegetative structures (stems, leaves, roots, bark, and buds) [155] or caterpillar host trees. The sporadic collection of these products may have little effect on the long-term stability of the exploited tree populations [155]. However, due to the high demand in domestic and international markets, the intensive annual harvest of a high-value NTFPs, coupled with unsustainable harvesting techniques, can gradually eliminate host species or providers of high-value NTFPs from the forest [168]. In many villages in Cameroon, long distances must be traveled to collect NTFPs that were once readily available near settlements, indicating the degradation of forest ecosystems that supply NTFPs, contributing to improving livelihoods and diversifying the income sources of local populations [156]. Moreover, the collection of the majority of NTFPs often occurs in the same forest area, competing with other anthropogenic activities, including land clearing for agriculture, establishment of palm plantations, and timber logging, which negatively impact the availability of NTFPs in different types of forests [119,155,156]. However, our results show that studies on forest degradation due to NTFP collection in the Congo Basin represent only 7% of conducted studies. Future studies on monitoring forest degradation through NTFP harvesting should characterize the high socio-economic value of the NTFP exploitation system for local populations by adjusting the rate of plant regeneration to the rate of harvesting [155].

3.2.7. Hunting

In tropical regions, both rural and urban populations are heavily dependent on animal protein from unsustainable hunting, which can lead to the transformation of forest ecosystems into what [168] describes as “empty forests”. Annual bushmeat consumption in the Congo Basin exceeds 4.5 million metric tons [169]. Yet, preserving an adequate diversity of specialized animal species is important for maintaining the ecological balance of forest ecosystems. Fruit predatory herbivores play an essential role in the dispersal and regeneration of tropical forests’ many tree species and carbon storage [170]. However, defaunation can compromise this ecological function [170]. In Central Africa, the low abundance or density of large mammals is an indicator of forest degradation, especially defaunation affecting primates and ungulates [171]. An experiment conducted in the Republic of Congo showed that hunting-induced defaunation reduced the average dispersal distances of mammal-dispersed tree species by 22%. Furthermore, forests subjected to hunting had significantly lower aboveground biomass than industrial timber concessions [65]. However, our results demonstrate that studies on forest degradation related to hunting represent only about 8% of the studies reviewed (Table 2).

3.3. Approaches for Characterizing the Degradation of Dense Tropical Forests in the Congo Basin

3.3.1. Forest Degradation Indicators and Measurement Approaches

Our analysis identified 23 indicators used to characterize forest degradation in the Congo Basin. Field data, followed by remote sensing tools, remain the most used approaches for monitoring forest degradation, assessing, respectively, 20 and 10 of the 23 identified indicators. Field data have been particularly effective in evaluating indicators related to forest carbon storage capacity and biodiversity status, such as variations in aboveground biomass and changes in tree species richness. Remote sensing, on the other hand, has been widely used to assess structural changes in forests, with canopy gaps, clearings, and changes in forest cover rate being the most frequently evaluated indicators. The integration of field data and remote sensing has proven especially valuable for assessing variations in aboveground biomass on large scales (

Table 3. Classification of measured indicators and type of forest attributes for assessing forest degradation in the Congo Basin by type of approach used in analyzed studies to measure them according to their citation index (If %) and index of relative frequencies (Ifr%) (mixed: combination of remote sensing and field data).

Indicators	Type of Forest Attributes	Mixed (If %)	Remote Sensing (If %)	Field Data (If %)	Sum of If % for All Approaches	Ifr (%)
Aboveground or below ground biomass or carbon	Carbon storage	58	5	13	76	100
Canopy gaps	Structure	8	34	4	46	61
Forest cover change	Structure	0	23	0	23	30
Floristic species: trees, seedlings, lianas, shrub	Biodiversity	0	0	20	20	27
Forest types	Biodiversity	0	14	4	18	23
Logging intensity	Production	8	0	9	18	23
CO2 or other greenhouse gas emissions	Carbon storage	8	5	3	16	21
Tree density variation	Structure	0	0	12	12	16
Edge effects	Structure	8	3	1	12	16
Gross primary productivity	Carbon storage	8	0	0	8	11
Intact forest landscape	Biodiversity	0	8	0	8	10
Canopy and tree Height variation	Structure	0	6	1	7	10
Basal area variation	Structure	0	0	7	7	9
Defaunation	Biodiversity	0	2	5	7	9
Regeneration of logged trees	Biodiversity	0	0	5	5	7
Logging damage	Carbon storage	0	2	2	4	5
Soil fertility	Protective functions	0	0	4	4	5
Mean and growth rate of tree diameter	Structure	0	0	4	4	5
Litter or necromass	Carbon storage	0	0	2	2	2
Wood density	Carbon storage	0	0	2	2	2
Timber stocks	Production	0	0	2	2	3
Tree volume variation	Production	0	0	1	1	1
Specific leaf area	Carbon storage	0	0	1	1	1

).

While field data provide detailed information on a local scale, remote sensing offers the advantage of rapid data collection over large areas. This capability has significantly contributed to the proliferation of scientific literature on forest degradation, particularly through the field of landscape ecology. Remote sensing is frequently used to assess landscape fragmentation and the extent of intact or undisturbed forests by detecting canopy openings and evaluating the forest cover rates [21]. Fragmented forests are often considered degraded due to increased edge effects, reduced connectivity, and decreased forest cover, which can lead to biodiversity loss and impact numerous ecological processes [172]. Yet, detailed studies on the impact of fragmentation on the various biodiversity forest attributes are currently lacking. An intact forest landscape is free from any signs of human activity or fragmentation, capable of maintaining all local biodiversity, including viable populations of wide-ranging species [86]. Such landscapes must have a minimum area of 500 km², a minimum width of 10 km, and a corridor with a minimum width of 2 km [81].

Mapping canopy openings created by road networks for selective logging has enabled the quantification of forests degraded by the expansion of industrial logging concessions in Central Africa [53]. As part of the “CoforChange” project in Cameroon, the Central African Republic (CAR), and the Republic of Congo, ref. [60] used Landsat images to define a specific index called the “canopy opening index”. This index is calculated for each 500 m pixel, measuring the proportion covered by bare soil. It has helped to determine the time required for the revegetation of abandoned road networks following industrial selective logging operations [60]. Additionally, in the CAR, the combination of linear features and reflectance contrast in the red, near-infrared, and mid-infrared spectral bands has been crucial for detecting logging activities [50]. In the Democratic Republic of the Congo (DRC), mapping the expansion of rural complexes and photointerpretation of clearings has distinguished the major anthropogenic causes of forest degradation, with agriculture being the most significant [36].

Given the persistence of clouds over dense tropical forests [86] and the very rapid regeneration [86], the more or less accurate detection and measurement of small, isolated point disturbances, such as the fall of an individual tree, requires the use of active sensor data, including LiDAR [103], RADAR images [173], or very high-resolution optical images acquired by drones [174]. In the Congo Basin region, ref. [173] used the Sentinel-1 RADAR sensor to develop an algorithm capable of detecting small disturbance events (maximum area 10 m × 10 m) in near-real time. Between 2019 and 2020, the DRC, for example, recorded more small-scale disturbance events representing 541 Kha out of 1431 Kha of the total loss [173]. All events with a mapped size between 0.2 and <0.5 ha could be associated with selective logging [173], yet this approach has recently been shown to miss +80% of wood extraction events [106]. Ref. [106] developed an approach based on Sentinel-1 RADAR images to intercept gaps of less than 500 m², representing 80% of disturbances induced by selective logging, 70% of which escaped previous approaches [124,173]. Yet, the quality of the method heavily relies on exceptional data acquisitions combining field and drone inventories.

After 5 to 10 years of industrial logging, gaps, skid trails, and roads are no longer detectable by conventional remote sensing tools based on relatively high-resolution optical satellite images, notably Landsat or Sentinel-2 [9,50,60]. From this perspective, indicators of fragmentation or the index of integrity of intact forest landscapes by monitoring the road infrastructure network can only achieve short-term and partial monitoring of forest areas mainly affected by industrial selective logging.

The integration of textural image analysis of very high spatial resolution optical images acquired by UAVs (unmanned aerial vehicles or drones) and satellite imagery is promising for long-term and large-scale spatial historical monitoring of the causes of forest degradation at different logging intensities [174,175]. However, the current challenges of this approach are linked to the increased source of bias associated with the interaction between the level of detail provided and variations in acquisition conditions, notably sun–screen–sensor angles [176,177]. To improve local-scale monitoring, a promising avenue lies in the regular acquisition of drone data (https://www.canobs.net/protocols, accessed on 17 April 2025) to accurately monitor the consequences of forest degradation, both in terms of structural indicators and forest composition [178]. In the Yangambi region of the DRC, using aerial photogrammetry techniques, the canopy height model produced from optical UAV images was similar to that derived from accurate LiDAR data [179]. In Gabon, in the Ivindo Forest concession, a map of canopy height variation was produced from LiDAR UAV data. However, this map was only used to select control plots in the validation of the map of potential degradation by selective industrial logging based on the variation in aboveground biomass estimated by the interferometry of RADAR images from the TanDEM-X satellite [103].

Biomass quantification is a key indicator in the evaluation process of REDD+ programs. However, scientific work using optical sensors has been limited to characterizing the disturbance induced by logging gaps and has not, at the same time, monitored the carbon balance or biomass lost [6]. In Africa, biomass mapping with a low level of uncertainty has been carried out using optical sensors with very high spatial resolution [175] and active sensors such as RADAR [180] and LiDAR [181]. However, there are still very few initiatives on the application of remote sensing to the monitoring of biomass variation in the characterization of anthropogenic causes of forest degradation in the dense tropical forests of the Congo Basin (Table 3).

In Gabon, ref. [103] produced a less accurate map for postindustrial selective logging aboveground biomass variations above 130 Mg·ha⁻¹ on steep slopes between 2000 and 2021. In addition, the aboveground biomass map for Africa by [182], coupled with the fire impact map, demonstrated how fire accentuates the edge effect on biomass variation [151]. However, aboveground biomass maps produced from high-resolution RADAR and optical multi-sensor satellite data may present underestimates for forests with aboveground biomass above 250 Mg·ha⁻¹ [182]. The use of optical sensors at low spatial resolutions for biomass prediction without considering the strata variability of actual biomass stocks is not appropriate for mapping biomass variations [183].

Finally, the choice of the indicator to be measured in relation to specific activities is fundamental for an accurate and sustainable assessment of forest management practices [5,6,20,22]. For example, the assessment of selective logging intensity quantified by the volume or number of stems removed per unit area of a hectare plays an important role in predicting carbon emissions [184], conserving biodiversity [43] or assessing the renewal capacity of the wood stock to be logged [39]. Remote sensing, particularly radar satellite imagery combined with field data, has proven to be an effective tool for monitoring logging intensity in the Congo Basin [106]. However, our results indicate that this indicator, often measured using field data, is only the sixth most frequently used in scientific studies on forest degradation in the region (Table 3). This highlights the need for further advancements in remote sensing techniques to strengthen the link between forest structure indicators and other key aspects of forest degradation, such as logging intensity and biodiversity functional traits.

3.3.2. Relative Importance of Different Causes of Forest Degradation

It is noticeable that the same indicators are often the most frequently measured to monitor all anthropogenic activities. The most measured indicators provide information on the evolution of three types of criteria or properties of the state of a forest: the carbon storage capacity, primarily assessed by the variation in carbon or aboveground biomass changes, forest structure, generally evaluated by the density of clearings or canopy gaps, and biodiversity, mainly evaluated by the tree species richness changes. The use of these indicators predominates in assessing the environmental impact of industrial logging and shifting agriculture with slash-and-burn.

Building on these commonly used indicators, we now explore their application in assessing the environmental impacts of different anthropogenic activities. Comparing the variation in canopy gaps, aboveground carbon loss, and tree species richness, resulting from artisanal logging, selective logging, and shifting agriculture with slash-and-burn against undisturbed old-growth primary forest (

Table 4. Aboveground carbon loss (%), tree species loss (%), and canopy gaps loss (%) due to industrial and artisanal forms of selective logging, small-scale agriculture, or slash-and-burn agriculture (SD: standard deviation; Q1: 25th percentile; Q3: 75th percentile).

Aboveground Carbon Loss (%)
	n	Mean	SD	Min	Q1	Median	Q3	Max	Ref.
Industrial forest logging	13	14.79	22.21	−44	7.1	17	31.09	45.39	[62,64,93,102,108,113,114]
Artisanal forest logging	3	34.82	11.97	23.81	28.45	33.09	40.33	47.56	[114,119]
Shifting and burn agriculture	17	73.16	16.88	26.75	66.1	73.65	83.33	96.35	[119,135,136,139]
Tree Species Loss (%)
	n	Mean	SD	Min	Q1	Median	Q3	Max	Ref.
Industrial forest logging	25	4.27	21.07	−41.52	−6.45	4.35	14.97	64.29	[48,49,52,63,64,70,100,105,108,109,111,113]
Artisanal forest logging	4	16.33	15.68	−1.7	5.82	17.75	28.26	31.52	[111,117,119]
Shifting and burn agriculture	13	30.37	30.67	−17.39	3.03	40.22	48.91	76.19	[48,52,70,119,135,139]
Canopy Gaps Area Loss (%)
	n	Mean	SD	Min	Q1	Median	Q3	Max	Ref.
Industrial forest logging	17	8.98	13.63	0.5	4	5.2	9.8	60	[36,38,41,72,76,86,89,102,104,105,106,110]
Artisanal forest logging	1	10	-	10	10	10	10	10	[41]
Shifting and burn agriculture	3	61	28.58	29	49.5	70	77	84	[36,41,86]

), the results show that shifting cultivation has the greatest impact on both canopy gaps, carbon loss, and tree species richness in the 0–60-year regeneration stages (respectively, 61%, 73%, and 30%). Interestingly, while the impact of artisanal logging on canopy gap is relatively low (10%) and, consequently, potentially difficult to detect, its impact on carbon and tree species richness is four times higher than the one estimated in industrial forest logging concessions (Table 4).

The separate analysis of the magnitude of forest degradation caused by various anthropogenic factors remains a significant challenge, as illustrated by the works of [13,84,185]. This complexity arises from the difficulty in specifically linking the impacts of human activities to degradation indicators without considering the temporary or permanent changes in land use, complicating the distinction between deforestation and forest degradation [6,124].

To better understand these challenges, several studies have attempted to quantify forest degradation in the Congo Basin, with varying approaches to distinguish between deforestation and degradation. The results of [13] indicate that the annual rate of forest degradation in the Congo Basin is 0.09%, which accounts for 53% of the annual deforestation rate, of which only aspects of deforestation are discussed and not forest degradation. On the other hand, ref. [185] assessed forest degradation by detecting clearings smaller than 0.5 ha without addressing their potential causes. More recently, ref. [41], by analyzing the number of canopy gaps detected, provided a new perspective by dissociating the respective impact of deforestation and forest degradation causes in their analysis of canopy openings, thus providing a clearer overview of the various anthropogenic activities.

3.3.3. Considering the Temporal Scale

In contrast to deforestation, monitoring forest degradation requires regular monitoring over time of changes in forest properties [5]. Because it is essential to distinguish between what regenerates and what does not [124]. After agriculture, studies carried out in most Amazonian and West African rainforests indicate that specific leaf area and wood density are fully restored within 20–30 years but that it takes more than a century for biomass, floristic species richness, and composition to be fully restored [186]. Moreover, for selective logging, it is crucial to ensure the renewal capacity in composition and diversity of the resources harvested [39,187]. However, due to the lack of available information, there are still many uncertainties regarding the spatiotemporal variation in the recovery rate of different indicators on the state of tropical rainforests in the Congo Basin following various types of anthropogenic disturbances.

Despite the importance of long-term monitoring in understanding forest recovery, many studies overlook the temporal dimension in their analyses, focusing solely on the immediate impact following logging [67,69,73,76,85,99]. Among the studies that do integrate this factor, short-term monitoring (0–5 years) is the most used temporal scale for assessing the primary indicator of forest degradation, namely canopy gap detection (Figure 4). This predominance is partly explained by the rapid regeneration of canopy gaps and the technical capabilities of remote sensing tools, which mainly rely on passive (optical) satellite sensors with high and medium spatial resolution [9,60].

However, mid-term monitoring (10–20 years) is more prevalent in studies assessing aboveground carbon (AGC), belowground carbon (BGC), and changes in floristic composition (Figure 4), primarily relying on forest inventory data from permanent monitoring plots or chronosequences of post-disturbance forest succession [9,60]. However, permanent plots are often small-scale, located in protected areas, and their distribution is heterogeneous, with concentrations in certain regions and vast uncovered areas, especially in the Democratic Republic of Congo, which holds at least 60% of the Congo Basin’s forest cover [33].

The integration of very high-resolution optical imagery from UAVs with available high-quality historical datasets offers a promising solution for the long-term monitoring of the Congo Basin’s dense tropical rainforests. These historical datasets include national forest inventories, aerial photographs from 1950s missions over the Central Congo Basin, archived at the Royal Museum for Central Africa in Tervuren, Belgium [188], and declassified Corona mission imagery from the 1960s, obtained by the CIA–NASA agencies of the United States [189]. Leveraging these resources could significantly enhance our ability to track long-term forest dynamics and assess forest degradation trends with greater precision.

While long-term monitoring remains limited in the Congo Basin, studies conducted in most Amazonian and West Africa rainforests regions offer valuable insights into the varying rates of recovery for different forest parameters. Not all indicators of forest condition recover at the same temporal rate after from agriculture forest land use. For example, aboveground carbon (AGC) recovers very slowly following agricultural activities [186].

Similar trends have been observed in the Congo Basin, where small-scale slash-and-burn agriculture leads to a prolonged recovery of AGC. However, in contrast to agricultural conversion, forests subjected to industrial selective logging tend to recover their aboveground carbon rapidly (Figure 5a). In CAR, the aboveground biomass was more than 100% reconstituted after 24 years in forests subject to both low- and high-intensity selective logging [62]. Faster recovery has been reported in Gabon, where 97% and 87% of biomass was restored within just one year in FSC-certified and non-certified concessions, respectively [64]. The slow recovery of AGC in forests affected by small-scale slash-and-burn agriculture could be further exacerbated by repeated fallow cycles [136].

While carbon stocks can recover relatively quickly after selective logging, biodiversity follows a different trajectory after small-scale slash-and-burn agriculture, requiring much longer periods to return to pre-disturbance levels (Figure 5a,b). In the forests of Ituri, in the Democratic Republic of Congo, tree species richness in an industrial logging concession harvested 17 years prior was found to be higher than that of an unharvested mixed primary forest. In contrast, forests that regenerate after 10–15 and 40–60 years of agricultural use retain less than 42% and 3% of tree species, respectively, compared to undisturbed primary forests [52]. Similarly, in the forests of Upper Guinea and Central Africa, selectively logged forests contain less than 65% and abandoned farmland 76% of the tree species found in unlogged forests, even after 20 years [70].

In the context of sustainable forest management, it is important to understand the factors influencing post-disturbance recovery. Ref. [184] proposed a predictive model for biomass and carbon regeneration times, integrating key parameters such as logging intensity (volume or biomass of wood removed per hectare) and logging damage rate (biomass lost due to tree mortality). Such models could enhance selective forest logging monitoring efforts by providing more accurate recovery dynamics across various forest attributes and disturbance regimes.

The impact of selective logging on biodiversity in the Congo Basin forests varies among species and ecological communities. For animal biodiversity, logging tends to decrease the density of certain species, such as great apes, duikers, and bush pigs, while favoring others, including squirrels, insectivorous birds, and frugivorous bird species [58]. In terms of plant diversity, low-intensity logging influences forest regeneration by promoting high wood density tree species while altering the regeneration of light-demanding timber species [42]. However, high intensity logging significantly affects plant diversity by modifying species composition across different forest strata [44]. A comprehensive evaluation of forest degradation caused by industrial logging should consider multiple ecological attributes beyond forest structure and carbon storage. Integrating biodiversity, species composition, and ecosystem functioning into impact assessments would provide a more accurate understanding of the overall consequences of forest logging activities.

4. Conclusions

This analysis highlights that research on forest degradation in the Congo Basin does not yet sufficiently focus on the most significant anthropogenic drivers of degradation in the region, particularly small-scale agriculture and illegal or artisanal logging for timber and fuelwood. While industrial logging is well documented, it is not a major driver of degradation. Most indicators used to assess forest degradation primarily focus on changes in forest structure and carbon storage, but they often overlook the temporal scale, particularly in evaluating long-term effects.

Remote sensing remains the preferred method for characterizing degraded forests, especially for detecting ephemeral impacts such as canopy gaps. However, integrating more field data and leveraging technological innovations in remote sensing are necessary to assess the state of tropical forests in the Congo Basin more holistically. UAVs offer the resolution and spatial coverage needed to distinguish different types of degradation, with the advantage of precise temporal monitoring through multiple short-interval flight campaigns. For long-term monitoring, UAV data could be combined with high-quality historical datasets, such as national forest inventories, aerial photographs from 1950s missions over the Central Congo Basin stored at the Royal Museum for Central Africa in Tervuren, Belgium, or archival imagery from the CIA–NASA Corona mission from the 1960s.

These advancements could help redirect research towards the most critical yet least studied anthropogenic drivers of degradation, ultimately improving sustainable forest management policies to reduce emissions and biodiversity loss. A more refined approach could involve defining degradation thresholds for different types of selective logging by integrating additional indicators, such as post-harvest biomass variation relative to harvesting intensity and linking canopy metrics to forest structural and functional traits.