Analysis of farmers’ perceptions of bench terracing innovation in the eastern and southern Ethiopian highlands

ABSTRACT

Land degradation threatens environmental and agricultural development in the 21st century. To alleviate this problem, bench terracing has been implemented in eastern and southern Ethiopia. This paper investigates how farmers perceive the attributes and effectiveness of bench terracing in Ethiopia. A Multi-stage sampling techniques were applied to select 384 sample households. For this study, data were collected through primary and secondary sources, and the collected data were analyzed using descriptive statistics and content analysis methods. Primary data were collected using semi-structured questionnaires, focus groups, and key informant interviews; secondary data came from local authority reports. We found that bench terraces restored damaged land and improved crop yield where they were aptly implemented and maintained. The findings also disclose that 57.3% of farmers perceived that bench terracing was more cost-effective; 60.7% responded that it is compatible with the socio-cultural context; and 59.8% perceived Its outcomes are observable to others. However, when a farmer lacks sufficient social, human, or financial capital holdings and capabilities, it often fails. We conclude that the technology was adopted through a multifaceted process, promoted or hindered by both its attributes and effectiveness. Policy-makers and Planners should center those restraints on designing, implementing, and maintaining bench terracing.

GRAPHICAL ABSTRACT

KEYWORDS:

• land degradation
• environment
• innovation
• effectiveness
• crop yield

Public Interest Satement

In this article, we report farmers’ perceptions about bench terracing innovation in eastern and southern Ethiopia. Our research discloses that it is largely preferred for sloppy agricultural land in Haramaya and Soro districts of Ethiopia in response to land degradation caused by natural and anthropogenic factors. It works well to encourage ecosystem restoration and expand agricultural lands. Farmers recognize it as an echo that rewards them for their efforts. About 60% of respondents perceived that it was cost-effective, compatible with their sociocultural context, and that results could be observed among other farmers. On the contrary, most farmers (81%) agreed that designing &implementing within their own capabilities was difficult. In order to strengthen their capacity and increase their motivation to use technology, farmers must actively participate in all parts of technological development. Their opinions’ documentation can be used as additional proof to support appropriate policies and programs

1. Introduction

Agriculture is a foundational pillar of the Ethiopian economy and is the major source of raw materials for capital investments. It contributes 27.5 billion dollars, or 32.7% of the GDP of Ethiopia, employs 79% (two-thirds) of the population, and generates 79% of foreign currency through exporting products (NBE National Bank of Ethiopia, 2020; UN United Nations, 2020). It is rapidly changing as a result of environmental policy changes and increased relationships among innovation actors (scientists, agents, and farmers), which press smallholders on how to manage information and knowledge in agriculture (Spielman et al., 2016). The Ethiopian government has also invested heavily in innovations to improve the situation of smallholders. Smallholders cultivate 95% of Ethiopia’s farmland, and they produce 90% of the country’s agricultural output (Dadi et al., 2020; Gelaw, 2017), implying that smallholders have a significant contribution to Ethiopia’s agriculture sector. However, notwithstanding their contribution, smallholders continue to face several challenges that hinder agricultural productivity and food security (Admassie & Abebaw, 2021; Sileshi et al., 2019).

Land degradation is the major threat to environmental and agricultural development in the 21st century in the Ethiopian highlands. About 14.3 million hectares of land in Ethiopia (about 50% of the Highlands) is severely degraded (Gashaw, 2015). Land degradation accounts for 25%, 22%, and 14% of the terrestrial areas in Ethiopia, Kenya, and Tanzania, respectively (Kirui & Mirzabaev, 2016), inferring that Ethiopia has faced more severe land degradation than some East African countries. Soil erosion by water is the major form of land degradation in Ethiopia under various land uses. The average annual rate of soil loss in Ethiopia is projected to be 12 t ha⁻¹y⁻¹, and it can exceed greater than 300 t ha⁻¹y⁻¹ on steep slopes (Gashaw, 2015). The highest rate of soil loss happens on cultivated lands. Although the magnitude varies, soil erosion in Ethiopian highlands on cultivated lands ranged from 42 t ha⁻¹y⁻¹ (Hurni, 1993) to 179 t ha⁻¹y⁻¹ (Shiferaw & Holden, 1999), and areas with a high to very high erosion risk (between 30% and 50% slope) have a hilly and rugged landscape. On severely cultivated cereal crop fields in Ethiopia, this could rise to 300–400 t ha⁻¹y⁻¹ (Jaleta, 2020). The extent of land degradation over the last few decades continues to result in annual agricultural production losses of 2–5% of GDP (Schmidt & Tadesse, 2019), and the inability of Ethiopian agriculture to feed the population is partly owing to the collective effect of land degradation in the highland part of the country (Hussein, 2021). This implies that degradation increases the threat to the livelihoods of people in Ethiopia. About the cost of land degradation, different estimations indicate that it consumes a large amount of national income. e.g. between 2005 and 2014 alone, the cost of degradation of productive land in Ethiopia was about 35 billion USD, which was higher than 2 billion USD in Malawi, 11 billion USD in Kenya, and 18 billion USD in Tanzania (Kirui & Mirzabaev, 2016).

Concerning the extent of land degradation and its impact on environmental and agricultural development, since the 1970s, nationwide state-led SWC practices have gained momentum in Ethiopia. Afterward, a watershed-based approach took hold in the 1980s (Hurni et al., 2016). However, the interventions were found to be insufficient and unsustainable. This is due to the earlier SWC program’s top-down approach to scaling up environmental protection, covering large areas, and building the physical assets of the poor (Gebregziabher et al., 2016; Zeleke et al., 2016) rather than farmers’ perceptions and interests (Assefa et al., 2019). To reverse the earlier failure of the SWC approach, the experiences gained helped the Ministry of Agriculture (MoA) develop a national soil conservation and rehabilitation program (SWC program), called sustainable land management (SLM), in 2008, which has been funded by various donors. The SLM program aims at promoting SWC measures such as bench terraces, soil or stone bunds, trenches, cutoff drains, drainage canals, and check dams (Schmidt & Tadesse, 2019). BT was implemented by farmers in this effort and was implied to be technically well-made after 2008 (Admassie & Abebaw, 2021; Yaebiyo, 2018). Areas with soil erosion, dense populations, food shortages, high unemployment, and small land holdings should be used more for bench terracing than other SWC measures (Derrick et al., 2019; WOCAT World Overview of Conservation Approaches and Technologies, 2019). The steeper the slope of the plot, the more likely it is that the farmer will adopt bench terraces (Legesse et al., 2021; Sileshi et al., 2019). The studies reveal that the use of BT achieved multiple functions and services in improving environmental quality, such as reducing runoff, improving land productivity, increasing crop yield, improving biodiversity, and creating aesthetic landscapes (Mesfin et al., 2018; Wei et al., 2016; Yaebiyo, 2018). With this implementation, a variety of vegetables and crops could be grown, fruit trees could be planted, and soil erosion and flooding from hillsides could be stopped (Yaebiyo & Emuru, 2018). Wei et al. (2016) and Gebretsadik et al. (2017) found that BT was effective in creating job opportunities and could promote ecosystem restoration and increase crop yields.

Despite years of intense scaling efforts done by state partners to promote the adoption of this technology, the rate of adoption by farmers is very low. Thus, land degradation remains a significant development obstacle in Ethiopian agriculture. The factors that contributed to the failure of the scaling efforts of such interventions were that regional-level decision-makers developed annual agricultural plans without much participation from key professionals in the field and neglected local knowledge and farmers’ interests (WB World Bank, 2019). This approach, given the limited chances for farmers’ intrinsic values, constraints, and the entire farming system, expects farmers to adopt technology mechanically rather than with purpose. The conventional emphasis on linear processes—moving agricultural knowledge and innovations from scientists to extension agents who then reach smallholders—is an overview of complex processes that are tinted by non-linear learning processes, feedback loops, and other complex interactions that occur among heterogeneous actors (Diriba et al., 2019; Hurni et al., 2016; Spielman et al., 2012). It was also labor-, time-, and effort-intensive and entailed higher implementation costs (Derrick et al., 2019; Karaya et al., 2020) than other measures, like soil and stone bunds. Economic, institutional, and human-specific factors affect technology adoption (Mwangi & Kariuki, 2015). Meijer et al. (2015) revealed that the adoption level of innovations among smallholders is determined by their knowledge, attitudes, and perceptions. Therefore, smallholders remain vulnerable to food insecurity and poverty in Ethiopia (Admassie & Abebaw, 2021).

The status of BT aptly implementing and regularly maintaining is very low, and in some sites, BT structure is knowingly destroying. For this, some empirical studies in Ethiopia (Gebreslassie, 2014; Gebretsadik et al., 2017; Kosmowski, 2015, Mesfin, 2016; Yaebiyo, 2018) focused on the technical evaluation of BT. Yet, farmers’ perceptions of BT and the urgent need to address BT constraints have not been closely studied and understood. This current study aims to investigate farmers’ perceptions of BT attributes and effectiveness in the East Hararghe and Hadiya zones of Ethiopia using the innovation systems framework. New innovation thinking recognizes the presence of multiple actors and sources of agricultural knowledge and innovation. This provides a broad framework for investigating agricultural technological change on the one hand and heterogeneous actors’ engagement in knowledge generation, exchange, and practice on the other (Spielman et al., 2016). Lundvall (2016) argued that innovation is an interactive learning process that needs more innovative systems for agricultural technological transformation. The major research questions are: How do farmers perceive BT based on certain innovation attributes? How do they value its effectiveness in the local context? What are the major perceived constraints for farmers investing in BT?

2. Methodology

2.1. Study area

The East Hararghe zone is located in the Oromia National Regional State, 525 kilometers east of Addis Ababa. It lies between 8°–9° N and 41°–44° E. Haramaya is one of the 24 districts in this zone. This district lies between 9°5’ − 9°31“ N and 41°55” − 42°19’ E (Appendix Figure A1). The elevation ranges from 1400 to 2337 m.a.s.l. Its mean annual temperature is about 17.7°C. More than 43% of the total rain falls in the three months of June, July, and August (Appendix Figure A2). The total population is about 331,013 in 2020, with 168,582 males and 162,431 females. There were 47,288 household heads counted. Agriculture is the main economic activity. Sorghum, maize, wheat, barley, and teff are among the cereal crops grown here. Also, khat, coffee, pulses, and oil seeds like horse beans, field peas, lentils, ground nuts, and linseed are produced as cash crops (CSA, 2011).

Hadiya zone is located in the SNNPR State, 234 kilometers south of Addis Ababa. It lies between 7° − 8° N and 37°30’ − 38°10“ E. Soro district is one of the 20 districts in this zone. This district lies between 7° 13”-7° 41’ N and 37° 20’-37 ^o 40’E (Appendix Figure A1). The elevation ranges from 1733 to 2673 m.a.s.l. Its agroecological situation takes in woynadga (42.42%), dega (33.33%), and kola (24.24%). Its mean annual temperature is 16°C. More than 47% of the total rain falls in the 3 months of June, July, and August (Appendix Table A1). The total population is about 239,357 people. About 39,892 household heads were counted. Agriculture, considered mixed farming, serves as the main economic basis for the majority of the people in this district (CSA, 2011).

2.2. Sampling and sample size

This study is based on a one-point-in-time survey of 384 rural households in East Hararghe and Hadiya zones, Ethiopia. Multi-stage sampling techniques were applied for sample selection. In the first stage, East Hararghe and Hadiya zones have been purposefully selected based on the severity of land degradation. In these areas, some of the dominant indicators of land degradation shown by Tsegaye et al. (2014) and Adimassu (2015) are climate change, loss of biodiversity, groundwater levels having dropped, gullies and sand dunes having grown, disturbance of ecosystem services, and reduction of cropland and forestland. Accordingly, they were targeted by the SLM project. In the second stage, the two SLM project woredas, Haramaya and Soro, were selected at random from among the four woredas (Haramaya and Kerssa in the East Hararghe zone and Soro and Gibe in the Hadiya zone). In the third stage, eight SLM project kebeles (Table 1) from Haramaya and Soro woredas were randomly selected out of 10 and 13 SLM project-focused kebeles, respectively, using a lottery system.

Table 1. Distribution of sample districts, kebeles, and households

Woredas	Kebele (PA)	User household		Non-user household		Total sample
		Total HH	Sample	Total HH	Sample
Haramaya	Biftu-Gada	340	29	1,178	39	68
	Ifa-Oromia	163	14	1,421	47	61
	Haro-Adi	50	5	800	26	31
	Tuji-Gebissa	165	14	1,003	33	47
Soro	Bona-Dibro	346	30	337	11	41
	Arara	355	30	351	12	42
	Sundussa	448	38	384	13	51
	1^st Hankota	355	30	384	13	43
Total		2,222	190	5,858	194	384

Note: Kebele is known as the peasant association, which is the fifth-lowest administrative unit in Ethiopia.

Haramaya woreda has a total population of 331,013 people among 47,288 total households. Soro woreda has a total population of 239,357 people, with 39,892 households. The sum of the two woredas’ households is 87,180. As the population size is defined, the sample size is calculated using the Kothari (2004) formula. A total of 384 sample household heads were randomly selected for this study. In both strata, a probability proportional to size sampling technique was employed for each kebele in the year 2022 (Table 1).

2.3. Methods of data collection

The qualitative data type was collected in this study using both primary and secondary data sources. A semi-structured questionnaire, focus group discussions (FGDs), and key informant interviews (KIIs) were used to collect primary data. Secondary data were collected from journals and local reports. Survey data on the socio-economic and institutional aspects of the sample households across the sampled kebeles was collected. Eight enumerators, all from their respective eight sampled kebeles, and two supervisors, each from two districts, were recruited. They were given intensive training on data collection procedures, interviewing techniques, and the detailed contents of the questionnaire for one and a half days. The questionnaire was initially written in English and then translated to Afaan Oromo and Haddiyssa to allow enumerators to better realize the questionnaire and properly administer the interviews. The questionnaire was pre-tested with non-sample households to ensure that it was aptly prepared in terms of completeness.

In addition to the household survey, FGDs and KIIs were conducted to gain insights into the community profile and understand the major constraints they faced in the adoption and sustainable utilization of BT during the last 5 years. In general, 20 KIIs and 8 FGDs (one in each kebele) were established. Each FGD comprises 6–8 persons. All interviews were captured using notebooks and audio recorders.

The characteristics of FGDs (60 persons (40 males and 20 females) from different socioeconomic statuses were purposefully selected) in this study were wisely chosen based on their age, gender, experience, farm location, and wealth status. Here, the relationship between the number of women being 20 and the BT is that all women are users of the BT and divorced or widowed in their marital status. Thus, in each FGD, 2–3 women are included. Checklists were created that went in-depth on the characteristics of BT, its effectiveness, the obstacles farmers faced when investing in BT, and other relevant issues. These made it simpler to gather in-depth qualitative data and triangulate data from household surveys. KIIs were drawn from local elders, development agents (DAs) working in the sample kebeles, district natural resources management and SLM project focal persons, and regional agriculture and rural development bureaus. To reduce potential errors that could happen during data collection, the researcher closely monitored the interview process. Scientific studies and statistical information were also used as secondary data from the literature.

2.4. Methods of data analysis

Descriptive statistics include mean, standard deviation, percentage, and frequency distribution table were applied to define the socio-economic and institutional characteristics of the sample households. The mean of BT users and non-users were compared using inferential statistics (i.e. t-test), and (Mann-Whitney U-test), and show interdependency (i.e. Chi-square test) between BT adoption categories.

A five-point Likert scale (strongly disagree, disagree, neutral, agree, and strongly agree) was used to analyze farmers’ opinions on the constraints of investing in BT. Farmers who strongly agreed or simply agreed were considered to have perceived the constraints, while others did not. Cronbach’s alpha was used to test the reliability of the internal consistency of a measuring instrument (e.g. a questionnaire), as multiple Likert scale statements have been used to determine whether the scale is reliable or not. Hinton et al. (2014) stated that Cronbach’s alpha could be read across four different reliability types, such as excellent (0.90 and above), high (0.70–0.90), moderate (0.50–0.70), and low (0.50 and below). The higher Cronbach’s alpha values, the greater the consistency across measuring instruments. The Mann-Whitney U test was used in this study to evaluate the Likert response data. Since Likert scale responses are ordinal, not all differences between responses are quantitatively equal. Thus, it is often asserted that parametric tests that involve interval data, such as the t-test, are statistically unsound. To assess the differences in ordinal data, non-parametric tests like the Mann-Whitney U test are preferred (Gombolay & Shah, 2016).

In addition to descriptive analysis, the qualitative data from the FGD and KII were analyzed using content analysis. The quantitative approach has some drawbacks, which are addressed in many studies using qualitative content analysis. According to Gheyle and Jacobs (2017), content analysis techniques were used to examine farmers’ perceptions of innovation. The major themes in this study were BT’s attributes, effectiveness, and constraints to investing in BT perceived by respondents. The attributes were advantage, compatibility, complexity, trialability & observability of BT, using attributes from Rogers (2003) diffusion of innovations theory. The effectiveness of BT includes the restoration of degraded lands, and improving crop production based on real-life situations. The third was constraints on household livelihood assets such as social, human, financial, natural, and physical capital, based on McLeod (2010) livelihood assets.

3. Results and discussion

3.1. Household characteristics

3.1.1. Descriptive statistics of categorical variables

In this section, we presented the calculated results, interpretations, and discussions of the findings. Of the total 384 sample household heads, 80.5% were male and 19.5% were female. The majority of BT users—91.1%—were male-headed households. The Chi-square test shows that there is a statistically significant relationship between adoption decisions and the sex of the household heads at 10% level of significance.

The survey results indicate that the majority (59.4%) of household heads participated in different social groups. From this, 91.6 % of households used BT. Social group fosters social networking. FGDs revealed that using it, farmers can reciprocate labor, finance, and spread local knowledge related to the use of BT.

Credit is vital for farmers to make timely purchases of agricultural inputs and invest in BT. It was found that only 55.5% of the respondents have reported receiving credit at least once since the past five years from formal cooperatives and microfinance. FGDs showed that farmers did not take credit from formal institutions because of the household’s remittances abroad, their lack of access to adequate money, and their inability to pay back the credit in the agreed period of time, which may lead to a shocking penalty.

The severity of erosion is determined by the slope of the farmlands. The technical classification of slope in terms of its percentage is flat land (0–2%), gentle land (3–5%), moderate land (6–10%), hill (16–30%), steep (31–60%), and very steep (>60%) slopes (WOCAT World Overview of Conservation Approaches and Technologies, 2019). Sampled household heads were asked to classify their farmlands into diverse slope classes based on their indigenous knowledge. From the sampled households, 24.7%, 22.6%, 39.7%, and 12.8% have heads classified into gentle, moderate, hilly, and steep slope lands, respectively. Farmers who have lands (hills and steep terrain) more vulnerable to soil erosion.

Because of soil erosion, many farmers report that their land productivity is decreasing with each passing year. The majority—58.9%—of respondents perceived land degradation as a serious problem, at least on one of their farmlands. From this, 88.4 % of households used BT at least in one of their farms. The severity of soil erosion inspires farmers to invest more in BT. The Chi-square test result showed that social group, credit, slope, and perceptions have statistically significant relationships between adoption decisions at 1% level of significance.

The major sources of regular SLM information in the study area, as said by DAs and SLM project actors, were communications with neighbors and relatives, model farmers and community leaders, and DAs. Mobile phones and radio were also substantial informational tools. It was found that 53.4% of households had access to regular SLM information (Table 2). The Chi-square test revealed that there is statistically significant relationship between adoption decisions and regular SLM information access at 5% level.

Table 2. Descriptive statistics for dummy variables differentiating users from non-users

Variables	Category	Users (n=190)		Non-users (n=194)		Adoption decision
		N	%	N	%	X^2-test
Gender	Female	17	8.9	58	29.9	26.8*
	Male	173	91.1	136	70.1
Social group	No	16	8.4	140	72.2	161.6***
	Yes	174	91.6	54	27.8
Infon. Access	No	24	12.6	155	79.8	174.5**
	Yes	166	87.4	39	20.2
Received credit	No	15	7.9	156	80.4	204.3***
	Yes	175	92.1	38	19.6
Land tenure	No	76	40	91	46.9	25.2
	Yes	114	60	103	53.1
Slope of farm	Gentle	4	2.1	91	46.9	38.34***
	Moderate	23	12.1	64	32.9
	Hilly	114	60.0	39	20.1
	Steep	49	25.8	0	0.0
Perception of LD.	Disagree	22	11.6	136	70.1	135.76 ***
	Agree	168	88.4	58	29.9

Source: Results from sample survey data (2022).

Note: *, ** and ***show significance at 10%, 5% and 1% levels, respectively.

3.1.2. Descriptive statistics of continuous variables

The survey results indicate that the education level of the sample respondents is low: their average level of formal education was 3 ± 2 years. The average education level was 4.4 ± 2 years for BT users and 2.08 ± 1 year for non-BT users (Table 3). There is a significant mean difference between the two groups in education level at 1% level of significance.

Table 3. Descriptive statistics for continuous variables differentiating users from non-users

Variables	Users (n=190)		Non-users(n=194)		Total		Adoption decision
	Mean ± STD		Mean ± STD		Mean ± STD		t-test
Education level (years)	4.38	2.38	2.08	1.24	3.23	2.87	22.06***
Active family (no.)	3.98	1.44	3.86	1.31	3.92	1.40	5.37
Contact with DAs (month)	1.08	0.87	0.46	0.50	0.77	0.77	8.45***
Farm size (in ha/timad)	1.17	0.79	0.22	0.42	0.73	0.79	14.62**
Distance of farm (minute)	9.45	2.49	11.02	4.86	9.77	5.07	−16.60***

Source: Results from sample survey data (2022).

Note: ** and ***show significance at 5% and 1% levels, respectively.

The households were also requested to respond about contact with the DAs. In the past two years, DAs have contacted 55.8% of respondents once or twice each month. The average monthly contact was found to be 1.08 ± 0.8 days for BT users and 0.46 ± 0.5 days for non-users. The mean monthly DA contact difference between the two groups was found to be statistically significant. The average farm distance from home in walking was ten minutes. The distance was 9.45 ± 2 min for users and 11.02 ± 4 min for non-users. There is a significant mean difference between the two groups in farm distance at 1% level.

Survey results indicate that agriculture is the mainstay of the study area, and all the sample households have access to land. The average land holding size in the study kebeles was about 0.73 ha, ranging from 0.125 to 3.00 ha. The majority (83.1%) of households possessed ≤1 ha of land. Only 16.9% own ≤3 ha of land. The average land size for BT users was 1.17 ha & 0.22 ha for non-users. There is a significant mean difference between the two groups at 5% level. FGDs revealed that respondents do not have sufficient land to feed their families. The major reason quoted for the declining trend in land holding size over the years is the sharing of land with new family formations. These all imply that BT users are more educated, have frequent extension contact, have farm fields closer to home, and have larger land than non-users.

3.2. Perceived attributes of BT by farmers

In this part, we focused on assessing farmers’ perceptions of the bench terracing innovation attributes. It is unclear in the study area whether reputable BT is cost-effective at the household level. Regarding comparative advantage; it was found that 57.3% of sampled households perceived that BT was more cost-effective; this occurred when they correctly implemented and regularly maintained a bench structure. This result is consistent with Mesfin (2016) finding, which showed that bench terraces in Tigray were more cost-effective. In contrast, Table 4 reveals that 82% of households perceived that BT required a lot of time and effort for construction, and that it took a while to see results (there were no immediate rewards), as perceived by 70% of them. The latter implies that farmers’ short-term gains are not guaranteed by BT. In line with this finding, Schmidt and Tadesse (2014) reported that Ethiopian farmers must maintain an SLM structure for at least 7 years before receiving a higher value of total crop production on their own farmland. Also, Ervin and Ervin (1982) noted that while some innovations may stop the degrading of the land, they do not immediately address the needs of farmers. However, Kant (2018) argued that a new technology improves crop yield or income, saves time, labor, and money, or is less risky than a previous one.

Table 4. Farmers’ perception of the attributes of bench terracing

Attributes	Descriptions	Frequency	Percent	% of Cases
Comparative Advantage (CA)	Cost-effectiveness	220	57.3	CA=57.3
	Time and effort intense	313	81.5	CA=75.8
	Long-term rewards	269	70.1
Compatibility (CO)	Fit current needs	205	53.4	CO=60.7
	Fit experiences	250	65.1
	Fit lifestyle (culture)	244	63.5
Complexity vs. Simplicity (CS)	Hard to design	307	79.9	C=80.9
	Frustrating	315	81.8
	Easy to learn	228	59.4	S=59.4
Trialability (TR)	Relaxed to scale	263	68.5	TR=63.6
	Conceivable to test	248	64.6
	Apt to the environment	221	57.6
Observability (OB)	Shareable with others	220	57.3	OB=59.8
	Opening to see results	238	62.0
	Describable outcomes	231	60.2

Source: Results from sample survey data (2022).

Note: ** and ***show significance at 5% and 1% levels, respectively.

According to Rogers (2003), compatibility is the degree to which an innovation is trustworthy with the prevailing values, past experiences, and needs of the adopters. As shown in Table 4, households perceived that the BT did not conflict with their needs, past experiences, or way of life (culture). This suggests that BT practices and farmers’ existing culture is closely correlated and well-suited to the study area. In other words, farmers’ perceptions and, consequently, adoption behavior may be influenced when the new ideas and new practices (innovation) surrounding farmers do not fit with local norms and farming systems.

In relation to this, one of the key informants, (a 57-year-old man), said the following:

Whatever profit they are going to get from technology, if it conflicts with their existing cultures, beliefs, and cultivation techniques, they, as a community, will immediately reject it. He was worried about how SLM’s intervention in their village in 2014/15 would affect their existing cultures, beliefs, and farming practices. Nonetheless, with strong networks with the project focal persons and DAs, who are aware of what is acceptable in the community, he also considered their kind support, such as the delivery of training, fertilizer and seed varieties, and BT technology, and even brought all the assistance that they trusted with their consultations. He thus understood and accepted it right away because it did not conflict with their cultures, religious dogmas, or farming practices.

The above is an example of how the adaptability of technology to a local socio-cultural context is very important for the sustainable utilization of technology. This result is consistent with other studies (Mwangi & Kariuki, 2015; Riemer, 2018), which also found that a lack of adaptability of technology to local context and belief systems is an incompatibility that constrains a farmer’s needs, farming experiences, and cultural circumstances. As such, contextualizing and fitting technologies to the farmer’s context are very important to enhance the farmer’s optimistic view and use of bench terracing.

BT’s attributes of complexity vs. simplicity were assessed in terms of design difficulty, implementation difficulty, and learning curve complexity in relation to household capacity. As stated by the survey results, 79% & 81% of households (Table 4) thought that BT was hard to design and frustrating to implement within their own capacities, respectively. Implying that complex technologies are less likely to be implemented by farmers, thus they need expert support. Although there were 3 to 5 DAs at each kebele, the BT practices that entail engineering knowledge were poorly implemented by farmers in the study area. The result is in line with the findings of Gedefaw et al. (2018) and Yaebiyo (2018), who stated that bench terracing was complex to design, implement, and plan without the assistance of intermediaries. However, 59% of households perceived that this technology was simple to learn if they were consistently exposed to awareness creation and training-provision programs. As part of the government extension program, the government and NGOs delivered training jointly in the study area. But, focus group discussions showed that few farmers have actually participated in farmer training programs, which limits their effectiveness. e.g. female household heads experienced more difficulty accessing expert support and training as compared to their male counterparts. About this, 48-year-old woman key informant recalls the following:

Being a woman limits access to extension services and training in relation to BT, as she said. She lost her husband eight years ago. DAs once visited her home and discussed ways to increase land productivity and livelihoods. After her husband passed away, those DAs are not coming to her home, not even the woman agent. Thus, training is not available to her, and she constructed BT by paying money to daily laborers.

The above case shows that access to training and agricultural extension services for adopting BT is gendered. Women (single mothers) and households with training limits experience trouble adopting BT. This finding is consistent with other empirical studies, such as Omer et al. (2021), who found that there were very few female-trained farmers (7 out of 60). They agreed that female participation in training was lower than that of male participants. Similarly, Seyoum (2014) stated that due to their workload at home and the cultural settings of their communities (e.g. male agents are afraid of contacting female-headed households who are divorced or widowed to access services), female farmers were unable to participate in training and had no access to extension services.

Farmers were questioned about whether BT was simple to scale from small to large areas, testable, and appropriate for the specific environment. The household heads believed that BT was scalable, testable, and slope-specific, respectively, in 68.5%, 64.6%, and 57.6% of cases (Table 4). This implies that farmers are more likely to adopt technology that can be tested and scaled. Likewise, Doss (2003) stated that the ability of a farmer to test something out on a small scale before fully adopting it is a leading determinant of new technology adoption. Trial periods also allow adopters in the study area to test BT in a local setting, but they are not always promising, depending on the testability and scalability of the technology, how it is funded, and the availability of the support systems. In this regard, Mr. Jemal, a 29-year-old supervisor at the kebeles (PAs) of Biftu-Gada, Haro-Adi, and Ifa-Oromia, for instance, stated the following:

In the Haramaya district eight years ago, the SLM project taught farmers and other key stakeholders how to perform BT through training and demonstrations as he said. Farmers have therefore put this structure in place on their private and communal lands. Mainly, Biftu-gada kebele became a model kebele for all kebele, and he, as a supervisor in these areas, has received awards from district officials. Even the motorcycle that he is riding nowadays is the result of his participatory leadership approach.

The above example shows how small-scale training and demonstration events can increase the likelihood that farmers will adopt an innovation on a large scale. Bizoza (2011), Mwangi and Kariuki (2015), and Yaebiyo (2018) stated that BT can be tested on a small scale is more likely to be scaled by farmers.

Farmers who perceived the innovation as compatible with their environment were more likely to adopt it because they saw it as a positive path to overcome risks (Mwangi & Kariuki, 2015; Riemer, 2018). The survey results in this study showed that 57.6% of households perceived, BT innovation is well-structured for their specific environment, which includes hills and steep slopes (Table 4). This is due to the study areas’ location in Ethiopia’s eastern and southern highlands. They are, on average, 2100 meters above sea level. This means that the slope rise covers 10–30% and 30–60% (5°-9° & 17°-31°), of hilly and steep slopes, respectively, and is measured by a clinometer. This is in line with the findings of Kosmowski (2015), higher altitude (2090 m–2939 m) in Ethiopia is one of the environmental elements that contribute to terracing. Mesfin (2016) argues that, terracing is a practice that can be used in the northern Ethiopian region with land slopes greater than 10%. While, as Azene (2011) suggests for Rwanda, soil bunds should be recognized on soils with slope categories between 10% and 15%, bench terracing (16% and 40%), and land above 40% that is suitable for forestation. These all imply that farmers with sloping terrain are more likely to use BT than farmers with flat terrain. However, highly mountainous areas (slope ≥50%) are not apt for constructing BT as they expose infertile subsoils, which can lead to soil overload and landslides.

As indicated in Table 4, the observability of BT was assessed in terms of its ability to share knowledge and experience with other farmers, its ability to see its results, and its describable and tangible outcomes. About 57.3%, 62.0%, and 60.2% of households perceived BT as shareable with others, opening to see results, and having describable outcomes, respectively. Hence, in total, 59.8% of households approved BT as observable. This implies that the BT outcomes could be observable to other farmers using farmer-to-farmer extension. e.g. a 31-year-old man who was a key informant, Mr. Haileyesus, stated the following:

In the SLM project interventions in the Soro district in 2014/15, the Bona Dibro kebele received priority. Farmers in this kebele were trained, and a structure was then built along the Lintala micro-watershed that is adjacent to Bona and Arera kebele. Farmers in the Bona kebele acquired arable land and produced crops that can be seen both on-site and off-site as a result of the BT, while farmers from this kebele were brought to Arera as role models to scale-out their benefits. Thus, it makes sense that early adopter households could inform other farmers about the outcomes of their BT structure. It is owing to the scaling-up strategy used by the SLM project to implement SLM practices.

The above case shows how shareable, communicable, and visible innovations among rural households can increase the probability that farmers will adopt them. The result is congruence with other studies by Posthumus (2005), Wei et al. (2016), and Perkins (2018), who found that bench terracing outcomes onsite and offsite that can be observed by late adopters are more likely to be adopted by farmers. These all imply that the characteristics of bench terrace technology are the preconditions for adopting it.

3.3. Farmers’ perception of effectiveness of BT

In this section, we assessed how the effectiveness of BT and farmers’ opinions of BT influence farmers behavioral intentions to adopt it. Among BT user households, effectiveness of BT was assessed in terms of restoring degraded lands, and increasing crop yield.

Rehabilitating degraded land: The majority of user households approved that BT would help them restore degraded land. The Chi-square analysis disclosed that there is a significant association between farmers’ perceptions and BT’s capacity to restore degraded lands at 1% level of significance (Table 5). Hence, the increased effectiveness of BT in restoring degraded lands and farmers’ perceptions had an effect on their intentions to adopt BT. The finding of Mesfin et al. (2018) that BT is effective in restoring degraded lands is in line with this result. This is because NGOs are crucial to Ethiopia’s efforts to restore degraded land (Zeleke et al., 2016). DA. Addise, a 27-year-old man, key informant, calls his experience as follows:

Table 5. Farmers’ perception of the effectiveness of bench terraces

Variables	Category	Users (N=190)		Non-users (N=194)		X2 -test
		Number	%	Number	%
Rehabilitate degraded land	No	4	2.1	194	100	368.322***
	Yes	186	97.9	0	0.0
Increase crop production	No	97	51.1	194	100	125.305**
	Yes	93	48.9	0	0.0

Source: Results from sample survey data (2022).

Note: ** and ***show significance at 5% and 1% levels, respectively.

The degraded areas along the watershed are a major focus for various SLM project donors. Although it makes sense in theory, I advise protecting less degraded areas from further harm because doing so is less expensive than restoring degraded areas.

In addition, FGDs revealed that BT was effective in creating employment opportunities. BT could lead to the opening of jobs for women and young people in enterprises that produce coffee and beekeeping, and fruit plants like banana, avocado, and mango. Also, physical workers were paid 35 to 50 birr per day by SLM project. Similarly, Gebretsadik et al. (2017) stated that BT was effective in creating job chances.

Increment of crop production: In this analysis, less than half of the user sampled respondents perceived that BT was increasing crop yield (Table 5). BT, on the other hand, made crop production easier by turning unproductive hillslopes into useful land, making fertilizer application simpler, and increasing soil fertility at the lower end of the structure. The Chi-square analysis indicated that there is a significant but weak association between farmers’ perceptions and the effectiveness of BT in crop yield at the 5% level of significance. This is consistent with Posthumus (2005) finding that BT reduces the detrimental effects of slope on production without materially increasing agricultural production at the household level.

In relation to this, Mr. Lambamo, a 62-year-old man key informant, said the following:

SLM project support along the Lintala micro watershed in 2014/15 for training and restoring degraded areas was highly regarded. I initially embraced technology and got started with BT farming. At the time, my friends and relatives were afraid of terrace farming, because they thought it might cause landslides that would cause a lot of land to be lost. I did, however, continue terrace farming and regularly maintain it. As a result, the land finally started to pay me for my efforts. On my farmland, I have 30 rows of BT that are 120 meters long, with grass planted in each structure row, from which I have mainly benefited. The fertility of the soil is maintained, and I cut grass roots to sell to other farmers and feed my own three cows. I once sold grass root cuttings for 42,000 Birr, but I sell three times a year, so 42,000 X 3 = 126,000 Birr. In addition to the above income, my 2 ha of farmland were protected from soil erosion, which in turn increased wheat and fava bean production. I suggest that if farmers use BT to care for their land sustainably, it will pay off and offer them with a better income to improve their welfare.

The aforementioned example demonstrates how SLM interventions and the use of BT in combination with biological and agronomic land management practices can improve the efficacy of BT (Figure 1). According to Mesfin et al. (2018), building BTs and applying organic manure and compost are two highly advantageous ways to turn the most degraded and unproductive hillslopes into agricultural lands.

Figure 1. Bench-terrace practices in Soro and Haramaya districts.

3.4. Perceived constraints of investing in BT by farmers

As indicated in Table 6, statements on a five-point Likert scale were used to elicit farmers’ thoughts on the limitations of investing in BT and assess the reliability of the scale. The reliability coefficient in this study is 0.60, (Appendix Table A2) implying a moderate level of internal consistency in the reliability instrument. The Mann-Whitney U test was also used to test the hypothesis and see if there is a statistically significant difference between BT users and non-users’ perceptions of BT quality related to livelihood assets.

Table 6. Perceived constraints of bench terraces by farmers

Perception statements	5-Point Likert Scale (Cronbach’s Alpha =.599)					Mann-Whitney U Test	Decision
	SA	A	NO	D	SD
It is more social capital intensive	102(26.6)	173(45.1)	47(12.2)	50(13)	12(3)	−5.24***	The null hypothesis of equal distributions has been rejected.
It is more human capital intensive	170(44.3)	149(38.8)	36(9.4)	27(7.0)	2(0.5)	−7.63***
It is more financial capital intensive	164(42.7)	179(46.6)	41(10.7)	0.0%	0.0%	−6.86***
It is more in need of natural capital	128(33.3)	197(51.3)	50(13.0)	5 (1.3)	4(1.0)	−7.13***
It needs frequent physical repair	106(27.6)	193(50.3)	83(21.6)	2 (0.5)	0.0%	−3.32**

Source: Results from sample survey data (2022).

Notes: SA=Strongly Agree, A=Agree, NO= No Opinion, D=Disagree, SD=Strongly Disagree.

Social capital: BT is a labor-intensive structure because of its nature and hilly and rugged landscape, according to 71.7% of sample respondents. Farmers perceptions of BT quality be regarding social capital necessities; non-BT users have a larger mean rank (220.16) than users (164.25) (Appendix Table A3). A statistically significant difference was found (Z = −5.24, p < .001) that is less than 0.05 (Appendix Table A4). As a result, we have significant evidence to reject the null hypothesis. Therefore, the non-BT users take larger values, suggesting that they have higher constraining values. It is implied that non-BT users have complained about how labor-intensive BT is, which may prevent them from adopting this technology. The results are congruence with those of Samuel (2017) and Gedefaw et al. (2018), who found that BT in the Ethiopian highlands requires a lot of labor and that a lack of it is the main obstacle to farmers putting this erosion prevention measure into practice. As stated by Etsay et al. (2019), over 75% of respondents claimed that the labor-intensive nature of SWC prevented its adoption. According to Bizoza (2011) and Karaya et al. (2020), social capital in the form of reciprocal labor input required to build and maintain BT is critical to farmers. As a result of the study area’s high migration rates, labour force is a constraining factor to land management, so returns to labor, must be considered a significant solution. For instance, Ms. Mishame, a woman of 39-year-old key informant, stated the following:

You won’t have access to labor groups if you don’t live with a male household. My husband went to the Republic of South Africa before four years. Until now, I did not hear his voice or know where he was. Some people who were deported from Tanzania told me that ‘he was arrested in Tanzania’. Before my husband went there, we had 0.5 ha of arable land and 0.125 ha of degraded land used for animal rearing. The person who paid for his transportation there received a transfer of 0.5 ha of arable land. The SLM project made it possible to construct BT on my 0.125 ha of degraded land, and it is now the owner’s responsibility to maintain BT sustainably. It needs interhousehold labor reciprocities among households to solve the shortage of labor. One of the social associations (Debo) is vital to solving those problems. Four years ago, my husband was trustworthy with labor-sharing systems. But now my husband’s friends are not inviting me into the groups since they may undermine my physical ability as a woman, so I am not getting the desired crop yields from the structure. I recommend that the government bring my husband home and that the SLM project target incapable women farmers.

The aforementioned instance reveals how single mothers have exclusive access to mutual support for adopting bench terraces. For women (single mothers) and households that frequently experience labor shortages, constructing and implementing BT can be difficult. It is in line with the finding of Wordofa et al. (2020) that farmers who have access to labor implement SWC measures that need a large labor force.

Human capital: BT is knowledge-intensive (adult education and capacity-building training on the topic of sustainable land management (SLM) technologies), as said by 83.1% of respondents. Farmers perceptions of BT quality are related to the need for human capital; non-BT users rank higher mean (232.03) than users (152.14) (Appendix Table A3). A statistically significant difference was found (Z = −7.63, p < .001) (Appendix Table A4). Hence, non-BT users rather than users frequently complained about BT. Due to its increased need for human capital, which adversely influenced adopting this technology. This shows that expanding formal education and capacity-building training locally are essential for thinking about and implementing BT. The result is congruence with the findings of Bizoza (2011) and Yaebiyo (2018) that to promote BT, farmers need more training to determine its technical effectiveness and sustainability.

Financial capital: BT required more financial capital (construction and maintenance costs) than other SWC measures, according to 89.3% of households. Farmers perceptions of BT quality are influenced by their need for financial capital; non-BT users had a higher mean rank (227.33) than users (156.94) (Appendix Table A3). A statistically significant difference was found (Z = −6.86, p < .001) (Appendix Table A3). Thus, those who do not use BT place more limits on available money. It is implied that BT is costly because it depletes the limited resources that farmers have access to, which makes it less likely that they will adopt this technology. The result is in line with those of WOCAT (World Overview of Conservation Approaches and Technologies), 2019), Sileshi et al. (2019), Legesse et al. (2021), and Karaya et al. (2020), which reveal that BT is a more costly structure than other SWC measures like soil bunds and stone bunds.

Natural capital: Based on the 84.6% of sample households, it is evident that BT heavily depends on factors related to the environment, such as soil depth, slope in degree or percent, and highly prone landforms to erosion and flooding. The inherent properties of natural capital include slope, soil depth, texture, stoniness, and biodiversity. Land cover, temperature, soil quality, and sedimentation (manageable properties) (Orr et al., 2017). Farmers perceptions of BT quality in relation to requests for natural capital; non-BT users have a higher mean rank (228.83) than users (155.41) (Appendix Table A3). A statistically significant difference was found (Z = −7.13, p < .001) (Appendix Table A4). This indicates that complaints about natural capital are higher among non-BT users than among users. This directs that natural capital significantly influenced adoption of BT. This result is in line with Perkins (2018) finding that field experts must check the soil depth and advise farmers if it is suitable for BT. Thus, farmers should be aware of why SLM projects prioritize contingent natural capital.

Physical capital: About 78% of household heads concur that BT necessitates more frequent physical maintenance. Farmers’ perceptions of BT quality are related to the needs for physical maintenance; non-BT users have a higher mean rank (209.62) than users (175.62) (Appendix Table A3). A statistically significant difference was found (Z = −3.32, p < .005) (Appendix Table A4). Hence, the restrictions on BT’s frequent maintenance are higher for non-BT users. Implied that the adoption of BT was driven by the need for frequent repairs. The result is consistent with Mesfin (2016) in Ethiopia and Derrick et al. (2019) in Rwanda, who found that BT requests regular maintenance. Without proper repair, the structure will be harmed by a variety of natural and human-made forces, which could result in the collapse of the entire terrace (Wei et al., 2016). Belachew and Hailemicael (2019) contend that traditional SWC measures are superior to BT unless there is frequent maintenance.

BT often requires funds beyond a farmer’s reach. As a result, NGOs must contribute to building the social, human, and financial capital of farmers. However, owing to the FGDs, some farmers who have received project intervention have refused it, claiming that they prefer to use the degraded land for grass planting and animal rearing. Conversely, the majority of grouped farmers firmly appeal to the project to intervene on their degraded lands. Hence, it is vital to raise awareness among households and treat degraded areas. According to perceptions, sample households are essentially prepared to invest in BT if those restrictions are reduced. State farmers and public-private actors must invest in promoting and scaling BT innovation.

3.4.1. Limits in protecting bench terraced sites

Despite the BT’s optimistic and practical social benefits and ecological improvements in the study area, there were obstacles for farmers and the state to sustainably maintain and protect the installed structure.

Animals are rearing over BT structure, grasses and weeds are heavily invading on the structure, riser breaks are being ignored, and risers are being destroyed for cropping (Figure 2). They suggested removing weeds and vines from the BT sections, avoiding cattle from crossing the risers, and repairing any minor riser breaks as soon as possible.

Figure 2. Weaknesses in protection of bench-terraced sites.

This implies that the FGD members understand not only the extent of the problems they have associated with BT structure protection and sustainability but also their solutions. This case also shows how farmers’ perceptions of BT affect its effectiveness and long-lastingness. This is in line with the findings of Samuel (2017) and Gebretsadik et al. (2017) that livestock interference, destruction of bench risers, and heavy rainfall damaged the BT structures. The following statements shed light on the dilemma faced by farmers in our study area: e.g. a key informant, a 49-year-old man named Mr. Abera, stated the following:

Expecting benefits if the BT structure is not regularly protected is not a good idea. He cited a proverb to support his claim, which reads, ‘It is expecting cattle spinoffs where cattle do not exist’. I had 1 hectare of farmland acquired from my parents. My land had gradually lost its ability to produce crops due to overcultivation and erratic rainfall. My 0.5 ha had suffered severe degradation and was no longer suitable for growing crops. Those degraded lands were turned into productive lands thanks to the remarkable support of the SLM project and DAs. Soil erosion and the land’s slope are thus reduced. I produced 400 kg of wheat on the restored lands alone in the 2019 production year, 300 kg barely in 2020, and 400 kg of wheat on the BT structure in 2021. The main issue I have encountered is that some members of the farming community are secretly rearing their animals on my BT structure at specific times, such as in the early morning (2:30–3:30 AM local time) and in the late afternoon (10:30–11:30 PM local time), and when I visit markets and other locations. This action against the BT is shattering my structure and causing degradation. I believed that this was because Kebele lacked adequate law enforcement agencies to deal with farmers who broke the law. Truly, the watershed management committee is aware of them but cannot stop them from taking this action. I suggest that the state-farmer groups to support institutions to ensure its sustainability.

The above case reveals how a lack of enforcing institutions for maintaining installed technology in a local context affects its long-lastingness and leads to a landslide. This is in line with the studies by Derrick et al. (2019) and Mesfin (2016), who found that a lack of protection of BT could result in serious damage.

4. Conclusions and recommendations

This study assessed farmers’ perceptions of bench terracing attributes and effectiveness in Ethiopia. The attributes of BT determined farmers adoption decisions. It is clear that it is more cost-effective in situations where it is rightly implemented and maintained on a regular basis; compatible with farmers’ sociocultural contexts; trailable in small-scale areas; and its results onsite and offsite benefits can be observed. Farmers’ decisions to adopt or reject BT were also influenced by its profitability (It restored damaged lands, and increased crop yield). Farmers’ investment on BT was however limited by the low level of household livelihood assets, such as social, human, and financial capital holdings. It is extremely labor-intensive, expensive, complex, and requires a lot of training, making its design and implementation within their capabilities challenging. The awareness to maintain and protect installed structures among households has its limitations, such as when animals rear over risers, grasses and weeds grow unnoticed, small breaks occur unnoticed, risers are ruined for cropping, and crops are handled ineffectively at each project site.

We derive three broad recommendations from the study’s findings, including (1) that agricultural research institutions, local governments, and other development actors—including farmers—cooperate to modify the attribute of BT. This should be accomplished by emphasizing the renaissance of indigenous terraces and examining the background knowledge of households and how they were truthful with each other and could pass on practices over generations. Farmers’ active involvement in all attributes of technological development could achieve this. (2) BT should be integrated with indigenous land management practices (e.g. they are originally practiced by the local people for a long time, such as planting indigenous trees and root grasses on the outlets of the BT structure, manure application, and leaving crop residues) to maximize its profitability. This could be accomplished by building farm households’ social, human, or financial capital holdings and NGOs could be encouraged to target raising farmer awareness and using holistic approaches. (3) State-farmer groups should strengthen institutions through the optional implementation of two measures. First, farmers who do not sustainably maintain their installed bench terraces must be given the opportunity to negotiate using their customary laws. Second, these households should consult a local arbitrator (locally authorized elders called Jaarsa in Afaan Oromo, or Lommencho in Haddiyssa) before bringing a dispute before formal authorities or courts.

Data Available Statement

Analyzed datasets in this paper are available from the corresponding author on reasonable request.

Acknowledgements

The authors also acknowledge Haramaya and Soro district agriculture and natural resources offices, kebele development agents, enumerators, and SLM project personnel for their valuable information and support during the survey. We also want to direct our gratitude to the farmers for their unreserved cooperation in sharing their opinions.

Disclosure statement

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

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/27658511.2023.2293261.

Additional information

Funding

This work was supported by funding from Haramaya University, Oromia regional state, Ethiopia., Office of Vice President for Research Affairs.

Notes on contributors

Alemayehu Temesgen Gebremikael

Alemayehu Temesgen is a Ph.D. student in Rural Development and Agricultural Innovation at Haramaya University, Ethiopia. He obtained a BA degree from Dilla University and an MSc degree in Rural Development and Agricultural Extension from Haramaya University. He has teaching, administration, and socio-economic research experience. He is certified by Globalics and Africalics, entitled “Innovation heading for Socio-economic Development” and “Innovation and Capacity Building Systems”. His research interests include topics related to youth roles and climate change; technology communication and adoption; and agricultural innovation & environmental protection. These contribute to the development of national policies & programs that can effectively ensure environmental safety and food security

Appendix

Table A1. Max, min, and mean annual temperature and mean monthly rainfall (1912–2022) of Soro district.

Month	J	F	M	A	M	J	J	A	S	O	N	D
Mean Monthly RF. (mm)	6.4	35.4	39.4	70.9	125.1	103.8	130.9	116.6	133	29.4	20.5	1.3
Mean Max Annual Temp. (^0C)	22	21.9	21.3	21.9	20.6	20.3	18.3	18	19.6	20.4	21.3	21.5
Mean Min Annual Temp. (^0C)	11.8	11.7	11.4	11.8	11	10.5	10.3	10.2	11	11.6	11.6	12.5
Mean Annual Temp. (^0C)	16.9	16.8	16.4	16.9	15.8	15.4	14.3	14.1	15.3	16	16.5	17

Source: Extracted from National Meteorological Services Agency (NMSA), 2022.

Table A2. Reliability statistics.

Cronbach’s Alpha	N of Items
.599	5

Table A3. Mann-Whitney test: ranks.

Bench terracing innovation	Users and non-users	N	Mean Rank	Sum of Ranks
Social capital intensive	0	194	220.16	42712.00
	1	190	164.25	31208.00
Human capital intensive	0	194	232.03	45013.00
	1	190	152.14	28907.00
Financial capital intensive	0	194	227.33	44101.50
	1	190	156.94	29818.50
Natural capital intensive	0	194	228.83	44392.50
	1	190	155.41	29527.50
Physical capital intensive	0	194	209.62	40667.00
	1	190	175.02	33253.00
Total		384

Source: Results from sample survey data (2022).

Table A4. Mann-Whitney test: frequency of bench terraces user and non-user households.

Source: Results from sample survey data (2022).

Figure A1. Geographic map of the study area.

Figure A2. Max, min, and mean annual temp. And mean monthly rainfall (1983–2022) of Haramaya district.

Source: Extracted from National Meteorological Services Agency (NMSA), 2022.