Smallholder adaptation pathways and resilience in African forest-agriculture landscapes: a review of climate impacts and economic implications Open Access

Climate change poses existential threats to smallholder farming systems across Sub-Saharan Africa, with profound implications for forest-dependent livelihoods, ecosystem services, and rural economies that underpin continental food security (IPCC, 2023; FAO, 2022; Ayanlade et al., 2022). African forest-agriculture landscapes face distinctive vulnerabilities arising from the intersection of climatic stress, land-use pressures, and institutional constraints, which demand integrated analytical frameworks that transcend conventional sectoral boundaries (Bryan et al., 2013; Duguma et al., 2019). Forest-savannah transition zones, agroforestry mosaics, and forest-fringe farming communities occupy critical positions within both agricultural production systems and forest ecosystem dynamics across the continent, yet their unique adaptation contexts remain inadequately theorised within African forest economics scholarship (Appiah and Guodaar, 2022; Chomba et al., 2020).

Smallholder farmers, typically cultivating less than 5 hectares and predominantly using family labour (Lowder et al., 2021; Samberg et al., 2016), constitute over 80% of agricultural producers across Sub-Saharan Africa and are primary actors in forest-agriculture landscape transformations (Jayne et al., 2022). These farmers engage simultaneously in agricultural markets, forest product extraction, and ecosystem service provision, creating complex livelihood portfolios that mediate climate vulnerability in ways that are poorly captured by single-sector analyses (Sunderlin et al., 2005; Angelsen et al., 2011). In Ghana's forest-savannah transition zone, smallholders integrate cocoa, food crops, and tree resources into diversified farming systems facing intensifying climate pressures (Asare et al., 2014; Abdulai et al., 2018). Understanding how African smallholders perceive, evaluate, and respond to climate change within forest-agriculture systems is fundamental to forest economics, rural development policy, and climate governance across the continent (Nyantakyi-Frimpong and Bezner-Kerr, 2015).

The Knowledge-Attitude-Practice (KAP) framework has emerged as a dominant paradigm for examining climate adaptation behaviours among African farming populations (Belay et al., 2022; Liao et al., 2022; Assan et al., 2020). Originating in health behaviour research, KAP assumes sequential progression from cognitive awareness through evaluative orientations to observable actions (Kwol et al., 2020). In African contexts, the model suggests that farmers' understanding of climate change phenomena lays the foundation for attitude formation, which subsequently motivates the adoption of adaptive practices (Antwi-Agyei et al., 2021; Ochieng et al., 2020). While intuitively appealing, this linear conceptualisation fails to capture the complex interdependencies that characterise adaptation decisions in African forest-agriculture systems, where institutional constraints, tenure arrangements, and collective action dynamics substantially shape behavioural outcomes (Awazi and Tchamba, 2023; Djoudi et al., 2016).

This systematic review addresses a critical gap in African forest economics scholarship by synthesising empirical evidence on smallholder farmers' knowledge, attitudes, and adaptation practices with explicit attention to forest-agriculture landscape contexts across the continent. The review advances beyond descriptive synthesis to interrogate dominant theoretical assumptions, identify methodological limitations that constrain current understanding, and propose an integrative framework that positions adaptation within African forest economics debates on ecosystem services, land-use governance, and livelihood resilience (Minang et al., 2014; Chomba et al., 2016). By prioritising evidence from Ghana, West Africa, and the broader Sub-Saharan region, the review contributes regionally grounded insights to address the persistent Northern-centric bias in climate adaptation scholarship (Carr and Thompson, 2014; Nagoda and Nightingale, 2017).

The review pursues three interconnected objectives aligned with African forestry economics research priorities. First, it systematically maps the empirical landscape, examining smallholder adaptation knowledge, attitudes, and practices across African forest-agriculture contexts, identifying geographic, thematic, and methodological patterns, with attention to underrepresented regions, including Ghana's forest-savannah transition zone. Second, it critically evaluates the evidentiary basis for KAP model assumptions in African contexts, examining whether knowledge translates into attitudes and whether attitudes predict practices, with institutional and governance factors as mediators. Third, it advances a novel conceptual synthesis integrating behavioural dimensions with forest economics concerns, offering actionable implications for African climate policy, extension programming, and forest governance reform.

The Knowledge-Attitude-Practice model provides theoretical scaffolding for substantial climate adaptation research across Africa; however, its application to forest-agriculture contexts warrants critical examination grounded in African empirical realities (Belay et al., 2022; Gebrehiwot and van der Veen, 2013). Mayo's original formulation assumed linear causation from knowledge acquisition through attitude formation to behavioural practice, an assumption subsequently challenged by research demonstrating persistent knowledge-action gaps across African farming systems (Khatibi et al., 2021; Meijer et al., 2015). In African climate adaptation, farmers frequently implement practices without formal climate knowledge, responding to immediate production pressures and drawing on indigenous knowledge systems rather than abstract climate concepts (Nyong et al., 2007; Nkomwa et al., 2014). Conversely, African farmers with a detailed understanding of climate may fail to act due to resource constraints, institutional barriers, or tenure insecurity, regardless of individual knowledge or attitudes (Agrawal, 2010; Mertz et al., 2011).

Protection Motivation Theory (Rogers, 1983) offers complementary mechanisms explaining how threat perceptions and coping appraisals drive protective behaviours among African smallholders. The theory distinguishes threat appraisal (perceived severity and vulnerability) from coping appraisal (response efficacy, self-efficacy, and response costs), suggesting that motivation to adapt requires both an adequate perception of threat and a belief in practical, achievable responses (Grothmann and Patt, 2005; van Valkengoed and Steg, 2019). Applied to African forest-agriculture contexts, this framework illuminates why farmers may perceive climate threats but fail to adopt forest-based adaptations if they doubt the effectiveness of the practices, question their capacity to implement them, or face prohibitive response costs arising from insecure tree tenure (Bugri, 2008; Kalaba et al., 2013). Research in Ghana's cocoa-forest landscapes demonstrates that even when farmers recognise climate risks to tree crops, tenure uncertainty and policy constraints substantially limit the adoption of agroforestry (Asare, 2019; Ruf et al., 2015).

The Theory of Planned Behaviour (Ajzen, 1991) further specifies that attitudes, subjective norms, and perceived behavioural control jointly determine behavioural intentions, which in turn predict actual behaviour. This framework emphasises social influence dimensions that are largely absent from individual-focused KAP applications but are critically important in African communal farming contexts (Meijer et al., 2015). Within African farming communities, peer networks, traditional authorities, and extension agents shape acceptable adaptation responses, potentially overriding individual knowledge or attitudes through normative pressures (Mango et al., 2017; Wuepper et al., 2018). Research on African agroforestry adoption consistently shows that neighbour effects and social learning substantially influence the uptake of practices beyond individual characteristics (Bandiera and Rasul, 2006; Ndah et al., 2024). The integration of these theoretical perspectives suggests that adaptation in African forest-agriculture systems operates through multiple interacting pathways that require institutional analysis alongside behavioural modelling (Agrawal, 2010; Pelling et al., 2015).

African forest-agriculture landscapes present distinctive adaptation contexts inadequately captured by conventional agricultural or forestry frameworks developed primarily in Global North settings (Chomba et al., 2020; Reed et al., 2016). Smallholders across Sub-Saharan Africa simultaneously engage with crop production, livestock husbandry, tree cultivation, and forest product extraction, creating diversified livelihood portfolios that spread climate risk while generating complex management trade-offs characteristic of African forest-farm interfaces (Sunderlin et al., 2005; Mbow et al., 2014). Adaptation decisions in one domain cascade through interconnected system components: shifting to drought-tolerant crops may reduce labour available for forest product collection; intensifying agroforestry may compete for food-crop area; expanding irrigation may affect downstream forest hydrology and community water access (Ngigi, 2009; Dewees et al., 2011).

The economic dimensions of forest-agriculture adaptation in Africa extend beyond direct practice costs to encompass opportunity costs across livelihood activities, risk redistribution within diversified portfolios, and temporal trade-offs between short-term coping and long-term resilience building (Wunder et al., 2014; Angelsen et al., 2011). Forest products serve as critical safety nets during agricultural production shortfalls in African rural economies, buffering climate shocks through accumulated natural capital and providing insurance functions that are particularly important for resource-poor households (Wunder et al., 2014; Kalaba et al., 2013). Ghana's forest-fringe communities clearly demonstrate this pattern, with NTFP collection intensifying during crop failures and providing 20–40% of household income during periods of climate stress (Appiah et al., 2009; Acheampong et al., 2019). Adaptation strategies affecting forest access or productivity thus carry implications for household risk management that individual-practice analyses, which overlook forest economics dimensions, may miss entirely.

Governance contexts fundamentally distinguish African forest-agriculture adaptation from purely agricultural settings and from forest-agriculture dynamics in other regions of the world. Forest tenure arrangements in Africa typically involve complex overlays of statutory, customary, and community-based claims, creating uncertainty that constrains long-term investments in tree-based adaptations (Alden Wily, 2011; Cronkleton et al., 2012). Protected area regulations, timber-harvesting restrictions, and tree-felling permit requirements frequently discourage on-farm tree retention and the adoption of agroforestry despite their adaptation benefits (Bugri, 2008; Acheampong et al., 2021). Community-based natural resource management institutions create enabling or constraining conditions for adaptation options, including tree planting, forest product extraction, and land-use modification (Agrawal, 2001; Persha et al., 2011; Agrawal, 2010). Institutional capacity for extension service delivery, credit provision, and climate information dissemination shapes whether individual knowledge and attitudes translate into the adoption of practices across African contexts (Davis et al., 2012; Ragasa et al., 2016). These structural factors suggest that behavioural models require integration with institutional analysis to explain adaptation patterns in African forest-agriculture systems.

This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines (Page et al., 2021), with particular attention to standards for environmental management research (Collaboration for Environmental Evidence, 2018). The review protocol was developed a priori, specifying research questions, eligibility criteria, search strategy, screening procedures, data extraction variables, quality appraisal methods, and synthesis approach. Protocol development drew on established systematic review methodology in climate adaptation research (Berrang-Ford et al., 2015; Ford et al., 2011) and forest economics (Hajjar et al., 2021).

Studies were included if they: (1) examined smallholder farmers cultivating less than 5 hectares, following standard definitions in African agricultural economics (Lowder et al., 2021; Jayne et al., 2022); (2) assessed knowledge, attitudes, perceptions, or practices related to climate change adaptation; (3) were conducted in forest zones, forest-savannah transitions, or forest-agriculture landscapes, with priority given to African and Global South contexts; (4) employed empirical primary data collection using quantitative, qualitative, or mixed methods; (5) were published in peer-reviewed journals between January 2018 and December 2026; and (6) were written in English. Studies were excluded if they: (1) focused exclusively on large-scale commercial agriculture; (2) examined climate mitigation without adaptation components; (3) were review articles, editorials, or conference abstracts; (4) lacked methodological detail sufficient for quality appraisal; or (5) reported findings from exclusively urban populations.

Systematic searches were conducted across three databases: Scopus, Web of Science Core Collection, and ScienceDirect. Database selection balanced comprehensive coverage of forest economics, agricultural science, and climate adaptation literatures with operational feasibility, following recommendations for environmental systematic reviews (Haddaway et al., 2018). Searches were executed on 15 September 2025. The search strategy combined terms across four conceptual domains: (1) population (smallholder* OR small-scale farm* OR peasant* OR subsistence farm*); (2) phenomenon (adapt* OR resilien* OR coping OR risk management); (3) context (climat* OR weather variability OR drought OR flood*); and (4) setting (forest* OR agroforest* OR tree crop* OR forest zone OR transition zone OR Africa OR Ghana). Boolean operators combine terms within and across domains. Search strings were adapted to database-specific syntax requirements. Supplementary searches included screening the reference lists of included studies and tracking seminal publications in African forest economics (Chomba et al., 2016; Minang et al., 2014).

Retrieved records were imported into the Rayyan systematic review software for screening (Ouzzani et al., 2016)—duplicate removal was performed using automated detection supplemented by manual verification. Title and abstract screening were used to apply the inclusion criteria, with conservative retention of potentially relevant records for full-text review. Full-text screening confirmed eligibility for all criteria, with exclusion reasons documented in accordance with PRISMA reporting standards (Page et al., 2021). Two reviewers independently screened records at both stages, with disagreements resolved through discussion and consultation with a third reviewer when consensus was not reached. Inter-rater reliability exceeded 0.85 (Cohen's kappa) at both screening stages, meeting established thresholds for systematic review reliability (Cooper et al., 2019).

A standardised data extraction form captured: (1) bibliographic information (authors, year, journal, country); (2) study characteristics (design, sample size, sampling method, forest context); (3) population characteristics (farmer demographics, farm size, crop types, forest dependence); (4) knowledge measures (climate awareness, adaptation knowledge, information sources); (5) attitude measures (risk perception, adaptation attitudes, self-efficacy); (6) practice measures (adaptation strategies employed, adoption rates, determinants); (7) institutional variables (extension contact, credit access, tenure security, organisational membership); (8) analytical methods (statistical techniques, qualitative approaches); and (9) key findings and author conclusions. Effect sizes were extracted or calculated where studies reported sufficient statistical detail for meta-analytic synthesis, following established procedures for behavioural science meta-analysis (Borenstein et al., 2009). In accordance with PRISMA 2020 Item 11 requirements, the following variable categories were explicitly operationalised. Dependent variables comprised: (a) adaptation practice adoption measured as a binary indicator or proportional adoption rate per practice; (b) adoption rates for specific strategy categories (crop-based, soil management, water management, and forest-based); and (c) composite adaptation indices where reported by study authors. Independent and mediating variables comprised: (a) climate knowledge score (standardised continuous score or proportion of correct responses); (b) risk perception and concern ratings (Likert scale scores or composite indices); (c) self-efficacy for practice adoption; (d) extension service contact (binary or frequency count); (e) credit access (binary); (f) tenure security (binary: formal title or documented customary right); and (g) farmer organisational membership. Where studies reported multiple operationalisations of the same construct, the measure most closely aligned with the primary research question was extracted.

Descriptive statistics for extracted variables: knowledge awareness prevalence ranged from 54.2% to 96.8% across 58 studies (mean: 78.1%, SD: 9.4%); adaptation practice adoption rates ranged from 11.2% to 71.8% (overall mean across practices: 38.6%); risk perception proportions ranged from 51.3% to 89.7% (mean: 73.1%, SD: 11.2%). Effect measures were selected as follows: prevalence proportions were used for awareness and adoption rate outcomes because most included studies reported single-arm cross-sectional prevalence without a reference group, rendering risk ratios inappropriate; Pearson r was used for KAP relational analyses because it is the effect size most consistently reported in observational behavioural studies and facilitates comparison with prior meta-analyses in this literature (van Valkengoed and Steg, 2019). Where studies reported regression coefficients or odds ratios, conversions to r were performed using established formulae (Borenstein et al., 2009, Chapter 7).

The study's quality was assessed using the Mixed Methods Appraisal Tool (MMAT) version 2018 (Hong et al., 2018), which provides category-specific criteria for quantitative, qualitative, and mixed-methods designs appropriate for the methodological heterogeneity in African adaptation research. Each study was rated across five methodology-specific criteria, generating quality scores from 0 (no criteria met) to 5 (all criteria met). Studies scoring below 3 were flagged for a sensitivity analysis to assess whether their exclusion altered the synthesis findings. Quality assessment was conducted independently by two reviewers, and discrepancies were resolved through discussion in accordance with established reliability protocols (Higgins et al., 2019).

Narrative synthesis organised findings thematically around knowledge, attitude, and practice dimensions, with particular attention to geographic patterns across African regions, institutional determinants, and theoretical orientations relevant to forest economics. Tables summarised study characteristics, adaptation strategies, and identified gaps with emphasis on African evidence. A meta-analytic synthesis was conducted, with sufficient studies reporting comparable effect sizes, using random-effects models to account for between-study heterogeneity with the DerSimonian-Laird estimator (DerSimonian and Laird, 1986). The DerSimonian-Laird estimator was selected over restricted maximum-likelihood estimation for its greater robustness with smaller numbers of studies and its use in comparable adaptation meta-analyses (van Valkengoed and Steg, 2019), facilitating direct comparison. Tau-squared is reported alongside I-squared as an absolute measure of between-study variance, following the recommendations of Borenstein et al. (2009). Heterogeneity was assessed using I-squared statistics and Cochran's Q test, with degrees of freedom and p-values reported for all Q statistics. Publication bias was evaluated using a funnel plot and Egger's regression test (Egger et al., 1997), with the intercept, standard error, and t-statistic reported. Subgroup analyses examined effect modification by geographic region within Africa, forest context, and study quality. Meta-analyses were conducted in R version 4.3.1 using the metafor package (Viechtbauer, 2010).

Database searches retrieved 3,124 records (Scopus: 1,412; Web of Science: 1,156; ScienceDirect: 556). Following duplicate removal (n = 489), 2,635 records underwent title and abstract screening, with 276 retained for full-text review. Full-text assessment excluded 209 records: 78 lacked forest-agriculture context, 52 did not examine smallholder populations, 38 were non-empirical, 26 lacked methodological detail, and 15 were published outside the specified timeframe. The final synthesis included 67 studies. Figure 1 presents the PRISMA flow diagram detailing the selection process.

The 67 included studies were published between 2018 and 2026, with publication frequency increasing over time (2018–2019: 11; 2020–2021: 17; 2022–2023: 25; 2024–2026: 14). Geographic distribution concentrated in Sub-Saharan Africa (n = 47, 70.1%), followed by South Asia (n = 11, 16.4%), Southeast Asia (n = 6, 9.0%), and Latin America (n = 3, 4.5%). Within Africa, East African countries dominated (Ethiopia: 14; Kenya: 8; Uganda: 6), while West Africa received comparatively limited attention (Ghana: 4; Nigeria: 3; Senegal: 2; Burkina Faso: 2). Forest-savannah transition zones, despite their ecological and agricultural significance for African food security, were explicitly examined in only 12 studies (17.9%), with none specifically addressing Ghana's transition zone.

Methodologically, quantitative designs predominated (n = 48, 71.6%), followed by mixed-methods approaches (n = 14, 20.9%) and qualitative studies (n = 5, 7.5%). Sample sizes ranged from 48 to 1,247 respondents (median: 384). Cross-sectional surveys constituted the dominant data collection method (n = 52, 77.6%), with limited longitudinal (n = 3) or experimental (n = 2) designs. Quality appraisal scores ranged from 2 to 5 (median: 4), with 11 studies (16.4%) scoring below the threshold of 3. Table 1 summarises the characteristics of the included studies.

Climate change awareness was assessed in 58 studies across African and Global South contexts. Awareness prevalence ranged from 54.2% (Burkina Faso) to 96.8% (Kenya), with substantial between-study heterogeneity (I-squared = 94.7%, Cochran's Q (df = 41) = 774.3, p < 0.001, tau² = 0.041). This level of heterogeneity indicates that the pooled estimate reflects a descriptive central tendency across a diverse body of evidence rather than a precise parameter; readers should interpret it in conjunction with the subgroup analyses reported in Section 4.8. Meta-analysis of 42 studies reporting comparable prevalence estimates yielded a pooled awareness rate of 78.4% (95% CI: 72.1–84.7%, random-effects model; z = 18.4, p < 0.001). Subgroup analysis revealed higher awareness in East Africa (82.3%) than in West Africa (71.6%; p = 0.023), potentially reflecting differences in extension service coverage and access to climate information across regions. Publication bias assessment showed a broadly symmetric funnel plot with minor asymmetry at the base. Egger's regression test was non-significant (intercept = 0.54, SE = 0.38, t = 1.42, p = 0.16), indicating no statistically significant publication bias.

Knowledge depth varied considerably across African study populations. While basic awareness of changing weather patterns was widespread, understanding of the causes and consequences of climate change, and of adaptation options, showed greater deficits. Only 34.7% of farmers across studies correctly identified anthropogenic contributions to climate change. Technical knowledge of specific adaptation practices averaged 56.2%, with notable gaps regarding soil conservation techniques (41.3%) and water management innovations (38.9%). Ghanaian studies have reported that while 76% of farmers perceived climate variability, only 48% could identify appropriate adaptation responses, suggesting an awareness-action gap that requires institutional intervention (Alidu et al., 2022; Antwi-Agyei and Nyantakyi-Frimpong, 2021).

Information sources for climate knowledge included radio (reported in 47 studies), extension services (39 studies), fellow farmers (36 studies), television (28 studies), and mobile phone services (18 studies). Studies consistently found that farmers who accessed multiple information channels had higher knowledge scores than those who relied on a single source (pooled effect: r = 0.31, 95% CI: 0.24–0.38, z = 8.6, p < 0.001).

Attitude measures were reported in 51 studies, employing diverse conceptualisations including risk perception, concern, self-efficacy, response efficacy, and adaptation intention—measurement heterogeneity limited meta-analytic synthesis for most attitude constructs. Risk perception was the most frequently assessed dimension (n = 38 studies). African farmers generally perceived climate change as a serious threat to agricultural livelihoods, with 73.6% (pooled estimate) expressing moderate to high levels of concern. Perceived vulnerability is positively associated with adaptation behaviour across studies (pooled r = 0.38, 95% CI: 0.29–0.46, z = 8.1, p < 0.001, I² = 71.3%, k = 24 studies). Ethiopian and Kenyan studies reported particularly high levels of risk perception, potentially reflecting recent experiences with severe droughts (Belay et al., 2022; Ochieng et al., 2020).

Self-efficacy for implementing adaptation practices varied substantially by practice type and farmer characteristics across African contexts. Farmers reported higher confidence in autonomous, low-cost adaptations (planting date adjustment, crop variety selection) than in technology-intensive practices (irrigation, mechanisation) or long-term investments (agroforestry establishment). Education level and extension contact emerged as consistent predictors of self-efficacy across African studies, suggesting that institutional support enhances perceived implementation capacity (Gebrehiwot and van der Veen, 2013; Maziya et al., 2024). The knowledge-attitude relationship showed a moderate positive association (pooled r = 0.42, 95% CI: 0.35–0.49, z = 14.1, p < 0.001, I² = 68.4%, k = 31 studies), providing partial support for the KAP model's assumption that knowledge underpins attitude formation. However, substantial residual variance indicated that factors beyond knowledge shape attitudes toward adaptation in African forest-agriculture contexts.

Adaptation practices were examined in all 67 studies, with considerable variation in practice categorisation and measurement reflecting diverse African agroecological contexts. Synthesising across studies, crop-based practices predominated adaptation portfolios across African forest-agriculture landscapes. Crop diversification showed the highest adoption rate (67.3% pooled estimate; 95% CI: 61.2–73.4%), followed by adoption of drought-tolerant varieties (52.8%), adjustment to early/late planting (48.4%), and crop switching (31.2%). These autonomous adaptations required minimal external inputs and aligned with existing farming knowledge systems, reflecting adaptation pathways accessible to resource-constrained African smallholders (Nyantakyi-Frimpong and Bezner-Kerr, 2015).

Soil and water management practices showed lower adoption rates. Mulching (34.7%), composting (29.3%), and conservation tillage (24.6%) were reported across studies, though adoption varied substantially by agroecological context and the availability of institutional support. Irrigation adoption remained below 15% across most African study populations due to infrastructure constraints, capital limitations, and water governance challenges (Giordano et al., 2012).

Forest-based adaptation strategies received limited attention despite their relevance to African forest-agriculture systems and livelihood resilience. Agroforestry adoption averaged 34.2% across 28 studies explicitly examining tree integration, substantially below its potential given documented benefits for microclimate regulation, soil protection, and income diversification (Mbow et al., 2014; Duguma et al., 2019). Non-timber forest product collection as a climate coping strategy was reported in 19 studies (average utilisation of 28.7% during climate-stress periods), confirming forests' safety-net functions for African rural households (Wunder et al., 2014). Woodlot establishment for diversification was examined in only 8 studies, revealing a significant gap in the understanding of the economics of forest-based adaptation. Table 2 and Figure 2 summarise adaptation strategies and behavioural dimensions identified across studies.

The hypothesised sequential relationships within the KAP model received partial empirical support across African studies. Knowledge-attitude associations were consistently positive (pooled r = 0.42, 95% CI: 0.35–0.49, z = 14.1, p < 0.001, I² = 68.4%, Q(30) = 95.1, p < 0.001, tau² = 0.018, k = 31), suggesting that climate understanding provides cognitive foundations for favourable attitudes toward adaptation. However, the knowledge-practice relationship showed weaker and more variable associations (pooled r = 0.28, 95% CI: 0.19–0.37, z = 7.6, p < 0.001, I² = 79.3%, Q(37) = 178.6, p < 0.001, tau² = 0.024, k = 38 studies), indicating that knowledge alone inadequately predicts adaptation behaviour in African contexts.

Attitude-practice relationships demonstrated stronger associations (pooled r = 0.47, 95% CI: 0.40–0.54, z = 13.9, p < 0.001, I² = 74.1%, Q(28) = 108.1, p < 0.001, tau² = 0.021, k = 29 studies), supporting attitudes as proximal determinants of adaptive action. Egger's regression tests for publication bias were non-significant for both KAP relational analyses (knowledge-practice: intercept = 0.48, t = 1.18, p = 0.24; attitude-practice: intercept = 0.31, t = 0.91, p = 0.37), indicating no statistically significant small-study effects. Mediation analyses, reported in 12 studies, found that attitudes partially mediated the effects of knowledge on practice. However, substantial direct effects persisted in most models, suggesting that pathways from knowledge to practice operate independently of measured attitudes.

Critically, institutional and structural factors explained greater variance in adaptation than individual-level KAP variables across African studies employing multivariate analyses. Extension service contact, credit access, tenure security, and farmer group membership consistently emerged as significant predictors independent of knowledge and attitude measures (Antwi-Agyei and Stringer, 2021; Atta-Aidoo et al., 2022). This pattern suggests that behavioural models emphasising individual cognition substantially underestimate the importance of enabling institutional environments for adaptation in African forest-agriculture contexts.

The synthesis revealed systematic gaps in the evidence base, warranting explicit documentation for African forest economics scholarship. Geographic concentration in East Africa (47.8% of studies) leaves a substantial underrepresentation of other African regions. West Africa's forest-savannah transition zones, despite their agricultural significance and acute climate vulnerability, featured in only 9 studies. Ghana contributed 4 studies, none of which explicitly examined the forest-savannah transition zone, despite its importance for national food security and cocoa production (Asare et al., 2014). This gap represents a critical priority for African forest economics research.

Methodologically, the dominance of cross-sectional surveys (77.6%) limits causal inference regarding KAP relationships. Longitudinal designs that tracked adaptation trajectories over time were rare (n = 3), precluding assessment of how knowledge and attitudes evolve with accumulated climate experience. Experimental or quasi-experimental designs testing the effectiveness of interventions were reported in only 2 studies, limiting the evidence on what works to support African smallholder adaptation.

Forest economics dimensions received inadequate attention across the African adaptation literature. Economic analyses of adaptation costs and benefits were reported in 14 studies (20.9%), most of which provided descriptive accounts rather than rigorous economic evaluations. The contribution of forest-based strategies to livelihood resilience was examined in only 8 studies, representing a critical gap given the documented evidence of forests' insurance functions for African rural households (Wunder et al., 2014; Angelsen et al., 2011). Table 3 summarises key theoretical, methodological, and geographic gaps identified through this review. Similarly, Figure 3 shows the geographic distribution of studies considered in this study.

Given the high I² values observed across all meta-analytic outcomes, sensitivity and subgroup analyses were conducted to examine sources of between-study variance and test the robustness of pooled estimates. Two pre-specified subgroup analyses were performed: (a) by geographic region within Africa and (b) by study quality tier (MMAT score 4–5 versus 2–3).

Subgroup analysis by geographic region for the awareness prevalence outcome yielded the following within-subgroup estimates: East Africa (k = 20): 82.3% (95% CI: 77.1–87.5%), I² = 87.2%; West Africa (k = 9): 71.6% (95% CI: 63.2–80.0%), I² = 89.4%; Southern Africa (k = 7): 76.4% (95% CI: 68.1–84.7%), I² = 78.3%; South Asia (k = 6): 74.1% (95% CI: 64.9–83.3%), I² = 83.1%. While within-subgroup I² values remain elevated, the between-subgroup Q test was significant (Q(3) = 12.4, p = 0.006), confirming that geographic region accounts for a meaningful component of total between-study variance. East African studies show systematically higher awareness prevalence, consistent with the comparatively greater density of government-supported extension systems and radio-based climate information services in that region.

Subgroup analysis by study quality tier for KAP relational outcomes showed that restricting analysis to high-quality studies (MMAT score 4–5, k = 23 for knowledge-practice) yielded a somewhat lower pooled estimate (r = 0.24, 95% CI: 0.14–0.34) compared with the full sample (r = 0.28), with modestly reduced heterogeneity (I² = 72.1 versus 79.3%). This confirms that the knowledge-practice gap persists in higher-quality studies and is not an artefact of methodologically weaker studies. The sensitivity analysis excluding all studies scoring below 3 on the MMAT (n = 11) did not alter any pooled estimate by more than 0.04 correlation units, indicating robustness of the synthesis findings. The persistence of high I² within subgroups reflects genuine contextual variability across African agroecological zones and institutional settings rather than methodological artefact, underscoring the importance of region-specific longitudinal studies to generate more precise estimates.

The meta-analytic finding that knowledge shows a substantially weaker association with adaptation practices (r = 0.28) than attitudes (r = 0.47) challenges the foundational assumption of linear KAP progression prevalent in African extension programming and climate adaptation policy. This knowledge-practice gap, observed consistently across Sub-Saharan African studies, carries profound implications for forest economics scholarship and intervention design. If knowledge does not reliably translate into action, information-focused interventions emphasising climate-awareness campaigns and technical training may prove insufficient to catalyse the behavioural changes required to build resilience in forest-agriculture systems.

A critical qualification applies to this interpretation. Because 77.6% of included studies rely on cross-sectional survey designs, the associations between knowledge, attitudes, and practices reported here are correlational in nature and cannot support causal inference. The direction of association cannot be established from this evidence base: farmers who have already adopted adaptation practices may actively seek out climate knowledge to justify or refine their decisions, inflating observed knowledge-practice correlations rather than confirming the theorised causal sequence. The three longitudinal studies and two experimental studies identified in this synthesis are insufficient to establish temporality across the range of African contexts examined. Future research must prioritise longitudinal panel designs and experimental or quasi-experimental evaluations to test causal mechanisms within the KAP framework in African forest-agriculture settings. The pooled r = 0.28 also carries a wide confidence interval reflecting high between-study heterogeneity (I² = 79.3%); the subgroup estimate for high-quality studies (r = 0.24, 95% CI: 0.14–0.34) may offer a more representative benchmark for contexts with relatively developed institutional infrastructure.

The magnitude of the gap in African contexts exceeds that reported in studies from the Global North (Khatibi et al., 2021), suggesting that structural constraints facing African smallholders more severely impede knowledge translation. Resource-poor farmers may understand adaptation options yet lack the capital to implement them, face tenure insecurity that discourages long-term investment, or confront institutional barriers that restrict access to inputs, credit, or markets essential for adopting practices (Agrawal, 2010; Mertz et al., 2011). This interpretation aligns with institutional economic perspectives emphasising that behaviour change requires not only individual motivation but also enabling environmental conditions (North, 1990; Ostrom, 2005). African forest economics scholarship must therefore integrate behavioural analysis with institutional assessment to explain and address adaptation deficits.

The consistent finding across African studies that institutional variables, including extension contact, credit access, and organisational membership, explain greater variance in adaptation than individual-level knowledge and attitudes represents a fundamental challenge to the predominant behavioural framings in climate adaptation research. This pattern, particularly pronounced in forest-agriculture contexts where tenure arrangements, forest governance, and collective action institutions substantially shape available adaptation options, demands reconceptualisation of adaptation determinants.

For African forest economics, this finding redirects analytical attention from individual farmer characteristics toward the institutional architecture governing forest-agriculture landscapes. Extension services, despite documented capacity constraints across African agricultural systems (Davis et al., 2012; Ragasa et al., 2016), emerge as critical conduits for adaptation not only through knowledge transfer but also by legitimating practices, facilitating access to inputs, and connecting farmers to support networks. Credit institutions, when accessible to African smallholders, enable investment in capital-intensive adaptations, such as establishing agroforestry, building soil-conservation infrastructure, and improving water management, which are otherwise beyond the reach of resource-poor households (Giordano et al., 2012). Farmer organisations facilitate collective learning, resource pooling, and market access, thereby enhancing individual incentives for adaptation and enabling landscape-scale coordination (Mango et al., 2017).

The forest governance dimension merits particular attention within African contexts. Tenure security for trees and forest products emerges across studies as a critical determinant of agroforestry and forest-based adaptation (Bugri, 2008; Alden Wily, 2011). When African farmers cannot capture long-term benefits from tree investments due to ambiguities in customary tenure, statutory restrictions on tree harvesting, or unclear rights to forest products, rational actors underinvest in potentially resilient livelihood components (Acheampong et al., 2021). Ghanaian research specifically demonstrates that tree tenure uncertainty substantially constrains on-farm tree retention and agroforestry expansion despite farmer awareness of climate benefits (Asare et al., 2014; Ruf et al., 2015). Policy reforms that secure tree tenure rights, streamline forest product regulations, and create incentives for sustainable forest-agriculture integration are high-priority interventions to enhance adaptation outcomes across African forest-agriculture landscapes.

Forest-based adaptation strategies emerge from this synthesis as significantly underutilised among African smallholder populations, despite substantial evidence of their contributions to livelihood resilience. Agroforestry adoption (34.2%) and NTFP collection during climate stress (28.7%) lag considerably behind crop-based adaptations, notwithstanding documented benefits including microclimate regulation, soil protection, diversified income streams, and insurance functions buffering agricultural production failures (Mbow et al., 2014; Wunder et al., 2014; Duguma et al., 2019).

This underutilisation pattern in African contexts likely reflects multiple converging constraints that existing interventions inadequately address. Tenure insecurity discourages long-term investments in trees when farmers cannot confidently expect to benefit from them (Cronkleton et al., 2012). Policy barriers, including tree-felling permit requirements and restrictions on the commercialisation of forest products, reduce economic returns to on-farm tree cultivation (Acheampong et al., 2021). Extension services that prioritise annual-crop agriculture over integrated tree-crop systems provide limited technical support for establishing and managing agroforestry systems (Franzel et al., 2004). Market underdevelopment in tree products reduces incentives for income diversification (Shackleton et al., 2011). Addressing these constraints represents a priority for African forest economics research and forest governance reform.

The underutilisation also reflects temporal mismatches between climate adaptation pressures and the realisation of agroforestry benefits. Trees require years to mature and generate substantial products, creating investment horizons that resource-constrained households facing immediate consumption pressures may find prohibitive (Meijer et al., 2015). Climate finance mechanisms that provide patient capital for tree establishment, combined with interim support that bridges the gap between planting and production, could address this temporal constraint (Minang et al., 2014). Payment for ecosystem service schemes that recognise the adaptation co-benefits of smallholder agroforestry represent promising but underdeveloped opportunities for enhancing forest-based adaptation incentives across African landscapes (Chomba et al., 2016).

The pronounced geographic concentration of evidence in East Africa (47.8% of included studies) relative to West Africa creates significant knowledge gaps limiting the generalisability of synthesis findings across African forest-agriculture contexts. West African forest-savannah transition zones, characterised by distinctive agroecological conditions, farming systems, and governance arrangements, remain critically underrepresented despite their substantial importance for regional food security and climate vulnerability.

Ghana exemplifies this evidence gap: it has contributed only four studies, none of which specifically examine the forest-savannah transition zone, which constitutes the country's primary food-producing region and supports millions of smallholder livelihoods. This absence is particularly problematic given Ghana's significance as a climate-adaptation research site, its ongoing forest governance reforms, and its strategic importance for cocoa production under intensifying climate pressures (Asare et al., 2014; Abdulai et al., 2018). Addressing this gap through targeted research in Ghanaian forest-savannah transition zones represents an urgent priority for African forest economics scholarship. The concentration of evidence also limits understanding of how institutional contexts shape adaptation dynamics across diverse African settings, highlighting the need for comparative studies that examine how adaptation patterns vary across these contexts.

This review advances African forest economics theory by demonstrating the inadequacy of compartmentalised analytical approaches that separate agricultural adaptation from forest system dynamics in livelihood analysis. The evidence synthesised here establishes that African smallholder livelihoods in forest-agriculture landscapes integrate crop production, livestock management, tree cultivation, and forest product extraction into unified livelihood systems where climate adaptation decisions in any component affect and are affected by other components (Sunderlin et al., 2005; Angelsen et al., 2014). This integration challenges forest economics models that treat forestry and agriculture as separate analytical domains and calls for frameworks that capture livelihood-level interdependencies across land-use components.

The proposed Forest-Agriculture Adaptation Typology offers an integrative framework that distinguishes adaptation modes along temporal (short-term coping versus long-term transformation) and spatial (farm-level versus landscape-level) axes. Type I adaptations (autonomous, farm-level) have the highest adoption rates in Africa but offer limited protection against severe climate stress. Type II adaptations (sustained farm-level investment) face adoption constraints due to tenure insecurity and capital constraints. Type III adaptations (collective short-term responses) require functional institutions for coordination. Type IV adaptations (landscape-level transformations) require governance innovations that transcend individual farm decisions. This typology repositions adaptation within debates in African forest economics on ecosystem service governance, landscape management, and institutional design for common resources (Ostrom, 2005; Agrawal, 2010).

Theoretically, the synthesis findings support the integration of behavioural economics and institutional analysis to explain patterns of adaptation in African forest-agriculture systems. The KAP framework provides useful heuristics for understanding individual cognition but requires contextualisation within institutional analyses that specify the enabling and constraining conditions for behaviour change. Transaction cost economics illuminates why knowledge fails to translate into practice when implementation costs exceed farmer capacity (Williamson, 1985). Common-pool resource theory explains how collective action challenges constrain landscape-level adaptations (Ostrom, 1990). Property rights theory explains how tenure insecurity discourages long-term investments in trees (Alchian and Demsetz, 1973). African forest economics scholarship should integrate these theoretical traditions to develop frameworks adequate to the complexity of forest-agriculture adaptation.

The finding that institutional factors consistently outperform individual-level variables in predicting adaptation has fundamental implications for the construction of forest economics theory. It suggests that behaviour change models premised on individual utility maximisation inadequately explain African adaptation dynamics, in which institutional constraints substantially bound available choice sets. Structural theories emphasising political economy, governance quality, and power relations may offer greater explanatory power than behavioural theories that assume autonomous individual decision-making (Pelling et al., 2015; Nightingale, 2017). Future theoretical development should examine how structural constraints interact with individual factors to shape adaptation trajectories across African forest-agriculture landscapes.

Forest economics scholarship should engage more substantively with African indigenous knowledge systems in theorising adaptation. The synthesis reveals that African farmers frequently adapt, drawing on accumulated experiential knowledge rather than formal climate science information, suggesting that indigenous knowledge constitutes an undertheorised resource for adaptation (Nyong et al., 2007; Nkomwa et al., 2014). Integrating indigenous knowledge perspectives with scientific understanding of climate, rather than treating them as competing epistemologies, offers promising directions for both adaptation theory and practice in African contexts.

To enhance the utility of the Forest-Agriculture Adaptation Typology for empirical research design, survey instrument construction, intervention evaluation, and policy monitoring, this section specifies concrete measurable indicators for each of the four adaptation types. The operational specifications are derived from the evidence synthesised in this review and are intended to provide directly applicable benchmarks.

Type I (Autonomous Farm-Level Adaptation) encompasses low-cost, individually initiated adjustments that require no external institutional support and are implemented within one growing season. Representative practices include changes in crop variety, planting date adjustments, and crop spacing modifications. Measurable indicators: (a) proportion of farmers adopting at least one Type I practice within 12 months of a perceived climate anomaly; (b) number of autonomous practices adopted per household per season. Operational threshold: adoption within one agricultural year, requiring no credit, external inputs, or tenure documentation. Data collection: structured household survey with one-year recall period using a practice checklist. Institutional prerequisites: none; accessible to all smallholders irrespective of tenure, credit, or organisational status, which explains the consistently high adoption rates for this type across the synthesis.

Type II (Sustained Farm-Level Investment) encompasses planned, capital-requiring investments with payoff horizons exceeding three years, typically involving tree planting, soil infrastructure construction, or water management improvements. Measurable indicators: (a) hectares under agroforestry or tree crops established within the past five years; (b) proportion of farm area with documented soil conservation infrastructure; (c) tenure security status coded as a binary indicator (formal statutory title or documented customary right = 1); (d) investment sustained across at least three consecutive seasons. Operational threshold: capital investment maintained for three or more growing seasons with observational evidence of establishment. Data collection: farm parcel mapping combined with household survey; review of tenure documentation, where available. Institutional prerequisites: secure tree tenure rights; access to medium-term credit, payment-for-ecosystem-service income, or savings-group finance; extension technical guidance for species selection and management.

Type III (Collective Short-Term Coping) encompasses community-coordinated responses to acute climate shocks requiring functional group institutions with governance rules enabling rapid mobilisation within one season. Measurable indicators: (a) farmer membership in an organisation with a documented climate response protocol or emergency resource-sharing arrangement; (b) proportion of households accessing communal resources (water, grain, or seed) during the most recent climate-stress season; (c) frequency of community adaptation meetings per season, with a threshold of at least two per growing season. Operational threshold: coordinated collective response activated in at least two of the past five climate-stress seasons. Data collection: household survey combined with key informant interviews with group leaders. Institutional prerequisites: functional farmer organisation, water user association, or community-based natural resource management institution with clear governance rules for resource access during climate stress.

Type IV (Landscape-Level Transformation) encompasses governance-mediated changes in land-use configuration, forest cover, or watershed management that operate across multiple farm units and require multi-stakeholder coordination over five or more years. Measurable indicators: (a) proportion of landscape area under an agreed land-use plan or community forest management agreement with multi-stakeholder endorsement; (b) directional forest cover trend from satellite imagery (Landsat 8/9 or Sentinel-2 composite; five-year baseline; minimum mapping unit 0.5 ha); (c) number of formal multi-stakeholder governance platforms with documented meeting records per year, threshold of at least four per year with cross-sector representation. Operational threshold: measurable, attributable change in landscape-level forest cover or land-use configuration sustained over five years. Data collection: remote sensing time-series analysis combined with governance document review and multi-level stakeholder interviews. Institutional prerequisites: district or catchment-level governance platform with legal mandate; national legal framework enabling community-based forest management with benefit-sharing; external facilitation for multi-stakeholder coordination during the establishment phase.

For policymakers across the African forestry and agricultural sectors, this review suggests that information-based interventions targeting farmers' knowledge, while necessary, are insufficient to catalyse comprehensive adaptation within forest-agriculture systems. Structural reforms that address institutional constraints warrant priority alongside extension programming. First, strengthen extension service capacity for climate-specific technical guidance that integrates crop, livestock, and forest components, rather than delivering siloed sectoral messages. African extension systems face documented capacity constraints, but their demonstrated importance for adaptation suggests that investing in capacity yields substantial returns for climate resilience (Davis et al., 2012; Ragasa et al., 2016).

Second, reform tenure arrangements to secure African smallholders' rights over trees and forest products, creating incentives for long-term investment in forest-based adaptation strategies that are currently deterred by tenure uncertainty (Alden Wily, 2011; Bugri, 2008). Ghana's ongoing land administration reforms offer opportunities to strengthen tree tenure security with direct implications for adaptation outcomes. Third, develop financial instruments appropriate for smallholder adaptation investments in African contexts, including patient capital for tree establishment, index-based insurance for climate risk management, and results-based payments for ecosystem service provision linking adaptation with conservation incentives (Minang et al., 2014). Fourth, support the development of farmer organisations and collective action capacity, recognising that landscape-level adaptations require coordination mechanisms beyond individual farm decisions (Mango et al., 2017). Fifth, invest in research to address identified evidence gaps, particularly in underrepresented regions, including Ghana's forest-savannah transition zones, where climate vulnerability is acute but empirical understanding remains limited.

This review identifies priority research directions for advancing African forest economics scholarship on climate adaptation: (1) Longitudinal studies tracking adaptation trajectories over time across African forest-agriculture contexts are urgently needed to establish causal relationships and examine how knowledge, attitudes, and practices evolve with accumulating climate experience. (2) Economic analyses quantifying adaptation costs, benefits, and trade-offs across practice types and livelihood components would inform efficient resource allocation and intervention targeting within African forestry programming. (3) Research specifically addressing Ghana's forest-savannah transition zone would fill the critical evidence gap identified in this synthesis, generating insights directly applicable to policy and extension programming in a nationally significant agricultural region. (4) Experimental and quasi-experimental evaluations testing extension approaches, financial instruments, and governance interventions would strengthen the evidence base for African forestry policy recommendations. (5) Comparative studies across African forest-agriculture contexts would identify context-specific versus generalisable adaptation dynamics, enabling appropriate localisation of interventions across the continent's diverse institutional and agroecological settings. (6) Interdisciplinary research integrating behavioural science, institutional economics, and forest ecology would develop analytical frameworks adequate to the complexity of African forest-agriculture systems facing climate change.

This systematic review synthesised evidence from 67 studies examining smallholder farmers' knowledge, attitudes, and climate change adaptation practices in forest-agriculture landscapes across Sub-Saharan Africa and the Global South. Meta-analytic findings reveal widespread climate awareness (78.4%, 95% CI: 72.1–84.7%; z = 18.4, p < 0.001; I² = 94.7%) inadequately translating into comprehensive adaptation, with crop diversification dominating practice portfolios while forest-based strategies, including agroforestry and NTFP collection, remain underutilised despite their documented contributions to livelihood resilience. Attitudes partially mediate knowledge-practice relationships, but institutional factors consistently explain greater variance in adaptation than individual-level behavioural variables across African contexts. The high between-study heterogeneity observed throughout (I² = 68.4–94.7%) means that all pooled estimates should be interpreted as descriptive central tendencies. Because the evidence base is dominated by cross-sectional designs (77.6%), the KAP associations identified here are correlational and do not establish causal sequences.

The review advances African forest economics scholarship by reconceptualising adaptation as embedded within forest-agriculture systems characterised by complex tenure arrangements, multiple governance authorities, and intersecting livelihood components. The proposed Forest-Agriculture Adaptation Typology integrates behavioural dimensions with temporal and spatial considerations, now operationalised through concrete measurable indicators for each of the four adaptation types (Section 6.1), offering a framework for research prioritisation and policy targeting relevant to African forestry and climate programming. Critical interrogation of KAP assumptions reveals that linear sequencing is inadequate for capturing the recursive, institutionally mediated adaptation processes that characterise African smallholder contexts.

Significant evidence gaps demand urgent research attention. Geographic concentration in East Africa leaves West African forest-savannah transitions critically underrepresented, particularly in Ghana, where empirical evidence for this ecologically vital zone is lacking. Methodological dominance of cross-sectional surveys limits causal inference. Economic analyses of forest-based adaptation contributions to African livelihood resilience are notably scarce. Addressing these gaps represents a priority for African forest economics research with direct implications for climate policy, extension programming, and forest governance reform.

Climate change will continue to reshape African forest-agriculture landscapes regardless of progress in mitigation, making adaptation imperative for the sustainability of smallholder livelihoods across the continent. This review establishes that supporting effective adaptation requires moving beyond information-focused interventions toward integrated approaches that address institutional constraints, secure forest tenure, develop appropriate financial instruments, and strengthen collective action capacity. African forest economics scholars are uniquely positioned to advance this agenda, bringing analytical tools and regional expertise essential for navigating the complex intersections of climate, forests, and rural livelihoods that define the continent's development trajectory.