Modelling nature-based tourism impacts on rural development and conservation in Sikunga Conservancy, Namibia

ABSTRACT

Community-based natural resource management and nature-based tourism often go hand in hand to drive conservation and economic development in sub-Saharan Africa. However, the complementarity of the two strategies is controversially discussed in the literature. Built on survey data from 200 households conducted in 2012 we analysed the trade-off between conservation and development objectives by means of a mathematical programming model representing the economy of a rural conservancy in Namibia. We found that in the scenario describing unrestricted resource extraction, local communities mainly benefit from fishing and utilising forest products. In comparison, the scenario representing the social optimum, implying sustainably managed fish stocks and appropriate diets for community inhabitants, shows that community households increase agricultural diversification and shift livelihoods towards tourism employment.

KEYWORDS:

• Nature-based tourism
• conservation
• community-based natural resource management
• nutrition
• mathematical programming

1. Introduction

Namibia promotes community-based natural resource management (CBNRM) via legally established conservancies with assigned rights to collectively manage their natural resources (Lapeyre, 2011). The key objective of the programme is rural community development through conservation that basically relies on the expansion of nature-based tourism (Silva & Mosimine, 2012). Tourism’s total contribution to Namibia’s Gross Domestic Product was 16% in 2015, and almost 17% of national employment is attributable to the tourism sector, nature-based tourism being ranked first (World Bank, 2015; WTTC, 2016). Nature-based tourism is defined as ‘all forms of tourism that rely on or incorporate visitations to natural environments’ (Poonyth et al., 2001:4).

In Namibia nature-based tourism activities encompass consumptive activities such as hunting, and non-consumptive activities such as game viewing, bird watching, enjoying scenery and recreational angling (Millennium Challenge Account, 2013). The Zambezi Region in particular is well known for its variety of fish species and a number of lodges have specialised in angling tourism operations making a big contribution to the region’s tourism business (Tweddle & Hay, 2012; NNF, 2013). However, fish stocks along the Zambezi River and nearby floodplains are increasingly under pressure due to continuous overfishing (Weyl et al., 2010; Tafida et al., 2011; Daw et al., 2012; Porter & Orams, 2014; Tweddle et al., 2015). Dwindling fish stocks observed across African wetland areas stress the need to broaden the conceptual framework of CBNRM, which is currently focusing on wildlife management (Mbaiwa & Kolawole, 2013; Kolding & Zwieten, 2014). Angling tourism could become an integral component of an all-encompassing conservation strategy (Porter & Orams, 2014; Tweddle et al., 2015). However, considering fish as a primary income and protein source for local people, corresponding management strategies must seriously reflect on food security issues and fair benefit sharing (Wynberg & Hauck 2014; African Development Bank, 2015).

Recently, conservancies have started on taking responsibility for their fish stocks through developing local management plans (Kuriyama et al., 2015; Tweddle et al., 2015; Gupta et al., 2016). This article focuses on nature-based angling tourism in a developing conservancy in Namibia’s Zambezi Region by examining at its impact on food security and fish stock development. The article raises the following research questions:

1. Does nature-based tourism improve local livelihoods?

2. Does nature-based tourism have the potential to reduce overfishing?

Several studies have explored the impact of nature-based tourism on the development of rural communities in developing countries (Boudreaux & Nelson, 2011; Coria & Calfucura, 2012; Ngwira & Mbaiwa, 2013; Adiyia et al., 2017). In particular, Van der Duim et al. (2015) analysed the emergence of different institutional arrangements in Eastern Africa and Southern Africa. Several studies found that employment in the tourism sector has improved income and livelihoods (Lapeyre, 2011; Nyaupane & Poudel, 2011; Liu et al., 2012; Ahebwa & Van der Duim, 2013; Naidoo et al., 2016). However, there are strong regional differences (Snyman, 2013). Mbaiwa & Stronza (2010) pointed out that employments in nature-based tourism within CBNRM programmes are limited to some community members, even though indirect effects on livelihoods might be non-negligible (Gartner & Cukier, 2011; Adiyia et al., 2017). In Southern Africa, CBNRM is possibly the most important tool for driving community development (Fabricius et al., 2013; Ngwira & Mbaiwa, 2013). In view of the recently debated linkages between conservation and development, a number of studies analyse the particular impact of nature-based tourism on conservation (Nyaupane & Poudel, 2011; Liu et al., 2012; Ahebwa & Van der Duim, 2013; Mbaiwa & Kolawole, 2013). Coria & Calfucura (2012) stated that nature-based tourism has a great potential to improve biodiversity conservation. In this context, CBNRM is suggested as a tool to achieve sustainable management of natural resources (Ngwira & Mbaiwa, 2013; Naidoo et al., 2016). CBNRM strengthened communities’ incentive to conserve wildlife (Boudreaux & Nelson, 2011). Porter & Orams (2014) argued that tourism has the potential to alleviate the pressure on fish resources by reducing fishing effort and promoting conservation strategies that render the fish environment more attractive for tourism. However, even in a CBNRM context, individual benefits from environmentally destructive activities can exceed benefits from tourism, thus stressing the need for considering the existing power relations and benefit distribution schemes (Mosimane & Silva, 2015).

The majority of the mentioned studies used qualitative research tools to explore the linkages between nature-based tourism, livelihood improvement and conservation. The quantitative analysis of the trade-offs, however, has largely become possible via mathematical programming (Kaiser & Messer, 2011; Williams, 2013; Klapwijk et al., 2014). The tool is applied to livelihood analysis (Chen & Önal, 2012; Majeke et al., 2013; Maruod et al., 2013; Tesso et al., 2013; Delgado-Matas & Pukkala, 2014), to dietary requirements and crop mix optimisation (Jati et al., 2012; Mamat et al., 2012; Adeniyi & Adasina, 2014; Niragira et al., 2015), and to natural resource management issues (Conrad et al., 2012; Dissanayake et al., 2012; Galatsidas et al., 2013; Gavina et al., 2014; Winter et al., 2015). To our knowledge, the approach has not yet been employed in the context of CBNRM. Following the classical optimisation approach by Hazell & Norton (1986), we constructed a mathematical programming model of a CBNRM economy to simulate different natural resource management scenarios in the study region. Section 2 describes the study area and data collection procedure, Section 3 explains the model, simulation results are presented and discussed in Section 4 and then Section 5 concludes.

2. Data

2.1. Study area

Sikunga Conservancy is located in Namibia’s Zambezi Region (Figure 1). The topographically flat area has a semi-arid climate with a mean temperature of 22°C and a highly variable annual rainfall, amounting to 550 mm on average (Mendelsohn et al., 2006). Climatic conditions are subject to the occurrence of three seasons: dry, rain and flood seasons. The natural environment consists of grassland, floodplains and Mopane woodlands (Mendelsohn et al., 2006). Even with relative favourable geographical conditions, around 50% of the region’s population lack an adequate dietary intake, thus making food security a national concern (WFP, 2016).

Figure 1. Sikunga Conservancy in the Zambezi Region, Namibia.

Sikunga was gazetted as a communal conservancy in July 2009 (NNF, 2013). The conservancy is located 60 km east of the regional capital Katima Mulilo and the Zambezi River forms the northern border of the conservancy. The conservancy covers an area of 287 km² and hosts approximately 2000 residents living in six villages (NASCO, 2013). Agriculture, fishing, extraction of forest products and tourism employment are the primary sources of income. Two lodges offer recreational angling operating with marginal consumptive fish use (catch-and-release fishing). Captured fish is normally released at rates of around 90% (Sweeney et al., 2010). Fishing is principally unrestricted, although recently Sikunga started to establish fish protected areas to provide breeding zones for overexploited stocks (NASCO, 2013; NNF, 2013; Cooke et al., 2016).

2.2. Data collection

Model data were obtained from a structured survey of 200 households, which cover 45% of households, undertaken in September and October 2012 (Morton et al., 2016). For each of the six villages, households were randomly sampled from a comprehensive household list proportionally to village size. A household is defined as a group of conservancy residents sharing the same roof. The non-stratified sampling method guaranteed a representative choice across villages. Replacements for absent household representatives were picked from a list of oversampled households. When possible, the household head was asked to participate in the survey. The survey covered a broad range of the household’s socio-demographics, networks, socio-economic activities, income sources, time allocation, consumption and expenditure, use of fish and forest resources, as well as livestock and crop management. For all transactions, the agent itself, the origin and the destination of goods produced and traded were recorded. In addition, we used secondary data on regional fish stocks, intrinsic growth rates, regional prices, lodge employments and nutrient compositions of selected food items to parametrise the model (see Appendix 1).

3. The mathematical programming model

3.1. Model structure

The model illustrates the multiple activities observed in the study region through a set of constrained linear and non-linear equations. The method is appropriate for problems related to the efficient utilisation of scarce resources where multiple activities compete for the same resource and trade-offs have to be balanced (Hazell & Norton, 1986; Kaiser & Messer, 2011). Referring to Lipton’s (1968) theoretical framework ‘The theory of the optimising peasant’, the model’s objective function assumes that households optimise their collective well-being subject to a range of binding and non-binding constraints. The selection of certain activities may cause the exclusion of alternative ones, reflecting the context-specific opportunity costs incurred. The model is a simplified representation of an interrelated management problem; it can be seen as a planning tool to determine the efficient resource allocation maximising the social welfare of Sikunga Conservancy. Likewise, the collective optimisation strategy can be interpreted as simulating the outcome of a cooperative game, where individual households form a grand coalition (Britz et al., 2013). The analogy of cooperative games and a grand coalition parallels the CBNRM conception, aimed at the cooperative management of the community’s common pool resources (Mukwada & Manatsa, 2012).

The model is programmed in the General Algebraic Modeling System (GAMS)¹ and is calibrated to the survey and supplementary secondary data. Appendix 1 presents the key parameters used for model calibration. Data from the questionnaire deliver the majority of model input. FAO (2012) and WFP (2015a) provide data on nutritional compositions of food items. In addition, Lapeyre (2011) and Sweeney et al. (2010) provide information on lodge employment; data from FAO (2012), Welcomme et al. (2010) and Hay et al. (2002) were used for estimating fish growth rates and initial stocks.

The model depicts the economic structure of the study region for a single year. A special feature of the model is seasonal time allocation, which is highly relevant for communities in the Zambezi Region (Kamwi et al., 2015). Model activities and constraints are specified per month to allow households to adapt activities subject to changing seasonal conditions. The model distinguishes between the wet season (October to February), the flood season (March to May) and the dry season (June to September).

The mathematical model formulation is as follows:(1)

Subject to:(2) (3) (4)

with , , , and

where  = community’s social welfare maximum (net income),  = level of activity in month by household (decision variable),  = price of a unit of activity ,  = unit cost of activity ,  = amount of input needed to operate activity in month ,  = available supply of resource in month ,  = amount of food/resource needed for activity in month  and  = minimum level of food/resource in month .

3.2. Model activities and constraints

Consistent with the community’s livelihood system, considered activities cover crop farming, fishing, livestock and milk production, harvesting (firewood, thatching grass and reeds), shopping, hiring labour and off-farm activities. Represented households are involved in multiple activities at different times and to different levels of intensity. Sources of income vary from month to month as the activity mix changes in the course of the year. Produced outputs might be used for either subsistence consumption or marketing. Corresponding to the observed cropping technologies, cultivation can be done by manual hoeing or using oxen. Livestock (cattle) is largely grazed on common pool pasture and forest area; and milk production is a function of herd size, lactation length and forage availability. Fishing depends on seasonally varying effort, and forest products are collected and harvested from community lands. Additionally, the model considers certain off-farm activities inside and outside Sikunga, and finally it distinguishes different job types.

Boundaries on the households’ resource allocation are created via inequality conditions. Upper bounds state that the total use of a resource cannot exceed its availability and are defined for most inputs. Goods and services can be met either by subsistence production or obtained from the market. Lower bounds are set for cash and nutrition minimum requirements. Nutritional requirements for energy, protein, fat, vitamin A, iron and iodine are set to account for the impact of different production activities on food security (WFP, 2015b, 2015c).

3.3. Model scenarios

We specified two basic scenarios; scenario one represents the reference situation and assumes unrestricted resource extraction. Scenario two represents a well-managed CBNRM economy and presumes a restricted fish harvesting policy, which implicitly protects the angling tourism sector and mimics the sustainability objective of Sikunga Conservancy. Today, the Zambezi Region indicates high rates of food insecurity (WFP, 2016). Our data analysis showed that people living in Sikunga do not meet their minimum dietary needs; this observation matches the unrestricted reference scenario one.

In scenario two, the fish stock is specified by a biological growth function (Equation (5)) following Clark (2006):(5) where  = annual net growth of the fish resource in year ,  = state of the fish resource at time step ,  = intrinsic growth rate of fish and  = carrying capacity of the ecosystem. Calculations are made for one year ( = 1), given the basic stock (y = 0). Total extraction may not exceed annual sustainable fish harvest level (steady-state equilibrium condition, Equation (6)):(6) Total fish extraction is the sum of subsistence fishing and the here insignificant consumptive use of the angling tourism sector. Equation (7) calculates the state of the resource at the end of the year:(7) where  = initial fish stock.

Scenario two determines the allocation of scarce fish resources and considers angling tourism as a means to generate income and improve stocks. It also ensures the minimum nutritional standards suggested by WFP (2015b). The scenario thus links nature-based tourism to the CBNRM economy to reveal the impact on local livelihoods and the conservation of fish stocks.

3.4. Model properties and limitations

The specified model is comparative-static, taking a one-year perspective in our context. While the model is able to simulate different scenarios with respect to risk perceptions, biological and economic parameters, such as tourist numbers and fish growth, in a comparative-static manner, the model cannot predict the dynamics of a long-term development path and its feed-back loops. The model version presented here mainly focuses on the average community household and how it is affected by the angling tourism sector. Because of the partial analysis, economy-wide linkages cannot be captured. Because we aim to demonstrate the effect of a social optimum in a CBNRM economy compared with the unrestricted reference situation, we abstracted from the distribution of benefits to individual social groups, an issue analysed in a related study (Morton et al., 2016). One inherent limitation of the method is that real agents’ behaviour diverges from optimality because they do not behave rationally and independently (Klapwijk et al., 2014); more readily they respond to other agents’ actions, as Roettgers (2016) showed for resource management in Sikunga Conservancy.

4. Results and discussion

The solution of the reference scenario one shows diversified livelihood strategies in the study region. Households obtain income from farm and non-farm activities, which average NAD$24 471 (Table 1). Subsistence agriculture is the most important activity, dominated by low-input low-yield maize farming. Nearly every household in Sikunga grows maize, harvesting around 362 kg per hectare. On average, households cultivate 2.2 ha of land, and more than 60% of community inhabitants own livestock, predominantly cattle.

Table 1. Household characteristics and annual income sources for the livelihood strategies in Sikunga.

Household members (number)	4.65
Farmland (hectare)	2.2
Net income from agriculture (NAD$)	2 835
Net income from fish (NAD$)	3 830
Net income from forest products (NAD$)	2 641
Off-farm income (NAD$)	10 960
Welfare payments and remittances (NAD$)	4 205
Expenditure on consumption goods (NAD$)	10 354
Total income (NAD$)	24 471

Fish is a key resource for the community because it contributes almost 20% to households’ social welfare. One in four households spends time on fishing, either for food or cash income. Accordingly, our model shows the highest opportunity costs for fishing in the rainy season (NAD$15 per day) and flood season (NAD$33 per day). Households even shift their labour away from lodge employment to fishing in the flood season due to the catch increase per unit of effort. In the dry season, forest resource harvesting returns the highest benefit and opportunity costs are NAD$20 per day. Forest products are mainly harvested and collected for subsistence purposes and generate approximately 10% of household income. Our findings are supported by Naidoo et al. (2016), who found that the benefits from natural resource extraction activities generally exceed those of tourism within Namibian conservancies; and even if tourism is well integrated into a rural community, traditional sources of income still remain more important (Liu et al., 2012).

Referring to the current problem of overfishing along the Upper Zambezi (Weyl et al., 2010; Tweddle et al., 2015), scenario one quantifies a total community catch of 103 500 kg per year, which is more than 500 kg per household. Around 80% of households’ fish extraction is caught during the rain and flood seasons. Simasiku (2014) found that a household in the Zambezi Region catches on average 370 kg fish per year. In terms of area, our analysis reveals an extraction of 90 kg fish per hectare, which is within the range of over-exploitation observed for tropical natural production systems (Welcomme et al., 2010).

More specifically, results show that the community’s fish extraction is highly sensitive to households’ labour–leisure allocation. A sensitivity analysis for labour indicates that if households increased their overall labour time (decreased leisure), their fish catch would increase proportionally. Because fishing displays the highest opportunity cost in the rain and flood seasons, households with excess labour capacity behave rationally by allocating their time to fishing. Without employment in the regional angling tourism sector, communities’ fish extraction would increase by nearly 20% (16 750 kg). The absence of the angling tourism sector would thus have negative effects; households would need to compensate for their lost income from angling tourism, by reallocating their labour resources to subsistence fishing, thus further worsening the problem of overfishing.

Regarding food security, model results for scenario one reveal that around 50 to 70% of important macronutrients and micronutrients are missing in daily diets (Appendix 2). Consequently, malnutrition is an urgent present-day problem in the community; this finding is in line with WFP (2016) records. The results of scenario one furthermore show that households significantly depend on the exploitation of natural resources for subsistence consumption and cash income generation, a situation that is seen as a major threat to biodiversity and future livelihoods (Ahebwa & Van der Duim, 2013; Porter & Orams, 2014). Unrestricted fisheries result in decreasing catch rates as well as decreasing species richness (Tweddle et al., 2015), again negatively impacting the chance to develop a sustainable angling tourism business. Finally, it is worth noting that the model solution reveals the opportunity costs of nature conservation, thus providing important information for developing sustainable regional management plans and guide values for interventions. Sound interventions in the context of the CBNRM agenda may balance the issues of employment creation and sustainable resource management.

Conditions specified in scenario two (CBNRM with restricted fishery) first of all lead to a reallocation of the community’s resource use. In order to conserve fish, the model calculates that the necessary reduction of subsistence fishing makes up about 65%. At the community level this corresponds to a catch of 39 000 kg per year or 195 kg fish per household. The model indicates that a representative fishing lodge causes the very low fish damage of 1654 kg per year; this reflects the very low consumptive use of angling tourism. A performed sensitivity analysis stresses that a diminished ecological carrying capacity of 10% would require a corresponding extra decrease in catch numbers by 18%. Thus, external effects on the ecosystem, such as climate change or increased pollution, should continuously be monitored in order to adapt sustainable management strategies.

Particularly significant in terms of planning and resource management is that the results of scenario two show a significantly higher marginal value of fish in angling tourism compared with subsistence fishing. Table 2 presents a comparison between subsistence fishing and a representative angling tourism lodge in Sikunga with respect to extracted quantities and derived fish shadow prices: 1 kg fish used for recreational fishing has a shadow price of NAD$715, whereas fish caught by subsistence fishers has a shadow price of only NAD$10 per kg. Our results are comparable with Sweeney et al. (2010), who found that the value of fish caught by local fishers has a value less than 1% of the value of fish caught and not released by angling tourists in the Zambezi Region. Model results indicate angling tourists’ high willingness to pay for fish as compared with local consumers. This presents a remarkable opportunity to conserve over-exploited stocks. Our findings are supported by Mbaiwa & Stronza (2010), who found that the net benefit from land allocated to touristic purposes is much higher than that from land in agricultural use.

Table 2. Total fish extraction and shadow price.

	Total fish use (kg)	Shadow price (NAD$/kg)
Households	39 000	10
Fishing lodge	1 615	715

The reallocation of land and labour obtained in scenario two has a positive impact on agricultural diversification. Results reveal that maize production is reduced by 30% in favour of protein-rich cowpea cultivation.² The shift is largely driven by the need to compensate the protein intake lost through the reduction of fish catches. More specifically, around 20% of the community’s fish intake (nearly 10 000 kg) is substituted by cowpea, which now covers one-third of farmers’ fields and shows an average yield of 148 kg per hectare. Households have to diversify crop production to satisfy their nutritional requirements. We find that as long as appropriate dietary substitutes for fish are available, the overall welfare can be maintained despite restrictions on fishing. Finally, we could derive the opportunity costs of an adequate diet; our simulation reveals that a nutritious food intake for an adult person (per day) would incur opportunity costs of about NAD$11. The derived costs indicate the likely dimension of a compensation scheme in favour of local fishers conserving fish stocks for the benefit of angling tourists. Model results could thus guide the negotiation of a necessary benefit distribution scheme to balance the restrictions set on subsistence fishing by the commitment to conservation.

Associated with a tourism-based development strategy, households shift their labour from fishing to angling tourism employment. As a consequence, past labour shortages faced by Sikunga’s fishing lodges mainly in the flood season can be better balanced. At the same time, the reallocation of employment towards angling tourism highlights the potential of the sector to support households who are required to reduce their fishing activity. This is in line with the findings of Mbaiwa & Stronza (2010), who argue that established tourism in Botswana has become the main livelihood activity in rural communities, replacing many traditional livelihood activities that damaged the environment. Promoting alternative livelihoods instead of fishing is one response to declining fish stocks (Porter & Orams, 2014). However, according to our model results only 6% of community households directly benefit from employment in the angling tourism sector. The expansion of nature-based tourism is promoted in many countries across sub-Saharan Africa, although this issue needs to be reflected critically because tourism-led development may lead to dependency, cultural disruption, socio-economic inequalities and revenue leakages (Sarrinen & Rogerson, 2013). Research from the Philippines demonstrated that fishers may even be against tourism development for fear of exclusion (Fabinyi, 2010). Nevertheless, angling tourism in Sikunga contributes approximately 20% to the community’s social welfare (NAD$1 120 000), an estimate that is supported by Adiyia et al. (2017) regarding off-farm activities. In addition, the model results indicate the importance of a holistic conservation strategy because households may allocate excess family labour to the harvesting of forestry resources. If activities are not monitored, this could lead to higher levels of deforestation, possibly negatively affecting regions attractiveness as a nature-based tourism destination.

Compared with the reference scenario one, the results of scenario two show nearly unchanged household incomes, stressing that households are not worsened as a consequence of the more restrictive fishery management. In the long run, households might even be better off through advanced fish species richness and the region’s increased attractiveness for angling tourism. A final sensitivity analysis points out that an additional fishing lodge in the Sikunga Conservancy would increase the overall welfare by 14% (NAD$685 000), and would decrease the community’s fish catch by 13%. However, this carries the risk of altering the social and cultural identity of rural communities because rural households exchange their subsistence activities for a cash economy (Kuenzi & McNeely, 2008). As a consequence, model results should be interpreted carefully because economy-wide trade-offs, future disruption of tourism, changed tourist preferences (Roe et al., 2004; Kuenzi & McNeely, 2008) and, finally, complex environmental–human interactions cannot be easily quantified by an equation-based comparative-static model.

5. Conclusion

We constructed a comparative-static mathematical programming model to assess the impact of nature-based tourism on community development and conservation. Two scenarios were specified: the reference scenario one, representing current weak resource management; and the hypothetical scenario two, representing a well-managed CBNRM economy with restricted fishing effort. Most notably, the model estimates the opportunity costs of sustainably managed fish stocks. Computed opportunity costs portray the losses incurred by local fishers as a result of the imposed restrictions on fishing effort; in other words, alternative incomes and a restructured farming system would be necessary to secure a healthy diet. The information on opportunity costs might guide negotiations on fair benefit-sharing agreements between the community and lodge operators.

Today, angling tourism in Sikunga is integrated into households’ livelihood system, and the community partly benefits from employment in the sector. However, most households still benefit more from natural resource exploitation activities, especially from unsustainable subsistence fishing, which considerably contributes to the regional problem of overfishing, and simultaneously threatens livelihoods, the angling tourism sector and biodiversity. As a consequence, the unrestricted fishery represented by scenario one indicates only a very weak, eventually negative, impact of nature-based tourism on conservation. On the contrary, representing a sustainably managed fishery, scenario two clearly reveals that angling tourism offers an alternative livelihood strategy to fishers, and moreover has a high potential to support households who are required to drive down their fishing activity to sustainable levels. In order to reduce overfishing, the community has to reallocate labour and land substantially; at the same time, this increases community’s agricultural diversification to meet dietary needs. Taking the partial analysis, economic welfare remains unchanged, and nature-based tourism has a high potential to reduce overfishing; however, positive side-effects such as diversification and sustainable stock management would bring about likely positive effects in the long run. Hence, research questions (1) and (2) can be answered positively: nature-based tourism has the potential to improve livelihoods and stop over-fishing if embedded in a socio-economic system, in which benefits are distributed fairly. Both impacts are interrelated, as has been demonstrated by model simulations.

The approach of mathematical programming can also be used in areas of wildlife protection and forest resource conservation in general. Therefore, it could be applied to other forms of nature-based tourism, such as game viewing and nature touring. A useful property is the assessment of the context-specific opportunity costs of conservation, essential information necessary for compensating traditional resource users for their incurred benefit loss. In addition, our model considers seasonality and therefore can be used to assess seasonality dependent livelihoods. In terms of planning and resource management, the derived economic and ecological values can provide important information for developing sustainable regional management plans, informing beneficiaries about fair compensation values and economic incentive costs to conserve biodiversity. The model could be extended to include additional macronutrients and micronutrients, as well as alternative protein sources such as chicken, to allow for a more in-depth nutritional analysis and a corresponding opportunity cost assessment.

Further research is needed to assess the long-term impacts of nature-based tourism. This requires the development of dynamic modelling tools such as agent-based models. Finally, a model depicting the linkages between product and factor markets and exploring the impact of external relations, such as a computable general equilibrium model does, would provide a more complete picture and could balance some of the limitations of a partial activity model.

Acknowledgements

This article has been written in the context of the project ‘SASSCAL – Southern African Service Science Centre for Climate Change and Adaptive Land Management’ (http://www.sasscal.org/). First of all, the authors express thanks to the people living in Sikunga Conservancy for their continuous cooperation. They thank Huon Morton, who provided essential comments on the work in various stages. They also thank Tererai Msakwa and Dr Matthias Beyer for further comments and data.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The project is funded by the German Ministry of Education and Research (BMBF) [01LG1201H].

Notes

1 GAMS is designed for the construction and solution of large and complex mathematical programming models. Various kinds of economic models including linear and non-linear optimisation as well as equilibrium modelling can be solved using GAMS. Revealed marginal values (shadow prices) by programming runs are a special software feature (Brooke et al., 1992).

2 Growing cowpea is currently marginal in the Sikunga Conservancy.

Appendix 1. Key model parameters

Table

Parameter	Value	Unit	References
Land
Sikunga Conservancy	287	km^2	NASCO (2013)
Grassland in Sikunga	208	km^2
Wetland in Sikunga	11	km^2
Woodland in Sikunga	63	km^2
Farmland in Sikunga	5	km^2
Crops
Price maize	1.75	NAD$/kg	Parviainen (2012)
Price cowpea	2.25	NAD$/kg	Parviainen (2012)
Yield maize	362	kg/ha
Yield cowpea	150	kg/ha	GRN (2008)
Labour maize (hoeing, ploughing, planting), October to December	7	Days/ha	FAO (2010), Mendelsohn et al. (2006) and Purvis (2002)
Labour maize (weeding), January to April	6.75	Days/ha	FAO (2010), Mendelsohn et al. (2006) and Purvis (2002)
Labour maize (harvesting), April	14	Days/ha	FAO (2010), Mendelsohn et al. (2006) and Purvis (2002)
Labour cowpea (hoeing, ploughing, planting), October to December	6.22	Days/ha	FAO (2010), Mendelsohn et al. (2006) and Purvis (2002)
Labour cowpea (weeding), January to April	4.26	Days/ha	FAO (2010), Mendelsohn et al. (2006) and Purvis (2002)
Labour cowpea (harvesting), March to April	10.24	Days/ha	FAO (2010), Mendelsohn et al. (2006) and Purvis (2002)
Cost of seeds	10	NAD$/kg	GRN (2009)
Seed planting rate maize	6	kg/ha	FAO (2010)
Seed planting rate cowpea	5.1	kg/ha	FAO (2010)
Livestock (cattle)
Price bull	4 000	NAD$/unit	Parviainen (2012)
Price oxen	2 466	NAD$/unit	Parviainen (2012)
Price cow	1 077	NAD$/unit	Parviainen (2012)
Price heifer	850	NAD$/unit	Parviainen (2012)
Price calf	550	NAD$/unit	Parviainen (2012)
Off-take rate	10	Percent	Mendelsohn et al. (2006)
Labour	1.52	Days/month/unit
Draft animal power for crop production (ploughing)	4	Days/hectare	Mendelsohn et al. (2006)
Charge for draft animal power	175	NAD$/day	Aagaard (2010)
Grassland requirement	12	ha/month/unit	Mendelsohn et al. (2006)
Cost of veterinary	80	NAD$/unit	Parviainen (2012)
Share of bulls (per herd)	6	Percent	Marius et al. (2012)
Share of oxen (per herd)	18	Percent	Marius et al. (2012)
Share of milk cows (per herd)	23	Percent	Marius et al. (2012)
Share of dry cows (per herd)	11	Percent	Marius et al. (2012)
Share of heifers (per herd)	23	Percent	Marius et al. (2012)
Share of calves (per herd)	19	Percent	Marius et al. (2012)
Stock (community)	2 205	Number
Price of milk	5	NAD$/litre	Marius et al. (2012)
Milk productivity of cows, wet season	4.83	Litre/cow/day	Gulelat (2002)
Milk productivity of cows, dry season	1.27	Litre/cow/day	Gulelat (2002)
Milking frequency by farmers	1.5	Hours/cow/day	Mapekula et al. (2009) and Somda et al. (2005)
Labour milk production, October to May	0.04 (19)	Days (min.)/litre	Gulelat (2002), Mapekula et al. (2009) and Somda et al. (2005)
Labour milk production, June to September	0.1 (48)	Days (min.)/litre	Gulelat (2002), Mapekula et al. (2009) and Somda et al. (2005)
Standard lactation length of cows	305	Days	Syrstad (1993)
Forest resources
Price of firewood	0.33	NAD$/kg	Palmer & MacGregor (2008)
Price of thatching grass	0.49	NAD$/kg
Price of reeds	0.72	NAD$/kg
Labour firewood, flood season	0.04 (18)	Days (min.)/kg
Labour firewood, dry season	0.02 (8)	Days (min.)/kg
Labour firewood, rain season	0.07 (35)	Days (min.)/kg
Labour thatching grass, flood season	0.07 (35)	Days (min.)/kg
Labour thatching grass, dry season	0.03 (16)	Days (min.)/kg
Labour thatching grass, rain season	0.14 (67)	Days (min.)/kg
Labour reeds, flood season	0.1 (49)	Days (min.)/kg
Labour reeds, dry season	0.05 (22)	Days (min.)/kg
Labour reeds, rain season	0.2 (97)	Days (min.)/kg
Biomass in north-east Namibia	2 300	kg/ha	Verlinden & Laamanen (2006)
Forest biomass in north-east Namibia	8 824	kg/ha	Barnes et al. (2005), GRN (2003), Palmer & MacGregor (2008) and Parviainen (2012)
Fish resources
Price	10	NAD$/kg
Labour, flood season	0.24(115)	Days (min.)/kg
Labour, dry season	1.03 (494)	Days (min.)/kg
Labour, rain season	0.52 (250)	Days (min.)/kg
Weight	200	Gram/fish	Simasiku (2014)
Annual growth rate	6.2	Percent	FAO (2013)
Carrying capacity in Sikunga Conservancy (Zambezi Region); yield for tropical natural fish production systems	1 814 500	kg	Hay et al. (2002) and Welcomme et al. (2010)
Fish stock in Sikunga Conservancy (Zambezi Region)	950 000	kg	Hay et al. (2002)
Hired labour and off-farm employment
Formal employment wage rate	46.1	NAD$/day
National farm workers’ minimum wage	23	NAD$/day	Parviainen (2012)
Hired in labour	98	Labourers
Formal employments	84	Labourers
Fishing lodge
Salary manager	8 512	NAD$/month	Sweeney et al. (2010)
Salary skilled labour	3 304	NAD$/month	Sweeney et al. (2010)
Salary unskilled labour	761	NAD$/month	Sweeney et al. (2010)
Salary part-time labour	400	NAD$/month	Sweeney et al. (2010)
Salary guide labour (skilled and unskilled)	3 023	NAD$/month	Sweeney et al. (2010)
Requirement manager	2	Employee	Sweeney et al. (2010)
Requirement skilled labour	17	Employee	Sweeney et al. (2010)
Requirement unskilled labour	9	Employee	Sweeney et al. (2010)
Requirement part-time labour	1	Employee	Sweeney et al. (2010)
Requirement guide labour (skilled)	5	Employee	Lapeyre (2011) and Sweeney et al. (2010)
Requirement guide labour (unskilled)	1	Employee	Lapeyre (2011) and Sweeney et al. (2010)
Fish catch (caught and not released by angling tourists)	1 218	kg/lodge/year	Sweeney et al. (2010)
Damage rate (released fish not survive treatment by angling tourists)	2.5	Percent	Sweeney et al. (2010)
Household requirements
Price rice	9.5	NAD$/kg
Price noodles	10.5	NAD$/kg
Price potato	7.86	NAD$/kg
Price of bread	8.3	NAD$/each
Price beef	14.33	NAD$/kg
Price chicken	18.81	NAD$/kg
Price egg	1	NAD$/each
Energy, protein, fat, vitamin A, iron, iodine of maize	See reference	kcal, g, μg, mg	FAO (2012) and WFP (2015a)
Energy, protein, fat, vitamin A, iron, iodine of cowpea	See reference	kcal, g, μg, mg	FAO (2012) and WFP (2015a)
Energy, protein, fat, vitamin A, iron, iodine of fish	See reference	kcal, g, μg, mg	WFP (2015a)
Energy, protein, fat, vitamin A, iron, iodine of milk	See reference	kcal, g, μg, mg	WFP (2015a)
Energy, protein, fat, vitamin A, iron, iodine of rice	See reference	kcal, g, μg, mg	FAO (2012) and WFP (2015a)
Energy, protein, fat, Vitamin A, iron, iodine of noodles	See reference	kcal, g, μg, mg	FAO (2012) and WFP (2015a)
Energy, protein, fat, vitamin A, iron, iodine of potato	See reference	kcal, g, μg, mg	FAO (2012) and WFP (2015a)
Energy, protein, fat, vitamin A, iron, iodine of bread	See reference	kcal, g, μg, mg	FAO (2012) and WFP (2015a)
Energy, protein, fat, vitamin A, iron, iodine of beef	See reference	kcal, g, μg, mg	FAO (2012) and WFP (2015a)
Energy, protein, fat, vitamin A, iron, iodine of chicken	See reference	kcal, g, μg, mg	FAO (2012) and WFP (2015a)
Energy, protein, fat, vitamin A, iron, iodine of egg	See reference	kcal, g, μg, mg	FAO (2012) and WFP (2015a)
Nutritional requirement energy	2 100	kcal/adult/day	WFP (2015b)
Nutritional requirement protein	52	g/adult/day	WFP (2015b)
Nutritional requirement fat	40	g/adult/day	WFP (2015b)
Nutritional requirement vitamin A	500	mcg/adult/day	WFP (2015b)
Nutritional requirement iron	22	mg/adult/day	WFP (2015b)
Nutritional requirement iodine	150	mcg/adult/day	WFP (2015b)
Thatching grass requirement	322	kg/HH/year
Reeds requirement	305	kg/HH/year
Firewood requirement	1 445	kg/HH/year
Purchases thatching grass (community)	36 000	kg/year
Purchases reeds (community)	20 000	kg/year
Purchases firewood (community)	2 000	kg/year
Purchases maize	239 429	kg/year
Purchases fish	5 437	kg/year
Purchases milk	2 892	Litres/year
Purchases rice	5 732	kg/year
Purchases noodles	2 571	kg/year
Purchases potato	1 654	kg/year
Purchases bread	10 084	Number/year
Purchases beef	2 324	kg/year
Purchases chicken	2 344	kg/year
Purchases eggs	3 045	Number/year

Note: Min., minutes; HH, household.

Appendix 2. Dietary characteristics per adult per day in Sikunga

Table

Macronutrients and micronutrients	Intake	Minimum requirement
Energy (kcal)	1 049	2 100
Protein (g)	51	52
Fat (g)	15	40
Vitamin A (μg)	160	500
Iron (g)	10	22
Iodine (mg)	59	150