Determinants of soil and water conservation practices in the West Hararghe zone of Eastern Ethiopia

Abstract

Soil erosion causes a loss of soil fertility, which reduces crop yield and leads to food insecurity. To curb the adversative effects of soil erosion, many efforts have been made at the global, regional, and national levels. However, it is not possible to generalize the factors for the adoption of soil and water conservation (SWC) at the global and regional levels. The current dynamism is of paramount importance to be investigated based on the real scenario of the study site. Accordingly, this study focuses on the objectives of identifying determinant factors to the adoption of SWC practices in the West Hararghe zone, Eastern Ethiopia. A multistage sampling procedure using structured questionnaires was used to select 250 respondents. The collected data were analyzed using a seemingly unrelated bivariate probit model (SUR BVPM). The study identified significant demographic, socioeconomic, institutional, physical, and other factors in the adoption of SWC, such as the use of a productive safety net program (PSNP) and the participation of local farmers in technology design and implementation (PinTDI). For example, using an additional one unit of PSNP services decreases the likelihood of adopting soil and stone bunds by 10.2% and 10.6%, respectively, in ceteris paribus. Our investigation empirically contributes to the prior study and policymakers as the extension system to the implementation of SWC lacks the ground-level context and reality that require appropriate policy to reverse the trend.

Keywords:

• adoption of soil and water conservation
• seemingly unrelated bivariate probit model
• soil and stone bund
• erosion

PUBLIC INTEREST STATEMENT

Soil and water conservation technologies can be defined as a reasonable use of land resources, the implementation of erosion control systems, and the practice of suitable cropping patterns to increase soil fertility and reduce land degradation, thereby athe meliorating livelihoods of the local communities. Until 1974 the use and practices of soil and water conservation did not started. Thus, in our country, the problem associated with soil and water conservation is becoming very serious, and policymakers and international donors are obligated to initiate different soil and water conservation measures starting in 1975 after the terrible drought and famine of 1974. Three main practices agronomic or biological measures, soil management strategies, and mechanical or physical methods are widely followed in our country. Thus, this study focuses on determinants of physical SWCP (soil bund and stone bund) by using a seemingly unrelated Bivariate Probit Model.

1. Introduction

Soil erosion is a common delinquent in all countries worldwide, which leads to land degradation and refrains of ecosystem services (Kong et al., 2018). Annually, each individual is estimated to lose 2.8–4.2 t of soil per year (Kopittke et al., 2019). Poor countries commonly face the challenges of recourse base deterioration and less food production capacity of agricultural land (Shiferaw & Holden, 1999). Specifically, environmental degradation and better adaptation to climate change shocks are key challenges to the production potential of Sub-Saharan African (SSA) countries (Shimelis et al., 2017). This is mostly due to the deterioration of natural resources, which is caused by both anthropogenic and natural factors. (Kumawat et al., 2020). Consequently, land productivity has worsened from time to time, which is primarily known as soil erosion, due to runoff, topographic variations, slope of the land, intensive cultivation, farming on steep slopes, and deforestation (Genene & Abiy, 2014; Kedir, 2020). As recent empirical evidence shows, Ethiopia loses an average of 12 tons/ha/year of total soil (Nega et al., 2022). Thus, soil erosion causes a loss of soil fertility, which results in a reduction of crop yield that causes food insecurity (Belay & Eyasu, 2017). This shows that insufficient production of food crops could be a cause for poverty and food insecurity in Ethiopia. As a result, SWC practices were opted to minimize the problem and were given as a crucial strategy in Ethiopia (Fontes, 2020).

Taking into account the soil erosion problem and the benefit of SWC on agricultural yield enhancement, SSA countries (Tengberg & Jones, 2000) and Ethiopia have been initiated and motivated to implement SWC practices for more than half of a century (Kedir, 2020). However, the existing SWC practices in many parts of developing countries, especially in some parts of SSA, are applied by indigenous knowledge and lack scientific methods. In different parts of Ethiopia, indigenous SWC was applied, e.g. in Southern Ethiopia (Konso area), Central Ethiopia (South Shewa), and Eastern Ethiopia (Harangue plateau) (Engdawork & Bork, 2014). The different approaches used to implement SWC in Ethiopia are indicated in Table 1. Since the early 1980s, through food-for-work incentives, new land conservation technologies have mainly improved in some degrading and food-deficit areas of the Ethiopian highlands. Based on the production problem encountered by serious erosion, the Sustainable Land Management Program (SLMP) was also initiated in 2008 to reduce land degradation, increasing land productivity, and improving farmer livelihoods in Ethiopia.

Table 1. Approaches to implement SWC from past to present time in Ethiopia

SWC implementation approach	Started year	End year
Food-for-Work (FFW)	1973	2002
Managing Environmental Resources to Enable Transition To more sustainable livelihoods (MERET)	2003	2015
Productive Safety Net Programs (PSNP)	2005	Present
Community mobilization through free-labor days	1998	Present
National Sustainable Land Management Project (SLMP)	2008	2018
Development-oriented program	1980	Present

Source: Haregeweyn et al. (2015).

Consequently, many adoption studies have been conducted nationwide to identify determinant factors for the adoption of SWC (Belachew et al., 2020; Mekuriaw et al., 2018; Sileshi et al., 2019; Teshome et al., 2016; Wordofa et al., 2020). Even though there are numerous studies in Ethiopia concerning the adoption of SWC, the empirical conclusions on determinant factors for the adoption of SWC vary based on the agro-ecology, socioeconomic, cultural, and physical context of the area (Adimassu et al., 2016; Bekele & Drake, 2003). Thus, it is not possible to generalize the factors of SWC as global and local. These studies gave different implications based on the context and methodology they used. This implies that based on the agro-ecology, socioeconomic, cultural, and physical context of a nation, nationality, and country, there is a gap concerning the implementation and adoption of improved SWC which requires further investigation¹.

Consequently, following the implementation of SWC practices in developing countries (De Graaff et al., 2008), two main reasons for failure to accept SWC practices were (1) inability to recognize erosion problem and its effects and (2) unfamiliarity with the advantages of SWC practices. Mekuriaw et al. (2018) also indicated two main reasons for farmers being reluctant to adopt ISWC in most highland parts of Ethiopia. According to the authors, the failures to adoption are as follows: “(1) they don’t perceive a significant advantage to their use; rather, they see that SWC structures reduce crop yields by narrowing formerly limited cultivable lands, and (2) lack of short- term profitable benefit.”

On the other hand (Bewket, 2007), theoretically indicated as the low adoption of SWC is the wrong extension approach, i.e., top to down approach which lacks the community participation and considering real farmers context in design and implementation of improved SWC. The other main factor indicated in recent studies that discourages farmers to adopt newly improved SWC is the that the technology is not compatible with the farming systems of the local farmers (Gedefaw et al., 2018). Thus, we want to further investigate the approach of the extension system used as a factor and other determinants like the use of safety net programs (PSNP users) in addition to the aforementioned socioeconomic, institutional, demographic and physical factors. Why the researcher wants to include PSNP users as a factor is that in Gemechis district of West Hararghe Eastern Ethiopia, about 8,312 HHs and 29,180 individuals are users of PSNP in the form of direct users and public work users. Whether using PSNP services or not have both positive and negative relationship to adoption of SWC. When farmers use PSNP as food-for-work (public work) by participating in works like natural resource conservation mainly watershed management to get food items, it has a positive relation to the adoption of SWC. On the other hand, when farmers rely solely on PSNP services without putting any effort in farm activities, they become waiters (expose to dependency syndrome) and fail to participate in SWC².

Generally, there are insufficient studies on the adoption of improved SWC practices (Wordofa et al., 2020). Thus, it is paramount to focus on an improved SWC, which is implemented by using clear scientific methods like appropriate measurements and area of the specific structures applied by the support of agricultural experts. Various studies have shown that there are many shortcomings in adoption studies. In particular, in developing countries such as Ethiopia, dynamism problems such as socioeconomic, demographic, physical, and institutional changes are occurring from time to time (Eweg et al., 1998). Although various studies have been conducted on SWC in eastern Ethiopia, these existing studies have not identified and investigated the best environmental and soil conservation practices. That is, failing to distinguish and separately examine indigenous soil and water conservation practices and improved SWC in Ethiopia and Africa. Therefore, an independent investigation of both traditional and introduced approaches will be needed to provide appropriate recommendations for policymakers and development actors.

This will contribute to three major issues to the following research questions:

1. What makes soil and water conservation practices currently unusable by farmers?

2. Does the updated information of improved SWC practices empirically recorded?

3. Does a contextual and timely focus adoption study of SWC is available to policymakers?

Therefore, our study aimed to address the research questions by focusing on quantitative and qualitative research approach because most of the recent articles published on the adoption study of SWC focus more solely on the quantitative research approach, which lack qualitative phenomenon or rationalization. Thus, it is important to apply both quantitative and qualitative research approaches to provide full and timely required information for future research and policymakers. So, the current study initiated to address the objectives of identifying determinant factors to adoption of SWC in Ethiopia.

2. Methodology

2.1. Description of the study area

This investigation was undertaken in Gemechis district of West Hararghe Zone Oromia regional state in eastern part of Ethiopia (Figure 1). Gemechis district is one of the potential areas for SWC implementation in the West Hararghe Zone.

Figure 1. Geographical map of study area

Source: Taken from Babu et al. (2023)

2.1.1. Location

The district is located 343 km East of Addis Ababa at latitude 9°8’0’’ (Babu et al., 2023). It is bounded by the Chiro district in the West and North, the Oda Bultum district in the South, and the Mesala district in the East.

2.1.2. Demography

The total population of the district is 240,442, of which 117,817 are males and 122,625 are females. The district also has 39,491 heads of households who are agricultural households. From this, 34,546 are male head households and the rest 4,945 are female head households.

2.1.3. Climate and agro-ecology

Gemechis is located within 1,300–3,400 m above sea level (m.a.s.l). The minimum and maximum annual rainfall is 800 mm and 1200 mm with an average of 850 mm. The Gemechis district receives adequate rain twice a year, once from March to May and the other from June to September, which is a bi-modal rainfall in nature. The minimum and maximum temperatures are 15°C and 30°C with the average temperature of 22°C. The district was categorized into three agro-ecologies. These are highland, midland, and lowland agro-ecologies that cover 15%, 45%, and 40% of the district, respectively. Sandy loam type of soil is dominantly found in the district (Geddafa et al., 2021).

2.2. Sampling design and sampling techniques

Multistage sampling techniques were used to select the samples from the study. At the first stage Gemechis district was purposively selected for the study on the basis of the intensity of soil erosion problem and the potential of soil practices. In the second stage, from 35 kebeles of the district, the kebeles predominantly practiced the technology purposively selected for the study. Accordingly, four kebeles were purposively selected because they are well known for their SWC structures. At the third stage, from the selected kebeles, the respondents were classified into adopter and non-adopter of the technology through stratified sampling technique. In the last stage, adopters (106) and non-adopters (144) were selected by using simple random sampling methods from the stratified respondents.

Therefore, for this study, 250 households were selected randomly as proportion to population size from both user and non-user among the four kebeles (Table 2). Thus, as Kothari (2004) indicated, this proportion to population size can be calculated as follows;

(1) $n=\frac{Z^{2}NPQ}{\left(N-1\right)e^2+Z^{2}PQ}=\frac{{\left(1.96\right)}^2\ast 39.491\ast 0.7\ast 0.3}{39,490\ast {\left(0.056\right)}^2+{\left(1.96\right)}^2\ast 0.7\ast 0.3}=249.1\approx 250$(1)

Table 2. Sample size determination from sampling frame

Kebeles	Number of HH Head	% of sample HH	Number of selected HH
Lega Lafto Medaria	678	19.4	48
Walenso defo	755	21.6	54
Waltane	780	22.3	56
Walargi	1282	36.7	92
Total	3,495	100	250

Source: Secondary data from Gemechis district, 2020.

where n = sample size, N = population, Z = confidence interval, P = probability of success, Q = probability of failure, and e = error for sample size N. According to Cochran (1977), perfection may be made by giving the quantum of crimes that are willing to tolerate in the sample estimates. Therefore, for this study, ±5.6% level of precession (e) was used to calculate the sample size at 95% level of confidence.

2.3. Methods of data collection

Quantitative and qualitative data collection methods were used to collect the data for this study. For quantitative data collection, a household survey was conducted, and 250 respondents answered the prepared questions in a scheduled interview in January 2020. For qualitative data, a checklist for KII and FGD was prepared and organized. In general, nine enumerators familiar with the local language and culture were recruited and trained to collect the survey data.

2.4. Methods of data analysis

The qualitative and quantitative data collected were analyzed using qualitative and quantitative analysis methods for data triangulation to ensure the reliability and validity of our data. The qualitative data collected were also analyzed qualitatively by using thematic analyzing methods. Information from the FGD and KII was collected through both audio recordings and field notes. The audio recordings were transcribed and written down completely in English without adding or reducing the ideas from FGD and KII. The textual information from the audio recordings and field notes was then cleaned by removing redundant ideas and unwanted information for the study (Babu et al., 2023). The ideas that had the same message or ideas were grouped under one umbrella and related to other information from the FGD and KII participants. After all these steps, the analyzed data were interpreted and reported for the study.

On the other hand, for quantitative method, we used descriptive statistics and econometric model (SUR BVP). Descriptive statistics such as mean and standard deviation were used to present the summary of quantitative variables through tabulation. For qualitative dummy variables, frequency and percentage were used to display the summary via tabulation and pie chart. Inferential statistics such as t-test and Chi-square test were also used to test the availability of significant difference between adopter and non-adopter of SWC in targeted area. Therefore, t-test was used to test the significant difference between continuous explanatory variables like, age, education level, family size, cultivated land size, farm experience, distance of farm from house, livestock holding, and frequency of extension contact. Chi-square test were also used to test the significant difference between dummy variables like, sex, secured land ownership right, credit services, PinTDI, perception, PSNP user, and slope of land.

2.4.1. Model specification

There are different models that are used to analyze the binary variables like logit, probit, and MNL or MNP. However, these models fail to capture interdependence among dependent variables (Belachew et al., 2020; Sileshi et al., 2019). Binary logit or probit models are used only for a single dependent variable with two possible outcomes (Wooldridge, 2002). On the other hand, MNLs are used for more than two binary dependent variables, and the outcomes are independent and mutually exclusive (Sileshi et al., 2019). From the econometric model, BVP was employed to analyze the factors affecting the adoption of improved SWC structures (soil bund and stone bund). BVP models have been used in numerous studies, including Nkamleu and Manyong (2005), Yesuf and Köhlin (2009), and Amare et al. (2012). BVP was selected due to the fact that dependent variables are dummy (adoption of soil bund and stone bund) and also the model can capture the issues of interdependency or mutual inclusiveness of the dependent variables (Belachew et al., 2020; Sileshi et al., 2019). However, in such conditions, logit and binary probit are unable to estimate where correlation exists (Sileshi et al., 2023). Normally, BVP model is an estimation of the seemingly unrelated (SURE) equation model (Greene, 2009). Where it is supposed that all covariates are exogenous and estimated with maximum likelihood (ML), BVP specification is seemingly unrelated (Jara-Rojas et al., 2013). Therefore, the current study used BVP specifically via SUR model. Thus, it is estimated as follows:

$\left\{\begin{array}{c}S{B}^{\ast }={X_1}^{\prime }{\beta }_1+u_{1}i\\ ST{B}^{\ast }={X_2}^{\prime }{\beta }_2+u_{2}i\end{array}\right.SBandSTB=1ifSBandSTB>0,0otherwise$

(2)

3 $\left(\begin{array}{c}{U}_{1}i\\ U_{2}i\end{array}\right)\text{~}N\left(\begin{array}{c}0\\ 0\end{array}\right)\left[\left(\begin{array}{c}1p^{12}\\ p^{21}1\end{array}\right)\right]$3

where SB* (Soil bund) and STB* (Stone bund) show the unobserved dependent variables, SB and STB show the observed variables 1 or 0 responses x_i is a vector of explanatory variables, β is a vector of ML parameters to be estimated, and (u₁i, u₂i) is a vector of error terms described by N, which is BVP standard normal distribution with a correlation of diagonal elements.

It is also possible to calculate the joint odds of adopting two or more practices. For example, the likelihood of a farmer combining two strategies is given by the following mathematical formula (Gong et al., 2021):

(4) $Jpsst={\sum }_{m\epsilon \left\{\delta m:Ys=1,Yst=1\right\}}\pi m$(4)

where Jpsst is the joint probability of adopting soil bund and stone bund, mε is the marginal effect, Ys is the probability of adopting the soil bund, and Yst is the probability of adopting the stone bund. This can be applied in a case where a researcher wants to estimate the complementarity between practices (Gong et al., 2021).

2.4.2. Marginal effects

Next to estimation of the parameters, it is important to calculate the marginal effects of the covariates in the conditional distribution because marginal effects can determine the magnitudinal change of conditional probability of the outcome variable by changing the value of explanatory variable, holding all the repressors constant at some value (Wambui et al., 2015). The marginal effect for the bivariate profit model is then given by Seyoum (2018). Thus, by following a procedure similar to that followed by Gong et al. (2021), the marginal effect of BVP model is derived as follows:

5 $\frac{\mathrm{\partial }Psst}{\mathrm{\partial }Cov}={\sum }_{m\epsilon \left\{\delta m:Ys=1,Yst=1\right\}}\frac{\mathrm{\partial }Psst}{\mathrm{\partial }Cov}$5

3. Results and discussions

3.1. Soil and water practices in Gemechis District

Soil and water practices currently practiced by Gemechis district farmers were investigated in detail (Table 3). The study descriptively indicates that traditional, improved, and non-users of any SWC farmers are available in study area. Thus, about 56%, 33.2%, 2.4%, and 8.4% are traditional SWC users only, both improved and traditional SWC users, improved SWC users non-user of any SWC respectively in Gemechis District of West Hararghe Zone (Figure 2).

short-legendFigure 2.

Current status of soil and water conservation practices in Gemechis district.

Table 3. Statistical summary of SWC status in the Gemechis district

Status of SWC implementation of sample respondents	Frequency	%
Not used any SWCs	21	8.4
Used traditional SWC only	140	56
	161 (Non-adopter)	64.4
Improved and traditional practices
Soil bund + traditional practices	36	14.4
Stone bund + traditional practices	15	6
soil and stone bund+ traditional practices	32	12.8
Soil bund only	3	1.2
Stone bund only	2	0.8
Soil and stone bund only	1	0.4
	89 (Adopter)	35.6
Total	250	100.0
User of Traditional SWC practices
Ditches/boy	69	27.6
Contour plowing	39	15.6
Cathara	56	22.4
Traditional terracing	34	13.6
Ditches and contour plowing	25	10.0

Source: Own computation, 2021.

As we found in our study, 14.4%, 12.8%, and 6% of respondents use simultaneously soil bund with traditional practices, both soil bund and stone bund with traditional practices, and stone bund with traditional practices, respectively (Table 3). Therefore, we simultaneously took the users of both traditional and improved practices as adopters 35.6% (89 HHs) and users of traditional SWC solely and non-user of any SWC as non-adopters 64.4% (161 HHs). In the study area, traditional practices such as ditches/boy, contour plowing, cathara, and traditional terracing were commonly used by local farmers.

As we found in our study qualitatively from the FGD group, those who didn’t practice any SWC were farmers because those their lands are flat and insensitive to any erosion. However, as we further searched for information from KII (one of kebele administrator) towards non-users of any SWC, they were pushed by district and kebele structures as they implemented it via a campaign on watershed management, although they were not doing it individually on their own plot.

3.2. Results of descriptive analysis

Continuous and dummy variables considered for this study are summarized in Tables 4 and 5, respectively. The study included approximately 40% of female and 60% of male HHs from both adopter and non-adopter HHs. From adopters of HHs, about 51% are female HH and 49% are male HH, and there is a solid significant difference at 1% significant level on adoption of SWC (Table 5). Therefore, from the descriptive analysis, adopter HHs are sufficiently mature (39.55 years) in age compared to non-adopter HHs (37.17 years) (Table 4).

Table 4. Statistical description of the continuous explanatory variables

Dependent Variable	Definition
Adoption SWC dummy (1 if adopted, 0 otherwise)	Adoption of at least one of SWC
Explanatory variables	Non-adopter		Adopter
Continuous variables	Mean	SD	Mean	SD	t-value	Sig.
Age of the HH (in year)	37.17	10.86	39.55	7.88	1.81*	0.07
Education level of HH (number of years in formal education)	3.81	1.72	4.42	1.88	2.6***	0.009
Family size (measured in AER)	3.42	1.1	3.83	0.89	3.01***	0.003
Cultivated land of the HHs (measured in ha)	0.52	0.32	0.72	0.31	4.83***	0.0001
Farm experience of HHs (measured in year)	21.64	0.89	28.73	12.04	4.73***	0.001
Distance of farm from house of HHs (KM)	0.86	1.46	0.78	1.08	0.455	0.324
Livestock holding in TLU	1.66	1.23	2.04	1.05	2.42**	0.016
Frequency of extension contact of HH (measured in number in a year)	3.86	2.64	4.17	2.27	0.93	0.352

***, **, and * symbols represent statistical significance level at 1%, 5%, and 10%, respectively.

Source: Own computation, 2021.

Table 5. Statistical description of the dummy explanatory variables

Variables	Dummy	Non-adopter (%)	Adopter (%)	Total	Chi-square test	p-Value
Sex of HH (1 =male or 0 = female)	Female	33.54	50.56	39.6	6.94**	0.008
	Male	66.46	49.44	60.4
HH have secured land ownership right (1 = yes, 0 = no)	No	50.93	47.19	49.6	0.32	0.571
	Yes	49.07	52.81	50.4
HH use credit services(1 = yes, 0= no)	No	60.25	57.3	59.2	0.20	0.650
	Yes	39.75	42.7	40.8
PinTDI	No	40	29.2	69.2	10.66***	0.001
	Yes	24.4	35.6	30.8
Perception of HH on soil erosion(1 = yes, if perceive as erosion is a problem now, 0 = otherwise)	No	54.66	31.46	46.4	12.4***	0.001
	Yes	45.34	68.54	53.6
PSNP user	No	19.6	17.6	37.2	8.86**	0.003
	Yes	44.8	18	62.8
Slope of land of head of HH (1= gentle slope, 2 = steep slope, 3 = flat slope, 4 = very steep slope, 5 = gentle and flat slope, 6 = flat and steep slope, and 7 = gentle and steep slope)|	Gentle	25.47	29.21	26.8	8.25	0.220
	Steep	25.47	21.35	24
	Flat	13.04	3.37	9.6
	Very steep	9.32	13.48	10.8
	Gentle and flat	13.04	16.85	14.4
	Flat and steep	9.94	10.11	10.0
	Gentle and steep	3.73	5.62	4.4

*** and ** symbols represent statistical significance level at 1% and 5%, respectively.

Source: Own computation (2021).

Those adopter HHs were also relatively more educated (4.42) than non-adopter (3.81) HHs (See Table 3). In terms of availability of family size, adopter HHs have a larger family size (3. 83 AER) than non-adopter (3.42 AER) HHs. Even the difference is statistically significant at 1% significance level.

The descriptive result shown in Table 4 indicates that adopter HHs have more cultivable land and farm experience than non-adopter HHs, which is significantly different at 1% significant level. It is also shown that adopter HHs have more livestock in TLU significantly than (5%) non-adopter HHs. The perception of farmers on the problem of soil erosion is the main factor to adoption of SWC (Shiferaw & Holden, 1998). In addition, we found a descriptively significant difference between adopter and non-adopter concerning perception to erosion, PinTDI (farmers participation in technology design and implementation), and PSNP user (using productive safety net program)(Table 5).

3.3. Results of econometric model (SUR BVP model) analysis

In this section the researcher presented the determinant factors to adoption of soil and stone bund in Gemechis district of West Hararghe zone. Table 6 summarizes the results of the seemingly unrelated bivariate model (SUR BVP). The model rationally fits the data with wald chi ²(66.52) test of statistics. The results from the SUR bivariate probit outputs indicated nine variables significantly affect the adoption of soil bund and stone bund from 13 variables included in the model. The researcher also computed the probability of adopting both practices jointly and failing to adopt both practices in study area (Table 6). Thus, the probability of adopting both soil and stone bunds is approximately 13.2%, and the probability of failing to adopt both improved practices simultaneously is 64.8%. When we separately see the probability of adopting soil bund and stone bund, soil bund (28.3%) is more adopted than stone bund (20%) in the study area.

Determinant factors to adoption of SWC practices.

Table 6. Results of SUR BVP in participating improved SWC structures

Factors	Adopter of soil bund		Adopter of stone bund		Marginal effect
	Coef.(SE)	Z(P>|z|)	Coef.(SE)	Z(P>|z|)	Soil bund	Stone bund
_cons	−1.772 (0.7394)	−2.40 (0.017)**	−2.585 (0.6959)	−3.72 (0.00)***
Age of HH	0.0103 (0.0115)	0.90 (0.371)	−0.0089 (0.013)	−0.71 (0.480)	0.003	−0.01
Sex of HH	−0.4856 (0.21)	−2.35 (0.019)**	0.1344 (0.2104)	−0.64 (0.523)	−0.128	−0.03
Level of education	0.0417 (0.0544)	0.77 (0.443)	0.1353 (0.0556)	2.43 (0.015)**	0.011	0.032
Farm experience	0.0223 (0.0092)	2.42 (0.016)**	0.0305 (0.0096)	3.19 (0.001)***	0.006	0.007
Family size in AER	0.1624 (0.0954)	1.70 (0.089)*	0.0932 (0.1011)	0.92 (0.357)	0.043	0.022
Livestock Holding in TLU	0.1283 (0.0880)	1.46 (0.145)	0.0576 (0.0949)	0.61 (0.544)	0.034	0.014
Land Size in ha	−0.5402 (0.5140)	−1.05 (0.293)	0.5059 (0.4978)	−1.02 (0.309)	−0.143	−0.12
Tenure Security	0.2623 (0.1942)	1.35 (0.177)	−0.0302 (0.203)	0.15 (0.882)	0.069	0.007
Credit user	−0.0308 (0.0178)	−1.73 (0.083)*	0.0089 (0.0090)	0.99 (0.321)	−0.008	0.002
Extension contact	0.0675 (0.0342)	1.97 (0.049)**	−0.0106 (0.015)	−0.70 (0.481)	0.018	−0.01
PSNP user^a	−0.3864 (0.2113)	−1.83 (0.067)*	−0.4424 (0.222)	−1.99 (0.047)**	−0.102	−0.10
Perception to erosion	0.6076 (0.1981)	3.07 (0.002)***	0.3820 (0.2043)	1.87 (0.062)*	0.160	0.091
Slope of land	0.0691 (0.0543)	1.27 (00.203)	−0.0245 (0.056)	−0.43 (0.665)	0.018	−0.01
PinTDI^b	0.8024 (0.2285)	3.51 (0.000)***	0.4874 (0.2353)	2.07 (0.038)**	0.212	0.117
Log likelihood −213.57 /athrho 0.672 Rho 0.586* Wald Chi^2 (26) 66.52 Prob >Chi^2 0.001 N 250	Probabilities of adopting improved SWC				Mean
	Probability of adopting solely soil bund (biprob1)				0.283
	Probability of adopting solely stone bund (biprob2)				0.200
	Probability of rejecting simultaneously both soil and stone bund (biprob00)				0.648
	Probability of not accepting soil bund but the probability of accepting stone bund (biprob01				0.069
	Probability of accepting soil bund but, the probability of rejecting stone bund (biprob10)				0.152
	Probability of adopting simultaneously both soil and stone bund, respectively (biprob11)				0.132

LR test of rho = 0: Chi² (1) = 24.87 Prob>Chi² = 0.0001.

***, ** and * denote significance at 1%, 5%, and 10% probability levels, respectively, AER shows adult equivalent ratio, PSNP user^a shows user of productive safety net Program, and PinTDI^b shows participation in SWC technology design and implementation.

Source: Own computation, 2021.

The probability of preferring a soil bund but not stone bund is 15.2%, which is more than the reverse of this statement, i.e 6.9% (the probability of choosing stone bund but not soil bund) (Table 5). This is contextually indicated that soil bund is more preferred by farmers and adopted than other improved SWC in Gemechis district of West Hararghe zone. This might be because the implementation of soil bund is more technically easy than stone bund. Overall, the sex of HH, level of education, farm experience, family size, credit user, extension contact, PSNP user, perception to erosion, and PinTDI were found to be significantly affecting the adoption of soil and stone bund in Gemechis district of West Hararghe. From these factors, eight variables were found to be significantly affecting soil bund, whereas only five variables significantly affect the adoption of stone bund (Table 6). But, as Table 6 indicates, four variables commonly affect both soil and stone bund in Gemechis area. Thus, we want to present the results as factors to adoption of soil bund, stone bund, and jointly influencing factors as follows.

3.3.1. Perception

Perception of erosion jointly influences the adoption of both soil and stone bunds positively and significantly at 1% and 10% significant levels, respectively (Table 6). Erosion and its consequences are the core problem to production and productivity in smallholder farmers in SSA countries. However, taking and perceiving it as a serious problem is also another issue that has got a recent research priority. We found that the probability of adopting an improved soil bund appeared to be increased by increasing of the farmers’ perception on erosion. Therefore, as farmer’s perception changes from the feel of no erosion problem to observe and feel to the presence of great erosion problem, the probability of adopting soil bund and stone bund is increased by 16% and 9.1%, respectively, ceteris paribus (Table 6). The prevalence of farmers’ perception on soil erosion is more probable to motivate them for participating on adoption of SWCs (Beyene & Feyisa, 2020).

Poor perception of farmers towards SWCs discourages farmers from adopting SWCs in the desired amount (Delibo, 2017). This shows that when a farmer considers the soil erosion problem in his area as a serious problem, he can focus on soil protection and erosion. This is to protect their own farm land from the adverse effects of soil erosion (Wordofa et al., 2020).

This implies that awareness creation and farmer’s capacity building are of paramount importance in the problem of erosion and its consequences to change the mindset of local farmers. Since perception is an important factor for the adoption of SWC, it is logical to policymaker and development sectors as priority plan may possibly be given for attitudinal build and awareness creation via short-term training and workshop.

3.3.2. Farm experience

Regarding minimizing the adverse effect of soil erosion in the study area, farm experience is found to have a strong and positive association to adoption of soil (P < 0.05), and stone bund (P < 0.01). In line with our study (Abebe & Bekele, 2014; Kassa et al., 2013; Wordofa et al., 2020), farming experience of HHs positively and strongly related to the adoption of SWC. The study conducted recently in South Nigeria also confirmed that an increase in farming experience may possibly permit farm HHs to properly apply SWC to increase the productivity of rice in the area (Ojo et al., 2021). Experienced farmers can judge the cost and benefit of investing in SWC and have the skill to apply the required farm practices to conserve soil from erosion. Thus, experienced farmers are more likely to adopt SWC practices than their inexperienced counterparts. Therefore, it is paramount important to encourage farmers exposure visit, farmers field school and field days to conduct experience sharing program among farmers in enhancing adoption of SWC practices.

3.3.3. PinTDI

PinTDI is another factor that positively and highly influences adoption of soil bund and stone bund at 1% and 5% significance levels, respectively. Participation in technology design and development is an important factor in the adoption of soil and stone bunds in Gemechis district of West Hararghe zone, Eastern Ethiopia. As we found via our study by making other factors constant, the likelihood to adopt soil bund and stone bund is increased by a unit increment of 21.2% and 11.7%, respectively. This shows that farmers participate in technology design and implementation, as the farmers fully know the techniques and science of application of improved SWC by themselves. However, most of the SWC implemented in Ethiopia before a decade obviously through top to down approach lack farmers’ participation and the grassroots level reality. This results in low adoption of improved SWC, which require appropriate government policy to reverse the trend.

Farm experience also jointly influenced adoption of stone bund and soil bund positively significantly at 1% and 5% significance levels, respectively..

3.3.4. Extension contact

Extension contact positively and significantly affect the adoption of soil bund at 5% significant level. An additional one-time frequency of contacting extension personnel may increase the likelihood to adopt soil bund by 1.8%, ceteris paribus. This is because, when a farmer have contact with extension agents, they may acquire different advisory services towards the technology. Farmers who have close contact with extension agents can develop awareness and understanding of the soil erosion problem and become encouraged to adopt improved soil measures (Wordofa et al., 2020). This is because when farmers deal with DAs and extension agents, they can obtain advice and huge knowledge from them. Similarly, those who gain awareness through extension workers on SWC are more encouraged to use improved SWC practice on their farm lands than the other farmers who didn’t get opportunity to interact with extension personnel (Abebe & Bekele, 2014; Wordofa et al., 2020). Despite this study, some studies reported that extension contact negatively affects the adoption of improved SWC structures (Kebede et al., 2016). Thus, having contact with an extension agent has a probability of enhancing adoption of stone bund in the study area.

3.3.5. Family size

Family size, which is converted and measured in adult equivalent ratio, also positively and significantly affects the adoption of soil bund in Gemechis area. Family size (measured in AER) also affects the adoption of soil bund positively and significantly at 10% significant level. Because adoption of SWC is a labor intensive work (Teshome et al., 2016), adult equivalent ratio of HHs becomes an important factor to be considered in the adoption of SWC. But, there are inconsistent findings concerning the relationship between adoption and family size of HHs. For instance, some studies have reported a negative relationship between family size and adoption of SWC (Amsalu & Graaff, 2006; Bekele & Drake, 2003; Belachew et al., 2020; Kassa et al., 2013; Shiferaw & Holden, 1998; Teshome et al., 2016). In contrast, others have shown a positive relationship between adoption of SWC and family size (Abdulai & Huffman, 2014; Abebe & Bekele, 2014; Mekuriaw et al., 2018; Ojo et al., 2021). It has been suggested that large family size with more labor endowments are more likely to adopt this effort-needed SWC technology (Abdulai & Huffman, 2014). Thus, our study confirms the issue that adult equivalent ratio is an important factor in adopting the science and techniques of improved SWC structures. The application of SWC requires hot force, sufficient knowledge and skills rather than multitudes of unproductive age (high dependency ratio) (from KII discussions).

3.3.6. Sex

The sex of HH negatively and highly influences the adoption of improved soil bund at 5% significance level in our study area. Being male or female is another important factor influencing adoption of soil bund negatively at 5% significant level. It is not surprising that sex negatively affects adoption of SWC, because women has a productive as well as a reproductive role in a family, which make them more responsible for feeding the family by conserving natural resource than male counterparts. The role of women is very high in increasing yield through investing and assisting in investment of SWC (Bekele & Drake, 2003). So, as a number of male headed HHs increases by a unit making other factors constant, the probability of adoption of soil bund and stone bund decreases by 12.8% and 3.2%, respectively. The absence of a significant influence on the sex of HH on stone bund might be due to equal gender participation on conserving soil and mitigating erosion in study area. Because, as we informed from the FGD group, more women HHs were participated on using of improved soil bund than male HHs since it is not as hard as stone bund. Thus, the current study is consistent with (Amsalu & Graaff, 2006; Belachew et al., 2020; Tesfayohannes et al., 2022). In line with this study (Bekele & Drake, 2003), the exposure of women to the problem of soil erosion may improve the perception of the whole household about the problem and thereby influence the decision process. However, some studies reported the positive relationship between gender and adoption of SWC structures. As an example, Kassa et al. (Abebe & Bekele, 2014) reported the positive and significant relationship between gender and adoption of SWC structure. Thus, it can be concluded that as the number of women increases, the adoption probability of improved soil bund increases.

3.3.7. Using credit

Using credit services negatively and significantly affect adoption of soil bund at 10% significant level. The likelihood of adopting soil bund may decrease by 0.8% as an additional unit of credit services used by farmers. This might be because farmers’ attention may goes to running a business and failing to consider the implementation of SWC particularly and farming activities generally. Similar to our study, Belachew et al. (2020) indicated that the farmers who access to credit and get credit services may not use it for the intended or planned activities like managing of SWC. This implies that great attention should be given by microfinance and credit and saving institutions to monitor farmers as they trustily implement pardon they get via loan on the aimed and planned activities.

3.3.8. Using PSNP

Using PSNP negatively and significantly affects the adoption of improved soil bund. However, the use of PSNP services negatively and highly influence adoption of stone bund at 5% significance level. On the other hand, using PSNP services commonly influence adoption of soil bund and stone bund negatively at 1% and 5% significance levels, respectively. Using PSNP is the other core factor that negatively affects adoption of SWC in our study. Therefore, using an additional one unit of PSNP services decreases the likelihood to adopt soil and stone bunds by 10.2 % and 10.6%, respectively, by making other factors constant. This might be because when farmers keep only PSNP services without paying any effort on farm activities, it makes them waiters (expose to dependency syndrome) and fail to participate in SWC. Thus, the finding provides a direction as policymaker and different development actors give due attention to PSNP, especially on conserving and sustaining natural resource management like SWC.

3.3.9. Level of education

Level of education also positively and highly influences adoption of stone bund at 5% significance level. However, some studies show that education level has a negative effect on SWC (Tesfay et al., 2018). However, education and soil have an association because educated farmers can read and understand well the impact of land degradation and the role of conserving both soil and water than illiterate farmers (Wordofa et al., 2020). Some authors have indicated that educated farmers are more likely to use appropriate SWC measures than uneducated farmers (Desta & Neka, 2017). Moges and Taye (2017) found that the level of education of farmers is strongly associated with their perception to invest in SWC technologies. We also found a positive and significant correlation between education and adoption of stone bund at 5% significant level. Therefore, by making all other factors constant, an additional 1 year school of education can increase the likelihood to adopt stone bund by 3.2%. Accordingly, our result substantiates the findings of recent studies conducted in Ethiopia that documented the positive and significant effect of education in fostering adoption of SWC measures (Desta & Neka, 2017; Belachew et al., 2020, Sileshi et al., 2019a; Delibo, 2017; Wordofa et al., 2020). Thus, educated farmers can choose the appropriate improved SWC structures to implement and can also judge the consequences of the erosion problem than uneducated farmers (Wordofa et al., 20120). Thus, education is an important thing to help farmers make the decision to adopt appropriate SWC technologies.

Education is significantly important to adopt SWC by following its science and techniques. As we had formerly discussed, education level of HHs most significantly affects the adoption of stone bund at p-value ≤0.05. Similarly the recent adoption studies of SWC conducted in Ethiopia reported that education level is directly related to adoption of SWC practices (Sileshi et al., 2019a; Belachew et al., 2020; Wordofa et al., 2020; Yirgu, 2022). Conversely, the study conducted in South Nigeria reflects as education has an inverse relationship with adoption of SWC in rice farming (Ojo et al., 2021). But, Mango et al. (2017) reported on their study as education raises the awareness of farmer and their chance to participate in important SWC measures.

Thus, it is clearly indicated in literature as educating local farmers is principally important through adult education and training to enable them protect soil erosion and apply SWC for sustainable use of land (Wordofa et al., 2020). This implies that education is an input to being aware and informed well about the effect of adopting SWC and the techniques to be followed to construct SWC on own farm.

4. Conclusions and recommendations

The existing SWC practices in many parts of developing countries, especially in some parts of SSA, are applied by indigenous knowledge and lacks scientific methods. In Eastern part of Ethiopia, particularly in Gemechis district of West Hararghe zone, both traditional and improved SWC were applied by farmers. However, the majority of farmers (64.4%) are specifically based on traditional SWC and fail to accept improved SWC (i.e. only 35.6% were adopter). Therefore, our study has briefly identified some of the reasons related to farmers’ refusal to adopt improved SWC technology, as well as their focus on traditional technologies and contributions, to research. So, the current study identified some factors which significantly and highly determine the adoption of improved SWC at 1% and 5% significance levels in Gemechis area. Perception to erosion, farm experience, PinTDI, sex of HH, extension contact, level of education, and using of PSNP services are found to highly influence adoption of improved SWC.

From the current study, we found that changing the perception of farmers on erosion problem and revealing its consequences in real-life scenarios is paramount important to adoption of improved SWC. Thus, to enhance the rate of adoption in the current situation, dynamism experienced and educated farmers are required to sustain and expand the adoption of improved SWC in the study area. Community engagement or farmer’s participation in technology (SWC) design and implementation also makes it easy, as farmers can easily accept the technology to implement on his own plot.

However, the current study indicates that participation of farmers in design and implementation (PinTDI) of improved SWC in most parts of study area are overlooked. This results in low adoption of improved SWC, which requires appropriate government policy to reverse the trend. On other hand, using of PSNP is the other core factor that negatively affects adoption of improved SWC in our study. So, using an additional one unit of PSNP services decreases the likelihood to adopt soil and stone bunds by 10.2% and 10.6%, respectively. This is due to farmers keep aid rather than doing for themselves and conserving natural resource for better production. Extension contact and its approach have also great influence on adoption of improved SWC, which may require government policy to give more emphasis to change extension system in demonstrating new technology for adoption purpose.

Therefore, our study briefly identified a few reasons associated with farmers’ refusal to adopt improved SWC technology and their focus on traditional ones and contributed to future research. These are: the technology is not based on the practical situation of the farmer when planning and adaptation; there is a lack of extension system; there is a lot of expectations related to PSNP; and there is still a knowledge gap. These are some of the indicators found by our research.

Thus, depending on the real context of study area, the current study provides the following four major recommendations for policymakers and development actors and for further research in future:

1. Government policy should pay close attention to the use and provision of PSNP to farmers, because when farmers only retain PSNP services without putting any effort into agricultural activities, they behave like server farmers (they are susceptible to dependency syndrome) and do not participate in SWC.

2. Short-term training and awareness raising should be provided to farmers on erosion and its consequences.

3. Involve farmers in the design and implementation of SWC practices.

4. Because the researcher only focused on the adoption of ISWC and not on the level of adoption of improved SWC, further research on the intensity of adoption of improved SWC is needed.

Abbreviations

BVP: Bivariate probit, FGD: Focus group discussion, HH: head of household; MNL: Multinomial logit, MNP: Multinomial probit, NGO: Non-governmental organization KII: Key informant interview

Acknowledgments

We would cordially thank all of the enumerators, respondents, and reviewers, as well as the Oromia Agricultural Research Institute and Mechara Agricultural Research center, for all of their support to conduct and accomplish this study. In addition, our special thanks go to Zonal and district experts who volunteered and supported us in collecting the data for the study.

Disclosure statement

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

Data availability statement

https://data.mendeley.com/datasets/7g9gzhjnvb/draft?a=dcfb658b-a767-447c-a93b-054ada5e8c7b

Additional information

Funding

The work was supported by the No fund for publication process Oromia Agricultural Research Institute [No].

Notes on contributors

Mideksa Babu

Mideksa completed his first degree in rural development and agricultural extension from Arba Minch University. He also completed his master’s degree in rural development and agricultural extension from Haramaya University. He served as the leader of the Agricultural Extension research team for over two and a half years.

Muluken Gezahegn

Muluken G. Wordofa (Ph.D.) is an Associate Professor in the School of Rural Development and Agricultural Innovation (Haramaya University, Ethiopia. He obtained his Ph.D. in Local Development and Global Dynamics (University of Trento, Italy) in 2015, Erasmus Mundus Double M.Sc. in Agricultural Development from the Dresden University of Technology (Germany) and University of Copenhagen (Denmark) in 2010, and B.Sc.

Eric Ndemo

Eric Ndemo Okoyo (PhD) is an Associate Professor in School of Agriculture and Environmental Science, Haramaya University. He has published more than 20 articles and 1 book in rural/ regional development, Environmental Studies and Gender.

Notes

According to Babu et al. (2023) Cathara is the Amharic word used in Ethiopia for the embankment made from soil to protect land from erosion.

Kebele: is also the term used in Amharic in Ethiopia to indicate small village or peasant association (Babu et al., 2023).