Supply and demand of ecosystem services among smallholder farmers in irrigated and rainfed farming, Kilombero, Tanzania

ABSTRACT

A sufficient supply of ecosystem services (ESs) in agriculture provides the basis for human sustainable development. Intensified large-scale farming has changed wetland ecosystems extensively by reducing both the resilience and capacity to support production of many ESs. Small-scale farming may also affect the generation of ESs where the impact often reflects the differences in farming practices. This paper explores the supply and demand of the ESs between management practices, irrigated and rainfed, of smallholder farmers in Kilombero wetland, Tanzania. We conducted interviews involving 30 households and two focus groups with five discussants for each practice, rainfed and irrigation. Generally, we found that the need for ES, especially food, water and flood control, in both farming practices, were exceeding the capacity to supply. In general, irrigation farming compared to rainfed farming was associated with higher levels of food production, increasing flood regulation and erosion control. However, the ES delivery and need were not uniform depending on the river discharge. The differences in supply and demand of ESs between farming practices suggest that society would benefit from investing in irrigation and regulatory infrastructures to minimize flooding risk and to build up the ecosystem’s natural capacity to produce services. Such practical policy-relevant measures could balance the gap between supply and demand of ESs in smallholder farming systems in Tanzanian wetland.

Edited by:

• Patrick O’Farrell

KEYWORDS:

• Ecosystem services
• supply and demand
• irrigation
• rainfed farming
• farming practices

1. Introduction

Wetlands are among the most productive ecosystems on the earth providing many important goods and services to human, ecosystem services (ESs) (Millennium Ecosystem Assessment 2005; Sandifer et al. 2015; Burkhard and Maes 2017). Important ESs provided by wetland ecosystems are water purification and supply, medicine and raw materials, nutrient regulation, sedimentsand flood controls (Clarkson et al. 2013; Berbés-Blázquez et al. 2016; Burkhard and Maes 2017). Moreover, wetlands are important for biodiversity, offering breeding sites and forage for aquatic insects, birds and fish (Van Jaarsveld et al. 2005). These systems also provide important cultural services (Plieninger et al. 2013) such as recreational and spirituals sites. Wetlands are ecologically sensitive habitats and are threatened globally due to human activities, particularly agriculture and current pressure from shifting climate (Mouchet et al. 2014; Burkhard et al. 2015; Díaz et al. 2015; Wei et al. 2021). About 35% of wetland has been lost from 1970 to 2015 (Ramsar Convention on Wetlands 2018), which has negatively affected ecosystem function and service provision, especially water-related ESs such as water quantity and quality regulation and flood control.

To investigate what ES that can be supplied from an area and the demand of these services, a common tool is to use integrated assessments of ES supply and demand (Burkhard et al. 2014; Mouchet et al. 2014; Zoderer et al. 2019), often adapted to landscape type and management practice (Zhang et al. 2007; Zoderer et al. 2019; Wei et al. 2021). In such assessments, it is important to separate supply of services – capacity and provisioning from the ecological system (Villamagna et al. 2013; Wei et al. 2017), from demand of services – social beneficiaries derived from the ecological system (de Groot et al. 2010; Yahdjian et al. 2015; Wei et al. 2017). The supply of an ES is not constant and uniform and normally affected by external drives such as temperature, rainfall, landscape configuration and management practices. For example, efficient management is often needed in wetland agriculture to prevent flooding and erosion of the fertile soils (Kato 2007). The Mekong delta (Vietnam) and some parts of Rufiji delta (Tanzania) are examples of areas associated with intensive and well-controlled irrigation rice production (Duvail and Hamerlynck 2007; Berg et al. 2017) and mitigating flooding (Duvail and Hamerlynck 2007). In addition, initiatives are taken to increase production on par with flooding management and enhancing soil fertility.

The relation between ES supply and demand has a strong impact on human well-being. From this perspective, understanding the different community’s perceptions of supply and demand of ES is crucial in order to secure sustainable food production (Sanogo et al. 2017; Nzau et al. 2018; Mbande 2022). Such understanding depends on geographical setting, socio-economic characteristics and market forces. Communities close to the resources influence the demand (Burkhard et al. 2014; Al-Amin et al. 2021), although preference and management practice may also affect demand and supply of ES (Costanza et al. 2017; Syrbe and Grunewald 2017). For example, irrigated commercial farming generally produces agricultural products that can meet their demand, whereas the demand from rainfed smallholder farming usually is higher than the supply (Mwakalila and Noe 2004; Filipski et al. 2013; Malinga et al. 2018). Although large-scale commercial farming has a tremendous impact on ES generation and landscape (Tscharntke et al. 2012; Malinga et al. 2018), also small-holder farming systems can result in overuse and land degradation as they lack the resources to overcome these problems. However, overuse and erosion could be mitigated by small-scale irrigation systems and low-intensive use of irrigation and fertilizes among smallholder farmers, enhancing the ability to increase food production (Alavaisha et al. 2021). Differences in quality or quantity (mismatch) between ES supply and demand (Geijzendorffer et al. 2015; Wei et al. 2021) may cause current demand or potential demand for specific ES to be unsatisfied. On the other hand, the supply of an ES should spatially match the demand, locally and regionally. Therefore, a service production area often has to be physically linked to the service benefit area (Burkhard and Maes 2017). For example, water retention ES supply prevents excessive water flows during flood events, providing direct benefits to people living in flood plains regions. Overall, a better understanding between supply and demand and potential mismatches could provide important information for ES decision-making and policy-making (Costanza et al. 2017).

In Tanzania, irrigation farming adjacent to wetland is growing as a means to prolong the faming season and ensure food productivity under climate change. It is estimated that 381,000 ha is under irrigated agriculture, with 359,000 ha dominated by smallholder and small commercial farms (http://www.icid.org/v_tanzania). Paddy rice is the most common product grown within schemes, and yields can be as much as two to three times those of rain-fed areas (Kaswamila and Masuruli 2004; Filipski et al. 2013). However, in flood plains, the production can be even higher (Kato 2007; Siima et al. 2012; Seki et al. 2018). The backdrop of this agricultural development is often a loss of natural wetland and associated ES. Natural wetland vegetation is important for supply of fresh water and erosion control. Therefore, farming in the flood plain may cause a trade-off between ES as it increases production (Msofe et al. 2019; Alavaisha et al. 2021) but reduces vegetation. For instance, intensified food production in short term may diminish ES from natural ecosystems in long term (Ndetei 2006; Mwakaje 2009; Popp et al. 2013). This is especially interesting in the case of flood regulation as related services have to be provided, which are directly linked to the area where the demand is located, for example, along the same watercourse or within the same watershed. In contrast to many other ES, flood regulating services cannot be imported from other regions (Syrbe and Grunewald 2017). Farmland inside and outside these schemes contributes to flood regulation and water retention which protects settlements further downhill.

Wetlands are under pressure from different farming practices affecting its ability to provide ES. In this study, we use the Kilombero Valley in Tanzania to study the supply and demand of ES in irrigated and rainfed small-scale agricultural systems. Because of the projected population growth and migration into the valley, there is a need to understand the supply and demand of essential provisional, regulating and cultural services for sustainable future planning of agricultural in the landscape. Specifically, the study addressed the following key questions: (i) What are the general perceptions of ESs supply and demand among farmers? (ii) Is there a distinct difference between service supply and demand in rainfed versus irrigation practices? And (iii) to what extent do the services contribute to communities’ livelihood in irrigated and rainfed farming?

2. Materials and methods

2.1 Study area

The study area is located in Kilombero Valley wetland, Kilombero district, southern central part of Tanzania (Figure 1). The Valley is characterized by the flood plain 260 km long and up to 52 km wide, covering an area of around 4,000–6,250 km² during inundation at high water level (Vanden Bossche and Bernacsek 1990). The Valley flood plain is recognized by Ramsar convention because of its uniqueness in terms of productivity enabling rich flora and fauna (RAMSAR 2002; Nindi et al. 2014). The ecosystem also supports wild life in Selous game reserve and Udzungwa reserve and national park. Natural vegetation is the evergreen forest of the Udzungwa Mountains, riverine forest and marsh grassland in the broad valleys, which are used for agriculture.

Figure 1. Figure indicating study sites at Kilombero Valley. Njage sub-village is located close to Kilombero wetland, and they practice irrigation farming, while King’ulung’ulu is located near to residential areas and Udzungwa reserve, and they practice rainfed farming. The two sub-villages are located at Njage village.

The valley has a sub-humid tropical climate and is characterized by two distinct rainy seasons: a long rainy season from March to May and short rainy season from November to January. Each year, beginning in the late November or early December, the river floods and overspills the banks, creating the large floodplain, which gradually recedes back to the main channels during May and June (RAMSAR 2002). This annual inundation created by the seasonal flooding is the key environmental driver of both terrestrial and aquatic ES generation. The seasonal nutrient inputs from run-offs and the flooded marginal areas are crucial for crop production in the area (RAMSAR 2002; Wilson et al. 2017; Bassi et al. 2018).

The valley has both rainfed and irrigation farming with well-established small-scale irrigations schemes. Flooding and application of fertilizers are the main source of nutrients in the farming areas. Rainfed systems have commonly one season farming, while irrigation farming has either two to three farming seasons. Major crops in the valley are rice and maize; other crops are cocoa, sesame sunflower, peas, banana and vegetables. Most villagers cultivate rice and sugarcane in the alluvial fans and maize on the hill slopes. The Valley is characterized by demographic pressure. Because of the initiative of Kilimo kwanza (agriculture first) and drought in highlands, farmers have moved to the valley to engage in irrigation farming, ranging from small to medium scale (Colin 2018). Currently, the larger landscape consists of a mosaic of agricultural fields and intensive farming, both within and outside irrigation scheme.

2.2 Sampling design and data collection

We used Njage and King’ulung’ulu sub-villages located in Njage village as cases to explore the supply and demand of various ES associated with the practices irrigation and rainfed farming (Figure 1). Prior to the assessment, a reconnaissance survey was conducted to identify key ESs among smallholder farmers (presented in the first column of Table 1). We also performed a pre-testing of interview tools to assess respondents understanding of the questions and to identify any problems encountered before actual interviews. In all sites, farmers had a long history of both farming practices. King’ulungulu sub-villages are located far from the irrigation scheme, and the majority of farmers are conducting rainfed farming, whereas in Njage sub-village, most farmers engaged in irrigation farming.

Table 1. Indicators for ecosystem services provided and needed in rainfed and irrigation farming practice. All indicators were measured using a scale from 1 to 5, where 1 = very low, 2= low, 3= moderate, 4= high and 5= very high level of ES, supply or demand.

ESs	Indicator
Provisional of fresh water	Levels of water availability for agriculture.
Provisional of food (maize and rice)	Levels of production of food in a year
Water regulation	Level of services in a year
Flood hazard regulation	Level of services in a year
Erosion regulation	Level of services in a year
Water purification	Level of services in a year
Recreation and tourism	Levels of recreation and tourism in a year
Esthetic value	Level of esthetic value in a year
Education and research	Level of number of educational excursions in a year

To assess the supply of provisional, regulating and cultural services, we adopted and modified the methods used by Nzau et al. (2018), Turner et al. (2000) and Sanogo et al. (2017). The methods included mapping perceptions and attitudes of the smallholder farmers on ES supply and their contribution to livelihood, their demand. Data were collected during field work between October and December, involving primary data collection through focus group discussions (FGDs) (Appendix S1) and household questionnaire surveys (Appendix S2). The study included two FGDs each with five individuals and 30 head of household for questionnaire interview, for each farming practice, respectively. Participants were selected based on their experience of farming management and understanding of local agroecosystem transformation in the valley.

Purposive random sampling was applied in village register to obtain participants for FGDs. Criteria for selecting participants for FGDs were (1) long-term farming practices, (2) basic understanding of socio-ecological dynamics and ESs and (3) being a member of conservation, or environmental-related committees at village level. Questions were focused around major wetland uses such as the wetland status and their potential need (demand) and supply of provisional, regulating and cultural services. Following the guidelines of Powell and Single (1996) and O.Nyumba et al. (2018), two FGDs of five mixed participants in terms of gender and community role were chosen for each site. Kitzinger (1995) suggests that a group of four to eight participants is sufficient to enrich discussions. In each group, the composition included the smallholder farmers, conservation/environmental committee members and village council members.

Selection of participants for household questionnaire interview was made through systematic randomly sampling among heads of household, listed in village register, that participated in small-scale farming. About 20% and 33% of female-headed households were involved from irrigated and rainfed farming management, respectively. Semi-structured household questionnaires were administered to each household to gather information under four major themes: (i) socio-economic information including demography, household structures and economic activities in the sub-village, (ii) provisional ESs, including level and quality supply and demand of freshwater, crops production (maize and rice), (iii) regulating services, their level and quality of supply and demand for hazard regulation, erosion control and water purification and (iv) cultural services accounting for the need and provisional of recreation and tourism, authentic value of landscape, education and research. Survey questionnaire was conducted at the respondent’s home.

2.3 Data analysis

Qualitative data obtained from FGDs and household questionnaire surveys were analyzed through content analysis (Weber 1990; Hsieh and Shannon 2005). Descriptive analyses were used to analyze farmers’ res-ponses on level of ES supply and demand. Frequencies and percentages distribution of questions responses were used to generate tables and figures. Demographic and socio-economic data included gender, age categories, level of education, household size and occupations and level of income were summarized. We used the Likert scale, where participants were asked to rate the extent of needed and supplied ES from a scale of 1 (very low) to 5 (very high). We then developed a gradient matrix showing the relative contribution of ES to livelihood from potential significant positive to negative contribution. The categorical response from level of supply and demand of ES was compared among farming practices using chi-square test. Statistical analyses were performed using Microsoft Office Excel package (Office 2016) and IBM SPSS statistics (Version 26). All results were considered statistically significant at P < 0.05.

3. Results

Socio-demographic characteristics

The farmers who practiced irrigation slightly differed in socio-demographic characteristics compared to those practicing rainfed farming (Table 2). More than 50% of farmers conducting irrigation were below 30 years old, whereas in rainfed sites the farmers were above 45 years old (80%). Farmers in the irrigated areas were higher educated, about 60% completed secondary education compared to 3%, in rainfed farming. Instead, a majority of farmers practicing rainfed farming had informal education (60%) based on traditional indigenous knowledge. To practice irrigation farming, at least reading and measuring skill are necessary to be able to estimate use of fertilizer and spacing in planting. The average family size was higher in rainfed farming, about 60% of families had more than six members and had low income compared to the household engaged in irrigation farming. Furthermore, a majority (about 70%) of the farmers practicing irrigation were migrants, resided less than 5 years as opposed to rainfed farmers were the majority (87%) have lived more than 10 years.

Table 2. Socio-demographic characteristics of respondents of Njage (irrigation farming) and King’ulung’ulu (rainfed farming), Tanzania.

Particulars	Categories	Njage (N = 30)	King’ulung’ulu (N = 30)
Age	15–30	53.3	6.7
	31–45	36.7	13.3
	46–60	10	46.7
	Above 60	0	33.3
Gender	Male	80	66.7
	Female	20	33.3
Education	Informal education	10	60
	Incomplete primary education	6.7	0
	Complete primary education	23.3	36.7
	Complete secondary	60	3.3
Family size	1–5	63.3	40
	6–10	36.7	46.7
	11–15	0	13.3
Monthly average income	50,000–99,999	13.3	6.7
	100,000–249,999	10	30
	250,000–499,999	16.7	60
	500,000–999,999	10	0
	1,000,000 and above	50	3.3
Size of farm (acre)	Less than 1	70.0	6.7
	2–3	30.0	50.0
	More than 4	0.0	43.3
Years lived in village	1–5	70	6.7
	6–10	16.7	6.7
	11–15	6.7	56.7
	Above 15	6.7	30.0

Supply and demand of ESs

The supply and demand differed among ESs and irrigated and rainfed agriculture (Figure 2, Table S2). The demand for the services food, freshwater and water regulation was higher than the supply in both rainfed and irrigation farming practices. However, the demand for food, water regulation, flood hazard regulation and erosion regulation was significantly higher (P < 0.05) in irrigation farming than in rainfed farming (Table S2).

Figure 2. Levels of supply and demand of ES in irrigated and rainfed farming practice at Kilombero Valley, Tanzania. The number (1 to 5) indicated by color are levels of ES supply and demand, where 1= very low, 2= low, 3= moderate, 4= high and 5= very high level.

The levels of food supply, water regulation, erosion control, flooding hazard regulation and water purification did not match the farmers’ demand (Figure 2). Irrigation farming is located in the lowland where flood regulation, erosion control and water purification services have high levels of demand (Table S1). Supply of fresh water and food was significantly higher (P < 0.05) in irrigation farming compared to rainfed farming (Figure 2, Table S2).

More than 86% of the crop-producing households reported that rice was the primary agricultural crop in both rainfed and irrigation farming. About 63% (irrigation) and 70% (rainfed) of the respondents indicated significantly higher (P < 0.05) levels of food production from irrigation compared rainfed practice, respectively (Table S2). Despite the high supply of produced food from irrigation practices, the demand for food was higher for both irrigation (83%) and rainfed farming (66%). The higher demand was mainly due to the interest to increase profit among irrigation farmers (Table S1). The average production in irrigation farming was from 30–35 bags (each bag 175 kg) of rice/ha while only 11–25 bags/ha from rainfed farming practice.

Across the wetland, water was extracted for irrigation, animal and domestic use. Respondents reported the supply of fresh water as low to moderate in both rainfed (70%) and irrigation farming (65%) (Figure 2). In the rainfed farming practices, water is supplied to crops during rainfall through seasonal streams and stored water in the forest canopy, unlike irrigation farming, where water is supplied throughout the year through canals and streams. Approximate 73% of rainfed farming respondents reported a very high demand for water, especially during dry seasons, compared to irrigation farming (52%).

Erosion control was also highly demanded, and more than 80% of respondents in irrigation farming were concerned about erosion of riverbanks and higher discharge during rainy season, which was much higher than for rainfed farming. Focus group opinions and field observations showed that all rainfed areas, particularly the vegetated ones, stabilize and trap sediments, increasing the supply of fresh water.

Both irrigated and rainfed farming provide opportunities for scholars and recreational services and tourism (Figure 2; Table S1). However, the demand for this was low among farmers except for tourism as it was considered an alternative source of income for communities near the flood plain. Generally, irrigation areas close to the flood plain provided more bird watching tourism than the rainfed, but here the main attraction is the forest and closeness to Udzungwa forest reserve. The natural beauty of the flood plain and mountain forest contributes to esthetic value of the landscape. Additionally, irrigation-farming practices were associated with educational visits by rice researchers. Rice farm field schools (shamba darasa) demonstrating intensive rice farming system were common in irrigation farming for both education and research.

The majority of irrigation households reported that their practices provided research opportunities for scholars as well as recreational services for households, which was not commonly highlighted by rainfed farmers. Irrigated landscape had been transformed to agriculture, leaving very few areas of natural beauty. Education and research activities were conducted mainly within irrigation farming. Much investment was also on rice research and water use/flow in irrigated areas close to the flood plain because of complex relationship between the flood plain biodiversity and agriculture.

ES contribution to livelihood

Food and fresh water were the most important provisional services for livelihood in both irrigation and rainfed farming practices (Table 3). Water was highly important for maintaining agriculture in irrigation systems during dry season and for domestic uses. Irrigated farming usually drains water from river for rice irrigation. Rainfed farmers were not farming during dry season, except for very few sites that retained moisture to support maize growth. Interviews reveled that high rice yield was important to secure food and livelihood in both management practices. Rice was the most preferred crop compared to maize, because of their high price and production, especially to communities whose livelihood depends on irrigation farming.

Table 3. Relative contribution of ecosystem services to community welfare at Njage and King’ulung’ulu in Kilombero Valley.

ESs	Irrigation					Rainfed
	(++)	(+)	(0)	(-)	(–)	(++)	(+)	(0)	(-)	(–)
Provision of fresh water	83.3	16.7					90	10
Provision of food	93.3	6.7					90	10
Water regulation	26.7	46.7	27				76.7	17
Flood hazard regulation	36.7	56.7	6.7				46.7	40	3.3
Erosion regulation	76.7	19	3.3				20	43
Water purification	73.3		26.7			80	20
Recreation and tourism	20	73.3	6.7			6.7	77	17
Esthetic value	6.7	93.3					17	83
Education and research	86.7	13.3				3.3	57	40

(++) Potential significant positive contribution, (+) Potential positive contribution, (0) Negligible contribution, (-) Potential negative contribution, (–) Potential negative contribution, Blank= no response.

Both water regulation (73%) and purification (73%) are significant for livelihood in irrigation farming practices (Table 3). Respondents reported that the services are especially important during rainy season, because of high discharge and turbidity. Interestingly, for rainfed farmers, water regulation services had a positive contribution to livelihood. Riparian vegetation and grassland were mentioned as important to reduce the impact of flooding on crops in both irrigation and rainfed sites. In addition, flood regulation also reduces erosion and maintains soil nutrients potentially increasing production and income for irrigation farmers (Table S1). For such regulation to be maintained, even during storm discharge, farmers like to see the vegetation in the flood plain and river banks protected.

4. Discussion

Overall, there was a difference in supply and demand of ESs between farmer practicing irrigated and rainfed farming. The supply of the provisional services food and water was not equal between practices, where irrigated areas had higher food production and better supply of water-related services compared to rainfed farming (Figure 2; Table S2). This is congruent with previous studies (Clover 2003; de Vries et al. 2010; Coates et al. 2013; Adam et al. 2016), showing that improved irrigation farming practices increases supply of food compared to rainfed, contributing to improving communities welfare. As demand exceeds ecosystem capacity to provide food in rainfed practices, this practice experiences recurrent food shortage. Such uneven balance of food production suggests that society should invest in irrigation infrastructures to produce more food and accommodate small scale farmers (Chiwaya 2013; Filipski et al. 2013; Mbande 2022). To decrease demand and balance the supply–demand gap is a way to improve livelihood as well as minimize the pressure to convert large areas to farming field (Villamagna et al. 2013; Alavaisha et al. 2021; Wei et al. 2021). Availability and supply of water can increase production through informal management of water involving less water use rice intensification system (Ceesay et al. 2006; de Vries et al. 2010; Ye et al. 2013; Alavaisha et al. 2022). During rainy season in our study, the water supply was sufficient for both practices, but in dry season, the supply was insufficient for rainfed farming. In contrast, other studies have indicated low production in irrigation farming practices due to flooding (Clover 2003; Mwakalila and Noe 2004), poor farming technology and insufficient education among farmers (Comas et al. 2012; Mlenga 2015; Marambanyika et al. 2021).

Regulating ESs are highly affected by different management practices at different spatio-temporal scales and is thus especially susceptible to successful implementation of the ES concept in environmental planning (Burkhard et al. 2014; Lindborg et al. 2017; Wei et al. 2017, 2021). This is particularly relevant to erosion control, flood risk regulation, water purification and flow regulation, where demands often are spatially separated from supply. In this study, the flood regulation and erosion control were demanded at higher levels for irrigation farming practices compared to rainfed, although the supply was insufficient for both practices. Similar findings were reported by de Groot et al. (2010) and Sojka and Bjorneberg (2002) where erosion control, water flow regulation and purification ES generally had low supply in catchment despite high demand. The difference in river catchments physical characteristics creates unique discharge regimes and discharge responses to precipitation (Postel and Thompson 2005; Senkondo et al. 2018), influencing management practices to supply ES. This was also shown in our study, where the capacity to protect against flooding risk and erosion had a higher demand in irrigated farms compared to rainfed as they are located higher up in areas less susceptible to flooding. The potential supply of regulating services is strongly dependent on the distribution of soil and land use, in turn affected by management regime. Irrigation farming in the low-land increases the stream flow but retains nutrients because the lowland is flat, compared to rainfed farms, being located slightly higher up in the catchment. In addition, more nutrients are trapped through riparian vegetation and sediments, thus contributing to flood plain fertility and protecting the water quality downstream (Alavaisha et al., 2019; Seeteram et al. 2019). Therefore, having suitable and well-managed vegetation in the flood plain along irrigation canals may enhance flood and erosion control and mitigate nutrient loss, especially during long rainfall. Moreover, enhancing food production through irrigation could also help protect ESs in other parts of the valley, as food production could be concentrated in irrigation farming areas, leaving the less fertile highland as conservation areas (Fischer et al. 2014; Ashley 2016; Tsiafouli et al. 2017; Alavaisha et al. 2021)

Regulation of stream flow to protect the irrigation infrastructure and crops and reduce risk of flooding is also important to minimize the risk of famine in flood plains (Coates et al. 2013). This has been discussed in several studies (e.g. Fukai et al. 1998; Berg et al. 2017; Al-Amin et al. 2021) highlighting the link of regulating services to human health and economy. Even though previous research on livelihood has reported that the use of irrigation is a question from wealthy practice, there are still many low-income farmers practicing irrigation. The situation can be explained by high input cost associated with management of flooding and erosion, in the absence of natural vegetation. Several farmers in our study expressed that they received government and donor subsidies by practicing irrigation, which partly minimized production cost, and encouraged farmers to further engage in irrigation. Irrespective of the risk of flooding and erosion farmers accept living along the flood plain, where ecological flood control, mitigation strategies and preparedness for disaster is closely tied to their livelihoods (Duvail and Hamerlynck 2007; Eakin and Appendini 2008).

Education, research, esthetic value and tourism had a slightly lower supply than demand. In particular, for research and education over 40% of respondents in both irrigation and rainfed farming indicated low capacity of the system to deliver these services (Figure 2; Table S2). Other studies also show that research and education are important components to improve management of ESs (Burkhard et al. 2014, 2015; Sanogo et al. 2017). Landscapes with a lot of irrigation farming was found to have low esthetic value, although other studies have reported opposite (Peña et al., 2015). One reason was transformation to monoculture, accompanied with loss features symbolizing natural beauty of the land (Lindemann-Matthies et al. 2010). Intensive single crop farming results in simplification of landscapes reducing natural vegetation and ecological functions in the ecosystem (Tamburini et al. 2016; Grab et al. 2018). Esthetic and recreational value can be understood in multiple ways and different features attract different people: e.g. species birds in flood plain close to irrigation farming and wild animal in the forest close rainfed farming. In addition, the identity of being the caretaker of a farm and its surrounding landscape was strong in both systems, and both were proud of the qualities of their landscape.

Farmers using irrigation were different in terms of socio-economic characteristics compared to farmers relying on rainfed crop production. We found that a majority conducting irrigation farming were aged below 30 years and have been living there less than 5 years. Several other studies have shown that young people are more motivated to engaged in commercial irrigation farming (e.g. Comas et al. 2012; Adam et al. 2016; Colin 2018). In Kilombero, improvement of education coupled with government policy of irrigation contributed to the increased engagement in irrigation farming compared to rainfed. Despite government policies to encourage irrigation, market forces and differentiation among farmers were also important factors. For instance, rainfed farmers were shifting to irrigation after accumulation of capital. This has been motivated by increasing demand for rice and horticultural products in the market, resulting in centers closer to the irrigation canals constantly growing over time (Colin 2018; Mbande 2022).

5. Conclusion

Overall, our study showed a variation in ES supply and demand between irrigation and rainfed farming systems. Demand for food exceeded ecosystem capacity to provide it in both irrigated and rainfed farming practice, although there was higher production in irrigation farming. Further, landscape configuration was an important aspect where areas in low land, dominated by irrigated farms were more productive yet susceptible to flooding. This also holds for other services like the erosion control and water regulation that were highly demanded in irrigated farms, compared to rainfed farms that were located upland. Therefore, the link between supply and demand, and farm location has to be considered to fully understand differences in perceptions between farmers.

The farming practices also showed potential for improving livelihood through increasing food production, regulating flood and erosion, particular in irrigation farming. This was related to socio economic characteristics of farmers, as the majority practicing irrigation were educated with higher income compared to those of practicing rainfed farming. Therefore, to balance between supply and demand of ES, the society should invest in irrigation and regulatory infrastructures that matches farmers socio-economical situation but also taking location and the landscape configuration into account. Enhancing the natural capacity of ecosystems to produce services could hence simultaneously improve livelihood of smallholder farmers.

Acknowledgments

The authors acknowledge support from the Swedish International Development Agency under Grant SWE-2011-066 and Sida Decision 2015-000032 Contribution 51170071 Sub-project 2239. We thank Simon Julius for assistance with questionnaires survey, village leaders, committees and communities at Njage village for their willingness to participate in this research and cooperation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/26395916.2022.2143571

Additional information

Funding

The work was supported by the Swedish International Development Agency (Grant SWE-2011-066 and Sida Decision 2015-000032).