False Promise or False Premise ? Using Tourism Revenue Sharing to Promote Conservation and Poverty Reduction in Uganda

Copyright: © Tumusiime and Vedeld 2012. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and distribution of the article, provided the original work is cited. Abstract Tourism and the sharing of the associated revenues with local people have been increasingly fronted as key instruments for maintaining protected areas (PAs) globally. This paper focuses on a tourism revenue sharing scheme employed in Uganda’s Bwindi Impenetrable National Park, involving rural farmers. We fi nd that the scheme faces diffi culties in integrating with the existing local historical, socio-economic, and institutional landscapes. Similar experiences from other cases suggest that these challenges are generic, and relate to lack of real local participation; an insignifi cant scale of economic returns to local people relative to costs; inept institutions in charge of planning, managing and evaluation efforts; and an institutional complexity that constrains most activities. We conclude that although tourism revenue sharing is an appealing concept, and its oft-quoted logic of promoting conservation and rural development is diffi cult to ignore, it is challenging to plan and implement in competent ways. We do not suggest abandoning tourism revenue sharing, but rather believe that a more concerted effort to overcome the mechanism’s economic and institutional shortcomings, as identifi ed in this paper, may be more appropriate. The overall fi ndings indicate that problems are not with tourism revenue sharing as an ambition, but with the diffi culties encountered in putting it into practice.


INTRODUCTION
Over the last two decades, tourism has increasingly been fronted both as a key rationale and as an instrument for maintaining protected areas (PAs) (Balmford et al. 2009).A long-standing discourse on conservation management envisages that conservation can 'pay its way' through tourism (Eltringham 1994), and that the sharing of revenues with local people will demonstrate the economic usefulness of PAs and secure local people's allegiance.
This tourism revenue sharing approach increasingly fosters 'hybrid environmental governance' in which the responsibility and the right to manage and conserve the world's biodiversity assets is shared between communities, businesses, NGOs, and states.Such a philosophy identifi es well with both neoliberal and market-oriented approaches to economic development and environmental management ('ecological modernisation'), and has been well received by international fi nancial institutions, national governments, and the private sector (Brockington et al. 2008).The principle of tourism revenue sharing is also at the heart of the win-win narrative that combines concerns of environmental conservation with those of local development.In this respect, arguments have been made for 'pro-poor conservation' (Roe and Elliott 2004) and recently 'pro-poor tourism' (Ashley and Roe 2003;Ashley and Mitchell 2005).The approaches of pro-poor conservation and pro-poor tourism are thought to have the ability to jointly promote conservation and poverty reduction, two most important societal goals as expressed in the millennium development goals (UN 2008).Sharing tourism benefi ts with the poor people living adjacent to PAs becomes pivotal, as it not only has the potential to contribute to the mentioned twin goals, but also offers a perfect fi t with the 1992 Rio Agreement that advocates an integration of the concerns of environmental protection and economic development based on free market principles (Stabler 1997).
Tourism revenue sharing is thus an important element in the current alliance between capitalism and conservation.However, evidence shows that the effectiveness of such policies is mixed (Parry and Campbell 1992;Mehta and Kellert 1998;Kellert et al. 2000).In the light of the waning resistance, within the corridors of mainstream conservation power, to the idea that capitalism (through interventions such as tourism) can and should help conservation to achieve its prime goal of saving the world, examination of the institutional arrangements around tourism revenue sharing becomes necessary (Walker et al. 2009).In this paper we analyse the case of the tourism revenue sharing scheme at Uganda's Bwindi Impenetrable National Park (hereafter referred to as Bwindi; Figure 1), with a view to identifying challenges that such schemes may face.This study is thus also a response to the calls for more research on 'actually existing neoliberalisms' (Brenner and Theodore 2002;Castree 2008).
From a classical perspective, institutions are viewed as the conventions, norms, and formally sanctioned rules of a society.
They regularise life, support values, and produce and protect interests (Vatn 2005).While people can create institutions, institutions also form us, and shape the way we interact with other people and with our environment.The institution of revenue sharing in Bwindi seeks, through the provision of economic incentives, to demonstrate to local people the economic importance of protecting biodiversity.This in turn may infl uence attitudes, values and norms, and engender support for conservation.However, the success of any intended institution depends on the context of its application (Vatn 2009;Muradian et al. 2010), and on other existing institutional arrangements that may be in place (Young 2002;Young et al. 2008;Corbera et al. 2009).In the language of institutional analysis, the former is referred to as the 'institutional fi t' and the latter as 'institutional interplay'.
In general, the fi t of an institution or governance framework relates to its appropriateness in achieving its stated goal.This concept has been developed and used by several authors, including Hanna et al. (1997), Berkes and Folke (1998), Young (2002), andFolke et al. (2007), and has included contributions and insights from the fi elds of political science, ecology, and institutional economics.The extent to which tourism revenue sharing secures local support for conservation may depend on, among other aspects, the suffi ciency of revenues given back to the local people (Gibson and Marks 1995).Household level benefi ts should offset actual and perceived costs (Murphree 2005).Local people usually experience significant costs in the name of conservation, notably strong restrictions on

Figure 1
Map of the study area, the Bwindi Impenetrable National Park access to park resources, and crop raiding by the protected wildlife (Tumusiime and Svarstad 2011).Such costs increase livelihood insecurity which may sink the already poor PA neighbours deeper into poverty (Tumusiime et al. 2011).Livelihood alternatives may be provided, but it is paramount that appropriate benefi ciaries are identifi ed and that these alternatives fi t within current livelihood means and lifestyles (Spiteri and Nepal 2006).
Furthermore, the extent of the effectiveness of any specifi c institution often depends on its interplay or interactions with other institutions (Young 2002), which defi ne the boundaries of the rights, responsibilities, and infl uences of the institution.Inter-institutional confl icts and unexecuted responsibilities commonly occur along these boundaries, where the interests and jurisdictions of multiple actors overlap (Mitchell 1990).The heterogeneity of the local benefi ciaries confi gures the local context (Agrawal and Gibson 1999;Archabald and Naughton-Treves 2001), where there is often a multitude of actors with varied objectives, mandates, and skills, and where forms and sources of power and legitimacy interact.These factors increase the likelihood of a confl ict between stakeholders and the institution of sharing tourism revenues.
Interplay occurs at both vertical and horizontal levels.Vertical interplay concerns the interaction of institutions or actors at different levels of governance, e.g., local, regional, and national levels (Young 2002;Young et al. 2008;Corbera et al. 2009).In countries such as Uganda, where decentralisation reforms have been made, tourism revenues are important to local as well as national governments.This, combined with the international importance of the biodiversity these PAs host, results in a vertical chain of interests, and thus vertical interactions are evident.Horizontal interplay relates to the interactions between actors at the same local, regional or national level.The interplay among these actors may result in mutual interference but where successful may foster meaningful synergies.
In this study, we examine the institutional challenges to tourism revenue sharing, focusing on the complexity of institutional fi t and institutional interplay.We use the scheme in Bwindi as a case study.We assess stakeholders' practices, perceptions of, and experiences with the scheme, and we draw on examples from other cases to demonstrate how the challenges identifi ed in Bwindi are generic and apply to many other areas.
Bwindi is an interesting case because it has a vibrant gorillatracking programme, and also a tourism revenue sharing scheme that has been operating since it was fi rst piloted in 1994.The park generates more revenues than most other protected areas in Africa.The tourism revenue sharing scheme is meant to share this revenue with the local people.This approach resonates with the national tourism revenue sharing policy and its goal of "ensure[ing] that local communities living adjacent to PAs obtain benefi ts from the existence of these areas, improve their welfare, and ultimately strengthen partnerships between the Uganda Wildlife Authority (UWA), local communities and local governments for sustainable management of resources in and around PAs" (UWA 2000b: 6).This is a well-intended proposition, but its realisation requires scrutiny.

METHODOLOGY AND DATA COLLECTION
The empirical data for this study was collected during fi eldwork in Uganda between September 2008 and January 2011 (10 months).First, unstructured interviews were conducted to examine the way in which local actors view conservation of the area and the consequences for them.Interviews were audiotaped and later transcribed.Gorilla tourism-in terms of an established revenue sharing scheme managed by a Community Protected Area Institution (CPI)-emerged as a key issue and became a point of focus in its own respect.The unstructured interview data collected at this stage informed Phase I of data collection, which was complemented by two further phases.
In Phase I, we conducted interviews with 60 ordinary people living in 12 villages directly bordering Bwindi.We thought it likely that variations in economic, socio-cultural, and ecological conditions might affect the ways people both experience and talk about life adjacent to the park.Hence, we selected some villages where we assumed tourism benefi ts from the park were relatively high because of gorilla tracking sites, and some other villages far from these sites.We selected some villages with relatively high costs in terms of crop raiding resulting from being park neighbors, and some villages without such problems.Finally, we included some villages that have access to forest resources in the form of 'multiple use zones' through agreements with the Uganda Wildlife Authority, while other villages in our study did not have such agreements.After identifying villages with the mentioned characteristics, we randomly selected 12.
Within each village we arbitrarily approached some interviewees and then applied the snowball method with the aim of obtaining a comprehensive picture of the variations in the ways in which people experience life adjacent to the national park.In each village, we continued to interview people until we felt that a saturation point had been reached, that is, where new narratives presented to us did not contribute much more new understanding about life adjacent to the park.
In Phase II, we limited interviews to the subject of tourism revenue sharing.In-depth, unstructured interviews were conducted with a total of 12 randomly selected individuals.Similar to Phase I, we tried to elicit narrations that were as 'undisturbed' as possible, but which focused on local people's views about-and experiences with-the tourism revenue sharing scheme.As part of a broader project, we also conducted semi-structured interviews with 190 randomly selected households (Phase III), in which we specifi cally examined local attitudes towards the PAs and their role in securing local livelihoods.In terms of tourism revenue sharing, the interviewees provided similar information in Phases I and II, as reported in this paper.
In addition to interviews with local people, we also conducted 28 key informant interviews with local political leaders, park staff, governmental bodies, and NGO representatives, locally and in Kampala.Furthermore, we collected and reviewed written sources about Bwindi, and involved ourselves in participatory observation during local meetings, regional workshops, as well as some tourist experiences such as gorilla tracking.We also visited several tourism facilities.
We applied a standard procedure for the analysis of qualitative data using the elaboration of codes and indices.A core aim was to gain a good understanding of the aspects that the interviewees themselves emphasised when talking about the tourism revenue sharing scheme.Collecting data from different categories of local people and key informants enabled us to gather a comprehensive data set, and contributed to data triangulation.

Study context
Tourism (along with agriculture and forestry) is on top of the list of Uganda's key economic growth sectors (MFPED 2010).According to the World Tourism and Travel Council, tourism contributed 9.2 per cent or USD 1.2 billion to the gross domestic product in 2008.Bwindi is a renowned tourist destination for gorilla tracking.Gorilla tourism alone accounted for over 50 per cent of the revenue that was generated for the UWA in 2008 (Walaga andMashoo 2009).Uganda has a policy designed to share revenues with local people in the areas surrounding parks such as Bwindi.
Tourism revenue sharing in Uganda can be traced back to the 1950s, when the British colonialists used it as a tool to elicit co-operation from native Ugandans settled in areas adjacent to the country's game reserves.At that time, the monetary benefi ts were delivered to the local districts, while rural households received a direct share of the meat from crop raiding animals that were shot by the Games Department (Naughton-Treves 1999).Although revenue sharing was maintained in independent Uganda, the national administration that followed the attainment of political independence in 1962 was not pro-conservation.And in 1975, during the President Amin era, a 'double production campaign' was introduced encouraging people to increase their use of forest resources in a bid to increase agricultural production and double the gross domestic product (Kigenyi 2006).
Since 1987, the country has been relatively stable in terms of political conditions, and biodiversity conservation policies have been pursued in a consistent and determined manner.Between 1991 and 1993, six forest reserves, including Bwindi, were elevated to national park status, partly due to external pressures, particularly from USAID (Ditiro 2003).The local communities living adjacent to Bwindi constitute some of the most densely populated areas in Uganda, exceeding 300 persons per sq.km in some places (UWA 2002).Historically, these communities have used Bwindi as a source of wild meat, timber, mining (especially for gold), honey, land for cultivation, and non-timber forest products such as medicinal plants and materials for subsistence crafts.The upgrading of the status of the forest reserve to a national park in 1991 increased the restrictions on access to these resources, leading to substantial losses in total incomes.This has been a long-standing source of dissatisfaction among local communities (Scott 1992;Docherty 1993;Wild and Mutebi 1996;Hamilton et al. 2000;Namara 2000;Blomley and Namara 2003;Blomley et al. 2010).

Stakeholders in tourism revenue sharing in Bwindi
There are many stakeholders with interests in Bwindi.The UWA is at present entrusted with managing the national park on behalf of the citizens and stakeholders.A local governance system operates for the tourism revenue sharing account, whereby in principle, the UWA remits 20 per cent of park entry fees every month to fund various community projects.
The Local Government Act of 1997 (Republic of Uganda 1999a) obliges the UWA to work with local government when dispensing the people's share of tourism revenues.The local government structure in Uganda is a fi ve-tier system, starting with a local council at the village level, and rising up-through the parish, the sub-county, and the county levels-to the district council.Each local government committee has a member in charge of the environment; the committee member provides an avenue for partnerships with the UWA in natural resource management.However, the Act recognises only the sub-county and district levels as 'local governments' and the rest are seen as administrative units.The UWA must therefore distribute the local share of tourism revenues through either the sub-county council or the district council, which in turn are supposed to distribute it to the lower levels.The revenue has traditionally been disbursed as a conditional grant to the sub-county local governments.
Prior to 2000, a Park Management Advisory Committee (PMAC) administered the funds.In 2000 this committee was phased out, as reported by Archabald and Naughton-Treves (2001), when the UWA formulated a CPI policy.Every parish that is immediately adjacent to the boundaries of a protected area has a secretary for the environment and natural resources, who is designated as a representative on the board that manages tourism revenues.The CPI policy stipulates that the UWA works with parish secretaries for the environment in managing tourism revenue sharing.Using members of local government as representatives of the CPI has the advantage of providing already institutionalised channels for involving and communicating with communities (Blomley et al. 2002).The CPI representatives work with the sub-county and district councils, and a specifi c local government committee.The production and environment committee is in charge of local natural resources (Figure 2; UWA 2000a).

Tourism revenue sharing: The Bwindi practice
As early as 1994, a national tourism revenue sharing policy for PAs was drafted and piloted at Bwindi.As Archabald and Naughton-Treves (2001) describe, under this arrangement the Uganda National Parks (UNP) was required to give 12 per cent of their total revenue to the local communities.In 1995, the UNP formally adopted revenue sharing as a wildlife management policy, and in 1996, passed it as legislation under the Uganda Wildlife Statute.But a political decision was made to change the amount from 12 per cent of total park revenues to 20 per cent of park entry fees.The change sought to increase local shares of the revenues.This worked well in areas with mass tourism and no restriction on visitor numbers, but meant a sharp decline for Bwindi surroundings, where ecotourism dictates that visitor numbers are strictly regulated.
The stipulation on entry fees only, meant that no revenue from gorilla tracking permits is shared, which for Bwindi is the main source of income.For example, if a foreign tourist bought a gorilla-tracking permit in 2010 for USD 500, gate fees accounted for only USD 30.Under the former agreement, USD 60 (12 per cent of USD 500) would have been put into the revenue sharing scheme, but after the 1996 legislation, only USD 6 (20 per cent of USD 30) was put into the scheme.This amounts to only 1.2 per cent of the initial USD 500, against the 12% that would have been shared with the locals prior to the 1996 legislation.In 2000, various proposals were made to revise the revenue sharing policy to the pre-1996 arrangement.However, these proposals were not supported by the UWA top management and Board of Trustees, who argued that they had other substantial costs to meet (Adams and Infi eld 2003).
It thus proved diffi cult to amend the legislation to increase the local people's share, but through the advocacy of the Cooperative for Assistance and Relief Everywhere, Uganda, a gorilla levy fund was established at Bwindi in 2006, which collects USD 10 from each gorilla permit.Of this, USD 5 is forwarded to the UWA head offi ce in Kampala to be shared amongst people living adjacent to other national parks, particularly those that do not generate suffi cient park entry fees.The remaining USD 5 is intended for the villages adjacent to Bwindi, in addition to the current 20 per cent of park entry fees.By September 2008, Bwindi park management reported that it was ready to distribute about USD 157,642 to the adjacent villages.These funds were accumulated by the USD 5 gorilla levy fund between August 2006 and June 2008.However, it is only in July 2010 that UWA started giving out this money and by January 2011 several villages were yet to get this money.This is clearly a rather questionable situation, and is contrary to the story touted to tourists that most of the revenues they bring to Bwindi go to the communities neighbouring the park.

RESULTS AND DISCUSSION
The premise of the revenue sharing scheme is that the shared revenues will contribute to poverty reduction and act as an incentive for participating households to support conservation.From the start of the revenue sharing scheme in 1996 to the end of 2009 (13 years), an estimated USD 178,902 has been spent among the local communities adjacent to Bwindi.This amounts to an average annual disbursement of USD 13,000.
Most of the revenue (over 80 per cent) was spent on community level projects (Figure 3).The evaluation of these projects varied between the respondents, but most local people expressed little appreciation for expenses on local administration, for example in the form of building council halls.Since 2006, the emphasis has shifted to individual household projects, but by the end of 2009, this accounted for only 20 per cent (~USD 34,000) of the revenues.Support has been given to goat keeping and, to some extent, potato growing, and tree planting.The reason that park management accepted the goat project is that most of the land has lost its productivity and no longer provides good yields.Goats can provide manure to replenish the soils leading to better productivity, and thereby addressing both food uncertainty and poverty in the frontline communities.Goats also offer an alternative to game meat.

District Councils and Natural Resource Committees at the district level
Endorse revenue sharing projects, and ensure and support their implementation

/ Tumusiime and Vedeld
Notwithstanding the successes demonstrated in this case study, we have identifi ed several key challenges to the promise tourism revenue sharing offers, relating mainly to problems of institutional fi t and institutional interplay (Table 1).

The fi t between benefi ts and costs
While tourist destinations may bring in signifi cant revenues, only a fraction of those revenues actually reach the local people.Most of our interviewees were not satisfi ed with the total shares given to local communities.To substantiate their dissatisfaction, they always pointed to the limited revenues allocated to them in comparison with what the UWA keeps for itself.One interviewee said: Now you wonder why you are close to the park.A tourist pays USD 500, and the community share is to be only USD 5.This is not enough.They should give people at least USD 100.Let us compare it to the preparation of a meal: If you were a cook, would you prefer to get the smallest or the biggest portion to eat? Now, we the park neighbours are the cooks of that meal, but what do we get?The smallest portion.Is that fair?(Interview No. 46) Most of the revenue from Bwindi is remitted directly to the UWA's central treasury.Literature shows this to be the norm for many protected areas, with a central authority collecting the revenues and then budgeting for the national PAs from a central pool (e.g., see Campbell et al. 2001).There are thus few incentives for individual parks to increase incomes, since annual allocations tend to be independent of performance.In the case of Bwindi, it would run with a surplus, if given an autonomous economy; however the current policy leads to less money available for local people to share.
In this research project, we investigated payments for gorilla permits.We estimate that at full capacity, the current eight habituated gorilla groups could generate an annual revenue of over USD 11 million through permit sales, if 90 per cent of the permits were sold to foreign non-resident tourists, 3 per cent to foreign resident tourists and 7 per cent to tourists of East African origin.At the current rate of revenue sharing, the communities surrounding the park would then be entitled to 20 per cent of the park entry fees, which is equivalent to an annual USD 139,776 (or 1.27 per cent of the total revenue collected from permit sales).
In the light of the low amounts of revenue earmarked for the local people at Bwindi, the UWA tries to avoid distributing money every year, and use the gap years to build up sizeable amounts that could meaningfully support community projects.For example, by the end of 2009, the revenue sharing scheme that started in 1996 had disbursed revenues only four times (in 1996, 2002, 2006, and 2007).The disadvantage of this practice is that it reduces the predictability of the fl ow of benefi ts and diminishes local trust in the system, which in turn may reduce local support for conservation.
The number of people living near the park boundary is high.As mentioned earlier, Bwindi is located in one of the most densely populated areas in Uganda, exceeding 300 persons per sq.km in some places.A typical village has between 100 and 150 households.In the current goat scheme, which was implemented in most parishes after 2007, an average village adjacent to a PA receives nine goats per year.It may therefore take 11 to 17 years before each household will receive a goat, refl ecting the rather insignifi cant scale of these incomes for most households.The value of one goat is about 20 USD.This turns out to be a very small fi gure.Even when received, the goat has limited ability to pull the poor farmers up to the premised level, i.e, above the poverty line.
Local people obviously realise the inconsistency and small scale of revenues offered by the scheme.And as many view these incomes as a compensatory measure for substantial costs How does revenue sharing fi t within the decentralised framework?How do stakeholders at different levels (local, national, and international) impact the process and outcomes?How can power be transferred at the different levels?
accrued because of their proximity to the PA, the subjective comprehension of the small scale is reinforced.Local perceptions of their share of tourism revenues can be summarised in the following comment made by one interviewee: We used to collect basketry materials… but all those were stopped.We cannot get anything from there.Not even medicinal plants… And you see, in our village, at most we have received nine goats only… yet it is long since the park was declared a no-go area.And from that when you look at the park, these crop raiding animals of theirs, their baboons that come from the park and fi nd us in our villages.(Interview No. 15) Not all the local people we encountered agreed with the nature of benefi ts given to them in form of goats.They instead had several preferences including cash.As Emerton (2001) warns, failure of such a scheme to match the nature of costs may undermine its compensatory impact.A deliberate effort is needed to channel more revenues to the local people in a form that meets the nature of their costs, if the programme is to genuinely contribute to poverty reduction, and increase local support for conservation, premised on local appreciation of the economic value of conservation.

Defi ning benefi ciary communities
It is often assumed that the closer a village is to the PA boundary, the higher the local costs linked to PA establishment and management will be.However, it is often diffi cult to agree on a reasonable boundary for communities in relation to compensating for costs incurred (Brown 2002).In Bwindi, the UWA defi nes 'local people' as the inhabitants of villages that share an immediate physical boundary with the PA.It is assumed that these are the most affected by the PA and that they have a higher potential to impact conservation quality in the PA.
Such delimitations do not pass unquestioned, especially by the individuals that are affected.Moreover, in the case of Bwindi, the construction of village boundaries is not related to village proximity to the park.A particular household in a village defi ned as not park-adjacent may actually be nearer the park boundary than a household in a park-adjacent village.However, the operating defi nition of 'local' excludes the former household though it may incur substantial costs in being closer to the park.

The problem of interplay
There are challenges in Bwindi related to both vertical and horizontal interactions between institutions.

Horizontal interplay Channeling benefi ts to target benefi ciaries
The delimitation of target beneficiaries is difficult to implement in practice.Tourism revenues are often generated under complex circumstances, warranting varied claims.Perceptions of rights, and historical and present use by different stakeholders are obviously crucial.But so are the differential costs incurred by communities and households.And even if or when these complex circumstances have been addressed, tourism revenue sharing at the local level can still be subject to various types of governance failure.
Costs incurred for park-adjacency typically vary substantially between park-adjacent villages.Interviewees reported crop raiding as the most signifi cant cost.Households closer to the park talked more about crop raiding, particularly the individuals living near habituated gorilla groups.In some extreme cases, we encountered interviewees who reported that gorillas come to raid their banana gardens, and chew on the banana plant stems, much to the amusement of the tourists and their guides.It is the UWA's policy that local people maintain a distance from the gorillas.PA management zealously implements this, in particular when tourists are guided to gorilla groups feeding from community land.As one respondent noted, in such an incident, "One has to wait until the tourists have fi nished taking photographs [of the gorillas feeding on private banana stems]… this may take an hour… and… imagine how much will be eaten away in an hour."(Interview No. 52) Such episodes are obviously provocative, and add to feelings of helplessness, apathy, and anger, which strongly infl uence local attitudes towards park management.The crop raiding cost, as expressed in our study, is confi rmed by earlier observations that found the cost to an individual farmer to be particularly high (Tukahirwa and Pomeroy 1993;Kiiza et al. 2004).Baker (2005) estimates that an individual farmer adjacent to Bwindi would lose USD 472 in the 10 years life of a banana plant.Given that most of these farmers live on less than a dollar a day, this is a signifi cant loss (Kidd and Giampaoli 2006).Reports of food shortage have also increased with increasing reports of crop damage (Namara 2000;Olupot et al. 2009).The challenge is how to specifi cally channel the tourism revenues to the local individuals suffering the most.
Governance failure at Bwindi is observed in this study mainly in the form of elite capture and favouritism, political entrepreneurship, and lack of information on how to access the accumulated revenues.

Elite capture and favouritism
In the case of Bwindi, local people cited several incidents of leaders of the committees in charge of distributing revenues using them for their own benefi t instead.For example, one interviewee reported that she has great doubts as to whether she would ever benefi t from the scheme: "When they have brought the revenues, they have given to others and not me.The parish chairman and the [CPI] representative just choose themselves; they select themselves and do not consider us, the ordinary people."(Interview No. 17) Such claims that the ordinary people were too frequently left out, were made in all the sample villages.There were also claims that the representatives tend to give the revenue shares to their immediate families and people who have bribed them.Occasionally, claims were made of a single family getting between 4 and 6 goats.One interviewee said this about the process: The revenue sharing process is not transparent.We have a problem with our local political leaders.For example, I remember an incident where the head of the committee to distribute goats was from a family of six people.That family got six goats, while other families got nothing… I told you that revenue sharing and the park as such would be benefi cial to local people if the distribution of benefi ts was fair… In the mentioned incident, there was somebody in this village who is a close friend to the head of the committee, and in this village he was the only person who received a goat.Yet in the chairperson's village, about ten or more people received a goat or something else.(Interview No. 3) There were several reports alleging that local people have frequently been supplied with poor quality (cheap) goats, some even carrying infectious diseases which then resulted in the death of the original herd.The supply of goats under the revenue sharing scheme is done through a tendering process, but the integrity of this process is reported to be low.Villagers suggested that the individuals supplying the goats are friends of the leaders.
The problem of local elite capture seems to be widespread among tourism revenue sharing programmes across the world.In and around Kenya's Maasai Mara, nearly all tourism revenues were reported to be appropriated by local elites.Only about 6.5 per cent of the revenues identifi ed for the local people went to them, and the rest was siphoned off through various 'administrative mechanisms' and by direct embezzlement (Thompson and Homewood 2002).

Political entrepreneurship
In Bwindi, it was claimed that the representatives of the CPI and parish chairpersons tend to spend the revenues on people within their own villages as a way of rewarding their electorate.From the literature, this problem of political entrepreneurship (Byrnes and Dollery 2002) is a common phenomenon of local governance, in which representatives use councils as fertile grounds to capture the attention of prospective voters.Moreover, in the case of Bwindi, placing the CPI at the parish level means that the CPI representative may often not originate from a park-adjacent village.When the scheme is used to reward and capture the attention of electorates, it seems that park adjacent communities can easily be disadvantaged.This was noted by several park adjacent household members, who again pointed to the mismatch between conservation costs and benefi ts.Ribot (2008) stresses the benefi ts of linking conservation projects to local government institutions, thus making them more accountable to local voters, rather than to donors or park authorities.But even if elections (fair and free) can be a means of securing accountability, the Bwindi case demonstrates that a lack of fi t can still exist between voters and those suffering conservation costs, when the politicians favor only certain segments of their electorate.In Bwindi this lack of fi t is refl ected by the following comment: We are told it is only people living adjacent to the park who should benefi t, but you fi nd a person near Butogota [a distant trading centre] who has never guarded crops against the baboon and doesn't even know how a baboon looks like, getting a goat… It (also) often happens and it hurts us so much to realise that it is the people from the village near the sub-county [head quarters] where the leaders are from that are receiving all the goats.(Interview No. 19) Such political entrepreneurship is not unique to Bwindi.In a similar situation in the Mikumi National Park in Tanzania, nearly 50 per cent of park-based support for communityinitiated projects has been spent on projects in villages not directly bordering the park, including one village as far as 60 km from the park border.In the same area, high-ranking politicians infl uenced the transfer of wildlife revenues to a very distant district that even received wildlife revenues from other conservation areas and the national wildlife protection fund (Nyeme and Nilsen 2010).

Lack of access to information
In Bwindi, local people reported a lack of face-to-face contact, both with high ranking leaders and their own representatives at the local council committee that controls revenue sharing.In particular, households living adjacent to the park reported the lowest access to representatives and information regarding revenue sharing.One respondent noted: … The revenue ends up in the hands of those who get the information fi rst… they grab the fi rst chance… Communication needs to be improved.Let people know in advance what is going to happen… We learn of some opportunities when it is already too late.(Interview No. 8) The formal core function of the CPI is to represent the local community and its interests with regard to protected area issues, and to act as the offi cial spokesperson for the communities.This suggests that the involvement of local people should be a central feature, but as we have seen, this involvement has been consistently low.
Representation, as sought under the CPI, usually necessitates the convention of meetings where the community and their representatives deliberate on important issues.However, as suggested by the accounts of the CPI representatives, and also the local people themselves, attendance at meetings is usually low.The reasons include the high costs of attending meetings and the perception that the gain from doing so is minimal.As stated by one respondent: … One wonders why we should spend our time attending such meetings… We ask ourselves what good is it for us to attend a meeting where we are not going to benefi t anything?I myself would rather stay behind and guard my crops against the (other) baboons.(Interview No. 28) The CPI representatives felt that villagers could be more involved in the activities of the institution, but made the point that CPI members cannot move within and between villages or convene meetings to discuss park-related issues due to inadequate resources for facilitation.No specifi c arrangements are made to address the sharing of the benefi ts.Instead, opportunities are used as they arise.As a CPI key informant put it: Truthfully, we have no avenues to meet the local people.I have nothing to give the people… They want money [sitting allowances] when called for a meeting.So I try to use the available avenues, like church services, or I ask the chairman, when he meets his people… to create time for me to talk to them.(Key informant No. 15)

Lack of local involvement and infl uence
While rhetoric regarding the involvement of local people continues to receive substantial attention in PA planning documents, the reality remains elusive.Local people are seldom consulted to discuss decisions.For example, when the Ugandan Parliament decided that 20 per cent of gate fees should go to local people, no dialogue was held with them.From the inception of revenue sharing in 1996, local people have been asking why only 20 per cent and not a higher percentage?Some local people believe that park decisions are the prerogative of their representatives; others are of the opinion that attendance at the meetings provides no meaningful infl uence on decisions that are of local importance.One elderly village member commented: We lose trust because we realise that our pleas are not considered, and that we have no say… When we mention our suggestions to UWA offi cials, they… assure us that they will communicate with other top offi cials and give us feedback.Unfortunately, they usually tell us that our suggestions are not possible because there is no legal provision for it in terms of a policy for compensation.Or they say that UWA cannot accept this and that.(Interview No. 28) It appears that local people are not fully aware of their rights and/or duties.They don't know which aspects of management they can infl uence, and tend to view involvement in revenue sharing as a privilege.But, given the history of park management in Uganda, it may not be surprising that a participatory culture is yet to develop.In general, the literature shows that the experience of exclusionary events following the establishment of PAs in Africa did not go well with the local people (e.g., see Peluso 1993).In Bwindi, strong restrictions on access to park resources and the initial militaristic approach to managing the area are reported to have polarised the stakeholders into two opposing camps-park management and local people-that are hostile to each other.Some of the people have been very violent and have caused UWA employees to live in perpetual fear for their lives (Namara 2000;Sandbrook 2006), and vice versa.
Local people are presently not yet fully prepared to participate in park politics in order to infl uence outcomes.There has also been an extremely ambitious shift from park management working alone, having replaced the civil foresters with both paramilitary and uniformed protected area offi cers, to a more inclusive process in which local people are invited to express their opinions in terms of the management of PAs in their proximity.
However, power asymmetries currently exist both within communities, and between communities and the park management, and from the literature, these asymmetries are observed even in areas revered as the 'best cases' of community-based natural resource management, such as the Annapurna Conservation Area Project in Nepal (Timsina 2003).These findings are consistent with Ribot (1999), Agrawal (2001), andPlatteau andAbraham (2002), who provide numerous cases of special interest groups (such as local elites and NGOs) dominating local decision-making processes, often to the detriment of the poorest among the local people.The reluctance to involve the local people in park management can also be traced to the tradition in conservation that is rooted in fortress conservation ideology.Park managers need to change their mindset and practices, and to view local people as equal partners.
The involvement of local people in revenue sharing has tended to hinge primarily on the effectiveness of their representation.However, in Bwindi we fi nd that the lowest level of representation in charge of tourism revenue sharing (the CPI) is at the parish level, which is quite far from the local people.As several park neighbours noted, the CPI representatives are from areas that are far from the park boundary and from the adjacent local communities.Hence, their legitimacy is being questioned.Invariably, the individuals serving on the local councils are among the local elites, and are not necessarily those most affected by PA issues.A deliberate effort to have resource users from relevant villages serve as representatives would be more legitimate and ultimately more effective.
Successful representation is enhanced when the represented are able to organise, influence, or even call back their representatives (Blair 2000).In general, however, local people tend to have little opportunities to voice their concerns.For example, one interviewee noted: Corruption can never disappear in the distribution of resources that park management sends us… There is no way we can fi ght it.These resources come from high levels and at every level about 50 per cent is deducted… There is nothing a local person can do.It is those in park management who should change the way things are done.(Interview No. 12) Others remain silent about the irregularities in distribution for fear of losing their revenue share.Furthermore, many perceive the irregularities to be so entrenched in the social system that they will never be corrected.They thus tend to settle for the little that they can get.For example, in one revenue sharing meeting we attended, a local opinion leader encouraged people not to openly demand accountability, as this may put at risk the possibility of receiving any future benefi ts.Acknowledging that the revenue sharing process has been beleaguered by corruption, he stated: … There are villages that we know and in fact their chairpersons are here.They gave (goats to) people they were not supposed to give but we just kept quiet about this… But we shouldn't continue talking about those before people, because one time we will say such before a different audience and we will miss out on even the little that we are currently receiving.(Meeting 3)

Vertical interplay Institutional confusion
Starting with the rise of the conditions imposed by international development agencies in the 1980s, state contraction, increased decentralisation, and local governance have been pursued as policy in many developing countries; this tendency is partly linked to an increasing demand for democratisation as local people increasingly seek to influence decisions (Ouedraogo 2003).Increasing decentralisation creates both opportunities and challenges.For one, it implies a form of compartmentalisation in legislation, but it also confers agency responsibility.In Bwindi, the management of the national park is included under the mandate of the UWA, a parastatal under the Ministry of Tourism, Trade and Industry.However, park management operations are also infl uenced by the Ministry of Local Government.The Local Government Act of 1997 mandates local governments to oversee the management of natural resources within their jurisdiction, including PAs.The Act also requires the UWA to work with the local government in disbursing the local people's share of tourism revenues.But in the view of the UWA, the facilitation of the local sharing of tourism revenues is part of the activities of the local government.Thus the UWA has taken no steps to remunerate or support local government councilors since, in principle, such actions should be supported by their mother institution, i.e., the local government.
The Wildlife Policy (Republic of Uganda 1999b) encourages local governments to establish committees to advise the UWA on the management of wildlife within their jurisdiction.The CPI could have served as such a committee, but the problem was that the local government continued to regard CPI as a UWA structure, and on the other hand, the UWA viewed it as a local government structure.As a result, the CPI ended up having no 'institutional home', neither belonging to the UWA nor to the local government; consequently it was not adequately funded, and it ultimately failed its mission.
By ensuring that the CPI was composed of members of the local government in charge of natural resources, including national parks, the UWA was of the opinion that no new responsibilities were created for them through the establishment of the CPI.Yet, from our discussions with the local government offi cials, more responsibilities were implied.
In view of this mismatch of expectations, there was a need for some form of bargaining to encourage the local government to take on what, in its view, were 'new' responsibilities.
The bargaining process becomes particularly relevant given the history of the management of PAs in Uganda.Before the elevation of the forest reserve to national park status, it was managed by the local government's Forest Department, which, among other things, sold concessions for resource extraction.When the UWA was formed in 1996, park management was transferred to this parastatal, which constituted a fi nancial and resource loss, as well as loss of control of assets for the local government.The transfer of new responsibilities to the local government thus depended upon an elaborate bargaining process with mutually agreeable conclusions.Two particular spiral effects of this institutional confusion were the lack of facilitation for the CPI, and the lack of monitoring and evaluation of CPI activities.

Lack of facilitation
As several local government leaders noted, there was no possibility for them to draw on their already limited resources to support an institution that was doing what they saw as the job of the UWA.This fi nding confi rms that "when many rural parishes have an annual investment budget of less than USD 200, the choice between sending a parish representative to a distant location for a CPI meeting and building a new classroom in the village primary school is not a hard one to make" (Blomley and Namara 2003: 287).The reduced ability to conduct and attend meetings CPI (Figure 2).Members only met when the UWA and/or a UWA partner institution sponsored the meeting.
The local government's reluctance to support the CPI can be further understood from the events that led to some of the CPI representatives ceasing to be members of the local government.When the UWA formulated the CPI in 2000, the reigning secretaries for production, who automatically became CPI representatives, underwent training.These individuals, like other members of the local council, are voted for and serve for four years.When local council elections were held in 2002 and some of the members lost their seats, they should automatically have ceased being CPI representatives.However, the UWA reports that it did not have funds to train the new members and thus retained the old members as representatives of the CPI, regardless of whether or not they were members of the local council.The next local government elections were supposed to be held in 2004, and the UWA planned to train the newly elected secretaries for production.However, the central government did not conduct further elections.This means that the current representatives have been serving beyond their term of offi ce and they thus have weaker and less legitimate ties with the local government structure.This strengthens the views presented that the institution belonged to the UWA and not to the local government.Even the CPI representatives themselves, according to their revelations, regarded themselves as UWA revenue sharing employees, and were identifi ed as such by the local people.The impression also exists that they operate only when the UWA says there is some revenue to be shared.

Lack of monitoring and evaluation
Because of these unclear institutional arrangements, even if the institution had continued to be used for channeling revenues to the local people, there was no monitoring or evaluation of its activities.As a result, the local people were unable to hold their representatives accountable which created room for irregularities in the distribution of resources (as described above).When the CPI was created, it was entrusted with the main mechanism through which park management sought to demonstrate to the local people the economic importance of protecting biodiversity, which was intended to improve people-park relations.However, neither the park management nor the local government did any monitoring or evaluation of CPI activities.As one key informant from one of the UWA's partner institutions noted: In effecting the revenue sharing scheme, the UWA more or less undertook its activities as routine to meet its statutory requirements of disbursing 20 per cent of the park entry fees to the communities living adjacent.And this cannot have positive results if they [UWA] do not monitor to ensure that the money disbursed is used properly.(Key informant No. 18) The planned bi-annual updates from the UWA to the subcounty were irregularly reported.Even where made, the subcounties did not follow up on how the revenue had been used within the villages.During our interviews with the local leaders at the sub-county level, they maintained: … It is not possible for the sub-county to monitor the use of the park revenue shares because when that revenue comes, the sub-county is supposed to remit it to the parishes and nothing [meaning no share of the revenue] is left behind to facilitate the monitoring.(Key informant No. 26) But the park management position, as reported by a key informant, is that the local government should have considered monitoring in the same way it does with other projects or programmes of the Government of Uganda, such as the Poverty Alleviation Fund (PAF).The reluctance of the local government may again point towards its refusal to accept the activities of the CPI as falling within its jurisdiction.
Furthermore, the obfuscating effect of institutional bureaucracies may be a problem, as in the case of the gorilla levy fund mentioned above, where a decision made in 2008 to distribute revenues from 2006 to 2008 is yet to be effected in some target villages.Such institutionalisation and slow progress has also been observed in other benefi t sharing schemes, e.g., in Cameroon (Mayaka 2002).

Multiple stakeholders and power distribution
As we have seen, the sharing of tourism revenues involves different stakeholders, with different forms and strengths of power.This creates challenges as Foucault notes, in that some of the stakeholders may not deliver if, as a result of differential powers, they fi nd their "margin of liberty" limited (Foucault 1988: 12).This was confi rmed by our study in Bwindi, where several stakeholders reported extremely constrained margins of liberty in dealing with tourism revenues.Neither local people nor their leaders have any liberty to make (fi nal) decisions on how local tourism revenues should be shared.Such decisions have to be sanctioned by the UWA following the requirement that funded projects should be environmentally friendly and consistent with PA conservation objectives (UWA 2000b).
Although CPI representatives are supposed to be the communities' watchdog, they often found park management resisting their involvement.This is not a new phenomenon, e.g., Namara (2006) reports an incident in Bwindi that occurred during the late 1990s.A park warden complained about a CPI representative who, according to him, wanted to 'play the role of a warden' by asking for records of revenue-sharing funds and by trying to intervene in an incident where a local person had been imprisoned for illegal resource extraction.Yet, these are roles that park management itself defi ned for the CPI (UWA 2000a, b).Besides, as noted by Namara (2006), it is rather curious that the institution to be watched (the UWA) took the lead in defi ning the roles of the watching institution (CPI).While this may raise questions regarding the roles that were identifi ed, it is certainly a cause for concern that the CPI representatives are side-lined and their actions viewed negatively by some in park management, even when these are the core activities listed in the CPI policy.Although these limitations were reported during the establishment of the institution, they have, nevertheless, continued with implementation and management, casting further doubt on the possibility for the CPI to really take care of local people's interests.

CONCLUSIONS AND THE WAY FORWARD
Tourism revenue sharing is an appealing concept and its oftquoted logic of promoting conservation and rural development is diffi cult to ignore.However, despite its implementation around several PAs in developing countries, the mechanism has yet to deliver adequately.The main problems relate to the challenge of forming an effective organisational and institutional architecture.This includes the need for participatory planning, profi cient implementation, legitimate monitoring methods, and control and adjustment of policies and practices.
In particular, it has proved challenging to ensure that the local people's share meets their expectations, and that it reaches the most deserving communities, and/or individuals within a community.It is necessary to set aside suffi cient revenues for the local people, and craft legitimate and competent institutions that adequately involve local people in the decision-making process on both the structure and the process of distribution and utilisation of the tourism sharing revenues.
Furthermore, even if they are not adamantly against it, bureaucrats, local government officials, and wildlife management authorities remain sceptical about involving local people in park management.This reluctance refl ects particular management cultures, specifi cally, attitudes, values, norms, and practices, as well as present power relations.Local people, for their part-based on past experiences-are not convinced that their involvement will produce meaningful outcomes.
The decentralised framework within which natural resources are governed offers both opportunities and challenges.We do not suggest abandonment of tourism revenue sharing, but we rather suggest that a more concerted effort to overcome the mechanism's shortcomings, such as identifi ed here, may be the more rewarding constructive path to follow.The problem is not with tourism revenue sharing as an idea and as a concept, but with the diffi culties in putting it into real-world practice.
Figu re 2 Responsibilities and the institutions involved in disbursing the local share of tourism revenue under the CPI arrangement

Table 1 Institutional challenges relating to the problems of fi t and interplay for tourism revenue sharing at Bwindi Impenetrable National Park
To what extent do the local people infl uence the tourism revenue sharing process?How can tourism revenue sharing combine with other local PA efforts to benefi t the local people?Vertical (governmental levels)