Introduction

When the spatial structure of an area is cast in terms of uneven development, it is often conceptualised by means of a “core-periphery model”. The “core” area consists of those places that wield the greatest economic power and have accumulated the highest levels of affluence. These are the places and regions for which the system creates development. The “periphery” lacks a stable power base, is characterised by a lack of wealth, and serves as supplier of capital, labour, or commodities to the core. These are regions through which the system creates development. Although most concepts or measures of uneven development would identify a continuum rather than a dichotomy (Scott 2012), core-periphery thinking provides geographers with a useful analytical framework, as it highlights that uneven development needs to be understood from a relational point of view (Brown et al. 2010). Indeed, the crux of the core-periphery model is that it seeks to explicitly grasp the vicious cycle of the “development of underdevelopment” characterising peripheries (Frank 1971).

This conceptualisation contrasts with the more traditional regional-geographical concept of a “periphery”, which is used to identify places or regions that are distant from a centre or situated on the fringes of a city, region, or country. Although both notions of “periphery” may empirically coincide, the use of this term in light of a core-periphery model is theoretically richer as it points to on-going processes of “peripheralisation”: the production of an accumulation base through uneven socio-economic relations across space (Kühn 2015).

Core and periphery very often remain more or less stable in space, and this in spite of major changes in socio-economic activities unfolding within them. The resources being extracted and the low value-added commodities provided by the periphery may change over time. However, this does not herald a fundamental change in a region’s peripheral status. Rather than specific socio-economic activities and processes, therefore, both core and periphery are defined by persisting, uneven exchanges. Although the core-periphery model is most commonly marshalled to conceptualise uneven development at the global scale (Knox et al. 2014), it has been deployed at various scales to grasp persisting patterns of uneven development (Friedmann 1966; Hoggart and Buller 2015; Labrianidis 2017). It can also be treated as a hypothesis to be tested (Arrighi and Drangel 1986), by examining whether core-periphery contrasts increase through time, or are rather being subdued, leading to “deperipheralisation” (Colmeiro 2018) or “non-peripheral integration” (Arrighi 1979). It is in this framework that, in this paper, we adopt the notion of peripheralisation to conceptualise and empirically document persistence and changes in the Beles basin of Ethiopia.

Ethiopia presents an interesting case in this regard, as the agriculture-based civilisation that developed in the northern highlands (McCann 1995) could use its topographic and socio-economic advantages to develop a state that remained independent (Scaetta 1935). As the only African country that was never colonised, Ethiopia could incorporate besides the northern highlands, the southern highlands, and lowland areas located at its eastern, southern, and western margins through the eighteenth and nineteenth centuries (Zewde 2002). It has often been claimed that especially the lowland areas of Ethiopia have remained peripheral in both senses of Kühn’s (2015) concept: both in terms of topography and distance, and in terms of how its development, its economic, and human rights have been defined in unequal power relations. Markakis (2011) for instance considers two partly overlapping concepts: the dichotomy between highlands and lowlands, and a more politically defined concept of concentration of power in the hands of central, highland elites that rule with the help of dominated peripheral elites and migrations of highlanders to the lowlands.

At the onset of the twenty-first century, the Climate Resilient Green Economy (CRGE) strategy has become the framework of Ethiopia’s developmental activities (Death 2015; Jones and Carabine 2013). The CRGE is based on four pillars: (i) crop and livestock for food security and farmer income while reducing emissions; (ii) forests for their economic and ecosystem services, including carbon stocks; (iii) electricity generation from renewable sources of energy; and (iv) energy-efficient technologies in transport, building, and industrial sectors (FDRE 2011).

Here we investigate the relevance of the notions of “core” and “periphery” in the geographically peripheral Beles basin, a region situated at the western margin of the Ethiopian highlands which is characterised by a very different environment and indigenous culture, and where recent important investments have taken place in the framework of the CRGE. Hence, we explored the regional human-environment interactions in this basin, as well as the important changes that occurred over the last 50 years. In recent decades, important changes have occurred in relation to road building, resettlement, international land deals, irrigation, and water transfers, which makes the region exemplary for remote area development (Pelican 2009; Saugestad 2001) in Africa south of the Sahara. After introducing the Beles catchment, we review the existing literature. First, the scene is set through a detailed presentation of the abiotic, biotic, historical, social, and cultural characteristics of the study area. Then, the processes of change (and persistence), particularly in relation to water, migration, infrastructure, and socio-cultural relations, are reviewed starting from the first account by Emperor Sertse Dengel’s chronicler (ca. 1590). Notably, we compare the current situation with that of around 1960, as described in “Land, economy and settlement of the Gumuz” (Kuls 1962). Based on the observed human, climatic, geological, and agricultural characteristics, we (i) propose a regionalisation of the Beles basin; (ii) identify knowledge gaps; (iii) investigate to what extent persistence and changes align with the CRGE strategy; (iv) assess the emerging or likely implications for land, water, biodiversity, cropping, and livelihood; and finally (v) analyse whether the changing conditions of the Beles basin at the beginning of the twenty-first century are translated into deperipheralisation.

Research methodology

We compiled an extensive review based on existing scientific papers (using Web of Science) as well as grey literature (reports, master theses, etc.) regarding the Beles basin. The latter were retrieved from the different co-authors’ thematical archives as well as through visits to district offices in the basin.

In addition to disciplinary investigations, multidisciplinary field visits were undertaken along the route Guba–Gublack–Manbuk–Gilgel Beles–Mandura–Chagni (Fig. 1), with lateral incursions at many places, including Gumuz settlements and commercial farms west of Gublack, the middle Beles basin around Pawe, and the slopes of Belaya Mountain. Additional visits were made to the upper part of the Beles basin which could be accessed through service roads prepared for the Tana-Beles hydropower installation, and further on foot.

Fig. 1
figure 1

Beles basin at the western side of the Ethiopian highlands. G.B. Gilgel Beles town

Full size image

To integrate and conceptualise the disciplinary and interdisciplinary information collected, three brainstorming workshops among co-authors took place on 25–29 January 2016 in Brussels, on 11–14 November 2016 in Dangur, and on 18–20 January 2017 in Bahir Dar and the Beles basin.

The Beles basin in northwest Ethiopia

Study area

The Beles basin (Fig. 1) drains 13,571 km2 of the Western Ethiopian escarpment and its foot slopes towards the Blue Nile. The basin, with elevations ranging between 540 and 2755 m a.s.l., is mainly inhabited by Gumuz people (Simons and Fennig 2017; Fig. 2), who are also the least defensible and resilient population group of the basin (González-Ruibal et al. 2007; Swedesurvey 2010). Until not so long ago (Kuls 1962), the edge of the escarpment marked the beginning of an entirely different world that was largely inaccessible, inhabited by the Gumuz people with very different culture as compared to the highlanders, and also a totally different agroecology and agricultural system.

Fig. 2
figure 2

Ethnic groups and languages in Beles basin and surroundings. After Simons and Fennig (2017) and field observations

Full size image

Physical geography

The geology of the Beles basin comprises extensive Tertiary trap basalts and strongly folded and faulted Precambrian rocks (Fig. S1, Fig. S2). Mesozoic sedimentary rocks make up thin intercalations between both major formations.

Annual rainfall shows a unimodal distribution and is well over 1000 mm year−1 (Fig. S3). Maximum temperatures rise high, up to 40 °C in early afternoon in the lower parts of the basin, with a daily contrast that is stronger than the seasonal contrast. Global warming tends to lead to slightly increased rainfall over the basin by the end of the twenty-first century (Barnes 2017).

The Beles River, originating from the south-western escarpment of Lake Tana water divide, is a perennial river flowing towards the Blue Nile. At the foot of the escarpments of the highlands and also of inselbergs, it is quite common to observe springs (Fig. S4). Few studies attempted to estimate flows at various points in the basin using rainfall-runoff models, whereby results indicate that the total runoff generated in the Beles basin would be approx. 5690 × 106 m3 year−1 which corresponds to a runoff depth of 407 mm year−1 (World Bank 2008).

Groundwater occurs in Quaternary unconsolidated deposits (within the upper few tens of metres) in the relatively flat lands, and in fractured rocks or along major regional faults at shallow to intermediate depth. The general groundwater flow is from elevated recharge areas to low-lying discharge areas (Kebede 2013). The presence of deep-seated geological structures, as well as that of ancient fluviatile deposits and several episodes of volcanism in the Tana area (Hautot et al. 2006; Prave et al. 2016), may favour inter-basin groundwater flow from Tana basin to Beles basin, which needs a critical investigation.

The larger part of the Beles basin consists of “dry combretum wooded grassland” (Friis et al. 2010). Characteristic species are different Combretaceae; further, and particularly close to the step to the highlands, very often Syzygium guineense, Gardenia (a sacred tree for the Gumuz who call it “kota”), and Protea are found. The understorey is a combination of herbs and up to 3-m-high grasses. Many of the species have developed under the influence of recurring fire and as a result, the trees have evolved very thick bark while most herbs developed perennial bulbs (IBC 2005). The Beles River is accompanied by narrow riparian forest strips (dominated by Syzygium guineense), and on its extensive gravel fields, Tamarix sp. settles easily. Extensive lowland bamboo stands (Oxytenanthera) accompany the entire western edge of Ethiopian highlands and also penetrate into the larger westbound valleys (Embaye 2000). On the Wombera, Dangur, and Belaya Mountains, the natural vegetation is a moisture-loving mountain forest, composed of Hagenia abyssinica and Podocarpus gracilior. The biophysical environment of the Beles basin is further described with much detail in the Electronic Supplementary Material.

Settlement and resettlement

The Gumuz used to live in the highlands but were through the centuries pushed towards the less accessible yet fertile bush-savanna lowland environment through the expansion of the Amhara and Agaw (James 1986; Swedesurvey 2010). They were also exposed to perpetual slave raids by highlanders and Sudanese, both for their own use and for sale to the Arab world (Ahmad 1999; González-Ruibal et al. 2007; Markakis 2011; Pankhurst 1977). Slavery only disappeared from Ethiopia in the 1930s–1940s (Edwards 1982; Miers 1997). The Gumuz of today still have vivid memories of the slave raids, which occurred as recently as the 1930s (Ahmad 1999; Swedesurvey 2010). Crop production in Beles basin took already place in 1586, as Emperor Sertse Dengel’s chronicler (ca. 1590) reported that after “taking lots of cattle and slaves” in a hot plain near a large water (presumably Beles River), “the king sent out his troops to pillage grain and burn houses” on the slopes of Mount Belaya.

Neighbouring ancient population groups within the Beles basin are living in higher elevated places: Agaw on Mt. Belaya and the upper catchment, Amhara in the headwaters, and Shinasha on Mt. Dangur and around Wombera (Fig. 2). According to oral transmission, the Agaw would have come to Mt. Belaya around 1600 CE. The Beles basin roughly comprises the districts of Metekel zone in Benishangul-Gumuz region, as well as parts of three districts of the neighbouring Amhara region. For an analysis of the complex historical relations between different population groups in the region and its neighbours, the reader is further referred to González-Ruibal et al. (2007), Abbute (2002), Markakis (2011), and to the Temesgen Gebeyehu (2018).

Resettlement to the Beles basin lowlands, which lie in the Benishangul-Gumuz region, is particularly linked to its land and water resources. The potential for irrigated agriculture is estimated at 115,000 ha (World Bank 2008). The establishment of roads and large commercial farms as of the year 2000 enhanced changes in many places in the basin. In 2010, a drastic change came with the operation of the Tana-Beles hydropower scheme that transfers large volumes of water from Lake Tana directly to the Beles River via a 12-km-long tunnel to generate 460 MW hydroelectric power (Clark et al. 2013). Since the 1920s and 1930s, such a hydropower scheme had been considered (Grabham and Black 1925; Hårsmar et al. 2016), and anticipating on large inter-basin water transfers, the Ethiopian Government organised infrastructure works and population resettlement to the middle Beles basin in the 1980s. The most recent change is the construction of the Grand Ethiopian Renaissance Dam (GERD, or hidase in Amharic) that will eventually lead to flooding of lower stretches of the Beles basin (Fig. 1).

Results

Persistence and changes in the Gumuz society

Gumuz is one of the ethnic groups in the northwestern part of Ethiopia, with ten clans, each having their own territory. The Gumuz language belongs to the Nilo-Saharan family (Ahland 2012). At the last census, the total population of Gumuz in Ethiopia was 179,348 (CSA 2008), as well as approx. 40,000 in Sudan (Simons and Fennig 2017). Kuls (1962) estimated that there were around 20,000 Gumuz in the Beles basin. Currently, an estimate of 80,000 Gumuz people residing in the Beles basin seems reasonable. The pressure of the more powerful Afroasiatic (Amhara and Agaw) people on the Gumuz territory continues in the twenty-first century (Abbute 2002). The Gumuz practise shifting cultivation and collect wild fruits, roots, seeds, and honey. Hunting (Fig. S7) and fishing complement the diet. Fishing takes place with nets as mentioned by Kuls (1962) but also increasingly through the use of poisons such as Millettia ferruginea seed, bark, and agrochemicals that are dangerous for the whole aquatic ecosystem. These fishers aim at harvesting all fish from a water body at a time for the dry fish market. Such poisoning with agricultural chemicals is seasonal and occurs mostly during Orthodox fasting seasons.

Traditionally, and nowadays still in the majority, the shifting cultivation also involves frequent shifting of villages that are constructed in light materials (Fig. 3). Permanent Gumuz settlements now have come into existence, which do not differ much from the “mobile” settlements described and mapped by Kuls, with the notable exception of “home gardens” that have become squared in relation to more densely populated settlements. Granaries are decorated with male and female genitals (Fig. S8); Kuls (1962) mentioned and illustrated other decorations such as gift bowls or breasts. Either there has been an evolution in decorations or most probably decorations (always related to fertility) vary among clans.

Fig. 3
figure 3

Views of Gumuz homesteads (photos J. Nyssen)

Full size image

In addition, the Gumuz depend on other subsidiary livelihood sources such as traditional gold mining, firewood collection, charcoal preparation, and handicrafts (Moreda 2015). The Gumuz people also nurture traditional medicinal plants to solve health-related problems. The livelihood of Gumuz communities depends mostly on nature because of marginalisation and low orientation to technology. The Gumuz practise polygamy and exchange marriage, i.e. the man proposes a female relative (generally a sister) as wife to the brother of his bride (James 1986; Zeleke 2010). Girls marry early and move to the homestead of their husband; as marriage is not allowed within the same clan, they generally move to a place that is quite far away. Going to stay with their husband means having to get to know his relatives, his clan, and a new environment. “This makes young wives shy, often scared” (Swedesurvey 2010). An in-depth study of gender issues among the Gumuz of the Beles basin was carried out by Zeleke (2010).

Male elders exercise traditional administrative and judicial responsibilities for all clan members (Swedesurvey 2010). In recent years, these elders, obusuma in Gumuz language, have also taken up government responsibilities, which has led locally to loss of legitimacy.

In remote places in the northern part of the lower Beles basin, and especially in its less accessible southern part, the ancient Gumuz shifting cultivation farming system has persisted. Currently, the customary land tenure system co-exists with the formal land policy formulated by the regional government, since land registration and certification is at its early stage. The Gumuz usually have large areas of communal land which is used as grazing field, for collection of firewood, honey, and wild fruits, or for slash and burn agriculture (Gebre 2003). There are two traditional types of land ownership: (i) community ownership and (ii) lineage groups’ ownership. In the case of the community ownership, all members of the community of a given area have the right of ownership on cultivable virgin lands, forests, grazing lands, and river banks, whereas under the lineage groups’ ownership, clan members and their descendants have the right of possession as well as passing to their family members. In the Gumuz community, this right is applicable as long as they live in the lineage group. If they leave the group, whatever the case, their possession right ceases (Jira 2008).

Collective land ownership rights are administered by the clan system (Ahmad 1999). The clan is considered the owner of the natural resources; individual members are accorded only usufruct rights. Therefore, the resources cannot be possessed or controlled by single members of the community. Elders select the appropriate location for a hamlet based on both vegetation and soil types. Inter-clan relations require recognition of each other’s territories and rights. Natural landmarks such as rivers, hills, big trees, rocks, roads, or footpaths mark the territory of a clan.

Gumuz and settlers’ agricultural systems and commercial farming

Patchwise, the centuries-old Gumuz agricultural system which was described in much detail by Kuls (1962) has been maintained until now. It involves shifting cultivation, and the main tool used for land clearing and preparation are fire (Fig. S9) and the tiba, a hoe of which the blade is inserted approx. at a 45° angle (Fig. S10), which must be unique worldwide.

Direct sowing of crops in a zero-tillage system is implemented, in which planting holes are made using a planting stick fitted with a sharp iron tip. Use of oxen-plough is difficult in the traditional farming system because many trees and roots remain after land clearing. “The irregularly limited farm plots are mostly quite far away from the settlements, and most of them are cultivated for only one or two years. After that, the land remains for a longer time as bush fallow” (Kuls 1962), locally called bukuna. Dominant crops in this agroecosystem include sorghum and sesame. Ginger, finger millet, teff, and maize are grown especially around Mandura. Sorghum is the main food crop of the Gumuz people, and it is largely grown nearby settlements, because it needs protection. The fires that widely burn the Gumuz lands in the dry season are not to be considered as “deforestation”, as stated for instance by Adisu (2010) and many others, rather they are considered as a natural process, related to the large build-up of dry biomass. van Breugel et al. (2016) showed that 64% of the area was burnt at least once in the period 2003–2013. Adaptation to fire occurs in grasses, shrubs, and trees, so natural fires must have occurred long before now (Gashaw et al. 2002). Currently, fire is also used as a land management technique, aiming at the removal of stover and dry grasses and provoking the emergence of shoots of fire-resistant species in the woodlands (Fig. S9). Settlers and some commercial farms have adopted the same technique of land clearing, even on permanently cultivated land. On the other hand, such practices are known to lead to decreased flora and (pedo)fauna biodiversity (Platt et al. 2012, and many others). In many tropical regions, debates are ongoing regarding pros and cons of wildfire. Indeed, the Gumuz have been encouraged to change from shifting to permanent cultivation and to establish in permanently settled villages. Land ownership became an important issue, particularly after the resettlement of highlanders, starting in the 1980s. The introduction and emergence of a modern ruling structure of the central government made the Gumuz to own land individually. The shift of communal land ownership to the individual ownership was high in the areas that bordered non-Gumuz people and settlers (Jira 2008). Parts of the lands that were under shifting cultivation appeared then as vacant and thus considered as potential land that was allocated for transnational or domestic investments without correct compensation to the community (Crewett et al. 2008). Starting from 2007 and following the “invest in Ethiopia” call of the Ethiopian government, forests and woodlands were occupied by commercialised agriculture (Table S1). The deforestation, in turn, resulted in the declining and disappearance of wild animals, roots, and wild fruits, which are important components of the Gumuz diet (Mebratie 2004). Anticipating on such major land seizures, the Gumuz individually and increasingly rent out part of their land to settlers from the highlands without the consent of the elders (Crewett et al. 2008; Teklemariam et al. 2016). Particularly along the roads, such farmlands are increasingly cultivated using oxen-plough or informally rented tractors. For instance, in the upper Aymela catchment (Fig. S11), the part along the road was recently deforested and ploughed by tractor. At catchment scale, such tractor cultivation leads to 50% higher peak discharge (hence flooding risks) and more than doubling of sediment load (Teklemariam et al. 2017a) (Fig. S12). The traditional agricultural system of the Gumuz protects the land relatively better from erosion than commercial farming, which is explained by the smaller share of cropped area and absence of ploughing that protects the root mat (Teklemariam et al. 2017a). Wherever possible, settlers try to establish small-scale irrigation using pumped water from Beles River and its tributaries. A major problem here is the difficulty to store tomato and onion produces, particularly given the high temperatures, which leads to strong price drops at peak harvesting time compared to that at early harvests (from 25 ETB in October and November, down to 5 or even 1 ETB per kg for tomatoes, in January to early May—1 ETB was 0.03 € on 28 February 2018). Here, tomatoes need to be marketed immediately after harvesting. Early growing and harvest incur additional costs by spraying fungicides against late and early blight diseases which are favoured by high humidity in the summer rainfall season (Alemayehu and Alemayehu 2017).

As mentioned above, large tractor schemes have been developed since around 2000, on which crops such as cotton, sesame, maize, or the biofuel tree pongamia (Millettia pinnata) are grown. Major companies in the lower Beles basin include S&P Energy Solutions (Teklemariam et al. 2017b). Large sugar estates were also established in the upper Beles basin around Jawi (using irrigation water) (Clark et al. 2013; Fantini et al. 2018). Maintaining the original vegetation in 50-m-wide strips on both sides of drainage lines is compulsory in such land rental contracts, but it could only be observed in few cases (Fig. S13).

The agricultural system of the Shinasha and Agaw, as well as that of more recent settlers, is one with permanently farmed fields, very similar to that of the highlands (qualified as “permanent upland system” by Ruthenberg 1980, and as “grain-plough complex” by Westphal 1975). On most of these lands, including the lower and middle slopes of the mountains, no soil and water conservation activities are done; the perception that there is still “enough land” leads to less care for the land. In contrast, on the most elevated places, where there is a shortage of land, stone terracing is done.

Along the escarpment, a rare insight is given in environmental changes, when comparing Kuls’ photographs that were taken around 1960 with repeat photographs illustrating the current situation (Fig. S14). While Adisu (2010) shows a photograph of this area after burning and qualifies the annual fires as deforestation, the tree cover seems in reality to have increased over time.

From Kusa to regulated Beles River

In terms of hydrology, the Kusa (as the Beles River is called in local Gumuz language) has strongly changed: a large “Tana-Beles” irrigation project was initiated in the 1980s, and a diversion dam was built (Fig. S15), but the planned tunnel that ought to bring the water from Lake Tana was not constructed (Eguavoen 2009). The irrigation scheme itself, intended for growing rice, was never formally implemented though locally pumping irrigation is done, and vegetables, as well as tropical fruits, are grown. Concomitantly with the works for the irrigation scheme, numerous numbered villages were established. For instance, the original name of the town of Pawe is “Village 7”. The idea of formally irrigating the area was put on hold until the water tunnel was ultimately opened in 2010 (Eguavoen 2009). At that time, priority was given to irrigation of sugar cane around Jawi (Fantini et al. 2018). The integrated planning and management (including the impacts on the downstream projects in Ethiopia and downstream countries that depend on the flows of the Nile) of water resources in the Beles basin is one of the researchable matters. To give a clue about the issues at stake in relation to GERD hydropower dam, which is partly fed by the Beles River, the planned command area of Tana-Beles project could be estimated at 115,000 ha. Considering that the required irrigation water has a depth of about 1.4 m per year (Bekele et al. 2009), a corresponding annual volume of 1.6 × 109 m3 would be abstracted from the discharge entering the GERD, which is about 3.3% of the average annual flow (49.4 × 109 m3) used for modelling the reservoir filling strategies (Wheeler et al. 2016). If confirmed, using such an amount of water for irrigation may negatively impact the hydropower production of the GERD.

Since May 2010, the Tana-Beles project has become operational after which 55% of the Lake Tana water outflow is transferred directly through Beles River, rather than following the Blue Nile gorge (Dessie et al. 2015). Consequently, the average additional discharge of 84 m3 s−1 in Beles throughout the year (Dessie et al. 2015), which has increased to levels beyond 100 m3 s−1 in 2015–2017 (Annys et al. 2018), is a stimulus for the creation of irrigation schemes, as was already done with the establishment of 14,000 ha sugar estates at Jawi, aiming at 50,000 ha (Fantini et al. 2018; Annys et al. 2018). It has also become difficult and in many places impossible to cross the river, leading to problems in reaching relatives and farmlands across the river; more than 250 people have been taken by floods after the hydropower operation started (Annys et al. 2018). Another effect relates to the changed hydrogeomorphology; changing a river from highly seasonal to permanently strong discharges has consequences on its planimetric morphology (meander wavelength, river width) and on its depth. The releases from the Tana-Beles hydropower tunnel are passed downstream along the Jehana River, a tributary of the Abat Beles, significantly changing its hydrological regime and morphology (Abebe 2016); in places where the river is not confined, its morphology changes from a braided to a largely meandering river. In the plains around Pawe, indicators on bridge foundations and embankments clearly show an incision (sediment scouring) by several metres (Fig. S16).

Towns get their water from springs such as the Ligitcha spring for Mandura and from boreholes (groundwater). In rural places, there are hand pumps, but springs and surface water (rivers) are also frequently used. The water supply and sanitation coverage in the region is very low. In 2004, it was only about 35% of the population that got potable water (Delesho 2006). Hence, besides the use of the land, water consumption is also a continuous source of conflict. For instance, the Gumuz people in Kota do not allow the settlers along the road to come and take water from their wells; on the other hand, companies that allow local people to take water from their installations are better accepted (Teklemariam et al. 2017b).

Recent migrations

Whereas Kuls (1962) and Swedesurvey (2010) mention occasional migrations of highlanders to larger settlements in the Beles basin as of the 1950s, more than 100,000 people have migrated here from Wollo (N Ethiopia), Kembata (S Ethiopia), and other highland areas in the early 1980s as part of (forced) resettlement programmes (Abbute 2002; Zanardi 2011). Some 18,000 Gumuz were displaced (Abbute 2002). In absence of the necessary medical services, the settlers suffered heavily from diseases such as schistosomiasis and malaria (Teklehaimanot and Fletcher 1990; Fletcher and Teklehaimanot 1989); many died or have gone back to their region of origin. Violent clashes with Gumuz at the time of overthrow of the military Derg regime around 1989–1993 also contributed to the return of settlers to the highlands (Markakis 2011). The current and ongoing wave of immigration is quite spontaneous, following the establishment of roads, agricultural investments, and towns. Employees of agricultural investment companies commonly try to obtain some cropland for themselves, for instance by renting it from Gumuz people. Several studies showed negative impacts of these resettlements on the host community (Agneta et al. 1993; Gebre 2003). In the densely populated Pawe district, there are no more Gumuz. The settlers now recently started to cultivate the slopes, which gives them a typical “Wollo” aspect (Fig. S17). As a response, the Agaw came down from Belaya Mountain to establish homesteads, in order to secure the space.

In recent years, the GERD dam building works have also attracted an important workforce as well as petty trade near the site and along the main roads. This has attracted more settlers to the region, and along roads, the population becomes increasingly mixed. In addition, an estimated 20,000 Gumuz who were living in the impounded area and 5-km buffer strip (both along Beles and Blue Nile) are being relocated to villages along roads (Veilleux 2014), which led to the expansion of towns such as Aycid.

Environmental changes

The most dramatic change is the clearance of the semi-natural vegetation and its replacement by artificial vegetation (cropland) as can be easily observed when comparing historical imagery over a few years (Fig. 4). The traditionally protected kota trees (Gardenia ternifolia) are the last ones to remain on the farmlands, but also these are frequently removed, among others through debarking.

Fig. 4
figure 4

The DigitalGloble images (left) show the “old” traditional land use with shifting cultivation (typically rendered by plots with different shades of green). The Sentinel 2-derived false colour composites (right; resolution: 10 m × 10 m) show the “new” land use with mainly open fields. Here, dark pink areas display remnants of natural (riverine) vegetation, while bright pink areas show newly converted large-scale agricultural fields. Dark green to black areas display recently cleared lands, which are ready to be cultivated. Brownish areas display sparsely vegetated areas (sources: DigitalGlobe imagery as displayed in ArcMap 10.4; Sentinel 2 imagery downloaded from https://scihub.copernicus.eu/dhus/#/home; own representation)

Full size image

In Benishangul-Gumuz region alone, the lowland bamboo covers an estimated area of 440,000 ha, though gradually it is disappearing from the landscape (INBAR 2010). In contrast to a 50/50 share of woodland and bamboo mentioned by Kuls (1962), a sampling of 12 traditionally managed landscapes (between 0.5 and 15 ha each) shows that bamboo occupies 4%, woodland and forest 65%, cropland 17%, and fallow 4% (Teklemariam et al. 2017a). In eight catchments under agricultural investment, there is only 3% bamboo coverage. Admittedly, Kuls’ (1962) observations were mainly done in the Mandura district, which is better endowed with springs and has a somewhat milder climate, but also there, bamboo is nowadays quite rare to be observed. Pressure on bamboo stands stems from intensive use as cash income and for construction purposes, and poor regeneration due to livestock feeding, in that order (Mekuriaw et al. 2011). The lowland bamboo of this area has been shown to have a high potential as ruminant feed (Mekuriaw et al. 2012). Excavation of shoots, which are used for human food, does have less effect on the overall bamboo stands.

Discussion

Regionalisation of the landscapes in the Beles basin in terms of core and periphery

As Kuls (1962) rightly mentioned, the escarpment of the Gojjam plateau forms a striking boundary between the Ethiopian highlands and the lowlands, both with regard to the natural environment and population groups, which fits with Markakis’ (2011) dual frontiers. Our findings show that the contrast between Ethiopian core and periphery demonstrated by Markakis (2011) has remained very actual in the Beles basin. The resettlements (Agneta et al. 1993; Gebre 2003) and sugar plantations (Fantini et al. 2018) tend to make the upper part of the Beles basin part of the core as a result of the continued crowding out of the Gumuz. The further installation of a transit road and a few (largely failed) commercial farms in the lower half of the basin do not allow to consider it as not peripheral. Hence, within the Beles basin, there are strong contrasts in population groups, infrastructure development, and agricultural systems. Changes in these variables spatially often coincide, so that more or less sharp boundaries fitting with topography and river systems can be drawn between the different geographical regions of the Beles basin. Such regional units are only meant for research purpose and not aimed at replacing, or even interfering with political or administrative boundaries which have their own rationale. The regional geographical units (Fig. 5) include:

  1. A

    The uplands, sometimes with sharp boundaries corresponding to escarpments, sometimes more gradually like in the upper Beles basin. The uplands are typically inhabited by Agaw, Shinasha, and Amhara; there is a permanent cereal-based cropping system. The northern and eastern areas of these uplands are part of Amhara region. The upper Beles valley has been virtually cut into two parts, and numerous people lost their lives, due to permanent high flows in Abat Beles (Annys et al. 2018).

  2. B

    The Mandura escarpment and footslopes inhabited mainly by Gumuz, Agaw, but also settlers. Gumuz have settled in permanent villages, and the permanent cereal-based cropping system prevails throughout, though the land is still burnt yearly.

  3. C

    The Tana-Beles resettlement area from where the Gumuz who were still present in the 1980s (Teklehaimanot and Fletcher 1990) have outmigrated over the last decades. The valley bottom is formally irrigated around Jawi, and a lot of small-scale irrigation elsewhere and permanent cropping expand towards the slopes. It includes the major towns Pawe and Gilgel Beles. The upstream part of these plains is part of the Amhara region.

  4. D

    The northern part of the lower Beles valley. This includes the districts Dangur and Guba, and is crossed by a major road towards Sudan and the Renaissance dam. Here numerous agricultural investment companies have been established, and many Gumuz have settled in permanent villages.

  5. E

    The southern part of the lower Beles valley. It is separated from the previous region by the Beles River that can only be passed by wading, which has become increasingly difficult since even in the dry season, the Beles flow is maintained at high levels. As the area is difficult of access, no major roads have been established and only a few domestic agricultural investments took place (Table S1); the Gumuz live their traditional lifestyle. This geographical unit comprises the northern lower parts of the Wombera and Bullen districts. Changes concern basic health and education infrastructure that the Ethiopian government established at numerous places, and some rural access roads, as well as potential conflicts for land with Shinasha living in the Wombera highlands. The local perception is that the Gumuz in this more isolated region with less pressure are “better off” than those in the more developed northern part of the basin. This hypothesis needs to be verified.

Fig. 5
figure 5

Proposed geographical units of Beles basin as coloured shades on a map of districts and roads. For legend of geographical units A to E, see text

Full size image

The modified hydrology and hydrogeology

There is a strong push to use the available water resources in the basin, mainly for irrigation, triggered by the inter-basin water transfers from Lake Tana to Beles basin, and development of commercialised agriculture including sugar cane (Fantini et al. 2018). Besides a study by the World Bank (2008) that estimated the total runoff in the Beles basin before the inter-basin transfers from Lake Tana, no studies are known that address the Beles’ basins water balance, which is crucial for the Tana-Beles water contribution to the GERD hydropower scheme, hence to the CRGE. In addition, a catchment-level study (Teklemariam et al. 2017a) shows that tractor cultivation leads to 50% higher peak discharges (hence flooding risks). Sustainable water resources use and development largely depend on a profound understanding of the different water balance terms (runoff, river discharges, evapotranspiration, rainfall, groundwater, etc.), which is nearly absent in the basin.

In response to the increasing number of towns and villages, more water supply schemes are sought. Groundwater of Beles basin is the least studied and the least understood. The general hypothesis is that there is good groundwater potential in Beles basin, but the surface/groundwater interactions and inter-basin groundwater flows need to be studied in detail. An emerging researchable, trans-border groundwater security needs also to be considered (Albrecht et al. 2017).

Land changes and environmental services

Few baseline studies have been carried out regarding forests and soils of the basin, their landscape relations, nutrient recycling processes, and how they are impacted by wildfire. Also, the isolation of Mount Belaya and Mount Dangur at the edge of the Beles basin might have affected gene exchange from the montane vegetation of the highlands; the extent to which this has already affected metacommunity dynamics needs to be investigated.

The impacts of the Beles basin becoming an investment and settlement hotspot for the last 30 years should be studied with regard to land use and cover, biodiversity, and ecosystem integrity. So far, only one impact study is known: it dealt with assessing the impacts of land-use changes on flooding and sediment transport (Teklemariam et al. 2017a), but land management issues have not been addressed in detail. These include direct sowing using a planting stick, and planting of lowland bamboo on critical slopes, which serves with its dense roots and rhizomes as soil and water conservation technique (Embaye 1998). As part of the CRGE, indigenous environmental conservation practices and their nexus with “modern” policy-based conservation measures, as well as the determinants of the adoption and sustained use of soil and water conservation measures, must be explored. Furthermore, research should be done into the possibilities of linking environmental conservation in the Beles catchment with the life expectancy of the GERD hydropower reservoir at its outlet, devising of instruments such as payment of environmental services (PES) schemes (Engel et al. 2008) to be arranged between communities and the Ethiopian Hydropower Corporation. Advantages and loopholes of the production, certification, carbon storage effects, and marketing of timber and non-timber forest products need also to be investigated (Burivalova et al. 2017; DeFries et al. 2017).

In addition, particular changes occurred due to strong increases in river discharge, and the change of the Beles River from a seasonal to a permanent regime; the hydrogeomorphology, particularly width, incision/aggradation, and meandering patterns, as well as riverine forest ecology needs to be studied.

Biodiversity and aquatic ecology

Whereas baseline studies on aquatic biodiversity exist at least for fish (Berie 2007), the collection of baseline biological and environmental data is of paramount importance for a better understanding of the aquatic resources of the region. In addition, further study is needed to assess the impact of the water transfers from Lake Tana on the aquatic ecology of Beles River. Before the construction of the tunnel from Lake Tana, Beles was a seasonal river. However, as a result of the continuous supply of water, the river changed to a perennial one, carrying about 2.6 times the previous flow in upper Beles and 2.2 times the previous flow in the lower basin (Abebe 2016). These changes have likely impacted on the downstream biophysical environment. Ecological risks associated with such water transfers are manifold. First, the increased flow already had major implications for channel integrity, which in turn might have affected the ecological functioning of the river banks, leading to a potential loss of the benthic biodiversity. Second, the physical and chemical properties of the river water have changed which might have impacted on components of both the pelagic and benthic aquatic food web. Third, the direct connection of the river with Lake Tana has likely resulted in the spread of alien fish species, floating aquatic plants, and invertebrate and microbial communities, including animal diseases and their vectors (Comrie-Greig 1986; Davies et al. 1992; Day et al. 1982; Herrmann 1983; Rasmussen et al. 2014; Sible et al. 2015). Reversely, the inter-basin water transfer may increase the supply of surface water and the water content of soils near the riverbanks, which might eventually lead to the formation of local wetlands. Such wetlands can in turn mitigate ecological water deficiency, regulate the water volume of rivers, and provide new habitats for fauna and flora (Chen et al. 2008).

We call for research on the biotic and abiotic characteristics of Beles and its affluents (not affected by the water transfers) and propose to develop comparative studies of the aquatic communities as well as the physicochemical parameters of the Abat Beles (which receives the Tana water) and Gilgel Beles river. Also, the potential for the development of aquaculture for fish and the benefits and threats related to these activities should be explored.

Sustainability of crop production systems

The agricultural systems that are briefly described in this paper should be further investigated, so that the multifaceted indigenous knowledge is thoroughly understood including its potential contribution to the CRGE. Stable shifting cultivation with the help of slash and burning is in ecological balance with the environment and does not irreversibly degrade the soil and vegetative resources, provided a sufficient length of fallow is allowed for soil and vegetation restoration (Okigbo 1984). But, an increasing population density necessitates a more intensive use of land. The consequence is extended cropping periods and shortened fallows that are no longer adequate to restore the productive capacity of soils (Erni 2015; Keck et al. 1994; Ribeiro Filho et al. 2013). Permanent cultivation systems run by settlers, Shinasha and Agaw, and by investors are also not conservation-oriented. High post-harvest losses and limited demand during the peak of the harvesting season discourage farmers to engage in vegetable production. Crop production systems can be sustainable when they generate economic profit, social benefits to the farm family and the community, and environmental conservation, simultaneously (Dobermann and Nelson 2013; Sullivan 2003). Particularly, before recommending best technologies and practices ensuring sustainable crop production systems for the basin, research is needed on reducing the large post-harvest crop losses that are caused by improper harvesting, threshing, transportation, and storage management; and optimising the value chain of the major crops grown in the basin. Further research should also be done into the domestication of wild edible plant species commonly used by the Gumuz (Awas et al. 2010), as well as modern propagation techniques for bamboo in order to control dieback (Embaye 1998; Poppens et al. 2013).

Land tenure and ownership: socio-economic development

Currently, in Gumuz society, if the lineage groups’ ownership is quite well respected, there is a shift from communal land ownership to individual ownership and commercialisation of land to investors. Particularly, lands that are only used periodically have been considered as vacant and were commercialised (Crewett et al. 2008). The indigenous life and the environment of the communities have further been affected by the continuous resettlement programmes from central Ethiopia as well as by internal developments that led to the conversion of forests to sedentary agriculture. All this calls for research on how the Gumuz people have adapted their indigenous lifestyles to such changed ecological, cultural, and socio-economic environments.

In addition, the water transfers from Lake Tana to Beles have major implications for the agricultural systems along the river. As in many dam-impacted areas, the traditional agricultural rainfed system undergoes a shift towards small-scale pump or river diversion irrigation and commercial agriculture, with an increased number of cropping seasons (Thomas and Adams 1999). This mostly contributes to higher household incomes, but inefficient water use can lead to the impoverishment of farmers (WCD 2000). A discrepancy exists in literature on which of both systems is most beneficial (Thomas and Adams 1999; WCD 2000). In addition, large-scale irrigation projects often transfer money to government workers, to new settlers (Kloos and Legesse 2010), or to investors in commercial agriculture (Bazin et al. 2011) and are not exclusively beneficial for indigenous people. While mainly the resettled highlanders practise irrigation agriculture in the area, the Gumuz ethnic group has no prior experience with this type of agriculture. Hence, there is a need to study the dam-induced impacts on agricultural systems and people’s livelihoods in the Beles basin. A distinction between the impact on the indigenous Gumuz people and the resettled highlanders will be made. To study these impacts, crop and hedonic pricing modelling will be done. While crop models estimate crop production by taking soil-plant-water components into account (Gitelson et al. 2012), hedonic pricing models take human capital limitation and economic considerations into account and estimate long-term impacts of dam-induced water availabilities on net household revenues (Seo et al. 2009). The results of both models will be empirically compared and will provide insights to further develop economic valuation methods.

With regard to private land renting between Gumuz and settlers, as well as larger land deals, investigating the institutional mechanisms that could allow win-win land deals is essential (Teklemariam et al. 2015). Trends in the natural resource valuation (e.g. land lease price determination, disparities between the formal and informal land rental prices, and the way forward to ensure agro-ecologically and economically viable land market) are also vital areas of research.

Finally, a true non-peripheral integration of the Gumuz, who live astride the Ethio-Sudanese border, needs to be investigated in relation to their empowerment to use their knowledge of both countries to develop trans-border commercial and societal relations (Arter 2001; Markusse 2004). The scope for cross-border institutions needs also to be investigated, both traditional such as in Nigerian borderlands (Kehinde 2010; Mortimore et al. 2006) or formal, in line with the emerging Convention on Cross-Border Cooperation (Niamey Convention) (African Union 2014).

Conclusions

Based on literature and field observations, this scoping study of the Beles basin balanced social, ecological, and economic considerations; we searched for persistence and changes in peripheralism, as well as the likely implications for the population of the basin, and for Ethiopia’s Climate Resilient Green Economy (CRGE) strategy (Death 2015; Jones and Carabine 2013).

We found that changes affect mainly the central Beles valley bottom (unit C on Fig. 5) with immigration of highlanders and ongoing agro-industrial development. The upper Beles basin is affected by the permanent high discharges of the river where crossing can only occur at the peril of one’s life. In the northern part of the lower Beles basin, many large commercial farms have developed (“land grabbing”; unit D). In the less accessible southern Beles basin (unit E), many Gumuz have continued their traditional lifestyle.

Development of hydrology, hydropower, agriculture-based industrialisation, biodiversity, and carbon sequestration are centre pieces of the CRGE development activities in Beles basin, which should be undertaken in respect of local culture and land rights. Particular knowledge gaps identified are water balances at broader scale; local biodiversity and ecosystem functions; the land and its ecosystem services, related to agricultural production, but also biodiversity; the reduction of post-harvest losses and increased shelf life of vegetables and fruits while water and land are well managed; and local peoples’ and farmers’ empowerment, in the perspective of improved agricultural production and biodiversity.

We also investigated to what extent persistence and changes in the basin align with the CRGE strategy. The changed hydrology of the basin, as a result of hydropower development, has further favoured the development of irrigation and sugar cane plantation, which will lead to water abstraction affecting hydropower production at the GERD. The establishment of commercial farms in the Beles basin leads to sediment production and siltation of the GERD reservoir, and to wide deforestation (Fig. 4), also in contradiction with the CRGE strategy, while the farms are generally not productive. A reasoned sustainable development and intensification of the Gumuz agrosylvopastoral system should be investigated as an alternative for commercial farming, taking the CRGE criteria as standards for evaluation.

We conclude that the conditions of the Beles basin at the beginning of the twenty-first century still align with the concepts of core and periphery. Particularly in the lower part of the basin, where the Gumuz live, we found no empirical evidence of “deperipheralisation” in the sense of “non-peripheral integration” (Arrighi 1979). The central part of the basin has become more integrated with the highlands’ economy, as a consequence of large settlement of highlanders. But for most of the Gumuz, we suggest that a non-peripheral integration requires their empowerment and the development of trans-border commercial and societal relations.