Following an extensive review of literature on rural livelihood diversification in sub-Saharan Africa, Loison (2015) reported two dominant approaches that scholars have frequently used to study household livelihood diversification behavior in the region: the “household economic model” (Taylor and Adelman, 2003) and the “livelihoods approach” (Chambers and Conway, 1991; Scoones, 1998; Ellis, 2000). The household economic model views the farming household as a dual-purpose (both producer and consumer) decision-making unit that makes production, labor allocation and consumption decisions that may be interdependent. A key assumption of this model is that such decisions are made to maximize utility. While conceptually sound, the household economic model fails to capture the reality of livelihood activity and strategy choices made by households in stressed environments (Ellis, 2000), and does not account for social relationships between household members that in many cases influence those choices (Ellis, 1998).

Therefore, with its people-centered focus, the livelihoods approach has dominated empirical studies on rural livelihood strategies and diversification in developing countries. The livelihoods approach is premised on the notion that all people have a range of assets (social, physical, financial, human and natural) which determine (i) the options available to them, (ii) the livelihood strategies they adopt, and (iii) their vulnerability to risks and shocks (Ellis, 2000). However, these assets gain their meaning and value through the prevailing social, institutional and organizational environments (DFID, 1999). While many of its components are difficult to measure and have been conceptualized in numerous ways, often using proxies by different authors, the livelihoods approach provides a more realistic picture of rural household-level diversification strategies, constraints faced in their pursuit and the diverse character of livelihoods. This study therefore uses insights from the livelihoods approach to examine household livelihood diversification among forest-fringe rural communities in the Budongo Forest landscape.

This study was conducted in rural villages sharing borders with Budongo Central Forest Reserve (BFR) and two community forests (Ongo and Alimugonza) located at the periphery of BFR (Figure 1). BFR is a natural tropical forest located in mid-western Uganda, between 1^°37′ and 2^°03′ N, 31^°22′, and 31^°46′ E, spanning an area of 82,510 Ha. The Southern, South-Western and Eastern parts of BFR border farms and settlements. Many of the sections of the forest bordering these farms and settlements have been placed under CFM (Figure 1). There are six registered CBOs (CFM groups) that have signed agreements to co-manage those sections of BFR with the state agency (National Forestry Authority). These are Budongo Good Neighbours Community Association (BUNCA), Nyantonzi-Kamusenene Environment Conservation and Development Association (NECODA), North Budongo Forest Community Association (NOBUFOCA), Siiba Conservation Environment and Development Association (SEDA), Kapeka Integrated Community Development Association (KICODA), and Karujubu Forest Adjacent Community Association (KAFACA).

Most of the farms in the area are of subsistence nature, dominated by annual crops such as beans, sweet potatoes, cassava, and groundnuts. Over the past two decades, sugarcane plantations belonging to out-growers that supply Kinyara sugar factory have also expanded in the area, especially on the southern edge of the forest (Babweteera et al., 2018; Jeary et al., 2018). The factory is located about 5 km away from the forest edge. Resources accessed from BFR support the daily subsistence, and occasionally, cash needs of the forest-fringe households (Tumusiime et al., 2010; Mawa et al., 2020b).

The two community forests (Ongo and Alimugonza) are natural tropical forests located in the Budongo Forest Landscape. Ongo community forest covers 172.32 Ha and is located on the South-Western side of BFR while Alimugonza community forest is 28 Ha and is located on the Eastern side. Both community forests provide resources that are important for the daily subsistence needs of households in the surrounding villages (Mawa et al., 2020a,2021). Each of the community forests is managed by a Communal Land Association (CLA). The CLAs are formed by community members drawn from villages that have historically accessed resources from the respective forests. Most households in these villages are sedentary subsistence farmers.

The Budongo forest landscape is ethnically diverse with over 50 ethnic groups. The Banyoro who speak the Runyoro dialect are the indigenous inhabitants. The others comprise of immigrants from other parts of the country, Democratic Republic of Congo and South Sudan. The widely spoken languages in the area are Runyoro, Kiswahili, Lugbara, and Alur.

We conducted a cross-sectional survey in 17 villages located in the Budongo forest landscape. Ten of the villages were in Kapeeka, Kibwona, Kyaguzi, and Nyabyeya parishes that border Budongo Central Forest Reserve. The remaining seven shared boundaries with the two community forests. The survey was conducted between September and December 2018. We sought clearance and permission to conduct the study from the Uganda National Council of Science and Technology, Masindi District Forest Service and National Forestry Authority. A multistage sampling strategy was used to select households for interviews. First, a list of households was obtained from the respective village Local Council Chairpersons. These lists were stratified based on membership to CBFM schemes as CFM or CLA member households and non-members. Following Krejcie and Morgan (1970), a total of 423 households were selected for the two strata using systematic random sampling technique. This sample comprised of 96 CFM member households, 196 non-CFM members, 40 CLA member-households, and 91 non-CLA members. We adapted the Poverty Environment Network (PEN) household questionnaires (CIFOR, 2008) to collect data on household livelihood assets, activities, strategies, income sources and household socio-demographics. Interviews were held with household heads or any other adult member of the household who was knowledgeable about the household livelihood assets and activities. The interviews were held in the local languages by trained research assistants.

Both conservation group member and non-member households undertook similar livelihood activities. These were dominated by on-farm activities such as food and commercial crop farming, poultry and small livestock farming. Over 40% of the households in the sample (Figure 2) were engaged in each of these activities. Rearing of large livestock (cattle) was not popular in the study area. Forest environmental income was dominated by subsistence forest product extraction that over 70% of the households in our sample engaged in. Commercial forest product (timber and charcoal) extraction was practiced by 13% of the households in the villages adjacent to BFR while only 2% of those located in villages adjacent to the community forests were engaged in it. Non-farm income was dominated by petty businesses (20 and 10% in the CFM and community forestry sites, respectively). Salaries and wages, just like large livestock farming were not major sources of income for the average household in our sample. About 40% of the households in both CFM and community forestry sites obtained part of their household income from off-farm wage labor (offered on other farms in the area) as shown in Figure 2.

For households in the CFM sites, subsistence forest resource extraction was the most important income source for both CFM-member households (Median = UGX 92,000) and non-member households (Median = UGX 89,846). This was followed by small livestock farming (Median = UGX 68,210 for CFM-member households and Median = UGX 35,571 for non-CFM member households), poultry farming (Median = UGX 22,000 for CFM member households and Median = UGX 19,090 for non-CFM member households), commercial and subsistence crops were the least important household income sources (Table 3).

For households in villages bordering the community forests, food crop farming was the main income source among CLA member households (Median = UGX 106,363). While the livelihood activities that non-CLA member households engaged in were similar to those observed in the CFM sites, CLA member households obtained significantly higher median incomes from subsistence forest product extraction (Wilcoxon rank sum test, W = 1,307.5; p = 0.001) and petty businesses (Wilcoxon rank sum test, W = 1,554, p = 0.020) as shown in Table 3.

The average CFM member household in our sample was engaged in significantly higher number of livelihood activities (Mean = 5.25; SE = 0.13) compared to non-CFM member households (Mean = 4.61; SE = 0.10) (p < 0.05). However, their average Shannon’s and Gini-Simpson indices were not significantly different from those of non-CFM member households (p > 0.05). In the CF sites, there were no significant differences in the average number of livelihood activities that CLA member households and non-members engaged in (p > 0.05). Non-farm livelihood activities contributed marginally to the total household income, especially among communities surrounding the community forests, where its average contribution was less than 10% (Table 4).

Over 55% of the households around both BFR and the two community forests practiced high to very high levels of livelihood diversification (Figure 3). This pattern was consistent for all households regardless of membership in conservation groups.

The number of years of residence of a household head and membership in other social groups in the village significantly influenced a household’s level of livelihood diversification in the villages bordering BFR. For households in villages surrounding the two community forests, a household’s dependency ratio had a statistically significant influence (p < 0.05) (Table 5). For households in the CFM sites, each additional year of residence of the household head in the village increased their Gini-Simpson index of diversification by 0.0023 (p = 0.025). Similarly, having a household member belonging to other social groups in the village increased the household’s Gini-Simpson index of diversification by 0.081 (p = 0.025). For households in villages bordering the CFs, an additional increase in a household’s dependency ratio increased their Gini-Simpson index of diversification by 0.0001 (p = 0.037). The rest of the predictors did not significantly influence a household’s level of livelihood diversification (p ≥ 0.05).

Total household per capita income was negatively correlated with Gini-Simpson index in villages bordering BFR (Kendall’s tau-b = −0.133; p = 0.010) among both CFM-member households (Kendall’s tau-b = −0.148; p = 0.042) and non-member households (Kendall’s tau-b = −0.133; p = 0.049). However, there was no clear association between total household per capita income and Gini-Simpson index of diversification among households in villages bordering the community forests (Kendall’s tau-b = −0.031; p = 0.601). This was observed among both CLA member households (Kendall’s tau-b = −0.082; p = 0.456) and non-CLA members (Kendall’s tau-b = −0.089; p = 0.212).

Our findings reveal a precarious set of livelihood activities that dominate the Budongo forest landscape, an area that has been a hub for field experimentation of community-based forest management approaches in the country. A typical household in the area, regardless of membership in conservation groups, tends to engage in several agricultural enterprises. These enterprises are largely subsistence in nature and typify the rural agrarian landscape in Uganda. The only notable difference between conservation group member households and non-members was the increased value of income from commercial crop farming accruing to the households participating in CFM (Table 3). This is attributed to the increased efforts of actors, especially Non-Governmental Organizations (NGOs), toward supporting alternative livelihood activities through various schemes in the communities adjacent to BFR. Provision of alternative livelihood schemes dominate decoupling interventions in the area. These interventions aim to increase options available to their beneficiaries and reduce household dependence on the forest for subsistence. For example, Budongo Conservation Field Station (BCFS) through its alternative livelihoods projects has promoted small livestock rearing and supported production of high-value non-traditional crops such as rice, cabbage and soybean for commercial farming (Babweteera et al., 2018). Other NGOs such as Village Enterprise and the Jane Goodall Institute have also actively implemented related alternative livelihoods schemes in the area.

Previous studies on the effects of the alternative livelihood schemes in the Budongo forest landscape have reported mixed results. On one hand, success stories have been documented e.g., for project that supported ex-hunters (Hsiao et al., 2013; Tumusiime et al., 2010) while failures and negative externalities in an earlier intervention meant to curb human-wildlife conflict in the area have also been reported (Webber et al., 2007). Lessons from these interventions could be used to achieve better livelihood outcomes. However, given their fragmented, time-bound, small and short-term nature of the predominantly donor-funded grants that such projects often rely on Wicander and Coad (2018), Kazoora et al. (2019), best practices may be too costly to implement within their personnel, time and budget constraints. Therefore, dedicated state support in terms of adequately funding community-based conservation and development activities and ensuring a minimum level of coordination for projects promoting alternative livelihoods schemes in these areas is required.

In the community forest sites, CLA member households obtained higher income from subsistence forest products sourced from the community forests and petty businesses. Subsistence forest resource extraction is therefore an important livelihood activity in the BFR landscape, which equally offers other lucrative livelihood activities mostly pursued by wealthier and ‘more-connected’ households in covert ways (Mawa et al., 2020b). Because it is largely extractive in nature, the high dependence on subsistence forest resources in the adjacent communities could jeopardize forest conservation goals in the long run, leading to forest degradation (Mawa et al., 2020a) or forest conversion to other land uses.

In such rural landscapes, that are characterized by significant entry barriers for the poorest segments to meaningfully benefit from high return non-farm or farm-based activities, diversification interventions targeting specific socio-economic segments of the rural population should be promoted with a view of enhancing access to sustainable economic activities, credit and markets (Asfaw et al., 2019). This would enable the households to secure better living standards by engaging in activities that generate cash, expand their asset portfolios and enhance their ability to diversify across both farm and non-farm sectors (Ellis and Freeman, 2004). Consequently, such households would be more resilient in the face of the risky farming decisions that characterize the rural space (Barrett et al., 2001; Ellis and Bahiigwa, 2003).