Stakeholder Roles and Perspectives on Sedimentation Management in Small-Scale Irrigation Schemes in Ethiopia

: Irrigated agriculture, particularly small-scale irrigation (SSI), is a mainstay for sustainable livelihoods in the developing world. In Ethiopia, SSI sustainability is threatened mainly due to excessive sedimentation. Stakeholders’ perceptions of the causes of sedimentation and how they sustain SSI under excessive sedimentation conditions were investigated in two SSI schemes in Ethiopia. A participatory rapid diagnosis and action planning was implemented, consisting of a literature review, participatory rural appraisal, and semi-structured interviews. Results show that farmers slightly di ﬀ ered in perception of excessive sedimentation drivers. Farmers reported design problems as the main cause of excessive sedimentation (64%), followed by poor operation and maintenance (O and M) practices (21%) and external factors (15%). In contrast, 62% of the interviewed engineers indicated erosion and irrigation technologies as the main causes of excessive sedimentation, while few reported poor design (13%). In addition to an intensive desilting campaign, farmers delayed the start of the irrigation season to avoid the intake of highly sedimented water. Local social capital and knowledge appeared to be more important than formal knowledge and blue-print institutions for dealing with sedimentation problems. Well-organized structure and extra time devoted by famers were vital for SSI sustainability. Integration of the farmers’ knowledge with that of the engineers could yield more e ﬀ ective ways to deal with sedimentation problems.


Introduction
Irrigated agriculture is a prime sector to ensure food security, alleviate poverty, and promote economic development in the developing world [1]. Small-scale irrigation (SSI) schemes in particular make a massive contribution to national economies in many developing countries, while also serving as an incubator for collective action [2]. Nonetheless, "traditional" SSI schemes are largely overlooked by states [3]. Governments prefer the development of more "modern" irrigation schemes, considering "farmer-led" irrigation schemes "inefficient", "unproductive" and "traditional" [4][5][6]. To date in Africa, however, the total area under SSI schemes is much larger than that under medium-and large-scale irrigation [3,7].
In Ethiopia, traditional SSI schemes accounted for 80% of the total irrigated land in 2018/2019 [8]. Ethiopia's largest region, Oromia Regional State (28.66 million ha), had 612 modern irrigation schemes in 2016/2017, compared to 9379 "traditional" and 63,523 pump irrigation schemes, according to data from the Oromia Irrigation Development Authority (OIDA) [9]. However, the contribution of all these schemes to Ethiopia's national economy has been much diminished due to the underperformance of the systems [10,11]. Indeed, most are either non-functional or operate far under their potential [12,13]. For instance, in Oromia Regional State alone, 109 (18%) of the modern schemes and 8,508 (13%) of the pump schemes were reported to be inoperative or semi-functional in 2017 ( Figure 1). Several explanations have been given for this underperformance, such as design failure and poor design, excessive sedimentation in the headwork and main canal, scouring damage, poor scheme management, and inferior institutional set-up [2,12,[14][15][16]. Ethiopia experience the most serious erosion in the world [17][18][19], one of the most significant adverse environmental problem in developing countries. Excessive sedimentation in irrigation systems gradually blocks the flow of irrigation water, causes water stress and unfair distribution [20], damages infrastructure, and may trigger the complete collapse of irrigation systems. Since irrigation is a major consumer of water [21,22], the loss of water due to damage to infrastructure by excessive sedimentation results in a decline in water availability and increased competition among different water uses and users. The management of sedimentation in irrigation systems requires large and continual maintenance and operation investments [23]. In farmer-managed irrigation schemes, excessive sedimentation places a huge maintenance burden on farmers, in addition to their other farming activities ( Figure 2). Irrigation canals with excessive sedimentation are generally collectively dredged, with farmers who are perceived as not doing their share of the work sanctioned. This, however, may generate farmer dissatisfaction and conflicts, possibly undermining collective action and social interaction. Ethiopia experience the most serious erosion in the world [17][18][19], one of the most significant adverse environmental problem in developing countries. Excessive sedimentation in irrigation systems gradually blocks the flow of irrigation water, causes water stress and unfair distribution [20], damages infrastructure, and may trigger the complete collapse of irrigation systems. Since irrigation is a major consumer of water [21,22], the loss of water due to damage to infrastructure by excessive sedimentation results in a decline in water availability and increased competition among different water uses and users. The management of sedimentation in irrigation systems requires large and continual maintenance and operation investments [23]. In farmer-managed irrigation schemes, excessive sedimentation places a huge maintenance burden on farmers, in addition to their other farming activities ( Figure 2). Irrigation canals with excessive sedimentation are generally collectively dredged, with farmers who are perceived as not doing their share of the work sanctioned. This, however, may generate farmer dissatisfaction and conflicts, possibly undermining collective action and social interaction. Farmers introduced irrigation systems and through decades of experience have developed ways of dealing with excessive sedimentation [6]. Nonetheless, states have tended to overlook this indigenous knowledge. Instead, they have turned their focus to "modernizing" schemes, in the conviction that modernization will deliver improved irrigation performance [4][5][6]. Yet, without technology appropriation by farmers (i.e., farmers' adoption and adaptation of modern technology to their own setting), modernization of farmer-managed irrigation schemes may actually aggravate the problem of excessive sedimentation. This is because farmers' knowledge about the sedimentation problem may be overlooked and inadequate information and resources/technologies may be available locally for the farmers to undertake operation and maintenance of the system, whereas farmers are the best sources of information and knowledge about their localities [24][25][26][27].
Weak institutions for the management of schemes and poor operation and maintenance practices followed by users are also among the major contributors to the problem of excessive sedimentation [2,12,14]. As excessive sedimentation in irrigation schemes is inevitable, strong institutions for scheme operation and maintenance can play a crucial role in reducing the problem. To craft strong institutions for appropriate management of excessive sedimentation, it is essential that local values, norms, and knowledge be considered, as well as the diversity of irrigators, while also ensuring participation of and consultation with all the concerned stakeholders [6, [28][29][30]. This is particularly so in a country like Ethiopia, where social capital is rooted in local groupings organized around religious, burial, and wedding ceremonies, community savings, and loan services. Such local associations serve as platforms for communication and conflict resolution, which are also highly valuable for the sustainable management of irrigation schemes [31]. For instance, farmers value the "traditional or informal" conflict resolution mechanism higher than the "formal" legal system. Often, they discuss issues of scheme management on indigenous social gatherings, such as wedding or burial ceremonies.
Despite the foremost role of local norms, values, and knowledge, as well as institutions and stakeholders, in managing excessive sedimentation in irrigation schemes, few studies address these aspects directly. Most research rather investigates sediment transport, focusing essentially on understanding sedimentation processes and modelling sediment transport [20,23,[32][33][34][35][36][37]. Improved operation and maintenance of the system and real and coordinated participation of concerned stakeholders are crucial in dealing with excessive sedimentation problems. The current study therefore looks at stakeholder perceptions of the problem of excessive sedimentation and their roles in its management. We applied a collaborative and participatory approach to analyze sedimentation management practices in two irrigation schemes in Ethiopia. The results of the analysis are presented, followed by a discussion of the influence of institutions and scheme modernization. Farmers introduced irrigation systems and through decades of experience have developed ways of dealing with excessive sedimentation [6]. Nonetheless, states have tended to overlook this indigenous knowledge. Instead, they have turned their focus to "modernizing" schemes, in the conviction that modernization will deliver improved irrigation performance [4][5][6]. Yet, without technology appropriation by farmers (i.e., farmers' adoption and adaptation of modern technology to their own setting), modernization of farmer-managed irrigation schemes may actually aggravate the problem of excessive sedimentation. This is because farmers' knowledge about the sedimentation problem may be overlooked and inadequate information and resources/technologies may be available locally for the farmers to undertake operation and maintenance of the system, whereas farmers are the best sources of information and knowledge about their localities [24][25][26][27].

Materials and Methods
Weak institutions for the management of schemes and poor operation and maintenance practices followed by users are also among the major contributors to the problem of excessive sedimentation [2,12,14]. As excessive sedimentation in irrigation schemes is inevitable, strong institutions for scheme operation and maintenance can play a crucial role in reducing the problem. To craft strong institutions for appropriate management of excessive sedimentation, it is essential that local values, norms, and knowledge be considered, as well as the diversity of irrigators, while also ensuring participation of and consultation with all the concerned stakeholders [6, [28][29][30]. This is particularly so in a country like Ethiopia, where social capital is rooted in local groupings organized around religious, burial, and wedding ceremonies, community savings, and loan services. Such local associations serve as platforms for communication and conflict resolution, which are also highly valuable for the sustainable management of irrigation schemes [31]. For instance, farmers value the "traditional or informal" conflict resolution mechanism higher than the "formal" legal system. Often, they discuss issues of scheme management on indigenous social gatherings, such as wedding or burial ceremonies.
Despite the foremost role of local norms, values, and knowledge, as well as institutions and stakeholders, in managing excessive sedimentation in irrigation schemes, few studies address these aspects directly. Most research rather investigates sediment transport, focusing essentially on understanding sedimentation processes and modelling sediment transport [20,23,[32][33][34][35][36][37]. Improved operation and maintenance of the system and real and coordinated participation of concerned stakeholders are crucial in dealing with excessive sedimentation problems. The current study therefore looks at stakeholder perceptions of the problem of excessive sedimentation and their roles in its management. We applied a collaborative and participatory approach to analyze sedimentation management practices in two irrigation schemes in Ethiopia. The results of the analysis are presented, followed by a discussion of the influence of institutions and scheme modernization.

Location and Description of the Study Area
Two irrigation schemes in Oromia Regional State in the Great Rift Valley Basin of central Ethiopia, an area seriously affected by land degradation and erosion, were selected: Ketar medium-scale irrigation scheme and the Arata-Chufa small-scale irrigation scheme ( Figure 3). The main reason to select these schemes was that farmers manage to keep the irrigation system in good working order despite the excessive sedimentation problems.
despite the excessive sedimentation problems.
In addition, the following criteria were applied in selecting the case study sites: (i) the scheme should be a gravity/diversion type, making use of river runoff; (ii) the scheme of interest should utilize a river as its water source; (iii) the scheme should be managed exclusively by farmers or an irrigation community; (iv) the users should have a relatively long period of experience in water and sediment management; (iv) the scheme should be functional for a relatively long period; (vi) management of the scheme should face relatively severe sedimentation problems; and (vii) a water user association (WUA) should be active in scheme operation and management.
Ketar is medium-scale irrigation scheme, located at 7°49 N and 39°02 E, covering 430 ha, with an average elevation of 2294 m above mean sea level. Having a total main canal length of 12.1 km, the scheme consists of three sections: Ketar 1 (Ketar Genet), covering 110 ha and providing water to 289 households; Ketar 2 (Ketar Golja), covering 200 ha and providing water to 415 households; and Ketar 3 (Hamsa Gasha), covering 120 ha and providing water to 370 households. Each section has its own independent water users' association (WUA). The scheme is affected by sedimentation problems both from the river and overland flow sources.
Arata-Chufa is small-scale irrigation scheme, located at 7°59 N and 39°02 E, covers 100 ha, with an average elevation of 1740 m above mean sea level. This scheme's two main canals have a total length of 1.19 km, and supply water to 10 irrigation blocks. The water users' association of the scheme is one of the well-organized WUAs in the country. The scheme is mainly affected by sediment from the River sources. In addition, the following criteria were applied in selecting the case study sites: (i) the scheme should be a gravity/diversion type, making use of river runoff; (ii) the scheme of interest should utilize a river as its water source; (iii) the scheme should be managed exclusively by farmers or an irrigation community; (iv) the users should have a relatively long period of experience in water and sediment management; (iv) the scheme should be functional for a relatively long period; (vi) management of the scheme should face relatively severe sedimentation problems; and (vii) a water user association (WUA) should be active in scheme operation and management.
Ketar is medium-scale irrigation scheme, located at 7 • 49 N and 39 • 02 E, covering 430 ha, with an average elevation of 2294 m above mean sea level. Having a total main canal length of 12.1 km, the scheme consists of three sections: Ketar 1 (Ketar Genet), covering 110 ha and providing water to 289 households; Ketar 2 (Ketar Golja), covering 200 ha and providing water to 415 households; and Ketar 3 (Hamsa Gasha), covering 120 ha and providing water to 370 households. Each section has its own independent water users' association (WUA). The scheme is affected by sedimentation problems both from the river and overland flow sources.
Arata-Chufa is small-scale irrigation scheme, located at 7 • 59 N and 39 • 02 E, covers 100 ha, with an average elevation of 1740 m above mean sea level. This scheme's two main canals have a total length of 1.19 km, and supply water to 10 irrigation blocks. The water users' association of the scheme is one of the well-organized WUAs in the country. The scheme is mainly affected by sediment from the River sources.

Climatic Conditions
Rainfall is bimodal in the study region. There is a long rainy season ("Meher") from June to September, a dry season ("Bega") from October to January and a short rainy season ("Belg") from February to May (Table 1). Maximum and minimum temperatures at the Ketar scheme are 27 • C and 8.5 • C, respectively. The temperature range is wider at the Arata-Chufa scheme, from a maximum of 35 • C to a minimum of 5 • C. In the Ketar area, mean annual rainfall is 800 mm, and it is 620 mm in the Arata-Chufa scheme vicinity (2012-2016 data). The dry and short rainy periods are the main seasons for irrigated agriculture with mainly cash crops planted. The long rainy period provides the main cropping season for cereals, which are widely planted under rainfed conditions. The dry and short rainy periods are the main seasons for irrigated agriculture with mainly cash crops planted. Cereals are also cultivated to a limited extent in the short rainy period under rainfed condition.

Irrigated Area and Layout
The Arata-Chufa scheme initially had an irrigated area of 100 ha. Land redistribution activities in 1994/1995 resulted in an average landholding of 0.5 ha (Table 2). At the time of this study (October 2016 to August 2017), the irrigated area had expanded to 120 ha, as irrigation had progressively attracted more users. Likewise, for the Ketar scheme, the initially planned area of 110 ha had expanded to 128 ha, also due to increasing demand. Beneficiary numbers had risen in both the Arata-Chufa and Ketar schemes, respectively, from 324 to 374, and from 280 to more than 680.

Farming System
The study area is characterized by a traditional livestock-based mixed-farming system, with both crop production and animal husbandry. Predominant rainfed crops are food grains and pulses, including wheat, barley, teff, maize, beans, and haricot beans. Teff, onion, potato, cabbage, carrot, and tomato are the main crops grown under irrigation.

Participatory Rapid Diagnosis and Action Planning Approach
A participatory rapid diagnosis and action planning approach [39] was implemented to identify the causes of excessive sedimentation in irrigation schemes, to analyze the perception of stakeholders on the cause of excessive sedimentation problems and their solutions to perceived causes, and to scrutinize how irrigation systems has sustained under excessive sedimentation conditions by the farmers. This consisted of the following steps: • A literature review of policy documents, such as the Ethiopia Growth and Transformation Plan, irrigation performance reports, and other written materials from government and non-government sources; • Semi-structured interviews with selected professionals, WUA members, and farmers to understand their roles in managing excessive sedimentation and their perceptions of the drivers of excessive sedimentation, as well as to understand operation and maintenance practices and farmers' involvement in the schemes; • A participatory rural appraisal (PRA) of both irrigation schemes, including transect walks, resource map, structured direct observation, cropping calendars, and stakeholder analysis.
One hundred semi-structured interviews were conducted with selected professionals, WUA members, and farmers (Table 3). Interview subjects were selected based on the location of their farmlands and their roles and responsibilities in scheme management. At Ketar 1, three farmers were interviewed, one from the headrace, one from the middle zone, and one from the tailrace. At Ketar 2 and 3, twelve and eleven farmers, respectively, were selected from the secondary block of each. In each field block of the Arata-Chufa small-scale irrigation scheme, two farmers (one from the headrace and one from the tailrace) were selected for interview. Interviews sought to gather the frequencies of maintenance, including sediment cleaning and responsibilities of the various stakeholders, while also investigating how the farmers dealt with the problem of sedimentation. Respondents were asked how they organized dredging, their views on problems related to sedimentation, factors they thought contributed to the problem, and solutions proposed.
The PRA provided insight into the available resources and opportunities and challenges presented by excessive sedimentation. Component structures of the irrigation schemes were catalogued and sediment hotspots were identified. The strategies employed by the farmers to keep the schemes function were analyzed. The types of crops grown and cropping patterns were also documented, alongside the irrigation technologies available and used. Finally, researchers acquainted themselves with operation and maintenance practices.

Upstream, Midstream, and Downstream Farmers
Many of the farmers interviewed at Ketar-upstream (29%), midstream (31%), and downstream (25%)-identified the earthen canal (main canal without a concrete lining) to be a main cause of excessive sedimentation (Figure 4). The majority of farmers (60% of upstream farmers) and (69% of midstream farmers) considered faulty design to be a major cause of excessive sedimentation. A small proportion (8%) of the midstream farmers attributed the problem to poor operation and maintenance. The majority of downstream farmers (75%) suggested design problems or faulty design as the main cause of excessive sedimentation, and one fourth (25%) pointed to external factors as the main driver. None of the interviewed farmers at the downstream scheme (Ketar 3) attributed the problem of excessive sedimentation to poor operation and maintenance practice.

Ketar and Arata-Chufa Farmers
The farmers interviewed at the two irrigation schemes reported different perceptions of the causes of excessive sedimentation ( Figure 4). Foremost driver mentioned by farmers at the Arata-Chufa scheme was an absent and non-functioning sediment trap (25%), while the majority (28%) of the farmers interviewed at the Ketar scheme pointed to the lack of a concrete-lined main canal. This was the fourth most mentioned factor by the farmers at the Arata-Chufa scheme. The minority of respondents at the Ketar scheme (10%) associated the cause of excessive sedimentation with poor operation and maintenance practice. In the Arata-Chufa, a small portion (19%) of the interviewees, unlike Ketar counterparts, reported external factors as a major cause of excessive sedimentation.

Farmers and Engineers
Well over half of the respondent farmers (64%) claimed design problems as a major driver of excessive sedimentation problem, while just a few of the interviewed engineers (13%) agreed that design issues were at fault ( Figure 5, Table 4). Nearly two thirds (62%) of the interviewed engineers attributed excessive sedimentation to external factors, particularly erosion of highland areas (37%) and lack of technology and materials (25%). One fourth (25%) of the interviewed engineers claimed poor scheme operation as major driver of excessive sedimentation, while a small portion (21%) of interviewed farmers claimed poor operation and maintenance practice to be a cause of excessive sedimentation problem.
This study found that the main way farmers dealt with excessive sedimentation was to mobilize and engage huge amounts of labor (among system users) for intensive sediment cleaning campaigns that were completed within just a few days (3-5 days). Dates for canal cleaning were chosen carefully, considering public holidays and the end of the wet season, to avoid having to repeat the job due to overland flows and backflows of sediment removed from the canal. Despite frequent dredging, farmers also used the technique of delaying water abstraction at the beginning of a new irrigation season (in other words, at the end of every wet season), to avoid entrance of huge amount of sediment together with high sediment content water. The process of delaying water abstraction has potentially reduced the sediment load at the beginning of the irrigation season. This process was managed by the WUAs, which had full autonomy to open and close the intake gate. They have also applied their

Ketar and Arata-Chufa Farmers
The farmers interviewed at the two irrigation schemes reported different perceptions of the causes of excessive sedimentation (Figure 4). Foremost driver mentioned by farmers at the Arata-Chufa scheme was an absent and non-functioning sediment trap (25%), while the majority (28%) of the farmers interviewed at the Ketar scheme pointed to the lack of a concrete-lined main canal. This was the fourth most mentioned factor by the farmers at the Arata-Chufa scheme. The minority of respondents at the Ketar scheme (10%) associated the cause of excessive sedimentation with poor operation and maintenance practice. In the Arata-Chufa, a small portion (19%) of the interviewees, unlike Ketar counterparts, reported external factors as a major cause of excessive sedimentation.

Farmers and Engineers
Well over half of the respondent farmers (64%) claimed design problems as a major driver of excessive sedimentation problem, while just a few of the interviewed engineers (13%) agreed that design issues were at fault ( Figure 5, Table 4). Nearly two thirds (62%) of the interviewed engineers attributed excessive sedimentation to external factors, particularly erosion of highland areas (37%) and lack of technology and materials (25%). One fourth (25%) of the interviewed engineers claimed poor scheme operation as major driver of excessive sedimentation, while a small portion (21%) of interviewed farmers claimed poor operation and maintenance practice to be a cause of excessive sedimentation problem. own techniques for frequent removal of weed grown in the canal cross-section as they believed it traps a significant amount of sediment load.

Days of Labor Invested by Farmers to Manage Excessive Sedimentation
At the Ketar scheme, 3150 and 3086 days of labor were required to dredge 2690 m 3     This study found that the main way farmers dealt with excessive sedimentation was to mobilize and engage huge amounts of labor (among system users) for intensive sediment cleaning campaigns that were completed within just a few days (3-5 days). Dates for canal cleaning were chosen carefully, considering public holidays and the end of the wet season, to avoid having to repeat the job due to overland flows and backflows of sediment removed from the canal. Despite frequent dredging, farmers also used the technique of delaying water abstraction at the beginning of a new irrigation season (in other words, at the end of every wet season), to avoid entrance of huge amount of sediment together with high sediment content water. The process of delaying water abstraction has potentially reduced the sediment load at the beginning of the irrigation season. This process was managed by the WUAs, which had full autonomy to open and close the intake gate. They have also applied their own techniques for frequent removal of weed grown in the canal cross-section as they believed it traps a significant amount of sediment load.

Days of Labor Invested by Farmers to Manage Excessive Sedimentation
At the Ketar scheme, 3150 and 3086 days of labor were required to dredge 2690 m 3

Time Invested by Farmers in Agriculture and in Cleaning Excessive Sedimentation
Farmers' participation in sediment management varied according to the severity of the sedimentation problem in their particular scheme. The work required to manage excessive sedimentation significantly influenced the labor input to produce a crop (Table 6). Table 6. Hours invested by farmers in crop production and sediment cleaning activities to produce onion on 0.25 ha, considering a cropping period of four months (data from farmer interviews).

Percentage of Time Invested by Farmers in Sediment Management (%)
Ketar Upstream farmers spent 15% more time on crop production and 65% more time on the management of excessive sedimentation than midstream farmers. Compared to downstream farmers, upstream farmers (Ketar 1) spent 20% more time on crop production and 90% more time managing excessive sedimentation. Midstream farmers spent 6% more time on crop production and 68% more time on excessive sedimentation management, compared to downstream farmers. Overall, Ketar farmers spent 12% and 53% more time, respectively, on crop production and excessive sediment management than the Arata-Chufa scheme farmers.

Role and Structure of Water Users' Associations (WUAs) in Management of Excessive Sedimentation
WUAs collect annual operation and maintenance fees. In the Ketar scheme, farmers paid an annual US$ 8.73 operation and maintenance fee. If they did not participate in sediment cleaning activities, they were fined US$ 4.36, with this amount increased to US$ 6.55 for a second day of nonparticipation. Of that amount, US$ 2.18 went to the local police, who were delegated to take the legal action. At the Arata-Chufa scheme, member farmers paid an annual operation and maintenance fee of US$ 4.36 (1 US dollar = 22.916 birr (June, 2017)) for 0.25 ha of irrigated land, and they were required to participate in maintenance activities. If they did not participate, they were sanctioned with a US$ 1.75 fine. Farmers who were not WUA members paid US$ 13.09 for access to water.
The WUA structure for the Ketar irrigation scheme was originally introduced by "external actors" upon establishment of the scheme (Figure 6). The current organizational set-up has, however, drastically changed; only the functions of WUA head, deputy head, secretary, and cashier existed and the farmers themselves had established the "farmers' collective", which was observed to play an important role in dealing with the problem of excessive sedimentation. This collective implements and manages a major desilting campaign and coordinates minor repair activities. It is made up of subgroups of maximum 20 members. These subgroups are fully autonomous and responsible for imposing sanctions on members who do not participate in sediment cleaning activities. The institutional set-up of the Arata-Chufa WUA was also established by "external actors" in 1985/1986, at the time the scheme was handed over to the beneficiaries (Figure 7). The WUA head, deputy head, secretary, and cashier were observed to still be active and engaged in scheme operation and maintenance. The "field block", though not part of the original structure, had been set up by farmers to monitor the operation and maintenance of each block. As such, the field block heads were the main bodies responsible for monitoring sediment cleaning activities in secondary and tertiary systems and managing water distribution to each field block. The institutional set-up of the Arata-Chufa WUA was also established by "external actors" in 1985/1986, at the time the scheme was handed over to the beneficiaries (Figure 7). The WUA head, deputy head, secretary, and cashier were observed to still be active and engaged in scheme operation and maintenance. The "field block", though not part of the original structure, had been set up by farmers to monitor the operation and maintenance of each block. As such, the field block heads were the main bodies responsible for monitoring sediment cleaning activities in secondary and tertiary systems and managing water distribution to each field block.
1985/1986, at the time the scheme was handed over to the beneficiaries (Figure 7). The WUA head, deputy head, secretary, and cashier were observed to still be active and engaged in scheme operation and maintenance. The "field block", though not part of the original structure, had been set up by farmers to monitor the operation and maintenance of each block. As such, the field block heads were the main bodies responsible for monitoring sediment cleaning activities in secondary and tertiary systems and managing water distribution to each field block.

Scheme Modernization and Management of Excessive Sedimentation
The annual operation and maintenance fees paid by the farmers were not sufficient for required maintenance or repair cost. Farmers reported undertaking only minor maintenance activities by their own, saying that lack of resources and technology limited their ability to do so (Table 7). Due to lack of technology appropriation, farmers depend on "external actors" for major maintenance activities that curtailed their autonomy in keeping excessive sedimentation under control. The dependence of the farmers on external actors for maintenance and repair particularly concern the farmers for two issues. One, they could not afford the maintenance cost of the scheme requested by local contractor.

Scheme Modernization and Management of Excessive Sedimentation
The annual operation and maintenance fees paid by the farmers were not sufficient for required maintenance or repair cost. Farmers reported undertaking only minor maintenance activities by their own, saying that lack of resources and technology limited their ability to do so (Table 7). Due to lack of technology appropriation, farmers depend on "external actors" for major maintenance activities that curtailed their autonomy in keeping excessive sedimentation under control. The dependence of the farmers on external actors for maintenance and repair particularly concern the farmers for two issues. One, they could not afford the maintenance cost of the scheme requested by local contractor. Two, as they file scheme maintenance requests to the lowest Government office, this often took longer to respond to the timely needed repair request of the farmers.

Opportunity Cost of Scheme Operation and Maintenance
The farmers at the Ketar scheme paid US$ 8.37 annually for operation and maintenance (O and M), whereas Arata-Chufa farmers paid US$ 4.36. O and M fee was based what normally the users have agreed and afford to pay. The opportunity cost incurred by the farmers for their labor to dredge sediment just of 50% and 20% of the main canal length, respectively, for the Arata-Chufa and Ketar scheme were US$ 3457 and US$ 13,594, respectively. This means that if the farmers should be paid from annual O and M fees to cover for cleaning of excessive sedimentation from the canal and did not contribute labor, the WUAs would encounter a budget deficit of 145% and 45% for the Arata-Chufa and Ketar scheme, respectively (Table 8).

Discussion
A few key differences were found in the perceptions of the farmers on the sedimentation problems. The difference in views reflects the farmers' awareness of the problems they were facing. For instance, the foremost mentioned driver by the farmers interviewed at the Ketar scheme was the lack of a concrete-lined main canal, while the majority of respondents at the Arata-Chufa scheme pointed to an absent and non-functioning sediment trap. This difference in views can be attributed to the fact that the main canal of the Arata-Chufa scheme was already lined with concrete at the most critical sediment hotspot, whereas it was still earthen at the Ketar scheme. The cross-section (width and depth) of the main canal was the other most reported cause of excessive sedimentation problem by the farmers at the Ketar scheme. The farmers mainly concerned with the cross-section of the main canal at the critical sedimentation sections. This is because the cross-section determines the quantity of sediment cleaned by the farmers. They remove the deposited sediment from the canal bed and weeds from the side banks of the canal at the same time. In doing so, they further dig the bed of the canal and trim the side banks of the canal. This combined activity results in a damaged canal cross-section: deeper, wider, and a changed longitudinal slope of the canal. The maximum width and depth of canal at the sediment hot spot were recorded as 3.2 m and 0.85 m. Though this was not too wide, it was difficult for the farmers to remove the sediment from such cross-section by manual labor only.
The other reasons mentioned by the farmers as the causes excessive sedimentation that related to their acute awareness of their specific scheme were; absence and non-functional sediment trap, source of erosion, and longitudinal slope of the main canal. The Ketar upstream and Arata-Chufa farmers emphasized the importance of having a working sediment trap. Ketar 1 and Arata-Chufa schemes were initially equipped with sediment trap and undersluice gate that serve to flush sediment back into the river, which was non-functional during interview period. The farmers indicated that with timely repair and improved operation of the sediment trap and undersluice gate, the problem of excessive sedimentation could be substantially reduced. This is because they had previous experience with the function of fully operational structures. With regard to sources of erosion, farmers at the Arata-Chufa scheme reported erosion outside the scheme as a major factor aggravating sedimentation problems.
Here they referred to farmers who used pumps to irrigate in a buffer zone of the river just upstream of the intake for causing much of sedimentation problems in their scheme. Ketar farmers, however, attributed excessive sedimentation to erosion of agricultural lands within the scheme. Contrary to other farmers, farmers at the Ketar 3 mainly identified gentle longitudinal slope as a cause of excessive sedimentation problems. This is due to the fact that most reaches of the main canal from Ketar 2 to Ketar 3 was laid in chute structure, which is not suitable condition either for the sediment to settle or for the growth of weed. None of the downstream farmers though mentioned weed growth in the canal as a major cause of sedimentation problem It is not surprising that engineers had somewhat different perceptions of excessive sedimentation than farmers. While farmers saw structural, technical, and external factors as the main drivers of excessive sedimentation, engineers attributed excessive sedimentation mainly to poor scheme operation and maintenance, as well as erosion of upstream areas. One design problem that engineers did note was the location of the intake, though the design issues cited by farmers related to the layout of the main canal (slope, cross-section, lining materials). With respect to technology, the engineers emphasized the potential of moving away from surface irrigation towards pressurized irrigation technologies (sprinklers and drip) as an option to address excessive sedimentation, while farmers demanded technologies for removing the sediment from the canal and night storage ponds and conveying irrigation water. In sum, most of the drivers of excessive sedimentation reported by farmers (too mild longitudinal slope, wide and shallow canals, absence of and dysfunctional sediment trap) were indeed consequences of poor design and operation and maintenance of the schemes. In this regard, the findings of the current study confirm evidence from previous work [2,12,14]. For instance, the longitudinal slope of the main canal at the sediment hotspot was calculated as 2.3% for the Arata-Chufa scheme and 1.4% for the Ketar scheme. This can be regarded as a very gentle slope, confirming farmers' claims. This sediment hotspot reach of the canal with very gentle slope reduces flow velocity and allows the sediment to settle. Designing for an optimum permissible velocity that neither allows sediment to deposit nor scours the canal bed could improve sediment transport in the canals.
At the Ketar scheme, the main sediment hotspot section was found at the upstream scheme (Ketar 1) 5 kilometers from the intake ( Figure 3). This section covers 2433 m (20% of the main canal). The main canal is collectively cleaned mostly once, but sometimes twice a year depending on sediment inflow load. The secondary and tertiary canals, which are adjacent to the field plots, were cleaned by the farmers individually. The work load of sediment management differed between the upstream, midstream and downstream farmers due to the difference in sediment inflow load ( Table 6). The majority of the sediment settled at the upstream (Ketar 1) scheme. Thus, upstream farmers spent more time on sediment management compared to midstream and downstream farmers. Furthermore, downstream farmers were least affected by the sedimentation problem, and contributed the fewest hours of labor to cleaning sediment as they irrigate with water stored at night storage pond. This reveals that the management of excessive sedimentation brings other issues to the fore regarding interactions between upstream, midstream and downstream farmers. Previous studies [2,40,41] argue that upstream farmers may have a comparative advantage over midstream and downstream farmers in terms of water availability; that is, more water may be available to upstream users, with less flowing to the middle and downstream zones. We point out, however, that this is not always the case. The current case study suggests that middle-stream and downstream farmers had similar water allocations to upstream farmers, but invested less time in management of excessive sedimentation.
The role of the WUAs in relation to sediment management tasks were to set the annual sediment cleaning and maintenance dates, monitor sediment cleaning activities, and communicate with the local government bureau to file requests for scheme maintenance and repairs that the farmers could not perform on their own. WUAs also play the role of enforcing the sanction that was set out in the WUA by-laws. However, it was found that these rules were hardly applied. For instance, in the Arata-Chufa scheme, the fines farmers paid depended on crop yields and market values in a particular year. If productivity was high, the sanction to be paid by offending farmers was increased; otherwise, it would be reduced. By-laws stipulating that farmers would not get water if they did not participate in maintenance were also softened, in particular, for women and elderly farmers or at least excused from participation in the heavy work of desilting. At the Ketar scheme, farmers followed their own rules for sediment cleaning. The farmers' group (formed by the farmers themselves) decided collectively what type and magnitude of sanctions to impose on those who did not participate. It was observed that there was relatively good communication and consensus within the groups, which made it easy for farmers to empathize with the situation of those who had not participated in cleaning activities. If the reason for not participating was deemed acceptable, the farmer was excused; if not, an appropriate sanction either in kind or in cash was imposed and the farmer generally paid it. Thus, local norms, values, and social capital seem to have played a substantial role in keeping these schemes functional for more than 30 years. Functions introduced by the "external actors", such as audit, loan, marketing, conflict resolution, and warehouse, were inactive. Instead farmers themselves formed positions like "farmers' group" "field block head". Farmers consider roles that still exists such as head, deputy head and secretary as "traditional role".
The majority of farmers were willing to pay the annual operation and maintenance fee, to contribute labor to clean the sediment, and to pay sanctions if required. Moreover, they considered the annual operation and maintenance fee to be fair. There were, however, various scheme components that the farmers could not maintain and repair on their own, mainly due to the modernization of the schemes. While farmers could do minor repairs of earthen canal works and dredge sediment from the main canal, they could not remove sediment from the night storage ponds. In 2016/2017, Ketar scheme farmers paid a local contactor US$ 13,090 to use heavy machinery to excavate the sediment from the night storage pond. At the time of the field work, Arata-Chufa scheme farmers were facing a shortage of water because they lacked the machinery and funds to pay for sediment to be cleaned from the night storage pond, which supplied 60 ha, or 60%, of the total irrigated area. This reflects the problem of a lack of technology appropriation, which studies have shown leaves users dependent on external technology and developers [5,6]. Farmers were willing to invest, and of course they annually contributed a huge amount of labor to manage excessive sedimentation, but the operation and maintenance fees paid were insufficient to cover major maintenance and repair costs.
The labor output in the Ketar scheme was higher than the Arata-Chufa scheme (Table 5). This difference could be attributed to the work processes implemented to manage the desilting activities at the different schemes. The Ketar scheme had a better system, which was more effective in utilization of the labor days invested by the farmers. Farmers were divided into groups numbering a maximum of 20 each. The desilting operation was then divided among the groups, with every 20 farmers responsible for about 100 m of the canal. At the Arata-Chufa scheme, sediment cleaning was carried out collectively in a process in which only a few farmers could be actively engaged in dredging at a time, while the remaining farmers stood aside and waited for their turn. It is very difficult to compare the labor output at the two schemes to experiences elsewhere in the country, as very little data exists on the quantity of sediment desilted by farmers from canals and numbers of labor days devoted to the task. However, we estimated the opportunity cost of those labor days as US$3457 and US$13,594 for the Arata-Chufa and Ketar schemes, respectively. Comparing these estimates to the total regional operation and maintenance budget for 2017/2018 ($1.2 million), we found that 0.28% and 1.11% of the regional budget would be spent to remove sediment from 50% and 20% of the main canals of the Arata-Chufa scheme and Ketar scheme, respectively. The region had 612 modern, 9379 traditional, and 63,523 pump irrigation schemes in that year (OIDA, 2017).

Conclusions
Excessive sedimentation is indeed one of the major causes of underperformance of small-scale irrigation schemes in Ethiopia. In this study, the stakeholders' roles and perspectives on sedimentation management in two small-scale irrigation schemes, Ketar (430 ha) and Arata-Chufa (120 ha), were analyzed using a collaborative and participatory approach. In these farmer-led irrigation schemes, farmers use their local knowledge and informal institutions to mobilize and engage huge amounts of labor for intensive sediment cleaning campaigns. In the Ketar Scheme, the farmers (1680 in 2016/2017 and 1646 in 2017/2018, respectively) spent on average 3 days per year on this campaign and the farmers in Arata-Chufa (260 in 2016/2017 and 252 in 2017/2018, respectively) on average 5.5 days per year. The upstream farmers spent between 12% (Arata-Chufa) and 23% (Ketar 1) of the total time invested in crop production on sedimentation management, compared to only 3 to 9% of the midstream and downstream farmers. On top of this input in labor, farmers pay annual operation and maintenance fees, US$ 8.37 in Ketar and US$ 4.36 in Arata-Chufa. In these farmer-led irrigation systems, the farmers mainly devoted extra hour of drudgery for desilting excessive sedimentation from the canal, but they have also used their knowledge, such as delaying the abstraction of irrigation water at the start of new irrigation season (end of wet season) to avoid entrance of excessive sedimentation to their scheme together with diluted irrigation water. They have also applied own technique of frequent removal of weed grown in the canal cross-section as they believed it traps a significant amount of sediment load.
Farmers' understanding of the drivers of excessive sedimentation reflected their close personal knowledge of the irrigation system and the sedimentation problems they faced. Farmers and engineers have different perceptions of the causes of sedimentation. The drivers of excessive sedimentation were indeed the consequences of poor or faulty design, poor operation, and maintenance practices and external factors like erosion due to degradation of the land and low-technology level of water conveyance systems. Farmers reported design problems as the main cause of excessive sedimentation (64%), followed by poor operation and maintenance (O and M) practices (21%), and external factors (15%). Contrary, the engineers indicated erosion and irrigation-technologies as the main causes of excessive sedimentation (62%) and only 13% on design problems. Though low-technology level contributed to the excessive sedimentation problem, lack of adaptation and adoption of the technology by the farmers have aggravated the problem The existing role and structure of the Water Users Associations are significantly simplified compared to the institutional set-up introduced by external actors, as most of the planned management layers and committees were not operational. Local social capital appeared to be more important than by-laws in enforcing O and M practices. It can be concluded that the cost of sediment management for the farmers is very high and requires new socio-technical solutions that capitalize on the existing local social capital, norms, values, and indigenous knowledge. The integration of the farmers' knowledge with that of the engineers could yield more effective ways to deal with sedimentation problems. To implement a sustainable intervention to rehabilitate excessive sedimentation problems in farmer-led irrigation systems, it should follow a proper technology appropriation by the farmers.