Introduction

Consumption of bushmeat – defined as wildlife hunted for human consumption [1] – is a pervasive threat to wildlife not only in Africa but worldwide [2]. Until recently, most studies and concern about bushmeat consumption in Africa were centered on the west and center of the continent because bushmeat is often considered a forest activity and there is a perception that East African savannah reserves are well protected; nevertheless, it is now regarded as a serious problem in East Africa too [3], with Tanzania in particular receiving a good deal of study [456789–10]. Our current understanding of this form of wildlife exploitation in Africa is centered on two distinct issues – what levels of protection are appropriate to limit bushmeat hunting? And can availability of alternative sources of protein affect bushmeat consumption? With respect to protection, there is some evidence that law enforcement does reduce offtake [11] and this has led to recommendations in the academic literature that governments should take on-ground law enforcement more seriously [12]. With respect to alternative sources of protein, there is considerable evidence that illegal bushmeat offtake is closely associated with economic factors [13], although this may vary according to urbanization and economic strata. At a broad level, then, it is generally agreed that increased patrols, arrests and fines in conjunction with much greater attention to social and economic problems may together have a chance of reducing bushmeat consumption on some parts of the continent [see 14].

Moving from the general to the specific, however, it is difficult to know how to change people's behavior without clear knowledge of the factors affecting bushmeat consumption [see 15]. For example, if bushmeat consumption is associated with low income, or lack of alternative protein sources, providing additional economic opportunities or new sources of protein might help alleviate the pressures on wildlife. If interest in bushmeat is linked to particular ethnic affiliations, education targeted at specific groups, and with appropriate cultural nuances, might be warranted. If bushmeat consumption is negatively associated with distance to protected areas, then focusing attention on villages closest to protected areas might be the most effective way of ensuring that conservation goals are met. All these implementation proposals must be evaluated in light of the fact that animal source protein is often a critical source of nutrition and health of residents in parts of rural Africa, and that hunting may carry cultural significance [16].

There is considerable agreement among studies over of the factors shown to affect bushmeat consumption in various parts of Africa, namely ethnicity, household size, household wealth, education, and distance from a protected area, but less consensus over the precise nature of these effects. For example, household size has positive effects on bushmeat consumption on Bioko Island, West Africa [17, see also 6] but negative effects in Gabon [18]. Household wealth is also typically associated with bushmeat consumption, with richer households consuming more in urban households in Rio Muni [19] although poorer households consume more in rural Equatorial Guinea [13]. Education is negatively associated with consumption of certain wild mammals in rural Gabon [18]. Bushmeat hunters enter protected areas from outside but to a decreasing extent with distance from their settlements [2021–22]. Finally, ethnicity is strongly associated with bushmeat consumption in parts of Rio Muni [23]. To refine our understanding of household correlates of bushmeat consumption in western Tanzania where bushmeat activities are prevalent [7, 10] we chose the Katavi-Rukwa ecosystem where declines in mammal populations are in evidence [242526–27]. We focused on Mpimbwe Division in particular, a rapidly growing area occupied by several ethnic groups and for which there is solid ethnographic background [2829–30].

We hypothesized that bushmeat consumption would be less common among the Sukuma, an agro-pastoralist ethnic community that has recently migrated to the area, as opposed to the more sedentary indigenous hunter-fisher-cultivators of mixed ethnic origins (primarily Pimbwe, Fipa and Rungwa), in part for cultural traditions (see below) and in part because of the availability of alternate sources of protein in livestock-keeping households. Additionally, we predicted that bushmeat consumption would be more common in poorer households, and in households headed by less educated men, and in larger households, due to greater need and fewer economic alternatives. Because preliminary analyses showed that the explanatory factors varied by ethnicity [9] we present ethnic-specific models which can potentially capture the different dynamics of immigrant and indigenous ethnic groups.

Methods

Background information

Katavi National Park (Fig. 2a) was established first as a game reserve in 1956, and as a national park in 1974; it was extended in size in 1998. Each of these land planning changes entailed population displacements in the Rukwa Valley – the indigenous Pimbwe (Fig. 2b) and Rungwa were evicted from their original villages at various periods starting in 1927, and later the immigrant Sukuma were forcibly removed from their dry season grazing areas at various times starting in 1998. The Pimbwe and Rungwa are renowned for being both hunters and fishers, economic activities that are part of their cultural heritage [34]. They originally occupied and hunted in much of the area now gazetted as KNP and, in the case of the Rungwa, in the RGR [35]. It is worth noting that their legal access to areas formally used for hunting and fishing is now much reduced as a result of protectionist conservation policies. The Fipa, original inhabitants of the Ufipa Plateau lying to the west of the Rukwa Valley, also consume bushmeat and are known for eating rodents [34]. Only the immigrant Sukuma have no strong reputation for being hunters, as they depend on an agropastoral production system. Therefore, there are good ethnographic justifications for separating Sukuma and non-Sukuma ethnic groups when examining patterns of bushmeat hunting in this area of Tanzania.

Our reference to Pimbwe, Fipa, and Rungwa as indigenous is for clarity, to differentiate them from the immigrant Sukuma, and does not suggest they have greater rights to the land in Mpimbwe. Note also that the area is typically associated with high levels of seasonal food insecurity [30], which are associated with high incidences of maternal anxiety and depression in both the indigenous and the Sukuma populations [36], although disproportionately so in the indigenous sample [37].

Fig. 1.

Study area showing Katavi National Park (NP) and Rukwa Game Reserve (GR) in western Tanzania (see inset). Black dots show location of the eleven villages in Mpimbwe Division at the time of the study, all of which are included in the sample. Villages lie between 5.7 and 19.1 kms from the border of one of the two protected areas. Bar at top right denotes 25 kms.

Data collection

Data were collected in Kiswahili from Mpimbwe between July and September 2008 during the dry season; every one of the 11 principal villages was visited (Fig. 2c). In each village approximately 20 households were selected randomly (by use of village lists) from the center of the village to capture the indigenous population, and 20 households from a randomly selected subvillage in order to ensure coverage of Sukuma households that are typically situated outside the central village area (Fig. 2d). A total of 435 households out of an estimated 5,895 in the Division were studied, of which 245 were indigenous and 190 were Sukuma. Villages lie approximately 6 to 19 kms from the border of one of the two protected areas (KNP and RGR, Fig. 1).

Prior to interviewing households, meetings were held with village officials to solicit consent and obtain household lists from which households were picked at random. Because bushmeat consumption is an illegal activity, and interviewees might be expected to see a cost (either real or perceived) in providing honest answers, we developed a questionnaire that started with general questions about land under cultivation, crop harvests, education level, household members, livestock and other material assets owned, and food items consumed within the last month. Only later in the interview were households asked to answer voluntarily questions about bushmeat use in the village, and household bushmeat consumption (see Response Variables, below). By avoiding questions specifically focused on hunting activity, interviewees were never asked about an illegal activity directly; moreover, availability of bushmeat in the village is public knowledge, as meat is generally openly available. Interviews were conducted with heads of the household, or, in rare cases when household heads were absent, with other permanent family members. There were no cases where interviewees chose to discontinue the interview when asked about bushmeat even though withdrawal was always an option; we did not ask where bushmeat came from.

Fig. 2.

(a) Entrance sign to Katavi National Park from Mpimbwe Division; (b) A Pimbwe household; (c) Mirumba village, the most westerly of the 11 villages sampled. Katavi National Park is the background; (d)) A Sukuma dwelling away from the village center. Photo credits: (a) and (c) Tim Caro; (b) and (d) Monique Borgerhoff Mulder.

Results

Discussion

The principal species hunted in Mpimbwe were impala (Aepyceros melampus), common duiker (Sylvicapra grimmia), warthog (Phacocherus africanus), buffalo (Syncerus caffer) and bushbuck (Tragelaphus scriptus) [44]. Our first main finding concerns bushmeat activity in relation to distance from the protected area boundaries. Unsurprisingly, we find that households nearer to protected area borders report more carcasses entering their villages regardless of ethnicity, although other factors for which we have no information are also responsible for inter-village variation, as seen from the relatively high standard deviations of the mixing distributions. With respect to bushmeat consumption, the same pattern is seen for the indigenous sample (but not for the Sukuma sample), insofar as greater use of bushmeat was reported in villages close to protected area boundaries. Similar findings regarding greater bushmeat consumption in villages close to protected areas have been made on other parts of the continent [e.g., 45], and point to the need for anti-poaching and outreach activities being focused on villages close to protected area boundaries.

Our second general finding relates to the role of various forms of wealth on perception of carcass availability and bushmeat consumption. In short, we did not support our hypothesis that poorer households would rely more heavily on bushmeat, but instead found an interesting ethnic effect. Among the indigenous sample, larger households, and households with more land under cultivation and more livestock reported greater carcass availability, and households with more land under cultivation, more livestock and more poultry reported greater bushmeat consumption. This may provide evidence that Pimbwe, Fipa and other indigenous populations of the Rukwa valley use their traditional sources of wealth to generate cash to purchase and consume bushmeat. The fact that assets holdings among the indigenous sample were negatively associated with perceptions about carcass availability and bushmeat consumption requires further explanation. Recall that assets include watches, mobile telephones, radios, tin roofs and pieces of non-traditional furniture (wooden beds), as well as items of capital value such as oxcarts, ploughs and beer drums. The asset-holding scale therefore differentiates households according to their “modern,” development-oriented, or pro-economic diversification stance. Conceivably such households may be less interested in bushmeat because they are more modernized. We are nevertheless wary of this interpretation because education level of household head, a likely indicator of less traditional leanings, did not enter into the best models. Rather, we speculate that among traditionally wealthy households bushmeat and external goods may be substitutes for each other.

The situation among Sukuma households is quite different. Whereas larger indigenous households reported more carcasses entering villages, larger Sukuma households reported fewer carcasses. Because Sukuma household size is strongly correlated with land, livestock and poultry wealth, this suggests that the wealthy Sukuma households are generally less aware of carcasses in the village and possibly less engaged in bushmeat consumption, quite contrary to the findings for the indigenous sample.

In other studies, the association between measures of wealth and involvement with bushmeat is similarly conflicting: some studies report positive associations [e.g., 23], some report no association [e.g., 13] and others report urban-rural-dependent associations [e.g., 19]. In Mpimbwe, bushmeat is a third of the price of domestic meat [10], and therefore land-poor households would be expected to avail themselves of this cheaper animal source protein. Yet this is not the case. Poorer indigenous households eat only maize meal or cassava, supplementing this with vegetables such as domestically or wild harvested greens, or beans, and much more rarely meat, fish or eggs [30]. While bushmeat is a cheaper option than domestic meat, it is still largely a luxury. In short, it is the wealthier indigenous householders who are supplementing their vegetable diet with the cheapest available animal protein, bushmeat, a pattern previously only seen in urban environments. Among the Sukuma, however, this pattern is not observed; among these immigrants both perception of carcass availability and possibly bushmeat consumption are greater among the smaller and less wealthy households. If this is a general result throughout the country, Tanzanian rural indigenous communities that experience economic growth may be having a negative impact on the nation's wildlife, an effect potentially exacerbated by the arrival of Sukuma immigrant families.

There are other interesting aspects of our findings that relate to protein availability, ethnicity, education, and multilevel analyses. Bushmeat consumption for the Pimbwe and Fipa appears to be a luxury for households already relatively rich in livestock and poultry protein, although this is not the case for the Sukuma. This suggests that supplementing protein intake may not be an effective intervention for lowering reliance on bushmeat within indigenous communities, unless the price can be brought below that of bushmeat. Our results for the Sukuma, however, are comparable to those of Loibooki and colleagues [5], who showed that livestock rich households around Serengeti are significantly less involved in illegal hunting than livestock poor households.

With regard to ethnicity, it is noteworthy that, in contrast to a number of other studies, we did not find support for the hypothesis that ethnicity affects bushmeat consumption (although it did affect perceptions of carcasses entering the villages, probably because indigenous households tend to form the core of the village, and hence the commercial hub where bushmeat is typically sold). Rather more interestingly, the precipitating factors for bushmeat consumption differ markedly between groups. Within Tanzania, these results can be compared with only one other study where ethnic differences in hunting preferences were explored: around Serengeti, ethnic groups differed in the reported size of animals killed, and the reasons for killing them [9]. Both studies indicate that a single ethnic group study of bushmeat consumption is inadequate to provide a full picture of the drivers of bushmeat offtake.

We found no evidence that level of schooling affects either the perception of bushmeat availability or its consumption. Education was measured only for the household head, suggesting that at least in the past primary and secondary school curricula provided little material to influence bushmeat consumption. The effects of primary and secondary school education, and park visits, on the present cohort of school children are current topics of investigation [46].

Finally we emphasize the need for studies of bushmeat drivers, particularly in ethnically and economically heterogeneous contexts, to use mixed effects model designs so that village-specific dynamics can be accounted for and made available for implementation activities. While we recognize that the large geographic spread required for such studies necessarily reduces the amount of time that can be spent with interviewees, we suspect (and have to some extent shown) that these behaviours are likely to be strongly spatially and ethnically structured.

Questions inevitably arise regarding the reliability of reported behaviour. First, an undetermined amount of underreporting may have occurred because bushmeat consumption is illegal. Subsequent interviews with hunters (not reported here) suggest this may have been the case (AM unpublished data) but it is unlikely to affect the relative contributions of causal drivers of consumption. Second, there is the false consensus effect, in this case the possibility that those who hunt overestimate the prevalence of hunting in the village, and those who do not hunt underestimate prevalence [47, 48]. Such effects may well bias overall estimates of offtake – downwards, for example if only a few individuals hunt heavily, upwards if the majority hunt even sparsely – but it should not affect analyses of variability within and between villages. Third, women were typically not interviewed unless there was no adult male household member available; it is possible that estimates would have been higher (women cook) or lower (women generally interact less with outsiders) if women been systematically surveyed. Fourth, responses could have been anonymized through randomized response techniques (RRT) [47, 48], but in our view the complexities of this approach would have exacerbated the potential sensitivity of questions regarding the eating of bushmeat, the illegality of which is not uniformly recognized by all villagers. Finally, an independent measure of illegal hunting activity per village based on focused interviews with hunters in the same village (AM, unpublished data) was correlated with the measure of number of carcasses reported here, (Kendall's tau = 0.50, p = .04) but not with bushmeat consumption (Kendall's tau = 0.38, p = .12). This suggests that the indirect question (number of carcasses) is a more reliable indicator of village bushmeat activity than the direct question regarding household consumption (which measures household-specific preferences).

Implications for conservation

Our study extends understanding of bushmeat consumption to a rapidly growing area of western Tanzania [29] known for its considerable wildlife offtake from protected areas [44]. We conclude that bushmeat activities are higher nearer to protected areas; that, contrary to our predictions, bushmeat consumption was less common in poorer indigenous households, challenging the idea that increasing the availability of alternative protein would necessarily reduce consumption of bushmeat unless its price became lower than that of bushmeat. We recommend that outreach programs designed to reduce bushmeat offtake in Mpimbwe be targeted at both hunters near the boundaries of protected areas and consumers located throughout the Division; that land-rich households should be a particular focus of attention; and that protein supplementation be given cautious consideration in solving the problem of bushmeat offtake.