Author: Donatien Ntawuruhunga1*, Enock Nyisingize1, Marie Therese Mukanyangezi2, Mathias Twahirwa3

East African University Rwanda Scientific Journal (EAURSJ)

Gender Differences in Climate-Smart Agroforestry Practices in Two of Rwanda’s Contrasting Agroecological Zones

Donatien Ntawuruhunga1*, Enock Nyisingize1, Marie Therese Mukanyangezi2, Mathias Twahirwa3

1Department of Business Administration, East African University Rwanda, P.O. Box 179, Nyagatare, Rwanda

2 Faculty of Business and Development Studies, Kibogora Polytechnic, P.O. Box 50, Nyamasheke, Rwanda

3 Faculty of Theology and Religious Studies, Protestant Institute of Arts and Social Sciences, P.O. Box 619, Huye, Rwanda

*Correspondence to: Department of Business Administration, East African University Rwanda, P.O. Box 179, Nyagatare, Rwanda. E-mail address: drdonatien@eaur.ac.rw  (Donatien Ntawuruhunga), ORCID: https://orcid.org/0000-0002-2945-129X

Abstract

While agriculture is a business, very little is known about gender differences in climate-smart agroforestry (CSAF) practices. The lack of understanding and documentation of gender dimensions and implications of CSAF in Rwanda could result in overlooking gender issues in CSAF interventions. This study aimed not only to address this knowledge gap but also to generate insights to inform gender-responsive CSAF interventions in Rwanda. Based on the survey conducted between April and September 2023, this study assessed the effects of gender differences on CSAF practices among 381 farmers in Bugesera and Rulindo AEZs. Data were collected through farm-level interviews and analyzed using descriptive statistics and logistic regression in Excel and R software. Results showed that most households and landholders were male farmers rather than female farmers, likely because, as in many parts of the world, women face particular disadvantages in accessing secure land rights. Also, farmers in the study areas practice CSAF using different approaches for various purposes. Additionally, the variance-covariance between gender characteristics and expected farmer CSAF adoption showed a positive covariance (0.01491228), implying that specific gender characteristics (such as female-headed households or specific roles) are positively associated with CSAF adoption and increase the likelihood of adopting these farming practices. For the cause-and-effect analysis, the study results showed that gender does not affect CSAF adoption in the target areas, suggesting that basic knowledge, training, and resources matter more than gender identity alone. An important implication is that policies should prioritize extension education and other tools rather than assuming adoption gaps are solely due to gender. Also, avoiding one-size-fits-all assumptions: policymakers must evaluate local data carefully, ensuring that a lack of gender disparity in one area does not mask other underlying socio-economic barriers.

Keywords: Climate-smart agroforestry; Agriculture business; Farmers’ resilience; Gender disparity; Logit regression

  1. Introduction

Agriculture provides most of the world’s food, creates employment opportunities, and helps eradicate poverty globally. In Sub-Saharan Africa (SSA), agriculture employs more than half of the workforce and therefore plays a critical role in poverty alleviation (Nchanji et al., 2022). However, SSA’s agriculture is significantly affected by climate change, resulting in prolonged dry spells, drought, heat waves, rising temperatures, cyclones, flooding, irregular rainfall patterns, and increased cases of pests and diseases, all of which affect overall agricultural production in the region (Nchanji et al., 2022).

Sub-Saharan Africa is the region worst hit by climate change, as a majority of its people depend on rain-fed agriculture as their primary source of income (Giller et al., 2021). In Rwanda, the economy is highly dependent on agricultural success (Prasad et al., 2016). About 80% of Rwanda’s population is directly or indirectly engaged in agriculture (Bank, 2015). Smallholder farmers account for 83.1% of Rwanda’s agricultural production and employ >70% of the labor force. Although poverty has declined in Rwanda over the past decade, the majority of those engaged in agriculture still earn incomes below the poverty line and remain food and nutritionally insecure (Prasad et al., 2016). The situation is often worse for female-headed households (Asante et al., 2010; Prasad et al., 2016). Thus, increasing agricultural productivity is key to the success of Rwanda’s economy and the well-being of its population (Prasad et al., 2016).

Women farmers are integral to climate-smart agroforestry (CSAF), often responsible for managing trees, especially during the early stages of establishment (Kiptot & Franzel, 2012). Women are also principal holders of knowledge and managers of traditional home gardens, and they make up about 60% of practitioners of innovative CSAF practices (Asante et al., 2010), such as the domestication of indigenous fruit trees and the production of dairy fodder. Thus, women’s knowledge of trees and genetic diversity, along with their roles as both suppliers and users of tree germplasm and genetic resources, makes them critical agents in scaling up CSAF practices to improve livelihoods (Meijer et al., 2015).

In Rwanda, climate change severely threatens smallholder farming, making climate-smart agroforestry (CSAF) vital for resilience. Because men and women hold distinct roles in labor, access to resources, and decision-making, adoption rates vary across contrasting agroecological zones like the semi-arid Bugesera and the mountainous Rulindo. Examining these gender dynamics is key to sustainable rural development. Throughout the literature, very little is known about gender and adoption of CSAF practices (Kiptot & Franzel, 2012), that is, ones in which perennial trees and shrubs are deliberately grown on the same land management system as annual crops and/or livestock. The lack of understanding and documentation of gender dimensions and implications of CSAF in Rwanda could result in overlooking gender issues in CSAF interventions. Our study aims not only to contribute to this knowledge gap but also to generate insights to address gender and CSAF issues in Rwanda. Based on the survey conducted between April and September 2023, we examined the challenges women face in adopting CSAF. Specifically, we investigated the following aspects: (i) gender differences in CSAF; and (iii) implications of gender differences on CSAF adoption. By investigating these research objectives, the study will develop strategies to address imbalances and promote gender equity in agricultural business.

  1. Theoretical framework

Analyzing the gender roles in CSAF adoption across contrasting agroecological zones in Rwanda—such as the semi-arid, drought-prone Bugesera district versus the high-altitude, erosion-prone northern areas like Rulindo—requires a robust theoretical framework combining “Feminist Political Ecology (FPE)”, the “Gendered Innovations / Access to Resources Framework”, and the “Technology Adoption Diffusion Model”.

Feminist political ecology is a feminist approach within political ecology that draws on various theoretical traditions, such as Marxism, post-structuralism, feminist geography, ecofeminism, and cultural ecology (Elmhirst, 2011). The theory uses intersectional frameworks within and between these traditions to explore ecological and political issues, including development, landscape, resource use, agrarian reconstruction, and rural-urban transformation. Its strength also lies in its capacity to direct attention to many of the most important questions of our age, such as poverty, social justice, the politics of environmental degradation and conservation, the neoliberalisation of nature and ongoing rounds of accumulation, enclosure and dispossession (Elmhirst, 2011). Scholars in this field argue that gender is an important factor in shaping access to, control over, and knowledge of natural resources (Owino, 2017), and that mainstream environmental policies often rely on gender-neutral assumptions that can overlook women’s knowledge of resource management. According to some scholars, such policies reflect a traditional decision-making framework that draws a binary distinction between “scientific knowledge” and “local experience”, leading to some development discourse presenting women as victims of environmental change rather than as political actors (Shayan & Juan, 2024).

The Gendered Innovations / Access to Resources Framework is a theoretical and analytical model that integrates sex, gender, and intersectional analysis into the scientific research, engineering design, and technology development processes (Schiebinger, 2014). Rather than just adding women to existing structures (“fixing the numbers”), it aims to “fix the knowledge” by treating gender analysis as a strategic resource for advancing equity, redefining structural barriers, and ensuring fair access to technological and social resources (Lindberg et al., 2015).

The Diffusion of Innovations theory, formulated by Everett Rogers in 1962, is the core theoretical framework explaining how new technologies spread through a culture or social system (Meryeme et al., 2025). It outlines five perceived attributes of an innovation, five distinct adopter categories, and a sequential decision-making process (García-avilés, 2020). The theory of diffusion of innovations research draws upon rational theories of organizational life adopted from sociology, management, and communication theory. It develops predictive accounts of the diffusion phenomenon that supposedly help technology implementers advance the dissemination of selected technologies. Overall, the diffusion of innovations tradition has sought to explain individual adoption decisions or intentions to adopt, which concern well-defined innovations and relatively homogeneous populations. The diffusion of innovation theory has mainly focused on the perceived features of technologies and the innovativeness of the organizations adopting them (Meryeme et al., 2025). Monsurat (2015) identifies five attributes of an innovation that influence its adoption: relative advantage over existing technologies, compatibility with the organization’s workflows and knowledge, implementation complexity, trialability, and observability of the innovation’s development both within the organization and among competitors. The individuals’ perceptions of these five characteristics predict the rate of adoption of innovations (Monsurat, 2015).

  1. Materials and methods

3.1. Study areas

Rwanda is a rural economy in East Africa (Ntawuruhunga et al., 2025a). Its major challenge is a growing population on fragile and eroding land, exacerbated by climate change risks (Republic of Rwanda, 2022). The study focused on two agroecological zones of Rwanda: Bugesera (eastern semi-arid savannah lowland zone) and Rulindo (temperate zone of the central highlands).

The lowland Bugesera is part of the eastern savannah semi-arid area between Latitude 1037′56″ S and 2013′9″ S and Longitude 29021′0″ E and 30018′0″ E. It has a surface area of 1337 square kilometers. The registered annual average temperature is between 26 and 290C. Its landscapes vary from 1,100 to 1,780 meters above sea level. Its annual precipitation varies between 700–1100 mm. Bugesera is part of the drier plains in eastern Rwanda, covered with savannah grasslands and scattered woody trees, notably acacia (Figure 1).

The upland Rulindo is part of the temperate zone of the central highlands between Latitude 1044’S and Longitude 29059’E. It is an upland area mostly characterized by mountainous landscapes. It has a surface area of 567 square kilometers. Its relief is formed of steep slopes with an elevation of 1470–2200 meters above sea level. It harbors food crop farming on hill slopes and vegetables in valleys, with woodlot trees of eucalyptus, some grevillea, and calliandra scattered on farmlands. The registered mean annual temperature is 190C., while the annual precipitation is 1243.3 mm (Figure 1).

This study employed a multi-stage stratified random sampling method to select the study areas. In Bugesera, soil fertility is depleted of major nutrients due to overexploitation, decreased organic matter, and high acidity (Mukashema et al., 2022). In Bugesera, agriculture is mostly rain-fed and subject to the caprices of meteorological patterns (Rwanyiziri and Rugema, 2013). Although endowed with abundant lakes, water is scarce for farming due to a lack of irrigation infrastructure and equipment to support needy farmers, especially in the dry season. Consequently, smallholder farmers have little means to increase productivity and shift out of poverty (ibid.). Climate change and concomitant trends in climate variability make agricultural production in this region unpredictable from one season to season. Droughts and extreme temperatures plague this area (Ntawuruhunga et al., 2025a). Rulindo is rural, with steep hills and valleys, and springs and rivers in canyons, which serve as traditional water sources (Ndolimana and Nahayo, 2021). Farmers in this area are predominantly vegetable growers (Ntawuruhunga et al., 2025a). Smallholder farmers in Rulindo grow staple and traditional crops, including climbing beans and maize, tea and coffee, and high-value horticultural and export crops, such as vegetables, spices, niche foods, and fruits and floriculture. Farmers rely on rainfall, while much of the water runs off the area’s mountainous landscapes, where floods are commonplace (Mutesi, 2024). As such, the two regions are vulnerable to food shortages thanks to dramatic climate stresses (Habiyaremye et al., 2022).

Figure 1. Map of Rwanda with study areas (Bugesera and Rulindo) (adapted after CGIS)

3.2. Sampling and data collection

This study employed a multi-stage stratified random sampling method to select the study areas, from which 381 farmers were sampled. In the first stage, we selected the study regions. We relied on the agroecological map of Rwanda divided into four major agroecological zones, according to Iiyama et al. (2018): the eastern plains, central plateau, highlands, and the area surrounding Lake Kivu. In this stage, we selected two zones (eastern plains and highlands) due to budget constraints, while considering their climatic disparities. In the second stage, two districts were selected from the two agroecological zones (Table 1), considering both heterogeneity and homogeneity in characteristics such as biophysical and climatic conditions, cropping patterns, water resources, irrigation systems, and livelihood conditions, among others. In the third stage, all the sectors (sub-administrative entities) were selected from each district. Then, in the fourth stage, we randomly selected 24 cells from each sector using stratified sampling. In this context, a cell serves as the fourth-level subdivision of Rwanda’s local administrative structure (Ntawuruhunga et al., 2025b). One cell may comprise several villages. In the fifth stage, fifty-eight to fifty-nine villages and forty-nine to fifty villages were randomly selected from each cell in Bugesera and Rulindo, respectively, using the Rwanda village statistics (NISR, 2023). In the last stage (the sixth stage), using the lists established by village leaders, we randomly selected three to four households from each village with no reference to farm characteristics. Using the Wonnacott and Wonnacott (1969) formula, a sample of 381 was obtained. 381 farmers were interviewed – 193 from Bugesera and 188 from Rulindo – from the sample drawn from the data published in the Rwanda Agricultural Household Survey Report 2020 (NISR, 2021). Before the study began, the questionnaire was pre-tested to improve survey quality and prevent the omission of important information necessary to achieve our research objectives. The questionnaire covered topics such as the local biophysical, climatic, and socio-economic characteristics as well as CSAF practices. The original questionnaire was built on an Open Data Kit (ODK) software uploaded on an Android mobile device (tablet) under the ODK Collect application. The built form on ODK included the Global Positioning System (GPS) capturing the farm coordinates, socio-economic information, bio-physical information, etc. Additional information was captured through field measurements using, e.g., a GPS area calculation application uploaded on an Android mobile phone for land area measurement, observations on the ground, and interviews with key informants (including local/village leaders, agriculture extension officers, and farmer groups).

Table 1. Sampled sites and size

Sites Households Farmer households Sample size (n) % sample
Bugesera 204,000 84,000 193 50.66
Rulindo 160,000 82,000 188 49.34
Total 364,000 166,000 381 100.00

We conducted interviews with household heads on their farms. We collected the data using a semi-structured questionnaire and personal field observations. In addition, we conducted three focus group discussions (FGDs) with 8–12 participants. The discussion sessions focused on understanding the gender roles in the practices and adoption of CSAF.

We also interviewed the local government (districts), the Albertine Rift Conservation Society (ARCOS), and the model farmers to complement our findings. Two interviews were conducted with local officials (districts), one interview with an ARCOS staff member, and two model farmers. These officials were contacted to acquire an overview of CSAF practices in the study areas.

We also used personal observation to collect information in the field. Using a diary and camera, we documented what we observed in the field about CSAF practices on the farms. This approach allowed us to observe and describe the CSAF structures and compare farmers’ utterances with our field observations (Castle et al., 2022). On farms, we directly observed agricultural practices, CSAF methods, and tree integration, while examining gender roles. We also gathered secondary data to complete the findings of this research.

3.3. Data analysis

This paper analyzed the data using descriptive and inferential statistics in Microsoft Excel and R software (Tokede et al., 2020). We applied descriptive statistics, covariance, and regression analyses. We used descriptive statistics to calculate frequency distributions and percentages to describe the respondents’ demographics and CSAF practice status and to determine the gender roles. Inferential statistics were also used. We used a binomial regression model to test the association between gender roles and CSAF adoption.

3.4. Model specification for gender roles on the adoption of CSAF

3.4.1. Correlation analysis

Pearson correlation coefficients were estimated to ascertain the relationships between gender characteristics and CSAF adoption. Correlations between variables were computed using the formula of Khanal (2015):

                                                                              (1)

Where:

  • and  are the sample means of  and ,
  •  and  are the sample standard deviations of  and , respectively.

3.4.2. Regression analysis

We performed a logistic regression model (Kabirigi et al., 2023), termed “logit,” to ascertain how predictor variables (independent) influence the outcome variable (dependent). According to Sperandei (2014), the logit regression analysis is performed to compute a statistic (odds ratio) that estimates how changes observed in the dependent variables (binary outcomes) are associated with changes in predictor variables. In this study, the outcome variable was the adoption of CSAF (1 = adopting or 0 = not adopting).  The input variable was farmers’ gender. The outcome variable ― the adoption of CSAF ― depicts heterogeneity in adoption among CSAF farmers (Kabirigi et al., 2023). The nature of the outcome variable imposed the kind of regression we had to perform (Ntawuruhunga et al., 2025a), and the data dictated the use of logistic regression. A logit regression can be binomial, ordinal, or multinomial (Kabirigi et al., 2023). Since the data used in this study show that the adoption of CSAF is a binary outcome, they were analyzed using a binomial logistic regression. In this respect, we coded the outcome variables as “1” and “0” (1 = adopting, 0 = not adopting). This coding was chosen as it leads to the most straightforward interpretation of results. It is worth noting that this approach enables the quantification of the model parameter perturbations affecting the probability of occurrence of a certain binary outcome (Morotti and Grandi, 2017). Employing the response variable “adopt CSAF”, the regression model is depicted as follows:

                             (2)

With:

                                                               (3)

Ceteris paribus, the parameter  represents the expected average change in the response variable for a one-unit change in the explanatory variable. Therefore, we model the adoption of CSAF as a function of gender characteristics.

Odds ratios are computed to report the strength of association between an event (response variable) and one or more input variables (predictor variables) (Kabirigi et al., 2023). The formula (Muhamadi and Boz, 2021) was used to compute the odds ratios for all predictor variables:

                                                                                            (4)

We transformed results into odds ratios (OR) reflecting the increase or decrease in odds associated ― ceteris paribus ― with the influence change of the explanatory variable (Norton et al., 2018). Its interpretation is that, all other predictor variables being held constant, how many times the likelihood for the farmer with a high level of CSAF adoption increases for a single predictor when the predictor is increased by one unit. Of note, odds ratios measure how strongly an outcome is associated with its predictor. These measures have been extensively used in research to analyze and interpret results from logistic regression models (Ohlmacher and Davis, 2003; Sprince et al., 2003; Jasinski et al., 2005; Rautiainen et al., 2009; McDonald, 2014).

One explanatory regression variable, notably gender characteristics, was used in this study for the binary logistic regression model. Various techniques are used to select variables that fit the model (Ntawuruhunga et al., 2025a). They include a stepwise approach that combines forward selection and backward elimination of predictor variables to be added or removed statistically, without disturbing the model prediction accuracy (Hosmer and Lemeshow, 2000). The reference group was female farmers.

  1. Results

4.1. CSAF farming in Bugesera and Rulindo

Figure 2 portrays the severity of soil erosion in these two separate AEZs (Bugesera & Rulindo).

  • (b)

Figure 2. Non-CSAF sites in research locations. Both sites are hit by severe soil erosion: (a) Bugesera, (b) Rulindo

4.2. Characteristics of sampled participants by gender and CSAF adoption

As shown in Table 2, most participants in Bugesera and Rulindo were male (35.2% and 31.5%, respectively). On the other hand, most participants did not adopt CSAF practices in Bugesera and Rulindo (29.4% and 26.8%, respectively).

Table 2. Gender and CSAF adoption among participants

Variable Sites Total
Bugesera Rulindo
Male 134(35.2%) 120 (31.5%) 254 (66.7%)
Female 59(15.5%) 68(17.8%) 127(33.3%)
Adopters 81(21.2%) 86(22.6%) 167(43.8%)
Non adopters 112(29.4%) 102(26.8%) 214(56.2%)

4.3. CSAF adoption patterns

Table 3 shows that farmer households in the study areas practice CSAF on farms using various techniques — home gardens, alley cropping, improved fallow, multipurpose trees, plantation-crop combination, shelterbelts/windbreaks, and silvopasture. Results showed that 18.9% of farmer households adopted home garden CSAF practices in Bugesera, whereas in Rulindo, the same proportion practiced shelterbelts/windbreaks CSAF. In contrast, in Bugesera, a small fraction (0.52%) of farmers adopted alley cropping as a CSAF practice, while in Rulindo, an equal proportion embraced silvopasture as their CSAF strategy. This result suggests that alley cropping is a less prevalent CSAF practice in Bugesera, and silvopasture is similarly adopted in Rulindo, albeit in a small segment of the farming community.

Table 3. Mean distribution of CSAF practices in study areas

  Bugesera

(n = 193)

Rulindo

(n = 188)

Total

(n = 381)

Mean (±SD)
Freq. (n) Freq. (n) Freq. (n)
Taungya – – – –
Home gardens 72 49 121 60.50(±16.26)
Alley cropping 2 36 38 19.00(±24.04)
Improved fallow 43 5 48 24.00(±26.87)
Multipurpose trees 34 13 47 23.50(±14.85)
Plantation crop combination 11 11 22 11.00(±0.00)
Shelterbelts/

Windbreaks

23 72 95 47.50(±34.65)
Silvopasture 8 2 10 5.00(±4.24)
Total 193 188 381 190.50(±3.53)

4.4. Variance-covariance of gender characteristics with expected farmer CSAF adoption

Variance-covariance of gender characteristics with expected farmer CSAF adoption showed a positive covariance (0.01491228).  A positive variance-covariance between specific gender characteristics (such as female-headed households or specific roles) and CSA adoption indicates that certain gender demographics move together with or increase the likelihood of adopting these farming practices (Table 4).

Table 4. Variance-covariance of gender characteristics with expected farmer CSAF adoption

Variable Gender Adoption
Gender 0.22280702 0.01491228
Adoption 0.01491228 0.24684349

4.5. Effect of gender demographics on CSAF adoption in study areas

We used a binary logistic regression model to assess the effects of gender demographics on CSAF adoption (Table 5). Using a predetermined 5% significance level, the results showed that gender demographics do not affect CSAF adoption in the study areas.

Table 5. Summarized logistic regression analysis results

Significant predictors OR SE Z P>|z|
Intercept 0.649 0.181 -2.377 0.017*
Gender 1.315 0.221 1.240 0.215

OR = odd ratio, SE = standard error

*significant at 0.05

  1. Discussion

This study investigated how gender differences impact the adoption of CSAF practices among rural farmers. As outlined in the introduction, very little is understood about gender differences in the adoption of CSAF practices, that is, ones in which perennial trees and shrubs are deliberately grown on the same land management system as annual crops and/or livestock. The lack of understanding and documentation of gender dimensions and implications of CSAF in Rwanda could result in overlooking gender issues in CSAF interventions. Gender is a vital determinant of agricultural success. Therefore, it must be investigated to understand CSAF adoption through a cross-site comparative analysis. Farmers’ CSAF adoption indicators were recorded as binary outcomes. A logistic regression model was used to determine the influence of gender differences on the adoption of CSAF practices in the study areas.

5.1. Characterization of surveyed farmer households

Results revealed that most respondents were male farmers (66.7 %), showing that the majority of households and landholders are male farmers compared to female farmers. This could be attributed to the fact that, as in many parts of the world, women are particularly disadvantaged in access to secure land rights. These results are consistent with FAO (2018), which emphasized that women are significantly disadvantaged compared to men regarding land rights. This situation comprises the whole picture of land rights, such as ownership, management, transfer, and economic rights. Accordingly, knowing the gender of landowners, farming practices, and adoption of the latest techniques in the context of climate change is important for identifying gender gaps in farming and how gender disparities affect agricultural productivity and rural livelihoods.

5.2. CSAF adoption and variance-covariance analysis

Results showed that farmers in the study areas practice CSAF with different approaches for various purposes — home gardens, alley cropping, improved fallow, multipurpose trees, plantation-crop-combination, shelterbelts/windbreaks, and silvopasture (Ntawuruhunga et al., 2025b). These sustainable farming approaches can significantly benefit farmers by boosting crop yields, increasing income, improving food security, and sustaining rural livelihoods (Buyinza et al., 2020). The variance-covariance of gender characteristics with expected farmer CSAF adoption showed a positive covariance (0.01491228). These results signify that a positive variance-covariance between specific gender characteristics (such as female-headed households or specific roles) and CSAF adoption moves together with or increases the likelihood of adopting these farming practices. Similar findings by Okumu et al. (2025), who found that access to agricultural information, input subsidies, loans and credit, and being a female-headed household increase the likelihood of adoption of CSA practices.  Also, Bogweh et al. (2025) emphasized that enhancing inclusive gender access to land and group-based approaches to information dissemination and capacity building would be relevant in enabling men, women, and young farmers to improve their adaptive and resilience capacities to climate change. From this perspective, gender dynamics should be considered in policy design and implementation of CSAF programs and policies to improve farmers’ resilience to climate change.

5.3. Effect of gender demographics on CSAF adoption

The results showed that gender demographics do not affect CSAF adoption in the study areas. In contrast, in their study examining whether there are significant gender differences in the adoption of CSAF practices in sub-Saharan Africa,  Hailemariam et al. (2024) found the opposite: that different factors are important in explaining the gender differential in the likelihood of CSA adoption. By extension, research on CSAF shows that while broad gender gaps often exist, certain practices and specific study contexts reveal that gender demographics do not significantly affect the adoption of particular techniques.

  1. Conclusion

In a small-sized country with high population pressure, small-scale farming involving climate-smart agroforestry (CSAF) is a viable strategy to sustain ecosystems and a mechanism to tackle poverty in rural areas. By adopting CSAF on farms, farmers can reap direct benefits regarding food and cash income, and receive considerable indirect environmental returns. To understand the influence of gender differences on farmers’ adoption of CSAF practices, we used a logistic regression model.

Findings indicate that most households and landholders are male farmers compared to female farmers, attributed to the fact that, as in many parts of the world, women are particularly disadvantaged in access to secure land rights. Also, farmers in the study areas practice CSAF with different approaches for various purposes — home gardens, alley cropping, improved fallow, multipurpose trees, plantation crop combination, shelterbelts/windbreaks, and silvopasture. Additionally, the variance-covariance of gender characteristics with expected farmer CSAF adoption showed a positive covariance (0.01491228). These results signify that a positive variance-covariance between specific gender characteristics (such as female-headed households or specific roles) and CSAF adoption moves together with or increases the likelihood of adopting these farming practices. Finally, the results showed that gender demographics do not affect CSAF adoption in the study areas, suggesting that basic knowledge, training, and resources matter more than gender identity alone.

The results of this study provide valuable insights that policies should prioritize extension education and other tools rather than assuming adoption gaps are solely due to gender. Also, avoiding one-size-fits-all assumptions: policymakers must evaluate local data carefully, ensuring that a lack of gender disparity in one area does not mask other underlying socio-economic barriers.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of Interest 

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data Availability

Data will be made available on request.

Ethics statement

Approvals from district mayors were received before fieldwork was undertaken, and informed consent was obtained from all participants in the study. All information was confidential and used for research purposes only. Anonymity was maintained using coded identifiers during analysis to guarantee all personal information would stay private.

Acknowledgments

The authors acknowledge support from local leaders in Bugesera and Rulindo Districts who permitted the study to be carried out in their jurisdictions, the district agronomists, extension workers, and farmers who spared their time to participate.

References

Bogweh, E., Agness, N., Catherine, N., Adolph, K., Victor, K., Ochieng, F., Mercy, O., Boaz, M., Cosmas, W., & Lutomia, K. (2025). Assessing gender disparities in farmers ’ access and use of climate ‑ smart agriculture in Southern Tanzania. Discover Sustainability, 6(337). https://doi.org/10.1007/s43621-025-01150-8

Buyinza, J., Nuberg, I.K., Muthuri, C.W., Denton, M. D. (2020). Assessing smallholder farmers’ motivation to adopt agroforestry using a multi-group structural equation modeling approach. Agroforestry Systems, 94, 2199–2211,. https://doi.org/https://doi.org/10.1007/s10457-020-00541-2

Castle, S.E., Miller, D.C., Merten, N., Ordonez, P.J., Baylis, K. (2022). Evidence for the impacts of agroforestry on ecosystem services and human well-being in high-income countries: a systematic map. Environmental Evidence, 11(10). https://doi.org/https://doi. org/10.1186/s13750-022-00260-4.

Elmhirst, R. (2011). Geoforum Introducing new feminist political ecologies. Geoforum, 42(2), 129–132. https://doi.org/10.1016/j.geoforum.2011.01.006

FAO. (2018). The gender gap in land rights. https://openknowledge.fao.org/server/api/core/bitstreams/4966d50c-233b-43a9-8fa7-8d43263dd082/content

García-avilés, J. A. (2020). Diffusion of Innovation. The International Encyclopedia of Media Psychology. https://doi.org/10.1002/9781119011071.iemp0137

Habiyaremye, C., Ndayiramije, O., D’Alpoim Guedes, J., Murphy, K. M. (2022). Assessing the Adaptability of Quinoa and Millet in Two Agroecological Zones of Rwanda. Frontiers in Sustainable Food Systems, 6. https://doi.org/https://doi.org/10.3389/fsufs.2022.850280

Hailemariam, A., Kalsi, J., & Mavisakalyan, A. (2024). Gender gaps in the adoption of climate- ­ smart agricultural practices : Evidence from sub- ­ Saharan Africa. March, 764–793. https://doi.org/10.1111/1477-9552.12583

Hosmer, D.W., Lemeshow, S. (2000). Applied Logistic Regression (2nd Editio). John Wiley & Sons, Inc. https://doi.org/10.1002/0471722146

Iiyama, M., Mukuralinda, A., Ndayambaje, J.D., Musana, B., Ndoli, A., Mowo, J.G., Garrity, D., Ling, S., Ruganzu, V. (2018). Tree-Based Ecosystem Approaches (TBEAs) as multi-functional land management strategies-evidence from Rwanda. Sustainability, 10(5), 1360. https://doi.org/https://doi.org/10.3390/su10051360

Jasinski, E., Morton, D., DeFries, R., Shimabukuro, Y., Anderson, L., Hansen, M. (2005). Physical Landscape Correlates of the Expansion of Mechanized Agriculture in Mato Grosso, Brazil. Earth Interactions, 9(16), 1–18. https://doi.org/10.1175/EI143.1

Kabirigi, M., Sekabira, H., Sun, Z., Hermans, F. (2023). The use of mobile phones and the heterogeneity of banana farmers in Rwanda. Environment, Development and Sustainability, 25, 5315–5335. https://doi.org/https://doi.org/10.1007/s10668-022-02268-9

Khanal, B. (2015). Correlation of climatic factors with cereal crops yield: a study from historical data of Morang district, Nepal. The Journal of Agriculture and Environment, 16. https://doi.org/https://doi.org/10.3126/aej.v16i0.19837

Kiptot, E., & Franzel, S. (2012). Gender and agroforestry in Africa : a review of women ’ s participation. Agroforestry Sy, 35–58. https://doi.org/10.1007/s10457-011-9419-y

Lindberg, M., Forsberg, L., & Karlberg, H. (2015). Gendered social innovation – a theoretical lens for analysing structural transformation in organisations and society Malin Lindberg, Lena Forsberg, Helena Karlberg. International Journal of Social Entrepreneurship and Innovation, 3(6), 472–483.

McDonald, J. H. (2014). Handbook of Biological Statistics (3rd ed.). Sparky House Publishing. McDONALD, J. H. Handbook of biological statistics

Meryeme, M., Cherqaoui, M., Issami, J., Meryeme, M., Cherqaoui, M., Issami, J., Innovation, T., & An, A. (2025). Determinants of Technological Innovation Adoption : An Overview of Modern Theories To cite this version : HAL Id : hal-04962090 Determinants of Technological Innovation Adoption : An Overview of Modern Theories . African Scientific Journal, 3(28), 617. https://doi.org/10.5281/zenodo.14904754

Monsurat, M. F. (2015). Perceived Attributes of Diffusion of Innovation Theory as a Theoretical Framework for understanding the Non-Use of Digital Library Services. Information and Knowledge Management, 5(9), 82–88.

Morotti, S., Grandi, E. (2017). Logistic regression analysis of populations of electrophysiological models to assess proarrythmic risk. MethodsX, 4, 25–34.

Muhamadi, S., Boz, I. (2021). Determinants of Rulindo tea farmers’ perception towards organic farming: a MASS::stepAIC() approach. Ciência Rural, 51(6), e20200695.

Mukashema, A., Ngoga, T. G., Nyandwi, E. Kanobana, M., Sihimbiro, F., Coroner, S. (2022). Spatial Distribution of Agricultural Land, Soil Fertility and Crop Suitability for Bugesera District. Technical report.

Mutesi, T. (2024). Knowing water better: Assessing water use and needs in Rulindo. FAO-Rwanda. https://www.fao.org/rwanda/news/detail-events/ar/c/1270875/

Nchanji, E. B., Kabuli, H., Nyamolo, V. O., Cosmas, L., Chisale, V., & Matumba, A. (2022). Gender di erences in climate-smart adaptation practices amongst bean-producing farmers in Malawi : The case of Linthipe Extension Planning Area. Frontiers in Sustanaible Food Systems, 6, 1001152. https://doi.org/10.3389/fsufs.2022.1001152

Ndolimana, E., Nahayo, L. (2021). Mapping Water Distribution in Grouped Settlements of Rulindo District, Northern Province of Rwanda. International Journal of Advances in Engineering and Management, 3(8), 1612–1627. https://doi.org/10.35629/5252-030816121627

NISR. (2021). Agricultural Household Survey 2020.

NISR. (2023). 5th Population and Housing Census. In National Institute of Statistics of Rwanda. https://www.dosm.gov.my/v1/index.php?r=column/cone&menu_id=bDA2VkxRSU40STcxdkZ4OGJ0c1ZVdz09

Norton, E.C., Dowd, B.E., Maciejewski, M. L. (2018). Odds Ratios-Current Best Practice and Use. Jama, 320(1), 84–85. https://doi.org/10.1001/jama.2018.6971

Ntawuruhunga, D., Estomii, E., Omari, H., John, R., & Lynch, K. (2025a). Farmers ’ knowledge , attitude , and motivation for adoption of climate-smart agroforestry in two contrasting agroecosystems of Rwanda. Trees, Forests and People, 19(December 2024), 100766. https://doi.org/10.1016/j.tfp.2024.100766

Ntawuruhunga, D., Estomii, E., Omari, H., John, R., & Lynch, K. (2025b). Impacts of climate-smart agroforestry practices on income and food security in two Rwandan agroecosystems. Scientific African, 29(June), e02785. https://doi.org/10.1016/j.sciaf.2025.e02785

Ohlmacher, G.C., Davis, J. C. (2003). Using multiple logistic regression and GIS technology to predict landslide hazard in northeast Kansas, USA. Engineering Geology, 69(3–4), 331–343. https://doi.org/https://doi.org/10.1016/S0013-7952(03)00069-3

Okumu, B., Radeny, M., & Cramer, L. (2025). World Development Sustainability Welfare effects and gender dimensions of adoption of climate smart agriculture practices : Evidence from Kenyan small-scale farmers ☆. World Development Sustainability, 6(November 2022), 100204. https://doi.org/10.1016/j.wds.2025.100204

Owino, M. D. (2017). Gender and Environmental Management : Who ’ s Role ? Who ’ s Responsibilities ? World Journal of Social Sciences and Humanities, 3(3), 61–63. https://doi.org/10.12691/wjssh-3-3-2

Rautiainen, R.H., Ledolter, J., Donham, K.J., Ohsfeldt, R.L., Zwerling, C. (2009). Risk factors for serious injury in Finnish agriculture. American Journal of Industrial Medicine, 52(5), 419–428. https://doi.org/https://doi.org/10.1002/ajim.20688

Republic of Rwanda. (2022). Environmental and Social Management Framework for Development of agroforestry and sustainable agriculture project under forestry investment programme (FIP) in Rwanda (ESMF).

Rwanyiziri, G., Rugema, J. (2013). Climate change effects on food security in Rwanda: Case study of wetland rice production in Bugesera District. Rwanda Journal, 1(1), 35–51.

Schiebinger, L. (2014). Gendered innovations : harnessing the creative power of sex and gender analysis to discover new ideas and develop new technologies. Schiebinger Triple Helix, 1(9), 0–17. https://doi.org//10.1186/s40604-014-0009-7

Shayan, T. R., & Juan, P. D. (2024). THE ROLE OF MASCULINITY AND FEMININITY CONSTRUCTS IN SHAPING PUBLIC DISCOURSE ON CLIMATE CHANGE. International Journal for Multidisciplinary Research, Review and Studies, 1(1).

Sperandei, S. (2014). Understanding logistic regression analysis. Biochemia Medica, 24(1), 12–18.

Sprince, N.L., Zwerling, C., Lynch, C.F., Whitten, P.S., Thu, K., Logsden-Sackett, N., Burmeister, L.F., Sandler, D.P., Alavanja, M. C. R. (2003). Risk factors for agricultural injury: a case-control analysis of Iowa farmers in the Agricultural Health Study. Journal of Agricultural Safety and Health, 9(1), 5–18. https://doi.org/10.13031/2013.12346.

Tokede, A.M., Banjo, A.A, Ahmad, A.O, Fatoki, O.A, Akanni, O. F. (2020). Farmers ’ Knowledge and Attitude towards the Adoption of Agroforestry Practices in Akinyele Local Government Area , Ibadan , Nigeria. Journal of Applied Sciences and Evironmental Management, 24(10), 1775–1780. https://doi.org/10.4314/jasem.v24i10.10

Wonnacott, T. H., Wonnacott, R. J. (1969). Introductory Statistics (5th Editio). John Wiley & Sons Inc.