This study adopts a systematic review with quantitative meta-analysis, aiming to rigorously synthesize the scientific knowledge published between January 2015 and March 2025 on the assessment of tropical soil health in Sub-Saharan Africa (SSA). The methodological approach is guided by the “Preferred Reporting Items for Systematic Reviews and Meta-Analyses” (PRISMA) recommendations [35]. Indeed, the review is based on a conceptual framework integrating three main dimensions of soil health: 1) physico-chemical indicators, 2) biological indicators, and 3) assessment tools and methodologies. These dimensions are analyzed in relation to sustainable management practices and agroecological specificities of the four Sub-Regions of Sub-Saharan Africa: West Africa, East Africa, Central Africa and Southern Africa.

A systematic search was conducted in the main international scientific databases: Web of Science, Scopus, ScienceDirect, SpringerLink, Wiley Online Library, MDPI, PLOS, Frontiers, as well as Google Scholar to complete the identification of relevant publications. The institutional bases (FAO, ISRIC, CGIAR, Soils4Africa) were consulted to integrate major technical reports when scientifically justified. The search equations combined keywords related to soil health and the target geographical area, using Boolean operators: (“soil health” OR “soil quality” OR “soil indicators” OR “soil biological indicators”), AND (“Sub-Saharan Africa” OR “West Africa” OR “East Africa” OR “Central Africa” OR “Southern Africa”), AND (“assessment” OR “evaluation” OR “soil quality index” OR “digital soil mapping”), AND (2015–2025). Additional research has specifically targeted biological indicators (e.g., “microbial biomass”, “soil respiration”, “enzyme activity”).

The selection criteria have been defined a priori in order to ensure scientific consistency. These criteria are presented in Table 1.

Table 1. Inclusion and exclusion criteria.

Only studies that met all inclusion criteria were retained for the final analysis.

The selection process took place in four stages:

1) Identification of references in databases;

2) Removal of duplicates;

3) reading of titles and summaries for a first sorting;

4) Complete read in full text for final validation according to the established criteria.

The application of the collection strategy leads to a pre-corpus of 220 documents. From the 200 references initially identified, including 150 from databases and 50 from complementary sources, 95 duplicates were eliminated, leaving 105 articles for initial screening. After a rigorous screening phase, 23 articles were excluded for non-compliance with the inclusion criteria, leading to 82 articles evaluated in full text. Of these, 34 were excluded for methodological or relevance reasons, allowing 29 studies to be retained for qualitative analysis and 19 studies for quantitative meta-analysis. Thus, a corpus of 48 documents (i.e. 21.81% of the pre-corpus). This methodical approach ensures transparency and reproducibility of the literature selection, while ensuring that qualitative and quantitative syntheses are based on robust and comparable data, thus providing a solid basis for the integrated assessment of soil health indicators.

A standardized extraction grid was developed using a spreadsheet to ensure the homogeneity of the information collection (Table 2).

The soil health indicators identified have been structured into three broad functional categories, in line with international conceptual frameworks for integrated soil assessment that recognize the physical, chemical and biological dimensions as complementary pillars of soil ecological functioning [1][2][5][6].

Physical indicators, such as bulk density and porosity, provide information mainly on the structural condition of the soil, its capacity for water infiltration, aeration and storage, which are essential elements for productivity and water

Table 2. Variables extracted from selected studies.

regulation. Chemical indicators, including soil organic carbon (SOC), total nitrogen (N), available phosphorus (P), pH and cation exchange capacity (CEC), reflect fertility potential, biogeochemical cycle function and nutrient availability for crops. Finally, biological indicators, such as microbial biomass and enzymatic activities, reflect the intensity of biological activity, the dynamics of organic matter decomposition and the ecological resilience of tropical agroecosystems. In order to identify regional trends, a frequency analysis was conducted to determine the indicators most used in the studies of each Sub-Region of Sub-Saharan Africa, thus highlighting methodological convergences, contextual specificities and possible imbalances in the representation of the physical, chemical and biological dimensions of soil health.

For studies reporting comparable quantitative data, including means, standard deviations and sample sizes, a quantitative meta-analysis was performed to estimate the effect of sustainable management practices on key soil health indicators, with a particular focus on soil organic carbon (SOC), which is recognized as a key marker of tropical soil fertility and resilience [36][37]. Effect sizes were standardized using Hedges’ g to allow comparison of results from different methodological contexts, and a random-effects model was adopted to account for inter-study heterogeneity [38]. Statistical heterogeneity was assessed using the I² index and the Cochran Q test, consistent with the recommendations of methodological reviews in ecology and soil science [39]. Subgroup analyses were conducted to compare effects by sub-region: West Africa, East Africa, Central Africa and Southern Africa, to identify specific regional trends and understand the influence of agroecological context and local practices [4].

In parallel, a thematic qualitative analysis was conducted to complement the quantitative approach and provide a contextualized interpretation of the results [40]. This analysis identified recurrent methodological challenges, such as variability in sampling protocols, lack of standardization of indicators, and analytical capacity constraints in Sub-Saharan Africa [6][41]. Analytical limitations, including the sensitivity of the methods to seasonal variations and soil types, were examined to identify areas for improvement. Emerging technologies such as MIR/NIR spectroscopy and DSM have been codified and ranked by relevance for integrated soil monitoring [13][14]. Finally, the themes were coded inductively, allowing for the emergence of common trends and opportunities for improvement, including the need to integrate local knowledge with analytical tools, to optimize composite indicators (SQI) and to strengthen the adoption of sustainable practices adapted to regional constraints [4][10][26].

This methodology combines systematic rigor, regional comparability and robust quantitative analysis. It identifies the dominant soil health indicators in Sub-Saharan Africa between 2015 and 2025, assesses their contextual relevance and proposes scientific guidance for the future improvement of assessment frameworks.

Analysis of the geographical distribution (Figure 1(b)) of soil health literature in Sub-Saharan Africa reveals a notable concentration in some regions, particularly in East and West Africa. Ethiopia stands out with 6 publications, followed by Ghana (3) and Kenya (2), while countries such as Mozambique, Benin, Côte d’Ivoire, Rwanda, Tanzania, Cameroon and Botswana appear with a more limited number of publications (1 to 2). This distribution (Figure 1(a)) reflects both the focus of research projects, the availability of scientific infrastructure, and the implementation of soil monitoring programs. The map highlights a spatial gradient of knowledge, with a strong representation of areas of intensive agriculture and historical research basins, while some regions remain under documented.

In terms of time, the annual evolution of publications between 2015 and 2025 shows a growing dynamic. Between 2015 and 2019, the number of publications remains limited (0 to 3 articles per year), reflecting an emerging interest in tropical soil health and the associated methodological challenges. From 2020 onwards, a clear increase is observed, peaking at 13 publications in 2025. This significant increase can be explained by several factors: the rise of research initiatives on food security, the increasing integration of sustainable agricultural practices and the active involvement of international funding programs. It also reflects the consideration of the Sustainable Development Goals (SDGs), which underline the importance of restoring and maintaining soil health to ensure the resilience of agricultural systems and the sustainability of ecosystems.

By combining spatial and temporal dimensions, this analysis provides a clear diagnosis of current scientific coverage and highlights areas and periods of high research activity. It highlights the importance of strengthening the harmonization

Figure 1. Temporal (a) and Geographic (b) distribution of publications between 2015-2025.

of methods and indicators, to enable cross-site comparison and support evidence-based policymaking. In addition, the increase in publications over the past five years illustrates a growing alignment of soil health research with global sustainable development priorities, reinforcing the relevance of studies to guide sustainable agricultural interventions and territorial planning.

The keyword cloud of the documents (Figure 2) highlights a clear structuring of the scientific corpus around a few dominant poles that reflect the methodological and operational priorities of research on soil health in Sub-Saharan Africa.

Figure 2. Clouds of Kays words from the documents in the corpus.

The visual pre-eminence of soil, quality, indicators, properties and assessment confirms that the literature is primarily oriented towards the quantification and standardization of diagnostics. The high frequency of the terms Africa, mapping, resolution and digital simultaneously reveals the rise of digital mapping and spatial modelling approaches, which have become central to extrapolating results and supporting territorial planning. The words land use, agriculture and farmers indicate that these assessments are mostly in productive contexts, where agronomic performance and sustainability guide the interpretation of the indicators. The sustained visibility of carbon, nutrients and fertility shows that biogeochemical regulation, in particular organic matter, is a pivot linking food security and climate issues. On the other hand, the smaller size of microbial suggests that, despite increasing recognition, biological dimensions remain less integrated than physicochemical parameters. Finally, the strong presence of certain countries, notably Ethiopia, reflects a geographical concentration of research efforts, revealing inequalities in infrastructure and funding at the regional level.

The quantitative analysis of the corpus (48 studies) highlights a clear dominance of physico-chemical indicators in the assessment of soil health in Sub-Saharan Africa (Table 3).

Table 3. Classification of soil health indicators.

Organic carbon is mobilized in about 36 publications, i.e. nearly 75% of the work, confirming its central role in simultaneously interpreting fertility, structure and sequestration potential. pH appears in 31 studies, total nitrogen in 24 and cation exchange capacity in 22, reflecting a strong reliance on standardized, comparable parameters compatible with regional mapping exercises. Physical variables such as bulk density (18) and available phosphorus (17) occupy an intermediate position. On the other hand, the biological dimensions remain marginal: microbial biomass is considered in only 9 studies, respiration in 7, enzymes in 6 and macrofauna in 5, i.e. less than a quarter of the corpus. At the same time, integrated approaches are progressing, with 26 publications offering composite indices and 19 using statistical weighting methods. These figures describe a field still dominated by soil chemistry, while the functional integration of living organisms remains in transition.

Indeed, the indicators identified can be grouped into three broad categories: physical, chemical and biological, in accordance with the international conceptual frameworks for integrated soil assessment. Physical indicators include bulk density, porosity, and texture, which reflect the structure of the soil and its infiltration capacity. Chemical indicators include soil organic carbon (SOC), total nitrogen, available phosphorus, pH, and cation exchange capacity (CEC), which are essential for fertility and biogeochemical cycles. Biological indicators, such as microbial biomass and enzyme activity, are crucial for the resilience and maintenance of soil biological function. Frequency analysis showed that SOC is the most widely used chemical indicator, while microbial biomass dominates in biological indicators. The distribution by subregion reveals a predominance of studies in West Africa (n = 22), followed by East Africa (n = 15), Central Africa (n = 13), and Southern Africa (n = 14).

The methodologies used vary from the classic field (systematic sampling, physico-chemical analyses) to numerical and spectroscopic approaches, including remote sensing and MIR/NIR spectroscopy [14][18]. Composite indices (SQIs) have been widely adopted to synthesize soil quality from multiple parameters, facilitating cross-site and inter-sub-regional comparison. Table 4 presents the methods identified.

An examination of the 48 studies shows a clear methodological hierarchy dominated by statistical tools and the aggregation of indicators. Principal component analysis is used in about 19 publications, mainly to reduce the redundancy of matrices including chemical, physical and sometimes biological variables, and to isolate the factors explaining the variability. The construction of synthetic soil quality indices is the most frequent approach (26 studies), reflecting the search for a directly interpretable result for decision-making; However, these indices remain sensitive to normalization and weighting rules. Scoring methods accompany this logic in ≈17 studies. Statistical comparisons between uses or practices, via ANOVA and associated tests, appear in 21 studies, while regressions and correlations aimed at establishing functional relationships are present in 23. Spatialization is progressing with digital soil mapping (14) and machine learning

Table 4. Methods used for soil health assessment.

(12), allowing regional extrapolation but highly dependent on the quality of training data. MIR/NIR spectroscopy remains emerging (8). Finally, participatory approaches remain limited (8) (5 of which are qualitative), indicating that local knowledge is still peripheral in quantitative frameworks.

Of the 48 methodologically usable studies, the dominant analytical infrastructure remains the conventional laboratory, which is used in about 40 publications (>80%). This work is mainly based on standardized physicochemical protocols (soil organic carbon, pH, total nitrogen, cation exchange capacity), guaranteeing inter-site comparability and metrological robustness. This predominance is explained by the international standardization of analytical methods and their integration into recognized evaluation frameworks, although these devices imply high operational costs and a low temporal density of observations.

Figure 3. Word cloud related to soil health assessment tools and technologies.

The lexical cloud presented in Figure 3 highlights the preponderance of geospatial and digital vocabulary in the corpus analyzed. The terms mapping, spatial and information occupy the largest visual area, which corroborates the methodological results showing that nearly a third of the studies use geographic information systems or predictive digital mapping approaches. The assessment of soil health thus appears to be closely linked to the ability to transform punctual observations from laboratory protocols (SOC, pH, nitrogen, CEC), dominant in more than 80% of publications, into continuous representations of the territory. This articulation between standardized measurement and spatialization reflects a shift in research towards decision-making frameworks at the landscape level.

The high frequency of the terms machine, learning, random and modelling confirms the rise of algorithmic approaches, consistent with the presence of about 12 studies integrating supervised learning models. This evolution signals a transition to a computational science where predictive performance depends on both the robustness of datasets and the sophistication of statistical architectures. In addition, the visibility of spectroscopy, MIR and NIR reflects the progressive development of rapid diagnostic technologies (≈8 studies), still in the minority but strategic to increase observational density at a reduced cost, subject to calibration libraries adapted to tropical soils. Finally, the presence of terms associated with production systems, such as forest, indicates that these tools are mainly applied to anthropogenic landscapes. Overall, the lexical hierarchy confirms a field oriented towards predictive spatialization, supported by AI and gradually enriched by spectral innovation, while remaining structurally dependent on robust analytical repositories.

The distribution of strategies and technologies /practices shows a clear dominance of conservation agriculture, cited in about 26% of the corpus (Figure 4).

Figure 4. Frequencies of sustainable soil management strategies and practices.

This pre-eminence reflects the strength of international programs promoting reduced tillage, permanent cover and rotations as rapid levers for improving carbon and structural stability. Organic amendments follow (23%), confirming the attention paid to the restoration of organic matter in highly depleted contexts. Agroforestry appears in nearly 17% of the studies, often associated with the maintenance of the SOC and microclimatic regulation. On the other hand, large-scale restoration approaches or integrated nutrient management remain less represented. This hierarchy reveals research that is more focused on the experimental plot than on the systemic transformation of agricultural landscapes. It also suggests a preference for short-term, measurable practices that are consistent with existing evaluation frameworks.

The main obstacle identified concerns the lack of methodological harmonization, mentioned in 44% of the studies (Figure 5). The studies use variable sets of indicators, different standardization procedures and rarely comparable thresholds, which limits regional syntheses. The cost of analyses represents the second major obstacle (33%), particularly for biological parameters. Data deficits and low density of observations appear in about a third of the corpus, constraining the robustness of spatial extrapolations. The limited transferability of the models confirms that many tools remain dependent on specific local conditions. Finally, the low place given to peasant knowledge indicates that evaluation remains predominantly technocentric. Together, these constraints draw a science that is analytically efficient but still fragile in its capacity for generalization and appropriation.

Figure 5. Constraints in assessing soil health.

The analysis of the corpus highlights a discipline engaged in a major methodological transition phase. Nearly 35% of the studies explicitly call for strengthening the integration of biological dimensions in order to overcome historical dependence on classical chemical variables. Work conducted in East Africa and highland areas highlights that microbial biomass, enzyme activity and functional indicators improve diagnostic sensitivity to land use change. This evolution responds to the need to better link soil quality and ecosystem services, already identified in previous conceptual frameworks.

At the same time, about 30% of publications insist on the rise of digital approaches. High-resolution predictive mapping and machine learning are transforming fragmented datasets into operational information at the regional scale. However, several authors point out that these performances are based on the availability of reliable analytical references. Without robust calibration, algorithmic sophistication risks amplifying uncertainties rather than reducing them.

The question of harmonization is another central axis, mentioned in nearly a third of the corpus. The development of standardized indices would facilitate inter-site comparability and strengthen the political usefulness of the assessments. Nevertheless, excessive standardization could ignore the pedoclimatic and socio-agronomic diversity of territories.

Spectroscopic technologies appear to be a promising intermediate solution. Several studies demonstrate their ability to simultaneously estimate many parameters with a substantial reduction in costs and processing time. However, their widespread use depends on sustainable investments in data infrastructures. In addition, nearly 20% of studies argue for a hybridization between formal science and peasant knowledge. The integration of perceptual indicators improves the local relevance of diagnoses and promotes the adoption of recommendations. Thus, the perspectives converge towards a more integrated evaluation, combining biology, digital technology and participation. The challenge is no longer just to measure better, but to produce indicators that are transferable, economically viable and socially legitimate.