Determining Factors Affecting Agroecological Practices’ Acceptance and Use in Mali, West Africa

Abstract

Land degradation issues and declining fertility are driving the need for agroecological practices. This research analysed the determinants of acceptance and actual use of five main agroecological practices (contour farming techniques, organic fertiliser, crop association, improved seeds and integrated crop management practices) by farmers in Mali. The extended Unified Theory of Acceptance and Use of Technology (UTAUT) was used to develop the conceptual model. Data were collected from 505 randomly selected farming households in the cotton and cereal production zones in Mali. Partial Least Square–Structural Equation Modelling (PLS-SEM) was used to estimate technology acceptance and use. The findings revealed that behavioural intention is significantly and positively influenced by the expected performance and social influence. The expected effort is a key influential factor of the behavioural intention to adopt organic fertiliser. Experience has a mediating effect on the relationship between social influence and behavioural intention to adopt improved seeds adapted to the agroecological conditions. The actual use behaviour is directly and positively affected by the behavioural intention, facilitating conditions and expected net benefit. These findings align with the UTAUT model, have useful implications for both farmers and decision-makers and offer directions for technical approaches to agroecological practices’ development.

1. Introduction

Feeding families while coping with land degradation and land use pressures is among the biggest challenges facing households in developing countries [1], particularly in sub-Saharan Africa, where land degradation, declining soil fertility and high input costs have a negative impact on agricultural productivity [2]. Improving agricultural productivity generally relies on the use of expensive chemical fertilisers, with risks that most farms are unable to bear [2,3,4]. From this perspective, questions arise as to whether the current production methods can produce enough food without further impacting the environment, at the risk of compromising the ability of future generations to meet their own food needs [5]. Other authors highlight the need to adapt current agricultural systems, given the strong potential for improving agricultural productivity in this area [6]. It should also be remembered that agricultural land extension possibilities remain limited due to the pressure already exerted by agriculture on natural areas, causing deforestation and its corollary impacts on agroecosystemic resources [7,8]. In Mali, land use pressure is very high, resulting in a removal of natural formations in favour of man-made formations, particularly under-cropped and fallow land. In fact, there is a perfect correlation between the dynamics of land use and population growth in Mali’s cotton and cereal production zones [9,10]. The prospects of reversing this trend remain very slim without a profound change in land use and agricultural practices [10].

One of the options advocated by professionals or practitioners in the agricultural sector (researchers/scientists, farmers) is the adoption of agroecological practices that aim to create integrated, structured and sustainable systems, but that do not conflict with conventional agriculture [1,11,12]. There is growing interest in alternative practices aimed at reducing the vulnerability of agroecosystems, including soil erosion, degradation of soil and water quality, excess or insufficient water, loss of plant and animal productivity and deterioration of ecological niches [1,11,12,13]. The present research was conducted from this perspective.

Agroecology offers real potential for improving the sustainability of production systems in Mali. It is important to note the lack of quantitative information on agroecological practices, which is not conducive to their promotion. There are a number of charities, associations and NGOs supporting agroecology through the implementation of agricultural development projects, but their quantitative data remain scarce [14]. Moreover, agroecology is a relatively unresearched area, making this research original and worthwhile. It is therefore more than opportune to conduct research at the level of farmers, who are most involved in the adoption of agroecological practices. Given the importance of farmers in agroecological practices’ cultivation, it is important to understand the determinants of their behavioural intention and use behaviour with these practices. Examining their behavioural intention and actual use behaviour will provide insights into agroecological practices. This will help improve knowledge on agroecological practices and allow recommendations to be made to decision-makers on their development in Mali.

From a theoretical perspective, the UTAUT model has been used extensively in research on behavioural intention and actual use behaviour [15,16,17,18,19,20,21,22,23,24,25,26]. Its applications cover several sectors (education, agriculture, food systems, applied sciences, transport, social media) and it is applied worldwide [22,27,28,29,30,31,32,33,34,35]. This model has been extensively tested for the prediction of technology adoption behaviour. The UTAUT model is user-friendly for explaining not only the acceptance of technologies but also actors’ actual behaviour in using them [16,26]. Based on that premise, the UTAUT model served as a theoretical basis for assessing the factors influencing behavioural intention and actual behaviour in adopting agroecological practices in Mali. Although the original UTAUT model has proven its effectiveness in behavioural studies, several authors have shown that it could be further improved with the introduction of new variables with a view to its extension [16,21,22,36]. Among these studies, less attention has been paid to the expected net benefit (Eb), particularly concerning the adoption of agroecological practices. The present research aimed to investigate the determining influencing factors of farmers’ behavioural intention and actual behaviour in adopting agroecological practices, in the process determining the impact of the expected net benefit on the original UTAUT model.

The different agroecological practices in this research are fivefold: (i) contour farming techniques (CTs), also known as “contour bunding” to improve infiltration and crop yields [37,38,39], (ii) organic fertiliser (OF) as a source of nutrients to improve the soil structure and fertility by improving the physico-chemical properties of the soil (facilitating ion exchange) and reducing soil erosion [40,41,42], (iii) crop associations (CRAs) or intercropping to help control pests and diseases, boost soil fertility, improve and stabilise yields and contribute to the food balance [43,44], (iv) improved seeds adapted to agroecological conditions (ISEED-AE), which are resistant in some respects to difficult climatic conditions, pests and diseases [45,46] and (v) crop rotation and integrated crop management practices (ICMPs) to protect crops and break the development cycle of certain pests or threats to crops [47,48]. Although there is a need to involve various stakeholders (policymakers, researchers, NGOs, farmers), the role of farmers remains essential in the large-scale use of agroecological practices [49].

The main objective of this paper is to provide insights into the critical factors that can improve farmers’ acceptance and usage of the stated five agroecological practices in Mali. Two main research questions are addressed: (i) What are the influencing factors of farmers’ behavioural intention and their actual use behaviour for the five agroecological practices? (ii) Does the expected net benefit significantly affect the original UTAUT model in improving farmers’ actual use behaviour for the investigated agroecological practices?

The findings align with the existing literature on the original UTAUT model. In addition, they highlight the direct and mediating impact of the expected net benefit, contributing accordingly to building up an extended UTAUT model, with a slightly better explanatory power in improving farmers’ acceptability and use of agroecological practices.

2. Literature Review and Research Hypotheses

There is abundant literature on theories relating to the acceptance of agricultural technologies. Most of these theories have been developed from an information systems perspective: the Theory of Reasoned Action (TRA) [50], Theory of Planned Behaviour (TPB) [51,52], Diffusion of Innovations Theory (IDT) [53], Theory of Technology Acceptance (TAM) [54,55], the combined TAM-TPB [56], the motivational model (MM) [57], social cognitive theory (SCT) [58] and the Unified Theory of Acceptance and Use of Technology (UTAUT) [15]. The multiplicity of models has led to the need to create a unifying framework based on existing theories of technology acceptance [15]. The original UTAUT model, whose validity and reliability have been demonstrated, is made up of four main constructs, namely expected performance (Pe), expected effort (Ef), social influence (Si) and facilitating conditions (Fc), which determine behavioural intention (Int) and actual use behaviour (Use) of technologies [15].

Acceptance of innovations represents a decision-making process that is manifested by acceptance and rejection factors and is thus a crucial concept for justifying the reasons why innovations do not succeed [59,60]. On the other hand, the use of a technology is associated with the behavioural intention to use that technology [61]. Behavioural intention reflects the strength of the potential user’s intention to make or support the decision to apply or use the technology in their mind [62]. Determinants of behavioural intention include subjective norms, which relate to the various social pressures that an individual experiences from groups of people or other people who are important in their life and willing them to act or adopt a specific behaviour [51]. Behavioural intention and actual behaviour have similar determining factors [54]. In addition, [51] identified behavioural intention as an immediate determinant of actual behaviour. In other words, the stronger the intention, the greater the individual’s probability of making an effort to adopt a behaviour and the greater the propensity to adopting that behaviour [63]. There is therefore a link between producers’ actual use of agroecological practices and their intention to adopt them. Consequently, the following hypothesis was formulated:

H1. 

Behavioural intention is a determining factor and positively influences the actual adoption behaviour for agroecological practices (CT, OF, CRA, ISEED-AE and ICMP) in Mali.

The research conceptual framework showing the hypothetical influential factors of farmers’ intention and actual use behaviour for agroecological practices are presented in Figure 1.

This framework shows that both intention and actual use behaviour are influenced by individual reactions that depend on technologies’ nature. According to this theory, four main factors (expected performance, expected effort, social influence and facilitating conditions) directly affect behavioural intention and actual technology use behaviour. The expected net benefit (Eb) is added in the context of this research in order to assess its influence (extended UTAUT model) on actual technology use behaviour [64,65]. Experience is selected as a moderating variable, as proposed in the original model by [15]. The main latent and moderating variables associated with the hypotheses are summarised in the following points.

• Expected Performance (Pe). Expected performance refers to the extent to which one individual believes that using a technology would or would not improve their results [15]. In this case, farmers need to know the extent to which the application of agroecological practices would enable them to protect or restore soil fertility and improve crop productivity. Thus, the authors [15] consider that performance is among the important factors influencing the adoption of technologies. Expected performance is the main factor determining behavioural intention to adopt a technology [66]. Expected performance affects and positively influences behavioural intention to adopt a technology [67,68,69]. In the context of this research, our hypothesis was the following:

H2. 

Expected performance has a positive influence on the intention to adopt the five practices (CT, OF, CRA, ISEED-AE and ICMP).

 

Expected Effort (Ef). The effort expectancy or expected effort corresponds to an individual’s perception of whether or not the new technology will allow them to save effort following its adoption [15]. In this research, expected effort refers to the ease or difficulty that the use of agroecological practices could generate for farmers. According to reference [70], it is a determining factor in the acceptance and use of technology. Other studies have shown that the expected effort has a positive influence on behavioural intention to adopt a technology [68,69,71]. Based on these elements, the hypothesis was as follows:

H3. 

Expected effort has a positive effect on behavioural intention in the adoption of the five agroecological practices (CT, OF, CRA, ISEED-AE and ICMP).

 

Social Influence (Si). Social influence refers to the degree to which an individual believes that important people or groups of people could persuade them to accept or adopt a technology [15]. This research aimed to study how the adoption of selected agroecological practices is influenced by specialist services (extension), other farms and other people important to them. Social influence is a determining factor and positively influences behavioural intention to adopt a technology [72]. Several other studies have indicated a positive influence of social influence on behavioural intention to adopt technologies [73,74,75,76]. Based on this literature, the hypothesis was as follows:

H4. 

Social influence has a significant and positive effect on behavioural intention to adopt the five agroecological practices (CT, OF, CRA, ISEED-AE and ICMP).

 

Facilitating Conditions (Fc). Facilitating conditions are defined as the degree to which one individual perceives that the organisational and technical infrastructure exists to support the use of one technology [15]. In the context of this research, the facilitating conditions refer to all the necessary facilities for farmers (technical and organisational support, own or external material and financial resources) that can improve the use of the agroecological practices chosen by producers. Facilitating conditions are considered to be a determinant of the ease or difficulty of carrying out a task or removing a barrier in the environment [68,69]. Several authors [11,77,78,79] have shown a positive influence on behavioural intention and technology adoption. Hence, the following hypothesis was formulated:

H5. 

Facilitating conditions directly and positively influence the adoption of the five agroecological practices (CT, OF, CRA, ISEED-AE and ICMP).

 

Expected Net Benefit (Eb). Perceived net benefit is proposed as a new independent variable in addition to the four previous variables proposed by [15]. Expected net benefit refers to the degree to which an individual believes that the new technology would allow a benefit greater than the costs incurred to implement it [64]. Referring to consumer theory, the utility derived from a product, which could be applied to agroecological technologies, corresponds to the sum of the utilities of each technology’s perceived characteristics [75]. Farmers are considered consumers of agroecological technologies and interested in related characteristics and their expected net benefits. In this research, the potential net benefit or utility includes, depending on the technology, cost reduction, improved income, environmental protection and health benefits. The propensity to adopt technology is high for farmers who perceive a potential net benefit from the technology [64]. Moreover, the net benefit was a determining factor in the adoption of technologies [65]. Our hypothesis was as follows:

H6. 

The perceived net benefit directly affects and positively influences the adoption of the five agroecological practices (CT, OF, CRA, ISEED-AE and ICMP).

 

Moderating Variables. In the formulation of the conceptual framework of the UTAUT model, ref. [15] indicated that experience is likely to moderate the impacts of behavioural intention and actual technology adoption behaviour. Ref. [76] indicates that experience plays a crucial role in the formulation of the intention to adopt conservation agricultural practices. In this research, the specific experience of the respondents with the adoption of practices is based on the hypothesis that with time, individuals who have accumulated a certain amount of positive experience will continue to apply good practices. The sub-hypotheses linked to the mediating variable are as follows:

HM1. 

Experience has a mediating effect on the relationship between the expected effort and behavioural intention to adopt agroecological practices.

HM2. 

Experience has a mediating effect on the relationship between the social influence and behavioural intention to adopt agroecological practices.

HM3. 

Experience has a mediating effect on the relationship between the facilitating conditions and actual use behaviour in adopting agroecological practices.

HM4. 

The expected net benefit has a mediating effect on the relationship between the behavioural intention and actual use behaviour in adopting agroecological practices.

3. Research Methodology

3.1. Study Area

Two administrative regions in Mali, Ségou in the cereal production zone and Sikasso in the cotton zone (Zone Compagnie Malienne pour le Développement des Textiles), were selected for this research. These two regions are representative of the agricultural production system in Mali because of the dominance of food crops and cash crops (cotton), supported by a strong integration of agriculture and livestock. One commune was chosen in each region, Cinzana in Ségou and Kléla in Sikasso.

In Ségou, the commune of Cinzana was chosen, latitude 13°10′00″ and 13°30′00″ N and longitude 5°40′00″ and 6°10′00″ W. The rural area of Cinzana is estimated at 11,100 km². The terrain is flat, with plains suitable for growing millet (Pennisetum glaucum), sorghum (Sorghum bicolor), maize (Zea mays) and livestock. The soils are loamy, sandy-clayey, silt-clayey and gravelly. The climate is tropical Sahelian, with average rainfall ranging from 650 to 750 mm, irregularly distributed and with inter-annual variability, with a rainy season (June to October) and a dry season (November to May). The average annual temperature is 28 °C, with extremes ranging from 22 °C in January to 33 °C in May. Cinzana is a multi-ethnic commune (Bambara, Sarakolés, Peulhs, Bobos, Somonos, Mossis) with a population of around 4896.

In Sikasso, the commune of Kléla was chosen, at latitudes of 11°32′30″ and 11°52′30″ N and longitudes of 5°30′00″ and 5°52′30″ W. The soils in Kléla commune varied from sandy-clayey to the silt type, and the climate is Sudano-Sahelian, with rainfall reaching 1000 mm per year. The terrain is rugged, with gravelly plateaus and depressions. The vegetation is dominated by species such as Néré (Parkia biglobosa), Baobab (Adansonia digitata), Caïlcédrat (Afzelia africana) and Karité (Vitellaria paradoxa). The population of the Kléla commune is estimated at 3198, divided between several ethnic groups (Sénoufo, Miankas, Bobos and Peulhs) (PDSEC 2016–2021).

The geographical locations of Cinzana and Kléla are presented in Figure 2.

3.2. Sampling Procedure

Multi-stage sampling techniques were used for this research. In the first stage, the commune of Kléla (Circle of Sikasso) and the commune the Cinzana (Circle of Segou) were selected based on their importance in cotton and dry cereal production, respectively. The commune of Kléla belongs to the Circle of Sikasso, and it is ranked first in cotton production among the agricultural production zones of the Sikasso sector. According to statistics from the Sikasso sector, the area under cotton of Kléla in the 2021–2022 cropping season was estimated at 25% of the total area under cotton, and Kléla had 26% of the area in the sector under maize. The commune of Cinzana is part of the Circle of Ségou, which ranks first in dry cereal production in the whole region (35%), according to the statistics from the 2021–2022 cropping season report of the agricultural services. It benefits from the proximity of a research station in the commune of Cinzana and the support of the agricultural extension services located in the town of Ségou, around thirty-five miles away. Cinzana therefore has a relative advantage over most other communes in terms of access to agricultural technologies.

In the second stage, an exploratory survey was carried out in 10 villages randomly selected in each zone, out of twenty villages in Kléla and fifty villages in Cinzana, in order to establish the sampling frame. A simplified survey guide was drawn up for this purpose in order to collect basic information on farm characteristics (number of farms, farming practices, and their adoption rates). Following this preliminary phase, four representative villages were selected in each zone for detailed surveys.

In the third stage, the Kothari formula was used to randomly select 505 farms, 247 farms in Kléla and 258 farms in Cinzana, using the following expression [80]:

$$n={\displaystyle \frac{Z^2\ast N\ast p(1-p)}{\left(N-1\right)e^2+Z^{2}p(1-p)}},$$

(1)

where the following are used:

 

$n$ = sample size of households within the eight selected villages;

 

$N$ = size of the target population, equal to 596;

 

$e$ = acceptable margin of error (threshold 0.05%);

 

$p$ = proportion of the population with the characteristics of interest (proportion of farms adopting at least one agroecological practice);

 

$Z$ = 1.96, representing the standard variate at a given confidence level (0.05).

In order to increase the level of inclusion of respondents, the same formula was applied to each village. This resulted in a minimum size of 481 farms for a total of 506 surveyed. The reason for this additional margin was to anticipate any errors that might occur during data collection and/or computer processing. After data processing, one observation was eliminated due to missing information. By doing so, 505 farms (84.73%) were retained for the final analyses (

Table 1. Sample size per agroecological zone and village.

Village	Total Number of Farms	Min Sample Size Required	Final Sample Size
Cotton zone (Kléla)
Dougoumousso	85	68	72
Kong-Kala	85	68	72
Siani	60	51	60
Nantoumana	43	38	43
Sub-total 1	273	225	247
Cereal zone (Cinzana)
Kondogola	133	96	97
Minangofa	63	53	53
Fambougou	54	47	48
Cinzana village	73	60	60
Sub-total 2	323	256	258
Pooled	596	481	505
%	100	80.70	84.73

).

3.3. Data Collection

Before the collection phase, a team of four experienced interviewers were trained on the questionnaire developed for this purpose. At the end of the training session, and prior to the data collection phase, a pilot survey was conducted on a neutral site where the questionnaire was pre-tested with 20 producers. The questionnaire was designed in French and translated into the local language during the interviews to overcome any language barriers. The questionnaire included information on socio-demographic characteristics, knowledge and perceptions of the main determining factors of acceptance and adoption of agroecological practices (CT, OF, CRA, ISEED-AE, ICMP). A Likert scale evaluation grid (1 = Strongly disagree, 2 = Disagree, 3 = More or less agree, 4 = Agree, 5 = Strongly agree) was designed. Data were collected between July and August 2022 face-to-face with the farmers. Farmers were free to participate once they gave their prior consent. Before each interview, the purpose of the research was explained to participants, and we reminded them of the voluntary nature of their participation before we asked questions. All participants provided their informed consent to participate in this research.

3.4. Harman’s Single-Factor Test

Harman’s single-factor test is important in quantitative research aiming to determine whether there is a common method bias [77,81]. The common method bias is determined by comparing the total explained variance with a recommended highest value of 50% as a threshold. On the other hand, the total explained variance should be lower than 50% [77]. Analyses revealed that the total variance explained by the single-factor solution varied from 28.369 to 47.147, indicating that the data were free from common method bias and therefore ready for analysis. The details of Harman’s single-factor test of the different agroecological practices are summarized in

Table 2. Results of Harman’s single-factor test.

Practice *	Component	Initial Eigenvalues			Extraction Sums of Squared Loadings
		Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
OF	1	7.376	28.369	28.369	7.376	28.369	28.369
CT	1	9.360	35.999	35.999	9.360	35.999	35.999
CRA	1	10.496	40.369	40.369	10.496	40.369	40.369
ICMP	1	11.426	43.945	43.945	11.426	43.945	43.945
ISEED-AE	1	12.258	47.147	47.147	12.258	47.147	47.147

* Practices. OF = Organic fertiliser, CT = Contour farming technique, CRA = Crop association, ICMP = Integrated crop management practices, ISEED-AE = Improved seed adapted to agroecological conditions.

.

3.5. Data Analysis

Descriptive statistics and Harman’s single-factor test were performed using SPSS 26 software while a Partial Least Square–Structural Equation Modelling (PLS-SEM) approach using SmartPLS 4 was applied to assess the causal relationships between intention and actual use behaviour for agroecological practices. PLS-SEM is very user-friendly and adapted to different sample sizes (small or large) and data normality problems, making it very popular [71,78].

The PLS-SEM analyses were carried out in two stages. The first stage involved assessing and analysing the suitability of the measurement model used for each of the agroecological practices. The parameters assessed for validity, which had to meet the acceptability criteria, were the following constructs: (i) factor loading, (ii) multi-collinearity (Variance Inflator Factor), (iii) convergent reliability, (iv) Cronbach’s alpha (Cronbach’s α) and (v) Average Variance Extracted (AVE). The results must be higher than the following: factor loading >0.70, multicollinearity coefficient VIF < 5, Cronbach’s alpha α > 0.70 and Average Variance Extracted AVE > 0.5 [79,82]. Factor loadings are weak and not acceptable if <0.4; acceptable if between 0.4 and 0.7; and strong if >0.7 [83,84], while composite reliability is achieved if higher than 0.5 (CR > 0.5) [85]. In addition, the Fornell–Lacker criterion stipulates that the factor loadings of each construct must be higher than the cross-coefficients; and the square root of the AVE for each construct should be greater than the correlation of this construct with the others [71,86].

The second stage analysed the fit of the structural model, where the bootstrapping method was appropriate. This method is applied to examine the significance of the indicator coefficients and to evaluate the coefficient of determination (R²) and the path coefficient [82,87]. It is used to determine the values of the β coefficients and the t-test using the SmartPLS 4 software in a structural measurement model [87].

An important aspect to be considered is the explanatory power, which is determined by the coefficient of determination (R²) [87] and assessed using three values: 0.25 when the explanatory power is low, 0.50 for an average explanatory power, and 0.75 when it is high. The minimum threshold for the R² is 0.10 [82,88]. It is also important to report the effect size function ($f^2)$ developed by [89] and the predictive quality ($Q^2)$ [82,90,91]. The expression for the effect size function ($f^2)$ is as follows:

$$f^2={\displaystyle \frac{{R^2}_{included}-{R^2}_{excluded}}{1-{R^2}_{included}}}$$

(2)

${R^2}_{included}$ and ${R^2}_{excluded}$ are the R² of the latent dependent variables when the latent predictor variable is used or omitted from the structural equation, respectively. The values of $f^2$ are compared to three thresholds (0.02: small effect size; 0.15: medium effect size; and 0.35: large effect size), which are similar to those of Cohen (1988) [92] in the context of operational definitions of multiple regressions [82,91].

The $Q^2$ function developed by [93] and [94] fits well with PLS [91]. Without loss of freedom, Q² represents the measure of how well the observed variables are reconstructed by the model and its parameter estimates [89]. When Q² > 0, the model has a good predictive quality, whereas a value of Q² < 0 means a poor predictive quality [82,91].

4. Results

The main findings of this research on the determining factors of the five practices (contour farming techniques, organic fertiliser, crop association, improved seed adapted to agroecological conditions and integrated crop management practices) are presented in three sections: (i) descriptive statistics on the main agroecological practices and farmers’ characteristics and agroecological practices, (ii) the measurement model and (iii) the structural model results using PLS analyses.

4.1. Main Agroecological Practices and Farmers’ Characteristics

The majority of farming households were male headed (99%) with a mean age of 53 years. A large proportion of them were married (97%) and 26% had attended formal education, with a mean of six years spent attending formal school. The main occupation of the surveyed farms was agriculture (95%), and the average population per household was 19. Membership of a farmers’ organisation accounted for 85%, while 49% of surveyed farms declared having received training on agricultural best practices. Overall, the findings showed that the adoption rate of agroecological practices during the 2021–2022 cropping season, which was the reference period, ranged from 19% (CT) to 92% (ICMP). For the purposes of this research, ICM practices included crop rotation, the use of biopesticides, legumes and microdosing.

The number of years of experience varied according to the zone and type of practice. The greatest experience was obtained with OF (19.2 years) and ICMP (17.5 years). Farms in the Kléla zone had more experience in adopting CT (7.8 years compared to 0.8 years in Cinzana), ISEED-AE (10.4 years compared to 6.8 years in Cinzana) and ICM (20.8 years compared to 14.4 years in Cinzana). Those in Cinzana had more experience in adopting OF (20.3 years compared to 17.9 years in Kléla), followed by CRA (11.7 years compared to 1.4 years in Kléla) (

Table 3. Agroecological practices and socio-demographic characteristics of surveyed farms.

Zones/Communes	Kléla (n = 248)	Cinzana (n = 257)	Pooled (N = 505)
Farmers’ characteristics and practices		Statistics
Socio-demographic characteristics
Gender of farm manager (1 = male)	1.00	0.97	0.99
Age of farm manager (years)	54.12 (14.14)	52.27 (14.15)	53.17 (14.17)
Main occupation (1 = agriculture)	0.97	0.94	0.95
Marital status of farm manager (1 = married)	0.99	0.95	0.97
Formal education (1 = formal education)	0.12	0.40	0.26
Number of years spent in formal school (years)	6.30 (3.36)	6.18 (3.08)	6.20
Household size (number)	23.5 (16.5)	16.1 (10.8)	19.7 (14.4)
Institutional factors
Membership of a farmers’ organisation (1 = yes)	0.98	0.71	0.85
Training on agricultural best practices (1 = yes)	0.62	0.37	0.49
Adoption of farming practices (1 = yes)
Contour farming techniques (CTs)	0.34	0.04	0.19
Organic fertiliser (OF)	0.87	0.94	0.90
Crop association (CRA)	0.08	0.41	0.25
Improved seeds (ISEED-AE)	0.53	0.53	0.53
Integrated crop management practices (ICMPs)	0.99	0.85	0.92
Specific experience in adopting practices (years)
Contour farming techniques (CTs)	7.6 (8.8)	0.83 (3.33)	4.2 (7.4)
Organic fertiliser (OF)	17.9 (11.9)	20.3 (13.1)	19.2 (12.6)
Crop association (CRA)	1.4 (4.3)	11.7 (13.9)	6.7 (11.6)
Improved seed (ISEED-AE)	10.4 (12.6)	6.8 (8.1)	8.6 (10.7)
Integrated crop management practices (ICMPs)	20.7 (11.2)	14.3 (12.5)	17.4 (12.3)

Note. The numbers in brackets are standard deviations. Source. Our surveys, July–August 2022.

).

4.2. Results of the Measurement Model

At the first level, the assessment of indicators’ reliability (values of the factor loadings) showed that for most of the practices, these values were greater than 0.7, as desired. For CT, 20 items (80%) had a value greater than 0.7, along with 23 items (92%) for OF, 24 items (96%) for CRA and ISEED-AE, respectively, and 25 items (100%) for ICM practices. Weaker values with item factor loadings less than 0.4 were removed. The variance inflation factor (VIF) was assessed and all values were between 0.2 and 4.1, suggesting the absence of a serious problem of multicollinearity with the data.

At the second level, the consistency of the internal reliability was determined by the values of Cronbach’s alpha and composite reliability. The overall examination (for all the practices) showed that all the values of Cronbach’s alpha (Crb-α) and composite reliability (CR) were greater than 0.7, indicating good internal consistency of the constructs. At the third level, the convergent validity analysis (AVE) showed that all the values were greater than 0.5, indicating a good convergent validity for all the practices (

Table 4. Measurement model results by type of agroecological practice.

Items	Performance Expectancy (Pe)					Effort Expectancy (Ef)			Social Influence (Si)			Facilitating Conditions (Fc)				Behavioural Intention (Int)				Actual Use Behaviour (Use)				Expected Net Benefit (Eb)
	Pe1	Pe2	Pe3	Pe4	Pe5	Ef1	Ef2	Ef3	Si1	Si2	Si3	Fc1	Fc2	Fc3	Fc4	Int1	Int2	Int3	Int4	Use1	Use2	Use3	Use4	Eb1	Eb2
Contour farming techniques (CTs)
Factor loading	0.820	0.805	0.646	0.723	0.655	0.866	0.814	0.750	0.781	0.813	0.771	0.854	0.728	0.847	0.423	0.847	0.796	0.725	0.659	0.817	0.656	0.778	0.866	0.884	0.891
Crb-α	0.784					0.742			0.702			0.705				0.755				0.788				0.730
CR	0.813					0.769			0.710			0.795				0.766				0.811				0.730
AVE	0.538					0.658			0.622			0.539				0.578				0.613				0.787
Organic fertiliser
Factor loading	0.863	0.760	0.788	0.837	0.776	0.909	0.935	0.857	0.852	0.756	0.831	0.891	0.929	0.886	0.915	0.833	0.770	0.777	-	-	0.781	0.854	0.829	0.958	0.937
Crb-α	0.865					0.865			0.757			0.927				0.708				0.770				0.887
CR	0.872					0.872			0.808			0.931				0.716				0.806				0.911
AVE	0.649					0.649			0.663			0.820				0.630				0.676				0.897
Crop association (CRA)
Factor loading	0.881	0.878	0.870	0.783	0.859	0.876	0.862	0.918	0.809	0.850	0.871	0.859	0.829	0.866	0.879	0.847	0.851	0.875	0.673	0.822	0.890	0.869	0.787	0.922	0.886
Crb-α	0.908					0.863			0.797			0.881				0.827				0.865				0.779
CR	0.916					0.873			0.800			0.888				0.835				0.883				0.796
AVE	0.731					0.784			0.711			0.737				0.665				0.711				0.818
Improved seed adapted to agroecological conditions (ISEED-AE)
Factor loading	0.882	0.865	0.875	0.881	0.875	0.917	0.907	0.856	0.841	0.902	0.918	0.807	0.898	0.890	0.914	0.896	0.907	0.903	0.590	0.722	0.795	0.858	0.628	0.955	0.936
Crb-α	0.924					0.873			0.865			0.901				0.846				0.747				0.882
CR	0.925					0.875			0.867			0.909				0.881				0.764				0.901
AVE	0.766					0.798			0.788			0.771				0.698				0.571				0.894
Integrated crop management practices (ICMPs)
Factor loading	0.710	0.868	0.840	0.829	0.721	0.856	0.830	0.810	0.879	0.890	0.886	0.809	0.782	0.735	0.785	0.789	0.787	0.842	0.799	0.791	0.880	0.858	0.684	0.946	0.930
Crb-α	0.853					0.778			0.862			0.783				0.818				0.818				0.863
CR	0.863					0.780			0.867			0.790				0.822				0.827				0.873
AVE	0.634					0.693			0.783			0.606				0.647				0.651				0.879

).

At the fourth level of the measurement model, the discriminant reliability was assessed with the Fornell–Larcker criterion. Overall, discriminant validity was achieved with a high square root of the AVE and a high inter-construct correlation for all the practices (

Table 5. Fornell–Larcker criterion results by types of agroecological practices.

Practices	Use	Int	Ef	Eb	Pe	EXP	Fc	Si
Contour farming techniques (CTs)
Use	0.783
Int	0.600	0.760
Ef	0.618	0.388	0.811
Eb	0.619	0.435	0.569	0.887
PE	0.611	0.442	0.633	0.581	0.733
EXP-CT	0.388	0.383	0.308	0.269	0.333	1.000
Fc	0.596	0.452	0.567	0.608	0.574	0.367	0.734
Si	0.578	0.509	0.457	0.518	0.486	0.304	0.608	0.789
Organic fertiliser (OF)
Use	0.822
Int	−0.007	0.794
Ef	0.110	0.324	0.901
Eb	0.448	0.210	0.360	0.947
Pe	−0.020	0.513	0.465	0.337	0.806
EXP-OF	0.111	0.055	0.049	0.156	0.053	1.000
Fc	0.596	0.058	0.227	0.655	0.137	0.063	0.905
Si	0.414	0.167	0.249	0.544	0.298	0.028	0.628	0.814
Crop association (CRA)
Use	0.843
Int	0.671	0.815
Ef	0.281	0.374	0.885
Eb	0.342	0.434	0.466	0.904
Pe	0.348	0.486	0.665	0.511	0.855
EXP-CRA	0.461	0.475	0.171	0.193	0.322	1.000
Fc	0.387	0.439	0.605	0.570	0.557	0.150	0.859
Si	0.465	0.418	0.477	0.524	0.483	0.239	0.698	0.843
Improved seed adapted to agroecological conditions (ISEED-AE)
Use	0.756
Int	0.626	0.835
Ef	0.376	0.541	0.893
Eb	0.491	0.656	0.695	0.945
Pe	0.424	0.613	0.712	0.665	0.875
EXP- ISEED-AE	0.378	0.414	0.285	0.299	0.354	1.000
Fc	0.524	0.622	0.576	0.742	0.520	0.279	0.878
Si	0.612	0.619	0.522	0.656	0.535	0.332	0.744	0.887
Integrated crop management practices (ICMPs)
Use	0.807
Int	0.643	0.804
Ef	0.426	0.535	0.832
Eb	0.525	0.554	0.544	0.938
Pe	0.594	0.714	0.693	0.657	0.796
EXP-ICMP	0.101	0.116	0.088	0.126	0.128	1.000
Fc	0.525	0.670	0.559	0.564	0.695	0.178	0.778
Si	0.589	0.670	0.490	0.630	0.693	0.160	0.703	0.885

).

4.3. Structural Model

Following measurement model validation, path analyses were performed for hypotheses testing for the five practices by assigning the indices a, b, c, d and e, respectively, for the specific hypotheses of CT, OF, CRA, ISEED-AE and ICMP. In view to testing the impact of the expected net benefit, this research analysed and compared the original and extended UTAUT models before and after their introduction. As the expected net benefit is not included in the original UTAUT model, hypotheses H6 and HM4 were tested only for the extended model.

 

Original UTAUT Model

The original UTAUT model results indicated that behavioural intention (INT) is a determinant of the actual use behaviour for CT (β = 0.382, t = 8.828, f² = 0.217) (H1a), CRA (β = 0.500, t = 11.017, f² = 0.324) (H1c), ISEED-AE (β = 0.183, t = 4.528, f² = 0.190) (H1d) and ICMP (β = 0.528, t = 10.446, f² = 0.270) (H1e). These findings revealed that the farmers’ actual use behaviour for CT, OF, CRA, ISEED-AE and ICMP was significantly and positively influenced by their behavioural intention. Therefore, hypotheses H1a, H1c, H1d and H1e are supported by the original UTAUT model. Facilitating conditions is a determining factor and positively influenced the farmers’ actual use behaviour for CT (β = 0.361, t = 8.790, f² = 0.189) (H5a), organic fertiliser (β = 0.594, t = 19.764, f² = 0.547) (H5b), CRA (β = 0.155, t = 5.385, f² = 0.039) (H5c), ISEED-AE (β = 0.222, t = 6.810, f² = 0.053) (H5d) and integrated crop management practices (β = 0.171, t = 3.195, f² = 0.028) (H5e), which indicated that hypotheses H5a, H5b, H5c and H5e are also supported by the original UTAUT model. The positive values of Q² indicated a good predictive quality of the variables used in the original UTAUT model. The R² values varied from 0.362 to 0.509, suggesting that 36.2% to 50.9% of variations in farmers’ actual use behaviour are explained by the original UTAUT model.

In regard to the influencing factors of behavioural intention, the original UTAUT results indicated that the expected performance (Pe) positively influenced behavioural intention (INT) to adopt CT (β = 0.169, t = 3.203, f² = 0.022) (H2a), OF (β = 0.461, t = 8.464, f² = 0.219) (H2b), CRA (β = 0.254, t = 4.797, f² = 0.050) (H2c), ISEED-AE (β = 0.500, t = 11.017, f² = 0.324) (H2d) and ICMP (β = 0.427, t = 7.845, f² = 0.148) (H2e). Effort expectancy significantly and positively affected behavioural intention to adopt organic fertiliser (β = 0.108, t = 2.307, f² = 0.012), thus confirming hypothesis H3b. The findings revealed that social influence (Si) is a determining factor and positively influenced the behavioural intention to adopt CT (β = 0.337, t = 7.807, f² = 0.124) (H4a), CRA (β = 0.187, t = 3.912, f² = 0.040) (H4c), ISEED-AE (β = 0.340, t = 8.336, f² = 0.161) (H4d) and ICMP (β = 0.340, t = 6.964, f² = 0.136) (H4e). Thus, hypotheses H4a, H4c, H4d and H4e are supported.

The mediating impact of experience on the original UTAUT model revealed that experience has a significant and positive mediating effect on the relationship between facilitating conditions and actual use behaviour for crop association (β = 0.183, t = 4.528, f² = 0.055), while a negative mediating effect was found with the relationship between the facilitating conditions and the actual use behaviour for CT (β = −0.114, t = 2.953, f² = 0.014), the social influence and the behavioural intention to adopt ISEED-AE (β = −0.158, t = 4.271, f² = 0.032), supporting HM3c, HM3a and HM2d, respectively. However, the values of the effect size indicate a weak mediating effect of experience on these different relationships.

 

Extended UTAUT Model

The results of the extended UTAUT model show a significant and positive influence of the behavioural intention and facilitating conditions on the actual use behaviour for CT, OF, CRA, ISEED-AE and ICMP. Consequently, hypotheses H1a, H1c, H1d and H1e, as well as H5a, H5b, H5c and H5e are also supported by the extended UTAUT model. The findings also revealed a significant and positive direct influence of the expected net benefit on the farmers’ actual use behaviour for CT (β = 0.294, t = 6.769, f² = 0.117) (H6a), OF (β = 0.105, t = 2.137, f² = 0.009) (H6b), ISEED-AE (β = 0.124, t = 3.398, f² = 0.023) (H6d) and ICMP (β = 0.265, t = 6.157, f² = 0.067) (H6e). The Q² predict values were positive and non-null for all the practices, which suggested a good predictive quality of data used in the analyses. The R² values ranged from 0.368 to 0.574, meaning that about 36.8% to 57.4% of variations in farmers’ actual use behaviour were explained by the extended UTAUT model.

Likewise, the extended UTAUT produced similar results for the influencing factors of the behavioural intention as the original UTAUT. In other words, farmers’ behavioural intention to adopt contour farming techniques, crop association, improved seeds adapted to agroecological conditions and integrated crop management practices is significantly and positively influenced by the expected performance and social influence, supporting hypotheses H2a, H2c, H2d and H2e. The expected performance and effort expectancy also influence the behavioural intention to adopt organic fertiliser, meaning that hypotheses H2b and H3b are supported by the extended UTAUT model.

The mediating role of the expected net benefit on the extended UTAUT was analysed. A positive and direct mediating effect was found after the inclusion of the expected net benefit, as shown in the results of the extended UTAUT model for CT, OF, ISEED-AE and ICMP, which support H6a, H6b, H6d and H6e. On the other hand, we found a negative indirect mediation effect on the relationship between the behavioural intention and the farmers’ actual use behaviour in adopting CT (β = −0.114, t = 3.129, f² = 0.022) (HM4a), while a positive indirect mediating effect was found on the relationship between behavioural intention and farmers’ actual use behaviour in adopting ICM practices (β = 0.105, t = 2.920, f² = 0.014) (HM4e).

Overall, the results of the actual use behaviour (Use) revealed greater R² values of the extended UTAUT model compared to the original UTAUT model for the different practices, indicating a greater explanatory power of the extended model. This suggests that farmers are interested in the expected net benefit when adopting agroecological practices, resulting in the improvement of the R² after the inclusion of the expected net benefit.

Detailed results of the structural model both for the original and extended UTAUT models are highlighted in

Table 6. Structural model results of Original UTAUT and Extended UTAUT by types of agroecological practices.

Path Hypothesis	Original UTAUT						Extended UTAUT
	Coeff. β	t Stats	R-sq	f-sq	Q2	Decision (*)	Coeff. β	t Stats	R-sq	f-sq	Q2	Decision (*)
Contour farming techniques (CTs)
H1a: Int -→ Use	0.382 ***	8.826	0.509	0.217	0.464	S	0.352 ***	7.931	0.574	0.189	0.534	S
H5a: Fc → Use	0.361 ***	8.790		0.189		S	0.198 ***	4.200		0.050		S
H2a: Pe → int	0.169 ***	3.203	0.351	0.022	0.331	S	0.170 ***	3.208	0.351	0.022		S
H3a: Ef → Int	0.066	1.456		0.004		NS	0.066	1.458		0.004		NS
H4a: Si → int	0.337 ***	7.807		0.124		S	0.336 ***	7.768		0.123		S
HM3a: EXP-CT x Fc → Use	−0.114 ***	2.953		0.014		S	−0.048	1.237		0.003		NS
HM2a: EXP-CT x Si → Int	−0.042	1.045		0.002		NS	−0.042	1.060		0.002		NS
HM1a: EXP-CT x Ef → Int	0.035	0.758		0.001		NS	0.035	0.757		0.001		NS
H6a: Eb → Use							0.294 ***	6.769		0.117		S
HM4a: Eb x Int → Use							−0.114 ***	3.129		0.022		S
Organic fertiliser (OF)
H1b: Int → Use	−0.046	1.374	0.362	0.003	0.354	NS	−0.065	1.823	0.368	0.006	0.359	NS
H5b: Fc → Use	0.594 ***	19.764		0.547		S	0.528 ***	11.707		0.236		S
H2b: Pe → int	0.461 ***	8.464		0.219		S	0.461 ***	8.464		0.219		S
H3b: Ef → Int	0.108 **	2.307	0.274	0.012	0.242	S	0.108 **	2.307	0.274	0.012	0.242	S
H4b: Si → int	0.000	0.005		0.000		NS	0.000	0.005		0.000		NS
HM3b: EXP-OF x Fc → Use	−0.013	0.429		0.000		NS	−0.003	0.091		0.000		NS
HM2b: EXP-OF x Si → Int	−0.027	0.614		0.001		NS	−0.027	0.614		0.001		NS
HM1b: EXP-OF x Ef → Int	−0.008	0.165		0.000		NS	−0.008	0.165		0.000		NS
H6b: Eb → Use							0.105 ***	2.137		0.009		S
HM4b: Eb x Int → Use							−0.005	0.137		0.000		NS
Crop association (CRA)
H1c: Int → Use	0.500 ***	11.017	0.516	0.324	0.368	S	0.464 ***	10.104	0.519	0.225	0.365	S
H5c: Fc → Use	0.155 ***	5.385		0.039		S	0.140 ***	4.054		0.025		S
H2c: Pe → int	0.254 ***	4.797		0.050		S	0.254 ***	4.797		0.050		S
H3c: Ef → Int	0.072	1.665	0.387	0.004	0.369	NS	0.072	1.665	0.387	0.004	0.369	NS
H4c: Si → int	0.187 ***	3.912		0.040		S	0.187 ***	3.912		0.040		S
HM3c: EXP-CRA x Fc → Use	0.183 ***	4.528		0.055		S	0.178 ***	4.149		0.005		S
HM2c: EXP-CRA x Si → Int	−0.005	0.122		0.000		NS	−0.005	0.122		0.000		NS
HM1c: EXP-CRA x Ef → Int	0.081	1.647		0.007		NS	0.081	1.647		0.007		NS
H6c: Eb → Use							0.056	1.224		0.004		NS
HM4c: Eb x Int → Use							0.064	1.241		0.006		NS
Improved seed adapted to agroecological practices
H1d: Int → Use	0.451 ***	12.809	0.439	0.190	0.369	S	0.451 ***	12.178	0.452	0.174	0.376	S
H5d: Fc → Use	0.222 ***	6.810		0.053		S	0.183 ***	4.225		0.011		S
H2d: Pe → int	0.278 ***	4.939	0.537	0.066		S	0.278 ***	4.940	0.537	0.066	0.523	S
H3d: Ef → Int	0.108 **	1.991		0.011		S	0.108 **	1.991		0.009		S
H4d: Si → int	0.340 ***	8.336		0.161		S	0.340 ***	8.335		0.025		S
HM3d: EXP-ISEED-AE x Fc → Use	0.063	1.628		0.005		NS	0.027	0.628		0.161		NS
HM2d: EXP-ISEED-AE x Si → Int	−0.158 ***	4.271		0.032		S	−0.158 ***	4.271		0.032		S
HM1d: EXP-ISEED-AE x Ef → Int	0.058	1.001		0.003		NS	0.058	1.001		0.003		NS
H6d: Eb → Use							0.124 ***	3.398		0.023		S
HM4d: Eb x Int → Use							0.027	0.628		0.001		NS
Integrated crop management practices (ICMPs)
H1e: Int → Use	0.528 ***	10.446	0.432	0.270	0.387	S	0.428 ***	8.548	0.467	0.163	0.409	S
H5e: Fc → Use	0.171 ***	3.195		0.028		S	0.117 **	2.167		0.012		S
H2e: Pe → int	0.427 ***	7.845	0.572	0.148	0.558	S	0.427 ***	7.845	0.572	0.148	0.558	S
H3e: Ef → Int	0.072	1.672		0.006		NS	0.072	1.672		0.006		NS
H4e: Si → int	0.340 ***	6.964		0.136		S	0.340 ***	6.965		0.136		S
HM3e: EXP-ICMP x Fc → Use	−0.018	0.541		0.001		NS	−0.015	0.476		0.000		NS
HM2e: EXP-ICMP x Si → Int	−0.005	0.132		0.000		NS	−0.005	0.131		0.000		NS
HM1e: EXP-ICMP x Ef → Int	−0.014	0.350		0.000		NS	−0.014	0.350		0.000		NS
H6e: Eb → Use							0.255 ***	6.157		0.067		S
HM4e: Eb x Int → Use							0.105 ***	2.920		0.014		S

*** p < 0.01, ** p < 0.05. Decision (*): S = Supported, NS = Not supported. Note. Further details of the original and extended UTAUT models before and after the inclusion of expected net benefit (Eb) are highlighted (Figure A1, Figure A2, Figure A3, Figure A4 and Figure A5) in Appendix A.

.

From these results (Table 6), the acceptance statuses of the different hypotheses based on the original and extended models are summarized in

Table 7. Summary results of acceptance status of hypotheses based on original and extended models.

Hypotheses			Acceptance Status *
Num	Relationship	Description	Original UTAUT	Extended UTAUT
H1	H1a: INT → Use	Behavioural intention is a determining factor and positively influences the actual adoption behaviour for CT	S	S
	H1b: INT → Use	Behavioural intention is a determining factor and positively influences the actual adoption behaviour for OF	NS	NS
	H1c: INT → Use	Behavioural intention is a determining factor and positively influences the actual adoption behaviour for CRA	S	S
	H1d: INT → Use	Behavioural intention is a determining factor and positively influences the actual adoption behaviour for ISEED-AE	S	S
	H1e: INT → Use	Behavioural intention is a determining factor and positively influences the actual adoption behaviour for ICMP	S	S
H2	H2a: Pe → Int	Expected performance has a positive influence on the intention to adopt CT	S	S
	H2b: Pe → Int	Expected performance has a positive influence on the intention to adopt OF	S	S
	H2c: Pe → Int	Expected performance has a positive influence on the intention to adopt CRA	S	S
	H2d: Pe → Int	Expected performance has a positive influence on the intention to adopt ISEED-AE	S	S
	H2e: Pe → Int	Expected performance has a positive influence on the intention to adopt ICMP	S	S
H3	H3a: Ef → Int	Expected effort has a positive effect on behavioural intention in the adoption of CT	NS	NS
	H3b: Ef → Int	Expected effort has a positive effect on behavioural intention in the adoption of OF	S	S
	H3c: Ef → Int	Expected effort has a positive effect on behavioural intention in the adoption of CRA	NS	NS
	H3d: Ef → Int	Expected effort has a positive effect on behavioural intention in the adoption of ISEED-AE	S	S
	H3e: Ef → Int	Expected effort has a positive effect on behavioural intention in the adoption of ICMP	NS	NS
H4	H4a: Si → Int	Social influence has a significant and positive effect on behavioural intention to adopt CT	S	S
	H4b: Si → Int	Social influence has a significant and positive effect on behavioural intention to adopt OF	NS	NS
	H4c: Si → Int	Social influence has a significant and positive effect on behavioural intention to adopt CRA	S	S
	H4d: Si → Int	Social influence has a significant and positive effect on behavioural intention to adopt ISEED-AE	S	S
	H4e: Si → Int	Social influence has a significant and positive effect on behavioural intention to adopt ICMP	S	S
H5	H5a: Fc → Int	Facilitating conditions directly and positively influence the adoption of CT	S	S
	H5b: Fc → Int	Facilitating conditions directly and positively influence the adoption of OF	S	S
	H5c: Fc → Int	Facilitating conditions directly and positively influence the adoption of CRA	S	S
	H5d: Fc → Int	Facilitating conditions directly and positively influence the adoption of ISEED-AE	S	S
	H5e: Fc → Int	Facilitating conditions directly and positively influence the adoption of ICMP	S	S
H6	H6a: Eb → Use	Perceived net benefit directly affects and positively influences the adoption of CT	-	S
	H6b: Eb → Use	Perceived net benefit directly affects and positively influences the adoption of OF	-	S
	H6c: Eb → Use	Perceived net benefit directly affects and positively influences the adoption of CRA	-	S
	H6d: Eb → Use	Perceived net benefit directly affects and positively influences the adoption of ISEED-AE	-	S
	H6e: Eb → Use	Perceived net benefit directly affects and positively influences the adoption of ICMP	-	S
HM1	HM1a: EXP-CT × Ef → Int	Experience has a mediating effect on the relationship between the expected effort and BI to adopt CT	NS	NS
	HM1b: EXP-OF × Ef → Int	Experience has a mediating effect on the relationship between the expected effort and BI to adopt OF	NS	NS
	HM1c: EXP-CRA × Ef → Int	Experience has a mediating effect on the relationship between the expected Ef and BI to adopt CRA	NS	NS
	HM1d: EXP-ISEED-AE × Ef → Int	Experience has a mediating effect on the relationship between the expected Ef and BI to adopt ISEED-AE	NS	NS
	HM1e: EXP-ICMP × Ef → Int	Experience has a mediating effect on the relationship between the expected Ef and BI to adopt ICMP	NS	NS
HM2	HM2a: EXP-CT × Si → Int	Experience has a mediating effect on the relationship between the social influence and Int to adopt CT	NS	NS
	HM2b: EXP-OF × Si → Int	Experience has a mediating effect on the relationship between the social influence and Int to adopt OF	NS	NS
	HM2c: EXP-CRA × Si → Int	Experience has a mediating effect on the relationship between the Si and Int to adopt CRA	NS	NS
	HM2d: EXP-ISEED-AE × Si → Int	Experience has a mediating effect on the relationship between the Si and Int to adopt ISEED-AE	S	S
	HM2e: EXP-ICMP × Si → Int	Experience has a mediating effect on the relationship between the Si and Int to adopt ICMP	NS	NS
HM3	HM3a: EXP-CT × Fc → Use	Experience has a mediating effect on the relationship between the Fc and actual use behaviour in adopting CT	S	NS
	HM3b: EXP-OF × Fc → Use	Experience has a mediating effect on the relationship between the Fc and actual use behaviour in adopting OF	NS	NS
	HM3c: EXP-CRA × Fc → Use	Experience has a mediating effect on the relationship between the Fc and actual use behaviour in adopting CRA	S	S
	HM3d: EXP-ISEED-AE × Fc → Use	Experience has a mediating effect on the relationship between the Fc and actual use behaviour in adopting ISEED-AE	NS	NS
	HM3e: EXP-ICMP × Fc → Use	Experience has a mediating effect on the relationship between the Fc and actual use behaviour in adopting ICMP	NS	NS
HM4	HM4a: Eb × Int → Use	Expected net benefit has a mediating effect on the relationship between the Int to adopt and actual use behaviour in adopting CT	-	S
	HM4b: Eb × Int → Use	Expected net benefit has a mediating effect on the relationship between the Int to adopt and actual use behaviour in adopting OF	-	NS
	HM4c: Eb × Int → Use	Expected net benefit has a mediating effect on the relationship between the Int to adopt and actual use behaviour in adopting CRA	-	NS
	HM4d: Eb × Int → Use	Expected net benefit has a mediating effect on the relationship between the Int to adopt and actual use behaviour in adopting ISEED-AE	-	NS
	HM4e: Eb × Int → Use	Expected net benefit has a mediating effect on the relationship between the Int to adopt and actual use behaviour in adopting ICMP	-	S

Acceptance status (*): S = Supported, NS = Not supported.

.

5. Discussion and Policy and Practical Implications

5.1. Discussion

The present research examined data on farmers’ actual use behaviour for five agroecological practices (CT, OF, CRA, ISEED-AE and ICMP) by building an extended UTAUT based on the original UTAUT model, introducing the expected net benefit variable. Overall, the extended UTAUT model indicates that actual use behaviour for the studied agroecological practices is significantly and positively influenced by behavioural intention (H1), facilitating conditions (H5) and expected net benefit (H6), which is consistent with the original UTAUT model [15,16,95]. This confirms the hypotheses relating to behavioural intention (H1a, H1b, H1c, H1d, H1e), facilitating conditions (H5a, H5b, H5c, H5d, H5e) and expected net benefit (H6a, H6b, H6d, H6e). These results revealed that the behavioural intention towards technologies, the facilitating conditions for their implementation and the expectation of net benefit from their adoption are essential factors in the acceptance and use of these technologies. As part of their research, [96] reported a significant influence of behavioural intention on actual use and concluded that behavioural intention improves actual use behaviour. Several authors have reported that facilitating conditions (H5) are determining factors in the adoption of technologies [11,77,78,79,97]. The positive influence of facilitating conditions indicates that farmers with the best conditions are more likely to adopt technologies. This is consistent with the findings of previous studies, which have shown that the more favourable the farming environment, the higher the probability of technology adoption [16,98]. Therefore, the better the conditions (technical and organisational, material and financial) available to farmers, the greater the use of CT, OF, CRA, ISEED-AE and ICM practices.

It could be noted that farmers’ actual behaviour in adopting the main practices reflects well-considered actions and not spontaneous and random actions resulting from chance. As the adoption of these practices would improve their net benefits (productivity, soil quality and fertility, environmental protection), farmers have an incentive to strengthen and/or increase their current level of adoption of the various agroecological practices. This is supported by [64,65], who found that the propensity to adopt technologies increases with a positive perception of the expected net benefit of a given technology. This implies that the expected net benefit is a determining factor in the adoption of CT, OF, ISEED-AE and ICM practices. The significant and positive influence of behavioural intention suggests that it plays a decisive role in the adoption of actual use behaviour for the main practices studied. In other words, willingness is a crucial element for a technology to be accepted and adopted by farmers [26,99].

The R² shows that between 36.8% and 57.4% of variations in the actual use behaviour are explained by the variables in the extended UTAUT model, compared to 36.2% to 51.6% for the original UTAUT model. Overall, the results showed a slightly better explanatory power with the extended UTAUT model, suggesting that the introduction of the expected net benefit has increased the R² value compared to the original UTAUT model. This improvement implies that the final benefit (gain) is an important element in farmers’ decision-making on whether a technology will be accepted.

The expected performance and social influence positively affect the behavioural intention to adopt CT, CRA, ISEED-AE and ICM practices. These findings confirm the hypotheses (H2a, H2c, H2d, H2e; H4a, H4c, H4d) in which behavioural intention (H1) to adopt CT, CRA, ISEED-AE and ICM practices is directly affected and positively influenced by expected performance (H2) and social influence (H4), respectively. This indicates that the usefulness of technologies and the influence of surrounding people are key determinants of adoption intention [100]. This implies that the more farmers are mentally predisposed to adopt the technologies, have high performance expectations and have a high belief in surrounding people and their way of thinking, the more they will accept using these practices, along with practices’ acceptance being supported when implementation conditions are improved. These findings are in line with those obtained by [101], who found a positive influence of the expected performance and social influence to adopt on the post-recruitment practices among fruit farmers in Johor, Malaysia.

Behavioural intention to adopt organic fertiliser is positively influenced not only by the expected performance (H2) but also by the effort expectancy (H3), thus supporting hypotheses H2b and H3b. In contrast to previous practices, a reduction in the drudgery associated with learning and applying organic fertiliser was found to favour the behavioural intention to adopt organic fertiliser. These findings are in line with [49,55,102], which indicated that the less constraining the use of a technology, the higher its performance will be thanks to its use. Alternatively, social influence (H4b) did not have a significant effect on behavioural intention to adopt organic fertiliser. Similar results were presented in [103], which found no direct relationship between social influence and behavioural intention to adopt technologies.

Experience has a direct mediating effect on the relationship between social influence and behavioural intention to adopt ISEED-AE (HM2d), but also the relationship between facilitating conditions and the actual use behaviour in adopting CT (HM3a) and CRA (HM3c). This indicates that farmers rely on previous knowledge from their own learning or other technical expertise in dealing with the use of different technologies [97]. In addition to its direct influence mentioned earlier, the expected net benefit mediates the relationship between the facilitating conditions and the actual use behaviour for CT (HM4a) and ICM practices (HM4e), suggesting the crucial role it plays in the adoption of technologies [49,65].

When investigating the adoption rates of the different technologies, a significant difference was found between Kléla (cotton zone) and Cinzana (cereal zone), particularly for CT and CRA, where the difference was stressed (Table 3). Indeed, the adoption rate of CT (34%) was higher in the cotton zone compared to the cereal zone (4%). In contrast, the adoption rate of CRA was 8% in the cotton zone compared to 41% in the cereal zone. The highest rate of CT in Kléla may be explained by a better technical skill of farmers in that area compared to Cinzana. This is supported by [104], which reported that farmers in the cotton zone had better technical skill because of the technical support provided by CMDT (Compagnie Malienne pour le développement des textiles). Moreover, some of the topo-sequences in the cotton zone, despite the generally low slope, are more exposed to soil erosion. Hence, farmers being better skilled at coping with soil erosion [105] may explain the better rate of adoption in the cotton zone. This implies that training plays a critical role in technology adoption. Inversely, for CRA, Cinzana is a cereal-based production system, making this area favourable, as CRA is more dominated by a cereal production system in Mali [106]. Cereal-producing farmers intercropped cereal with leguminous crops to diversify their production and to satisfy their cash needs. This corroborates [107], which reported that in a cereal production system, with leguminous crops, farmers can generate cash income, diversifying into easily marketable crops such as cowpea, soybean and sesame. In such situations, farmers in Cinzana who do not produce cotton are mainly interested in crop diversification for their cash needs. These findings are particularly interesting in regard to the influences that agroclimatic, economic and/or cultural factors might have in agroecological practices’ adoption. Farmers in the two agroecological zone seem to choose practices that better benefit to them depending on their experience and skills. The role of other socio-demographic factors such as formal education and household size in agroecological practices’ adoption should be noted. Farmers in the cereal zone, even though they have benefited less from training programmes and are less organised than the cotton zone’s farmers, might have used their relatively better level of formal education and experience to improve the adoption rate of CRA. This conforms with [108], which found a positive influence of the farm manager’s level of formal education on technology adoption. The household size is considered to be a critical variable in technology adoption as it presents a source of labour supply. Several authors emphasised the role of household size in the adoption of sustainable technologies [109,110,111]. These results call for targeted policy interventions taking into account the specificities of each agroecological zone.

Based on these research findings, it should be noted that agroecology as a scientific discipline is relatively ‘young’ in Mali, and it has received little public action inspiring its development, apart from a few well-intentioned initiatives mentioned in the National Agricultural Policy act (Loi d’Orientation Agricole). There is a need to establish fair and equitable means of developing agroecological practices, without taking a mutually exclusive approach with conventional farming practices, which so far, have benefited from more attention and support from agricultural policies. This may involve better communication on the various (socio-economic and environmental) benefits of agroecological practices to farmers, and supporting farmers by facilitating the necessary implementation conditions (access to credit, technical and organisational capacity building).

5.2. Policy Implications

Very little work has been performed on agroecological practices’ adoption in Mali. This paper analysed the driving factors of acceptance and use of the five main agroecological practices. The findings of this research contribute to the debate on the issues of soil fertility declining, and agroecological transition. So far, public policies have focused on subsidising synthetic inputs as the main means of improving farmers’ access to inputs, but this type of programme has sustainability issues in terms of both its effectiveness and its efficiency. The government subsidy programme is making a huge contribution in financing the chemical inputs’ subsidy, without any guarantee of the financial impact in terms of a return on investment, or of the environmental impact that the implementation of this programme could have in terms of (human) pressure on new land due to cultivation. A research programme to assess the subsidy programme’s impact and the strategies for a better integration with agroecological practices could help to improve crop productivity and minimise environmental risks. This programme should be an integrated one involving the best environmentally friendly combinations of chemical input use and agroecological practices investigated (CT, OF, CRA, ISEED-AE, ICMP). This will allow for highlighting all the potential associated with these agroecological practices in terms of soil and crop protection, improving soil fertility and crop productivity. Such a measure could increase the agricultural performance, and by doing so, the rate of adoption of agroecological practices.

Improving facilitating conditions (easier access to credit and/or equipment, technical and organisational capacity building) through an implementation programme of agroecological practices would improve their level of use by farmers, which would reduce the cost of access to conventional inputs and, above all, help to protect sustainable land use. Greater communication on the benefits (socio-economic and environmental) of adopting agroecological practices at farmers’ level is needed for sustainable land management and to maintain the diversity of ecosystem services. This aim could be achieved by implementing educational programmes to train farmers on the economic and environmental benefits of agroecological practices. Economic or market-based incentive programmes that subsidise inputs or agroecological cash crops may encourage farmers’ behavioural intention and usage of agroecological practices. Improving farmers’ group organisation (social influence) is also required; this is a way to increase their access and reduce the information gap on the diverse actions initiated for the promotion of agroecological practices. Additionally, the role of digital technologies could be explored in promoting agroecological practices among smallholder farmers in Mali. Such measures could be expanded to other geographical contexts with similar socioeconomic conditions such as the new cotton zones of Bougouni and Kita (with less saturated land and higher rainfall) and the central and northern regions of Mali with a lower level of annual rainfall.

5.3. Practical Implications

Based on the findings, specific implications of each practice could be initiated to improve farmers’ actual use behaviour.

5.3.1. Contour Farming Techniques (CTs)

The expected net benefit has a significant and positive direct effect on the actual use behaviour and a mediating negative influence on the relationship between the behavioural intention and the actual use behaviour for CTs. This suggests the critical contribution of the perceived net benefit to farmers’ motivation to adopt CT practices. Raising awareness on the advantages of CTs (soil protection or erosion control, moisture conservation for crops, yield improvement) could be a way to increase farmers’ motivation and to improve their usage behaviour for CT practices.

5.3.2. Organic Fertiliser (OF)

The results of the extended UTAUT model confirm and improve on those of the original UTAUT model by showing a significant and positive influence of the expected net benefit on actual use behaviour for OF. This suggests that actions to promote the use of organic manure could provide additional motivation to increase the use of this practice. These actions could include setting up an institutional support programme, similar to the fertiliser subsidy, aimed at improving the availability of organic fertilisers at a reasonable cost. This programme should be focused on setting up processing units to treat and add value to local products (domestic waste, industrial by-products). In addition, building the technical capacity of farmers in livestock intensification techniques and agriculture/livestock integration could be a way of making more efficient use of local resources (better use of animal faeces and crop residues, land use, soil fertility management), with a view to improving the availability of fertilisers and agricultural productivity.

5.3.3. Crop Association (CRA)

The introduction of the expected net benefit did not produce any significant difference from the original UTAUT model but the results do not call into question its contribution to the extended model. Indeed, although it did not have a significant effect, the results showed a slight increase in the explanatory power with the extended UTAUT model, suggesting a trend towards improvement compared to the original UTAUT model. The overall results revealed that farmers’ actual use behaviour for crop association is influenced by their behavioural intention and facilitating conditions. This shows that the economic benefit is not the only factor that can lead to the adoption of a technology; this can also stem from a real willingness on the part of farmers, and above all, occurs when the conditions for adoption are met [26]. This is why it is necessary to create the conditions for strengthening farmers’ technical capabilities and also to step up communication on the other benefits and the role of crop association in increasing biological diversity and activity (natural enemies, soil flora, other functional elements and biodiversity) and agroecosystem services and functions (pest regulation, improved soil fertility). This can form part of the promotion of agroecology, aimed at increasing the diversification of cropping systems and protection of biological diversity [47,112].

5.3.4. Improved Seed Adapted to Agroecological Conditions (ISEED-AE)

The results show that the expected net benefit had a significant and positive influence on the actual adoption behaviour for improved seeds adapted to agroecological conditions. Given the high vulnerability associated with climatic variability, the promotion of disease-resistant seeds from cultivars adapted to agroclimatic conditions is among the most plausible options for boosting agricultural productivity [46,113,114]. This requires the development of a vast forward-looking research/extension programme on the creation of varieties adapted to changing climatic conditions, while taking care to maintain genetic diversity (making the most of local cultivars) in order to minimise the risks of varietal erosion. Better dissemination via extension services and facilitating physical and financial access (subsidies) will improve farmers’ willingness to adopt, which will in turn increase their actual adoption behaviour. To make this vision a reality, farmers need to be better organised, informed and trained about the benefits of using improved seeds adapted to agroecological conditions.

5.3.5. Integrated Crop Management Practices (ICMPs)

The introduction of the expected net benefit had a significant and positive influence on farmers’ actual behaviour in using integrated crop management practices. This shows that the expected gain associated with adoption is crucial in motivating farmers to adopt integrated crop management practices. With better knowledge of the benefits of ICM practices (reduction in pest density, cost reduction through rationalisation of pesticide use, agronomic pest control practices), farmers will be able to improve their actual level of use [115,116,117].

6. Conclusions, Limitations and Future Research Direction

6.1. Conclusions

This research analysed the factors influencing the adoption of agroecological practices by farmers in Mali. The agroecological practices were contour farming techniques (CTs), organic fertiliser (OF), crop association (CRA), improved seeds adapted to agroecological conditions (ISEED-AE) and integrated crop management practices (ICMPs).

The findings revealed that the expected performance and social influence positively influence behavioural intention to adopt the main agroecological practices studied. In addition, expected effort emerged as one of the factors influencing the intention to adopt organic fertiliser. The analyses showed a mediating effect of experience on the relationship between social influence and behavioural intention to adopt improved seeds adapted to the agroecological conditions. The results also revealed that the behavioural intention, facilitating conditions and the expected net benefit directly affect and positively influence the actual use behaviour for the five agroecological practices, except for crop association, where no significant difference was found for the expected net benefit. In addition, a mediating effect of experience was found on the relationship between facilitating conditions and the actual use behaviour for CT and CRA. Farmers will significantly improve their current level of use of agroecological practices when they benefit from better conditions (easier access to credit and/or equipment, technical and organisational capacity building).

Overall, the extended UTAUT model indicates a slightly better explanatory power compared to the original UTAUT model, suggesting that the expected net benefit plays a key role in explaining farmers’ actual use behaviour for the agroecological practices considered in the present research.

6.2. Limitations and Future Research Direction

This study has provided useful insights that will help to improve knowledge of the acceptance and use of agroecological practices by farmers. Although this study covered the major production systems (cotton and cereal zones), all agroclimatic situations in Mali should be investigated in the future. Extending the research to other agroecological zones is necessary to cover the diversity of the whole country, as agroecology is context-specific, depending on geographical, historical and social situations [118,119].

The policy recommendations and practical implications in the discussion section should be taken into account to promote agroecological practices to their full potential in improving productivity in line with environmental standards and norms.