Shocks and rural livelihoods in the Okavango Delta, Botswana

Abstract

This paper describes the impacts that three shocks in the Okavango Delta, Botswana, have had on rural livelihoods: the desiccation of river channels, animal diseases, and HIV/AIDS. Primary data was collected from five study areas, using formal questionnaire interviews and focus group discussions. The paper reveals the adverse effects on rural livelihoods. It describes the way households have been exposed to poverty and vulnerability and the various ways they have coped or adapted, such as by re-allocating their labour, liquidating their assets to cover medical expenses and funeral costs, reducing the area ploughed for crops, hiring labour, digging wells and switching from flood recession agriculture to dryland farming. The Botswana government has provided safety nets to help households cope, but this paper recommends that people's responses to these shocks should be taken into account in future policy and programme formulation.

1. INTRODUCTION

Households of the Okavango Delta in Botswana are prone to risk as they depend on natural resources, which in turn depend on erratic rainfall and the variable water flow of the Okavango River (Wolski et al., 2006). They are often adversely affected by shocks, defined as large, unpredictable, irregular disturbances, as distinct from stresses, which are smaller, predictable, regular and sometimes continuous (Pearce et al., 1989). Shocks have a greater adverse effect on social, economic and environmental sustainability than stresses, since they destroy assets (Ellis, 2000; Wisner et al., 2004).

Households' responses to shocks include various ways of coping and adapting. According to Ellis (2000), coping strategies are ‘short-term responses to an unplanned crisis’ and are a form of ex post diversification. Coping strategies are not the same as adaptive strategies, defined as long-term responses to shocks aimed at improving livelihood security and poverty. According to Holzmann and Jorgensen (2000), poor households are more vulnerable to shocks since they are more exposed and have fewer instruments for risk management. Vulnerable households tend to be ‘defenceless’ because they are economically and socially dependent.

Knowing how households respond to shocks is of critical importance since it reveals what the poor can do to help themselves in these circumstances. There has been only limited research into these responses in the Okavango Delta (Kgathi et al., 2004), as is the case with research into such responses worldwide, because they tend to be ‘ephemeral and change quickly’ and therefore ‘remain unnoticed and understudied’ (Wisner et al., 2004: 112). It is useful to understand people's response strategies so as to guide the formulation of interventions aimed at helping the poor.

The study described in this paper examined the impacts of shocks on rural livelihoods in the Delta and how households respond to them. Its specific research objectives were to

• (1) analyse the socio-economic impacts of shocks to rural livelihoods in the Delta;

• (2) examine households' adaptive strategies in response to these shocks; and finally

• (3) determine the policy implications of the research findings.

The following section outlines the sustainable livelihoods framework, which is used as an organising tool for the discussion. Section 3 describes the study areas and Section 4 outlines the methodology for the study. Section 5 presents the research results. Section 6 summarises the discussion and considers implications for policy.

2. THE SUSTAINABLE LIVELIHOODS FRAMEWORK

The sustainable rural livelihoods framework links the broader socio-economic components of household assets, livelihood activities, outcomes of livelihood activities, and factors mediating access to livelihood activities (Scoones, 1998; Ellis, 2000; Farrington et al., 2004). According to Ellis (2000: 19), ‘a livelihood comprises the assets (natural, physical, human, financial and social capital), the activities, and the access to these (mediated by institutions and social relations) that together determine the living gained by the individual or household’. Activities are strategies or various ways in which households generate their livelihoods (Ellis, 2000). The means of securing livelihoods are usually diversified, and can be natural resource or non-natural resource based (Ellis, 2000). Examples of natural resource based activities in the Delta are arable farming, livestock farming, collection of veld products, basket-making, fishing and community based tourism. Non-natural resource based activities include formal employment, rural trade and social protection programmes (see Figure 1; also Kgathi et al., 2004).

Figure 1: A framework for livelihood analysis of rural households (Slightly adapted from Ellis (2000) for application to the Okavango Delta region): Notes: CBNRM – Community Based Natural Resources Management; ITK – Indigenous Technical Knowledge (ITK)

Households use a combination of assets to secure livelihoods. These comprise natural, physical, human, financial and social capital (Ellis, 2000). Different forms of livelihood activities require access to different forms of capital in order to generate a livelihood. Access to assets and adoption of livelihood strategies are mediated by social relations and institutions or organisations and are also affected by trends and shocks (Ellis, 2000; Figure 1). While institutions and social relations are factors endogenous to the household, trends and shocks are exogenous (Ellis, 2000; Farrington et al., 2004). Scholars of the new institutional economics define institutions as ‘rules, enforcement characteristics of rules, and norms of behaviour that structure the repeated interaction’ (North, 1989: 1321). Social relations are the way individuals and households are positioned in society. Such positioning is determined by factors such as religion, class, age, ethnicity and gender (Ellis, 2000). Trends include population change, economic growth, technological change and change in development policies, whereas shocks include drought, diseases (of crops, animals, and humans), pests, floods and civil war (Ellis, 2000).

According to Dercon (2001), the impacts of a shock will depend on its frequency and intensity. Long lasting and persistent shocks make coping very difficult (Dercon, 2001), hence they have greater adverse impacts on rural livelihoods. In the Okavango Delta, shocks include animal and human diseases, recurrent droughts and changing flooding patterns. The flooding of the Delta, which depends on rainfall in the catchment area, influences access to natural capital such as land for flood recession agriculture and water for human and animal consumption (Kgathi et al., 2005; Wolski et al., 2006). Desiccation of river channels and flooding are common shocks in the Delta and they result in adverse impacts on rural livelihoods. Droughts deplete assets such as natural and financial capital, hence adversely affecting the diversification of livelihood portfolios.

3. THE STUDY AREAS

Though this study focused on certain specific villages, the Okavango Delta – situated in Ngamiland District in northern Botswana – was the general study area. According to the 2001 census, the population of Ngamiland was estimated at 125 000, of which 76 per cent was based in the Delta area. This Delta, a globally renowned Ramsar site, is an important feature of Ngamiland. It is estimated that 95 per cent of the Okavango River water that flows into this alluvial fan is generated in the upper headwaters in Angola. The Delta and its wetland are a great attraction for tourism, the second most important economic activity in Botswana after diamonds (Kgathi et al., 2005). Botswana, with a per capita income of US$3430 (2003) has one of the fastest growing economies in Africa (World Bank, 2005), though its performance in human development has declined since the 1990s as a result of the HIV/AIDS pandemic.

The specific study areas, situated in both the lower and upper parts of the Delta, were the villages of Shorobe, Etsha 6, Seronga, Gudigwa and Sehitwa (Figure 2), whose main livelihood activities are arable farming, social protection programmes, livestock agriculture and formal employment (Kgathi et al., 2005). Table 1 shows that the 2001 population in these villages ranged from 732 in Gudigwa to 2629 in Etsha 6.

Figure 2: Map of the Okavango Delta showing the study areas (Source: Fieldwork, 2003)

Table 1: Population, number of households and sample sizes in the study areas

	Etsha 6	Gudigwa	Sehitwa	Shorobe	Seronga
Population (2001)	2629	732	1478	955	1641
Number of households (2001)	580	112	416	201	364
Number of households interviewed	36	18	28	17	30
Source: CSO, 2002a.

4. METHODOLOGY

4.1 Data collection

The primary data was collected from the five study areas, using formal questionnaire interviews, informal interviews and focus group discussions. The formal interview questionnaire survey was undertaken in August, September and October 2003, after a pre-test had been done in the village of Tubu, near the town of Gumare (Figure 2). A detailed questionnaire, with closed and open ended questions, was used to collect household data. Since the level of literacy in the study areas was low, individual questionnaires were administered to households face-to-face in Setswana, the national language. The number of households interviewed in the study areas was 129, the numbers for the individual study areas being as shown in Table 1. A two-stage stratified random sampling method was used to select the sample. A random sample of enumerations was selected, and within each enumeration area a systematic random sample was selected using a list drawn for the 2001 Census. The sampling unit was a household, and those interviewed were heads of households. The households were categorised into poor, middle and rich peasants, based on the assets they held and the income earned.

Information was also collected through six focus group discussions in the villages of Gudigwa, Sehitwa, Shorobe and Seronga from key informants from different parts of the village who shared similar characteristics (such as age, gender and social standing). A set of interview instructions was used to steer the interview process and help focus discussion on key themes. Each group discussion had a facilitator and a transcriber. Group discussions lasted for about two hours, depending on the dynamics of the group. The issues covered varied with the village, but were generally: 1) livelihood activities in the village and their ranking; 2) the drying up of river channels; 3) the killing of cattle to eradicate cattle lung disease; 4) drought; 5) damage of crops by wildlife; and 6) HIV/AIDS.

4.2 Data analysis

The quantitative household questionnaire data was coded, cleaned up and entered for analysis in Statistical Package for the Social Sciences (SPSS) (12.1). Analysis was in the form of frequency distributions, measures of central tendency, and variation. In addition, the Chi Square Test of Independence was used to test for association between variables. Qualitative data from informal interviews, focus group discussions and open ended questions were analysed thematically (Weiss, 1994). This involved reduction, display and summarising of data into themes.

5. RESULTS

5.1 Overview of the patterns of livelihoods

In recent years, new livelihood activities such as social welfare schemes, tourism, sale of veld products, and new employment opportunities have emerged in Ngamiland. The traditional livelihood activities of agriculture (both arable and livestock), fishing, and hunting and gathering still play a significant role, but their importance has been reduced, partly because the land they depended on (and its resources) is now used for other activities such as tourism and conservation. There has been a marked increase in the development of tourism in the Delta. The policies and regulations for its promotion have curtailed access to veld products and hunting, with some increase in commercial opportunities in craftwork, cultural tourism, tracking and guide-work. In particular, the poor socio-economic groups have been more negatively affected by these changes as they are more dependent on the collection of veld products and hunting (Kgathi et al., 2004). There is therefore an increase in vulnerability to shocks in the Delta and the rest of Ngamiland. A substantial number of households are now increasingly relying on social protection programmes such as food rations and the old age pension, which act as important safety nets.

Arable agriculture was ranked by 23 per cent of the households as the first most important livelihood activity (column A in Table 2). The second most important livelihood activity in the first most important category (column A in the table) was formal employment (mainly tourism related), which was ranked by 21 per cent of the households. This was followed by livestock farming and government protection programmes (excluding drought relief programme), which were both ranked third by 9 per cent of the households. However, when the ranking of the livelihood activities of the second most important category (column B) is examined, arable agriculture still takes a lead as the most frequently ranked livelihood activity (17 per cent of the households), followed by government protection programmes (11 per cent) and livestock farming (9 per cent).

Table 2: Livelihood activities of the Okavango Delta ranked by how households accord importance

	A	B	C
Livelihood activity	% of h/h ranking livelihoods as first n = 129	% of h/h ranking livelihoods as second n = 129	% of h/h ranking livelihoods as third n = 129	D Average % of a, b, and c
Arable agriculture	23.3	17.1	7.0	15.8
Livestock farming	9.3	8.5	5.4	7.7
Formal employment	20.9	7.0	0.8	9.6
Basket making	1.6	7.0	4.7	4.4
Remittances	7.0	3.1	4.1	4.7
Government assistance	9.3	10.9	9.3	9.8
Drought relief projects	4.7	1.6	5.4	3.9
Beer brewing	3.9	10.1	1.6	5.2
Others	11.6	5.4	7.8	8.3
Note: These columns do not add up to 100% because households mentioned more than livelihood strategy. However, the percentages are generally low because some of the households did not do the ranking of livelihood activities.

However, social protection programmes became the most frequently ranked activity among the activities of the third most important category (column C). Livestock farming and drought relief were ranked the third most important livelihood activities. Although livestock farming was once the most important livelihood activity in Ngamiland before the eradication of livestock in 1995/96, it is now significantly less important (Fidzani et al., 1999). When the frequencies of choices in the first, second, and third categories (columns A, B and C) are added together and averaged, arable agriculture is still perceived to be the most important livelihood activity (16 per cent) followed by government protection programmes (9.8 per cent) and formal employment (9.6 per cent) (Table 2). These frequencies should be considered the most accurate as they have been averaged from the first, second and third most important choices.

Other studies have also revealed that arable agriculture, formal employment and social protection programmes are the most important livelihood activities in Ngamiland (Fidzani et al., 1999). A comprehensive survey undertaken in Ngamiland in 1997 revealed that arable farming, salaries and social protection programmes were ranked as the first most important sources of livelihood in Ngamiland by 37 per cent, 17 per cent and 15 per cent of the households, respectively (Fidzani et al., 1999). The Every River Has Its People (ERHIP) survey undertaken in 22 villages of the Delta revealed that arable agriculture was the most important livelihood activity in these villages, ranked by 39 per cent of the households as the first most important livelihood activity, followed by formal employment, ranked by 18 per cent of the households.

5.2 Impacts of shocks and household responses

Common shocks in the Delta are animal and human diseases, recurrent droughts and changing flood distribution patterns. Animal diseases include cattle lung disease, contagious bovine pleuropneumonia (CBPP) and African animal trypanosomiasis (nagana), and human diseases include HIV/AIDS and malaria. In Ngamiland, droughts were experienced in the periods 1964–1965, 1982–1988 and 1995. During the 1982–1988 droughts, 28 per cent of the cattle died. Figure 3 shows that the average crop yields for the period 1982–1986 were lower than 100 kg/ha. In the study areas, 73 per cent and 65 per cent of the households perceived drought to be a constraint on arable and livestock agriculture, respectively. Drought mainly affects dryland farming, which is dependent on rainfall, but does not affect molapo farming (flood recession farming) which is more dependent on the Okavango River floods. There are more farmers involved in dryland farming than in molapo farming in Ngamiland. For instance, in 1997 and 1998 dryland farmers accounted for 73 per cent and 84 per cent of all the farmers, respectively, whereas molapo farmers accounted for the remaining 27 per cent and 16 per cent, respectively (Kgathi et al., 2005).

Figure 3: Maize yield variations in relation to precipitation in Ngamiland (Source: Bendsen & Meyer, 2003)

Crops being damaged by wildlife and livestock killed by predators were two of the main constraints on livelihood security in the study areas. The number of cases of livestock killings in Ngamiland ranged from 326 in 2000 to 1200 in 2003, and the cases of crop damage from 121 in 1999 to 196 in 2003. Elephants are the main cause of crop damage. There is a general concern that there are too many elephants in the North West District. This is mainly attributed to the international embargo imposed by the Convention on International Trade in Endangered Species (CITES) on elephant products in local and international markets (Darkoh & Mbaiwa, 2002). The following sections analyse in detail three particular shocks to rural livelihoods and the way households respond to them: the desiccation of river channels, cattle lung disease (contagious bovine pleuropneumonia), and the HIV/AIDS pandemic.

5.2.1 Desiccation of river channels

Although the Okavango River Basin is relatively pristine, a number of the Delta channels and floodplains have dried up at various periods (Figure 4), particularly in the past 30 years. This has adversely affected the livelihoods of those who depend on river water and the associated resources. Two reasons for the drying up of river channels have been identified: shifts in flow distribution patterns, caused by sedimentation, and a general decline in the inflow of the Okavango River and local rainfall observed between the 1970s and 2000s (Wolski & Murray-Hudson, 2006). The sedimentation process, caused by the low topographic gradient, may lead to blockages and eventually to avulsions – ‘the diversion of a channel to a new course at a lower elevation in the floodplain’ (McCarthy et al., 1992), a process that could also be triggered by tectonics. The desiccation of river channels and the formation of new ones is a natural process caused by the accumulation of sediments, resulting in the growth of vegetation that causes blockages and avulsions. Changes in flood distribution are of two kinds, causing two kinds of desiccation. Sedimentation makes the flow switch to another part of the river system, causing some parts to dry up. With a reduction in inflow and local rainfall, however, the flood water is still distributed throughout the system, but in reduced quantities.

Figure 4: Map of the Okavango Delta showing channels and flood plains that have dried up

The Thaoge River, which used to flow to Lake Ngami, became desiccated in the 1880s because of blockages caused by floating plant debris, leading to the formation of the Nqoga channel in the early 1900s (McCarthy et al., 1992). The Gomoti and Santantadibe rivers, which used to flow to the Thamalakane River, have dried up, adversely affecting the supply of water to Shorobe village. The Santantadibe last flowed to the Thamalakane in the 1980s, and the flow of the Gomoti has been sporadic since the 1950s. The desiccation of these rivers is attributed partly to the blockage of the Mboroga River, which has affected the flow of the water to the Gomoti and Santantadibe River channels (Bernard & Moetapele, 2003), and partly to the general decline of flood extent observed between the 1970s and 1990s (Wolski et al., 2003).

A number of rivers and floodplains dried up in the 1970s and 1980s because of the general decline in inflow and local rainfall: the Nhabe River, which used to flow to Lake Ngami near the study area of Sehitwa, the floodplains that supplied the study area of Gudigwa, and the Thaoge River that supplied the villages of Gumare, Nokaneng and Tsau (near the study area of Etsha 6).

5.2.2 Impacts of desiccation

The desiccation of river channels and floodplains has adversely affected access to water resources (for humans and livestock), molapo farming, fishing and the harvesting of veld products in the study areas. In Shorobe and Sehitwa, the drying up of the nearby rivers has resulted in a serious shortage of water for both humans and livestock. Edible veld products (tsita, tswii, mokotsumo, mochaba) and others used for various purposes (for example mokola palm, which is used for making baskets) became depleted, and can nowadays be described as economically scarce.

Focus group discussions clearly revealed that the drying up of the river channels in Shorobe was a disaster with more dramatic impacts than the 1996 eradication of cattle in the area, because only a few households owned cattle and those affected were partly compensated, whereas the desiccation of the river channels has affected everyone and there has not been subsequent compensation for the losses.

5.2.3 Adaptive strategies

The people whose livelihoods have been affected by the drying up of the river channels and floodplains have coped and adapted in various ways. As the river dried up, some of the households dug wells. Others had no means to cope and adapt and no access to safety nets, and so became poor and vulnerable. Our survey revealed the main strategies adopted by Sehitwa households as a reaction to the drying up of the Nhabe River and Lake Ngami, rated in the following order of importance: 1) digging wells (28 per cent), switching from molapo farming to dryland farming (28 per cent), and depending on the government for water supply (20 per cent) (Figure 5). It was necessary to dig wells to provide water for humans and livestock to drink.

Figure 5: Adaptive strategies to the drying up of the river in three villages

In Sehitwa people switched from molapo farming to dryland farming because the soils were suitable for this, but in Shorobe the soils were generally not suitable (being sandy and infertile) and here only 11 per cent of the households switched to dryland farming, while the rest continued to plough their molapo fields under dry conditions. Dryland molapo farming is a special type of dryland farming done in depressions where run-off rainwater may collect. Social protection programmes acted as a buffer against loss of livelihood owing to the desiccation of the rivers. In Shorobe a significant proportion of households (43 per cent) relied on government assistance for their water supply. In Gudigwa, 55 per cent of the households said they dug wells in response to the drying up of the nearby floodplains; in Sehitwa only 14 per cent did this, and 5 per cent in Shorobe. The high proportion in Gudigwa could be because most of the households there did not previously have wells as they had just been resettled.

5.2.4 Internal migration and local mobility

In addition to the above-mentioned adaptive strategies, some households out-migrated or moved to other areas because of the desiccation: 27 per cent of the households reported that some people had done this. The most striking movements, in the form of both internal migration and short-term mobility, occurred in Shorobe, Sehitwa and Etsha 6. In Shorobe, 59 per cent of the households interviewed said the drying up of the Gomoti River had acted as a push factor for some of the households to move as their shallow wells dried up, some to a number of localities close to Shorobe (which had water resources supplied by the government) and some to other villages such as Matsaudi and Matlapaneng. Table 3 illustrates the impact of the drying up of the Gomoti on the population of a number of localities near Shorobe.

Table 3: Trends in the population of Shorobe village and associated localities

	1981	1991	2001	Population growth rate (%): 1981–1991	Population growth rate (%): 1991–2001
Shorobe	539	758	955	3.4	2.3
Xabe	78	126	49	4.8	− 9.4
Shokomokwe	86	118	79	3.2	− 4.0
Gobomochaa	109	89	23	2.0	− 13.5
Xuxau	125	252	69	7.0	− 12.9
Makoba	59	51	41	− 1.5	− 2.2
Mochaba	82	105	151	2.5	3.6
Source: CSO, 1982, 1992, 2002b.
Note: Population growth rates estimated by one of the authors using the exponential formula: Pn = Poe^rn, where e is a mathematical constant with the value 2.71827, r is the population growth rate, and n is the inter-census period.

The 1981–1991 inter-censual growth rates for these villages were positive, except for Makoba. However, the growth rates for the 1991–2001 inter-censual period were negative as a result of the decrease in their population. For instance, the growth rates for the inter-censual period 1991–1991 for Xabe, Gobomochaa and Xuxao were − 9.4 per cent, − 2 per cent and − 12.9 per cent, respectively. As a result of such migrations and local mobility, the population of Shorobe grew by 2.3 per cent from 758 in 1991 to 955 in 2001, despite the fact that some of the Shorobe people had also out-migrated to Maun and other places. Informal interviews revealed that those who remained in the above localities when others out-migrated to Shorobe were mainly the poor households, as they could not afford the costs of relocation. On the other hand, those who out-migrated to Maun from Shorobe were mainly the economically active. This is confirmed by the fact that the village of Shorobe's population now has a very high proportion of people over the age of 65 (Kgathi et al., 2004).

In Sehitwa, 54 per cent of the households reported that some people had out-migrated as a result of the drying up of Nhabe River and Lake Ngami to a number of areas such as Tsau, Maun, Setlatla, Maila, Naune and Dobe, where groundwater or surface water could easily be obtained (Table 4). It was also reported that others had moved their livestock to other areas such as Kgomotshwaana and Habu, where there was plenty of groundwater. However, most of the respondents complained that in these areas predators were a problem for their livestock.

Table 4: Frequencies of households who said other households emigrated

Village	Frequencies	Number of households (N)	% frequencies
Gudigwa	2	18	11.1
Sehitwa	15	28	53.6
Shorobe	10	17	58.8
Etsha	8	36	22.2
Source: Fieldwork.

Studies on rural settlement in Botswana have revealed that scarcity of water is one of the reasons why people do not permanently settle in the lands areas (masimo) in Botswana (Silitshena, 1982). Lack of water drives them away from the lands areas and draws them towards the villages, where schools, health services, employment and social amenities can be found. It is expected that this kind of movement will continue in response to the ongoing desiccation.

5.3 Eradication of cattle lung disease

Besides the drought and the desiccation of river channels and the killing of livestock by predators, there are other factors that constrain livestock production in the Delta, such as animal diseases and lack of veterinary and marketing services. The livelihood survey revealed that the majority of the households (83 per cent) saw livestock diseases as a major constraint, followed by drought (65 per cent) and desiccation of river channels (30 per cent). Animal diseases have adversely affected livelihoods in the past 30 years in Ngamiland. Some of the diseases which have affected cattle are CBPP, trypanosomiasis and foot and mouth. For illustrative purposes, this section examines the impacts of the eradication of CBPP on rural livelihoods in the Ngamiland, and the coping and adaptive mechanisms adopted by households to mitigate these impacts.

5.3.1 Impacts of the killing of cattle

In an attempt to eradicate CBPP, all the 320 000 cattle in Ngamiland were killed in 1995/1996, to protect the lucrative European Union beef market, as the disease would spread to other areas (Townsend & Segwele, 1998). Restocking took place in 1997. In the study areas, 69 per cent of the households reported that they were among those whose cattle were killed. There was an association between the frequencies of those whose cattle were killed and the households' socio-economic status, at a 5 per cent level of significance (χ² = 16.8, probability less than .0005). For instance, all the rich households (100 per cent) said their cattle had been eradicated, as compared to 79 per cent of the middle and 53 per cent of the poor households. According to Ndozi et al. (1999), before the eradication of cattle in 1996, 70 per cent of the households interviewed in Ngamiland said they owned cattle, as compared to 30 per cent after the eradication in 1999.

The government compensated farmers in cash or cattle or both, depending on their choice (Fidzani et al., 1999). Informal interviews revealed that the majority of the households in the study areas, particularly those with small herds, received 100 per cent cash compensation, rather than being compensated with cattle. The killing of cattle had drastic effects on the livelihoods of the rural people in Ngamiland, the principal ones being the loss of draught power, milk, meat and social security. The eradication of cattle also reduced the number of people who depended on arable agriculture as a source of livelihood: in 1996, before the eradication, 52 per cent of the households ranked it as their first most important livelihood activity, compared to 37 per cent in 1997 (Fidzani et al., 1999). This decline is attributed to a reduction in the availability of draught power (Ndozi et al., 1999), which was mentioned as a major constraint on arable agriculture by 41 per cent of the respondents. The lack of milk and meat because of the eradication caused increased malnutrition in the under-fives (Fidzani et al., 1999). According to Ndozi et al. (1999), malnutrition due to lack of protein increased from 4.2 per cent in 1995 before the eradication to 8.8 per cent in 1996, 11.3 per cent in 1997 and 8.4 per cent, in 1998. The number of male agricultural labourers was reduced by 33 per cent in Ngamiland from 1325 before the eradication to 884 after it, mainly because herd boys lost their employment (Fidzani et al., 1999).

Focus group discussions and informal interviews revealed that the killing of cattle was one of the worst shocks experienced in the past 20 years (although its effects were less severe than those of HIV/AIDS and the desiccation of river channels in Sehitwa and Shorobe). However, it is important to note that the direct costs of the eradication of the disease were far lower than the accumulated costs to the economy would have been if the disease had not been controlled (Townsend & Segwele, 1998). Eradicating all the cattle in Ngamiland was the most realistic strategy to pursue.

5.3.2 Household coping mechanisms

The livelihood survey revealed that the coping strategies adopted to ameliorate the effects of cattle eradication in the study areas included reducing the area of land ploughed, resorting to donkeys as a form of draught power, switching to hoeing, and hiring draught power. In Sehitwa, where livestock farming is an important activity, more than 44 per cent of the households whose cattle were eradicated reported that they had resorted to donkeys (provided by the government), despite a general concern that donkeys are not as suitable as cattle for draught power because they are very slow, with the result that less land is being ploughed. Other households either switched to hoeing as a method of ploughing or gave up arable farming to engage in alternative activities, such as formal employment and beer brewing.

Although the government supplied the affected households with donkey draught power and temporary food rations, most households reported that the eradication of cattle had made them poorer and more vulnerable to poverty. They also reported that their arable production had declined because of the lack of draught power, that they had reduced access to meat, milk, and meat and milk products (for home consumption and sale), and had lost their revenue from the sale of cattle.

The way the cash compensation for the killing of cattle was used made a number of households in Ngamiland more vulnerable to poverty. Focus group discussions and informal interviews revealed that some of the households were registered as destitute persons after their cattle were killed, particularly those who had fewer cattle before the eradication. This category of households also tended to be those who had obtained 100 per cent cash compensation for the eradication of their cattle. The cash was in most cases spent rather than invested. According to Fidzani et al. (1999), 58 per cent of households received 100 per cent cash compensation, 24 per cent received 70 per cent, 13 per cent received 30 per cent, and 3.5 per cent did not know what type of compensation they had received. Although it was a realistic strategy for the government to eradicate all the cattle, the giving of cash compensation to some of the households should have been avoided.

5.4 HIV/AIDS

The HIV/AIDS epidemic, an issue of major concern worldwide, poses a major threat to households in the Delta and the rest of Botswana. Though it is a global problem, it is concentrated in the developing countries, particularly in sub-Saharan Africa (Cohen, 2001). HIV/AIDS differs from other shocks because it 1) has long-term impacts, 2) attacks the most productive cohorts, 3) puts heavy pressure on women as they play a leading role in caring for the sick and orphans, and 4) significantly reduces the gross national product (GNP) (Barnett & Blaikie, 1992; White & Robinson, 2000; Bayles, 2002). Shocks such as animal diseases and the desiccation of river channels do not have such devastating impacts on human welfare.

5.4.1 Magnitude and impacts of the epidemic

Barnett and Blaikie (1992) categorise households associated with the impacts of HIV/AIDS as ‘AIDS-affected’ or ‘AIDS-affected’, the former being those that are not infected but have used their resources to support households afflicted by HIV/AIDS (support of orphans), and the latter being households with members who are ill or have died from HIV/AIDS. They refer to households that are neither affected nor afflicted by HIV/AIDS as ‘unaffected households’. However, this distinction was not apparent in our study area as almost all the households had suffered from the HIV epidemic. Hence a more useful categorisation of households would be one that indicated the level of effect, e.g. the ‘most affected households’ or the ‘least affected households’.

The Botswana AIDS Impact Survey II of 2004 revealed that the percentage of people who are HIV/AIDS positive was 16 per cent in Ngamiland South and 13 per cent in Ngamiland North, as compared to the national figure of 17 per cent (CSO, 2004).

The livelihood survey revealed that 46 per cent of the households had chronically ill members in the past year, and some of them had already died. The perceived impacts of HIV/AIDS in the study areas included increased financial costs (46 per cent of the households), pressure on agricultural labour (22 per cent) and loss of employment (22 per cent). The epidemic has negative effects on financial resources because of the costs of transporting patients to health services, medical fees and funerals. There is pressure on the agricultural sector because the availability of labour has been reduced owing to the morbidity and mortality of prime age adults, and employment is lost because of the long-term illness of some of the household members, which is particularly serious in the case of breadwinners.

By implication, as adult mortality increases, regardless of the causes of death, so does the possibility of surviving dependants, especially children, being displaced from established family relations. The 2001 population census indicates that there were 40 994 orphans in Botswana, the highest incidence being in Ngamiland West, where orphans account for 4 per cent of the total population. In the research areas, registered orphans accounted for 4.2 per cent of the population in Gudigwa, 6.2 per cent in Etsha 6, 5.4 per cent in Sehitwa, 7.1 per cent in Seronga and 9.4 per cent in Shorobe.

Focus group discussions in Shorobe, Gudigwa and Seronga revealed that the HIV/AIDS epidemic was the most devastating shock they had ever experienced, not comparable with other shocks in Botswana, since it resulted in illness and death. All members of the focus group mentioned that they had experienced the death of either a household member or a close relative. In their opinion, shocks such as drought, desiccation of river channels and damage of crops by wildlife also had adverse effects on their livelihoods, but these impacts did not compare with those of HIV/AIDS.

5.4.2 Coping and vulnerability

Focus group discussions and informal interviews revealed various ways that households coped with HIV/AIDS in the study areas. The principal strategies included reduction in child-bearing, intra-household re-allocation of labour, liquidating assets such as livestock to cover medical expenses and funeral costs, seeking the support of the extended family and participating in burial societies.

Informal interviews in the study areas revealed that child-bearing is no longer an attractive option for women because of the risk of HIV infection. This demographic change has partly contributed to a reduction in fertility, as confirmed by national and district data from the 1981, 1991 and 2001 censuses. The total fertility rate (births per woman) in Botswana dropped from 6.6 in 1981 to 4.2 in 1991 and 3.27 in 2001 (Ramolemana & Shongwe, 2003). Despite the high mortality rates of the prime age population, the dependency ratios in Ngamiland have decreased from 124 per cent in 1991 to 83 per cent in 2001 (Majelantle, 2003). This could be attributed to low fertility rates and the increase in infant mortality.

Besides liquidating assets and sharing labour amongst households, another strategy used by some households was to abandon their livelihood activities when some of the household members who were actively involved in production activities became sick or died. Informal interviews in Sehitwa and Shorobe revealed that most households either reduced the area they ploughed or abandoned both arable and livestock agriculture altogether. The adverse impacts on livelihood activities resulting from HIV/AIDS have marginalised and impoverished a number of households in the study areas. According to several respondents in our informal interviews, the most affected were female-headed households and poor households in general.

The extended family still plays an important role in caring for the sick and orphans in Botswana. The focus group discussions revealed that most of the orphans in the study areas of Seronga, Gudigwa and Shorobe were taken care of by their relatives. This suggests that social capital is an important safety net for the afflicted households. However, the family institution is now under great pressure as a result of the increasing number of orphans, and the question is whether it will continue to cope in future.

Apart from the extended family, there are social support groups, which help afflicted households in the event of death. These include formal groups such as burial societies and informal social groups that provide financial, social and psychological support. Informal interviews in Shorobe and Sehitwa revealed that some of the ethnic groups such as Herero and Batawana have burial societies and, in the event of death, the members make financial contributions to the afflicted households, besides providing their labour and psychological support. Relatives of the deceased also provide financial, psychological and social support, and the level of such support normally depends on the degree of closeness to the deceased and the socio-economic status of the provider.

6. DISCUSSION AND CONCLUSION

6.1 Summary of shocks

The drying up of the channels and flood plains has adversely affected access to water resources (for humans and livestock), flood recession agriculture, fishing and veld products, and is perceived to be a worse disaster than the 1996 eradication of cattle. Households have responded by digging wells, switching from flood recession to dryland agriculture, depending on the government for water supply, and out-migrating to areas with reliable water sources. The findings support other studies in developing countries which reveal that out-migration is usually adopted as one of the strategies for ex post diversification (Ellis, 2000).

Another shock, the killing of cattle in Ngamiland to eradicate CBPP, has resulted in loss of draught power, milk, meat and social security. Households have responded by reducing the area ploughed, using donkeys for draught power, and resorting to hoeing as a method of ploughing. Government compensation of households was poorly handled: many consumed rather than invested the cash, thus increasing their vulnerability to poverty and destitution.

Furthermore, the HIV/AIDS pandemic has had the most devastating effects on livelihoods. It has put pressure on agricultural labour as a result of the morbidity and mortality of prime age adults, increased financial costs, and caused loss of employment. It has marginalised and impoverished a number of households in the study areas, the most affected socio-economic groups being female-headed households and other poor peasants. The principal coping strategies for households have included re-allocating household labour, liquidating assets to cover medical expenses and funeral costs, seeking the support of the extended family, participating in burial societies and enrolling in government social welfare support programmes.

It was revealed that the impact of HIV/AIDS is significantly different from that of other shocks since it results in illness and death, with very adverse impacts on livelihoods. The epidemic tended to affect the economically active population more than other population groups, and this is likely to have adverse impacts on the economy.

6.2 Implications for interventions

The research results on the impacts of desiccation of river channels and the strategies adopted by households to adapt to this problem could be used to construct scenarios for assessing the possible socio-economic impacts of desiccation of river channels in the Okavango Basin, and may be useful to researchers in other river basins for designing similar surveys.

The outbreak of animal diseases is a worldwide problem, and other countries can draw lessons from the Botswana experience. Although it was realistic to kill all the cattle before the disease spread to other parts of the country in order to protect the lucrative European Union beef market, this may be a difficult option to pursue for low-income developing countries with limited financial resources, particularly if the costs of compensation are high.

Although interventions aimed at addressing the impacts of other shocks such as the desiccation of river channels and drought could also be useful for addressing the impacts of HIV/AIDS, they may not necessarily be appropriate for mitigating its impacts, owing to the distinct nature of the epidemic. Those who support the view that general anti-poverty strategies are relevant to the mitigation of the impacts of HIV/AIDS (White & Robinson, 2000) are ignoring the uniqueness of the impacts of the epidemic, as revealed by this study. Policy makers need to take account of households' responses to shocks when making policy interventions, in order not to make wrong decisions, as the impacts of shocks are not necessarily similar.

Shocks tend to adversely affect environmental entitlements, which are important sources of livelihoods for the rural population. It is generally known that a decline in environmental entitlements further aggravates poverty and environmental degradation (Leach et al., 1999), and although poverty is not an underlying cause of this degradation, it accelerates the process (Barbier, 1991). Thus, the extent to which poverty causes environmental degradation depends on the options available to the poor to respond to the problem, but usually such options are limited and consequently the poor often resort to the overuse of natural resources (Barbier, 1991). The shocks and the subsequent poverty may result in the degradation of the biodiversity of the Delta, particularly if appropriate management measures are not taken.

Finally, diversification of livelihood strategies can be useful for poverty alleviation in Ngamiland. For instance, households that depend on natural resource based activities are more affected by drought or desiccation, whereas those that depend on non-natural resource based activities are less affected (Kgathi et al., 2004). The sustainable livelihood framework is therefore a useful tool for guiding interventions on poverty alleviation, as it also allows for the categorisation of livelihood activities and the factors influencing them. This research will be of use in the formulation of the Delta Management Plan, which aims at promoting the integrated sustainable management of the Delta's natural resources so that future generations can also derive benefits from these resources (ODMP Secretariat, 2006). As already stated, if the socio-economic sub-system is negatively affected by shocks, it is likely to negatively affect the bio-physical sub-system, further reducing its ability to provide goods and services.

Additional information

Notes on contributors

Julie Wilk

Respectively, Associate Professor, and Senior Research Fellow, Harry Oppenheimer Okavango Research Centre, University of Botswana; and Lecturer, Department of Water and Environmental Studies, Linkôping University, Sweden. This study would not have been possible without the generous financial support from the European Union (INCO-DC Programme, contract IC4-CT-2001-10040) and the University of Botswana. The authors' special thanks also go to Professor Piers Blaikie of the University of East Anglia for critically reviewing the earlier draft of this paper, and to their colleagues at HOORC, Keta Mosepele and Monica Morrison, for reviewing the final draft.