Problem statement: Over the last fifty years, Eastern African wetlands in general and the lower floodplains in particular, have experienced major changes due to climate change, land use changes, upper catchment impoundments, agricultural irrigation schemes and species introductions. The effects of these changes have led to the loss of livelihoods for communities dependent directly on wetland resources. Rapid biodiversity loss in these wetlands has created a need for better assessments of the ecosystem functions, socio-economic values and important biodiversity. In the past, several independent studies have been carried out in some wetlands but information necessary for mitigation, conservation and management has failed to reach policy makers because of the inability to communicate it in a suitable form that would be effective for this purpose. This project intends to coordinate such a team. In order to acquire the amount and variety of data required for the assessment of ecosystem functioning of wetlands, a team of researchers from various fields is required. In other words a multi-disciplinary team consisting of biodiversity experts, hydrologists, ecologists and socio-economics experts is essential.

A need for multidisciplinary studies on Eastern African wetlands: Wetlands are considered as the most productive ecosystems of the planet (Barbier 1994, Costanza et al. 1997, Bravard et al. 2000) and provide various services both at the local and the global scale. As a consequence, such areas are used by a large number of stakeholders exploiting a variety of resources. Some of these resource uses are in competition and the wetland conversion or management envisaged by the different stakeholders may be incompatible. For example, wetlands are among the ecosystems strongly targeted by biodiversity conservation programmes but concurrently coveted for irrigation projects. Irrigation programmes, when they are not well designed or technically too ambitious, can destabilise the local livelihoods that are based on complementarities between agriculture, fishing, livestock keeping and other natural resource use activities in the socio-ecological landscape. Even though considered as productive, biodiversity of these wetlands and especially tropical wetlands are not well known and the services provided by its ecosystems are not well described.

Moreover, the natural resource use of these ecosystems, traditionally exploited by various user groups, is in general regulated by complex access rules and land tenure systems. These rules and regulations often have a temporal dimension adapted to the natural flood regime. For example, in the Casamance estuary in Senegal, M.C. Cormier- Salem (1992) describes an environment structured into “aquatic terroirs”, exploited by the local communities through various ways and means depending on their state of flooding. The lower delta of the Senegal River, before the construction of the Diama dam, was similarly structured as a multi-use system (Duvail 2001): land is not appropriated exclusively by a single community but characterised by a multiplicity of rights for access, resource use and exploitation depending on the season. This type of complexity is not specific to coastal wetlands and also characterises the differentiated access rights in the Inner Delta of the Niger River. Here the balance of power between the various user groups is based on time-honoured covenants on the use of land and resources which go so far as to define „ethnicity‟ on the basis of resource use specialisation. C. & O. Barrière (1995) used the term “foncier-environnement” to describe this form of spatial organisation based on the partitioning of aquatic resources. In order to facilitate the understanding of these complex systems of land and resource tenure, the AFREPA („Association pour la Promotion des Recherches et Etudes Foncières en Afrique‟) has defined a typology of land tenure going from simple access to total control (including the right of transfer) (Le Bris et al. 1991, Le Roy et al. 1996). These access rules and the ways and means through which they are applied are complex, dynamic and usually unknown to the government institutions who often superimpose their own legislation onto the system by privatisation or by state control. However, some solid theory (Ostrom 1990) and various examples (Berkes et al. 1989) show that the common resource governance, widespread in rural Africa is economically efficient, socially acceptable and ecologically sustainable (Borrini-Feyerabend et al. 2004).

Complex, coveted, unknown, wetlands are also fragile. They depend strongly on the supply of freshwater from groundwater or from upstream and therefore on land use in their catchment. In the strongly seasonal environments of sub-Saharan Africa, the annual flood is a key element structuring the landscape and determines to a large extent the productivity of wetland ecosystems and therefore their associated livelihoods. Impeding the natural flows of water and sediments and altering natural seasonal patterns of river discharge has repercussions on the functioning of the wetland ecosystems.

Over the last fifty years, African wetlands ecosystems have experienced major changes for four main reasons:

–  Increased extreme events due to climate change (IPCC 2007), with both prolonged droughts and extreme rainfall and flooding. Predictions for Africa are an increased frequency of very dry and very wet years and an increase in the frequency and intensity of tropical storms in the Southern Indian Ocean (Boko et al. 2007) in a changing climate context change in rainfall pattern at the river basin scale land use change, typically deforestation, has impacted on run-off conditions and modified the hydrology with an increased risk for severe flooding (Bradshaw et al. 2007), a decline in dry season flows and an increase in sediment loads (Syvitski et al. 2005)

–  Hydro-dam construction which alters the hydrology of the rivers even more radically as the natural flood peak is reduced or suppressed altogether. Though dams have obvious advantages (development of irrigated agriculture, supply of electricity and water to urban areas, river regulation for navigation) the flood releases from the dams are often programmed according to sectoral needs (World Commission on Dams 2000) and the resulting reduction or suppression of flood peaks has negative impacts on downstream wetland ecosystems (Bergkamp et al. 2000). The creation of stagnant waters within a river system transforms the composition and structure of biological communities, directly by modifying migration patterns and indirectly by hydrochemical changes.

–  Numerous irrigated agriculture projects, in the rift valley for flower farming and on the coast for biofuel production, have converted or are planning to convert large tracts of land. Water abstraction for irrigation can potentially have strong impacts on downstream hydrology both quantitatively and qualitatively by the increased discharge of fertiliser and pesticides (Davis & Koop 2006, Johnson & Ebert 2000) while at the same time the exclusion of the multi-user groups from the vast tracks of land they traditionally exploited can destabilise the local economy.

In combination, these oncoming changes have the potential to destabilise the local wetland-dependent economies (Snoussi et al. 2007). To avoid the conversion of wetlands at the expense of the local users, decision-makers will need tools allowing them to ensure an equitable partitioning of water resources across different uses (local, agricultural, hydropower, environmental).

Integrated management of the wetland: Still, economic development does not automatically lead to biodiversity losses nor does it irrevocably cause a crisis in the local economies. It is possible to manage the water in ways that take into account the production objectives and, at the same time, are respectful of the water needs of the traditional users and the wetland ecosystems.

However, effective management of the wetland requires: – a better description of the biodiversity values of the wetland – a hydrologic model and a spatialised hydraulic model of the wetland. – a thorough understanding of the water needs of the ecosystem in quantitative terms. – a detailed analysis of the practices and socio-economic strategies of the various stakeholders and a quantification of the water demand.

Unfortunately, the impact assessments of the various projects are often conducted under time constraints that do not allow adequate description of the existing biodiversity and of the resource use calendars (fishing, agriculture, livestock keeping, gathering, etc.) under various flood regimes. Detailed descriptions and spatial models of the interactions between the hydrology, the availability of the natural resources and its biodiversity values and their use may contribute to improve the decision making process.

Traditional natural resource management: As the Tana River flows through semi‐arid lands its water is a highly precious commodity attracting different types of socio ecological landscape uses. Historically, complementary and mutually beneficial resource exploitation strategies between the various user groups have always regulated rights of access to key resources, especially during the dry season. Elders of both Pokomo and Orma groups would jointly perform the required rituals after which the Orma could access water‐points and floodplain pasture (Kagwanja 2003). In combination with low population density and sufficiently regular and extensive flooding these practices have allowed the different ecosystems to thrive and maintain the exceptional biodiversity. Thus 19th century explorers that travelled along the river described “an impenetrable jungle” and “beautifully foliaged trees covered with creepers fringed both banks of the river” (Gedge 1892, cited in Hughes 1984). Indeed, the forests, and especially the clay evergreen forest are considered a highly valuable resource even today (Luke et al. 2005). About a hundred plant species from the forests are commonly used by local people, especially for technology (43 species) and construction (34 species), traditional remedies (23 species) and food (15 species) and the main impact on forest structure is the use of large trees for canoes or beehives (Medley 1993), honey being particularly important to the Pokomo. Because of the shifting and meandering nature of the river and the forest use practices, the forest always has different stages of succession and is characterized by dynamic carbon storage (Glenday 2005). Though they need large trees, Red Colobus seem to prefer forest edges over mature forest (Mbora and Meikle 2004a) and the semi‐terrestrial Tana Mangabey are well adapted to a landscape mosaic with alternating small fields, bush and forest at different stages of maturity. According to Terer et al. (2004) the floodplain wetlands and in particular the oxbow lakes (recession agriculture of sorghum and millet, fishing, reeds for roof thatch, fresh water, grazing) are also considered of high value, as is the Tana River itself (water for various uses including irrigation, transportation, sand for building and as a protective barrier against bandits). The floodplains have a very important function also in the attenuation of flood peaks through the overspilling of the banks, storage on the floodplains and in the oxbow lakes, infiltration into soils raising the groundwater level, etc. thus protecting downstream areas from their destructive power. Though hunting is illegal in Kenya the rich wildlife associated with this mosaic of different forest types, wetlands, rangelands and small‐scale agriculture continues to provide animal protein. Elephants have virtually disappeared during the late 1980s, because of intensive poaching, but various ungulate species are still quite abundant in the area and human wildlife conflict, especially with buffalo, is a major complaint of the local communities when interacting with the Kenya Wildlife Service.

For the Rufiji system, as described in Duvail, Hamerlynck (2007) the river is part of the identity of the Warufiji. There is no “Warufiji” tribe as such as the designation covers a mixture of several lineages and ethnic groups, around the eponymous Rufiji River. These define themselves as floodplain farmers in contrast to the hill tribes that practice shifting cultivation in the coastal forests and with whom they have a “joking relationship” (Radcliffe-Brown, 1940, 1949). Floods are essential for the sustenance of floodplain fertility, and therefore of the farming system, and as vital for the productivity of most of the natural resources on which the communities depend: forestry, fisheries, wildlife and, more recently, grazing. The farming system is well adapted to the variation of the hydrological conditions and is rather sophisticated: the Rufiji Floodplain farmers have come to grips with the interplay between short rains crops, long rains crops, floods and recession agriculture and the subtle use of the topographical variability, which determines the nature of the soils and their flooding frequency. On the higher non-floodable levees, farmers plant perennial crops (mango, cashew, banana) and some annual rainfed crops (maize, sorgho, sesame). In the floodable area, the preferred farmland is on the Mbawila soils of the levees, consisting of loam and fine sand, relatively easy to work, well drained and suitable for a variety of crops but in general dominated by maize. These soils are most commonly interspersed with the lower- lying depressions with Finyanzi (dark heavy clay), which are suitable for rice. By cultivating a number of small plots (about 0.4 ha each) with different soil types and at different topographical levels, often at dispersed locations, each household harvests from about 1.5 ha of cleared floodplain land each year with double crops in years of good rainfall and/or adequate floods. Inside each plot, by making judicious use of their knowledge of the land, they also plant according to a risk avoidance strategy, intercropping maize with rice on the slopes of the depressions and using different varieties planted at different times. Thus, under a wide range of rainfall and flood behaviour, they are likely to achieve sustenance and in, favourable years, produce considerable surplus. Under such favourable circumstances the staple food crops (rice and maize) are also the main cash crops in Rufiji. For one given year, the success of these two staple foods is dependent on a subtle balance between rainfall and flood characteristics. The best agricultural scenario for a farmer is when there is conjunction between a good rainy season and a flood. Most of the “safety nets” (fisheries, hunting, gathering of wild fruits and honey) are also favourably influenced by the flood. Even though virtually all floodplain users identify themselves as farmers, agriculture accounts for only about 37% of average household income, which is supplemented by fishing, forestry (especially in drought years) and a host of other activities.

The importance of recognition of these traditional management systems is for the purpose of ensuring that modern management systems for these landscapes applies a bottom-up approach that ensures sustainability of conservation and an equilibrium in the balance between resource use and management. For co-management to be efficient there must be ability to agree upon, guarantee of equity and implementation of a fair share of management functions, benefits and responsibilities of a socio ecological landscape. This definition applied from „Sharing power’ by Borrini et al. 2004 emphasises on the need for trust among partners, consideration of diverse interest of communities and need to apply an iterative or participatory process and to build upon customary and local organisations.

Study sites: Two study areas Tana and Rifiji River Deltas (Kenya and Tanzania respectively) have been chosen for the present.  These will serve as models for study of other eastward flowing rivers. They have been chosen since both have experienced major changes in their hydrology and both are currently targeted by development projects that could greatly modify the functioning of the landscapes and the livelihoods of the local communities.

The Lower Rufiji Valley: The Rufiji is the main river in East-Africa (in terms of discharge, it is the 6th biggest river in Africa). The site is of interest for several reasons. From an analytical point of view, various data are available on this study site. Some studies on the agricultural practices (Bantje 1980, Havnevik 1993) from the 1970s by the BRALUP (Bureau for Resource Assessment and Land Use Planning). The IRD-IRA-IFRA project on the lower valleys has produced some hydrological, climatic, environmental and socio-economical data that can be exploited in an integrated modelling exercise.
From a management point of view, the Stiegler’s Gorge dam project, if implemented, would have a profound influence on the area. Most of the agricultural production of the area is flood-dependant and the sophistication of the traditional farming system has long been recognised by the research community (Havnevik 1993, Hamerlynck 2003). However, at the local administration and central government levels the predominant view is that the traditional system is ‘primitive’ and inefficient (Duvail and Hamerlynck, 2007). This was one of the reasons why, during Ujamaa villagisation, farmers were uprooted from their familiar agro-ecological zone, the floodplain. The dam construction plans would limit the peak flood discharge to 2500 m³/s, confining the river to its dry season channel and effectively excluding the continuation of recession agricultural practices. The economics of replacing these practices by irrigation are very negative, even under highly optimistic assumptions of irrigated surface areas and yields. As the main objective of the dam is the production of hydropower (for export to Southern Africa) the impacts on agriculture have not been sufficiently taken into account. The dam would also have very negative impacts on downstream ecosystems and on flood-dependent livelihoods (fisheries, mangrove exploitation, etc.). It would therefore seem to be necessary to rethink the dam design from a managed flood release perspective. The design of a managed flood release scenario that would maintain ecosystems and livelihoods downstream has to be based on sound knowledge of the current natural resource use practices and strategies (Duvail & Hamerlynck, 2006). In parallel, numerous irrigation projects are under development for the Lower Rufiji Floodplain, either for cotton or for sugarcane for agrofuel production. These are likely to compete for water and land resources with the traditional farming and fisheries practices.

The Tana Delta: The functioning of the coastal wetlands associated to the lower valley of the Tana have been strongly impacted by the alterations in its catchment. The study of how this came about and what it implies is highly relevant for the implementation of our research. The Tana is the largest river in Kenya. It originates in the Aberdare Mountains and the Mount Kenya and  subsequently flows through about a thousand kilometres of drylands in eastern Kenya where its average flow is around 150m³/s. Its river basin covers approximately 127,000 km². In its lower part, the river has built up a delta of about 1300 km². Five hydropower dams were constructed in its upper reaches: Kindaruma (1968), Kamburu (1975), Gitaru (1978), Masinga (1981) and Kiambere (1988). These 5 dams produce 3/4 of the power needs of Kenya and also supply the capital Nairobi with domestic water.

The Tana River Delta is currently confronted to a substantial reduction of river discharge and a decrease in the amplitude of the flood peaks, which is bimodal because of the two rainy seasons. The decrease in the amplitude of the flood peak is particularly significant since the completion of the Masinga Dam in 1981 (Maingi & Marsh, 2002). A new dam, downstream of the 5 existing dams, has been proposed at Mutonga-Grand Falls. This dam would further more attenuate the bimodal flood peak with highly negative impacts on fish production (Mavuti 1994 in Emerton 2003) and on the riverine forest (Maingi & Marsh, 2002). The livelihoods of the 200,000 inhabitants of the Lower Valley, dependent on recession agriculture, fisheries and livestock keeping, would be seriously affected (Emerton 2003). The attenuation of the flood peaks will also have impacts on the livelihoods of the nomadic livestock keepers and their estimated 2.5 million heads of cattle that currently use the pastures in the middle and lower reaches of the valley (CADP 1991 in Emerton 2003).

However, none of these impacts have been quantified in detail, nor spatialised. The ecological studies in the Lower Valley were primarily targeted at the floodplain forests upstream of the delta because of the presence of the red-listed endemic primates Tana River Red Colobus (Procolobus rufomitratus) and Tana Mangabey (Cercopithecus galeritus). Densities of Red Colobus are strongly correlated with forest quality, notably the presence of trees with a large diameter

(Mbora & Meikle 2004a). In its turn forest quality if strongly correlated with the flood hydrograph (flood height, frequency and duration) that determine forest regeneration and groundwater level (Hughes 1990).

A recent development is the proposal for a vast irrigated sugar cane project for agrofuel production, If implemented this will result in strongly reduced freshwater outflow to the ocean. The local economies will be strongly affected.