Gender dynamics and remittances in the adoption of sustainable agricultural practices in Nepal Authors: Giri Prasad Kandel, Mustapha Yakubu Madaki, Tereza Pilarova, Ayat Ullah, Prajal Pradhan, Ioannis Manikas, Miroslava Bavorova Published: 2025 ABSTRACT Agricultural systems in rural Nepal face significant transformation due to climate change and shifting household labour dynamics. Male out-migration, a key but underexplored driver of this change, disrupts traditional gendered labor divisions and reshapes agricultural decision-making. In smallholder farms, traditionally, men are more responsible for tasks like ploughing and harvesting, whereas women take on planting, weeding, and winnowing roles. In the context of male out-migration, women must take primary responsibility for managing farms and households. However, persistent social and structural inequalities continue to constrain their decision-making authority. The feminisation of agriculture has important implications for adopting sustainable agricultural practices (SAPs). Using survey data from 400 households and Poisson regression analysis, this study examines the effects of migration, remittances, and female-managed farms on SAP adoption. Our results highlight that household farms having migrated member(s), receiving remittance and female-managed farms are more likely to adopt SAPs. In contrast, a higher number of out-migrating females negatively affects adoption, reflecting women’s critical role in sustainable farming adoption. Their participation in women’s groups, which provide training and financial resources and their management of tasks such as seed selection, winnowing, and organic pest control, are essential to SAP implementation. As such, our study provides a deeper understanding of the positive role of females in SAP adoption. We advocate for policies that recognising intersectional vulnerabilities, supporting women’s groups, lead to increased adoption of sustainable agricultural practices.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-31848-8. Introduction Agricultural sector is increasingly vulnerable to the growing impacts of climate change, which pose serious threats to productivity, food security, and rural livelihoods. Climate-driven weather extremes and seasonal shifts such as erratic rainfall, shifting monsoon cycles, prolonged droughts, and more frequent floods and landslides have disrupted traditional farming calendars and reduced crop yields across many regions1. These climatic shocks are further compounded by rising temperatures, which accelerate soil degradation, increase pest and disease outbreaks, and reduce water availability for irrigation2,3. In response to these escalating challenges, sustainable agricultural practices (SAPs) such as organic farming, conservation tillage, changing crops and planting dates, agroforestry, and integrated pest management have gained increasing attention for their potential to enhance ecological resilience and mitigate climate risks. Nevertheless, adoption of these practices is not merely a technical matter: it is deeply embedded in socio-economic and gendered labor dynamics and shifting household structures, particularly in the context of rural out-migration4. International labor migration has emerged as a defining livelihood strategy for rural households grappling with limited employment opportunities and climatic shocks5–7. Although agriculture remains the backbone of Nepal’s rural economy7,8, labor within the sector has traditionally been highly gendered. Men performed tasks such as ploughing, terrace bund preparation, and harvesting, while women were confined to less physically intensive and economically undervalued activities like planting, weeding, and winnowing9–11. In Nepal, our study region, as of 2023, approximately 3.5 million people were working abroad, primarily in Gulf countries7. Remittances from migrants account for approximately a quarter of the country’s GDP7,12. Due to a growing trend of male out-migration in several countries including rural Nepal7, it significantly reshaped traditional gender roles, with women assuming greater responsibilities in farm management and household decision-making, contributing to the ongoing “feminization of agriculture”10,13–17. While this shift offers pathways to women’s economic and social empowerment, it also increases their labor burden and imposes emotional and psychological stress due absence of male members in the household18–25. Despite women’s expanded roles, structural barriers such as land ownership, credit, training, extension services, markets, and male-dominated institutions may constrain their capacity to implement innovative agricultural practices1,11,15,26. These social structure, labor and environmental shifts are occurring alongside increased financial inflows from remittances, which bring opportunities and constraints. Adopting SAPs, which often requires technical knowledge, upfront investment, and long-term planning, is intricately linked to intra-household power relations and gendered control of resources5. The transformative potential of remittances is similarly uneven. While remittances can reduce liquidity constraints and enable investments in agriculture, their impact depends on who controls and allocates these resources. In many households, men influence spending decisions remotely, often prioritizing immediate consumption or construction over long-term agricultural improvements1,27. Even when women gain financial control, social norms and fear of reputational risks can limit their autonomy in deploying funds toward innovation. As a result, the benefits of sustainable agricultural practice from migration-driven financial inflows remain contingent on shifting gender roles, household bargaining, and social structures. Male out-migration has been shown to significantly affect agricultural labor availability, decision-making, and farm productivity in Nepal13,14,22,28,29. Research by14,28, and29, highlighted that male migration leads to feminization of agriculture however such household often face economic vulnerability. At the same time, women in these household exercise greater autonomy in household and agricultural decision-making. Even though migration reshaped gender roles, it intensifies food insecurity due to labor shortages for agricultural production22,30. This highlights household composition and its critical role in shaping women’s empowerment. Women in nuclear households are more likely to act as primary decision-makers, with greater control over agricultural inputs, improved access to credit, and higher levels of participation in farmer groups, all of which are key to successful SAP adoption10,31. In contrast, women in extended households often remain subject to hierarchical control, particularly from mothers-in-law, which limits their agency in deploying remittances, accessing resources, or experimenting with new farming techniques1,11. Women’s empowerment in areas such as asset control, legal rights, and Information and Communication Technologies (ICT) access significantly enhances women’s decision-making power and SAP adoption32–34. Interventions such as training35 and literacy programs36 further catalyze adoption and even reduce migration pressure. Yet, due to structural gaps adoption rates among women is lower for yield-enhancing techniques, such as agroforestry initiatives37,38. In Nepal, households where women are involved early in agricultural decisions are 19% more likely to engage with environmental conservation agriculture26. This highlights that prior studies have examined migration’s economic impacts13,14,22,28,29 and feminization of agriculture10,13–17, none have empirically tested how intra-household control of remittances mediates women’s ability to adopt SAPs. This gap is critical because remittance usage (e.g., housing vs. farm investment) may reinforce or offset gender disparities in SAP adoption, a dimension overlooked in existing literature. Thus, our study seeks to fill this gap by examining how migration and remittances influence the adoption of SAPs in Nepal, with a particular focus on gender dynamics. Specifically, our study is guided by the following research questions: Does male out-migration influence the adoption of sustainable agricultural practices (SAPs)?What gender-specific challenges or opportunities do women face in adopting SAPs within the context of migration-affected households?Do remittances affect adoption of SAPs? Does male out-migration influence the adoption of sustainable agricultural practices (SAPs)? Does male out-migration influence the adoption of sustainable agricultural practices (SAPs)? What gender-specific challenges or opportunities do women face in adopting SAPs within the context of migration-affected households? What gender-specific challenges or opportunities do women face in adopting SAPs within the context of migration-affected households? Do remittances affect adoption of SAPs? Do remittances affect adoption of SAPs? By linking migration, remittance use, and gendered decision-making, this research offers an integrated analysis of the structural and socio-economic forces shaping sustainable agriculture. It uniquely bridges three often-siloed domains: gender, migration/remittances, and SAPs contributing to more inclusive and resilient agricultural development strategies. Theoretical framework We employed combined framework drawing from the Sustainable Livelihoods Framework (SLF)39–41 with Intersectionality Theory42–46 to examine how gender dynamics, remittance flows and the feminisation of agriculture influence the adoption of sustainable agricultural practices (SAPs) in rural Nepal (Fig. 1). The SLF posits that household livelihoods are shaped by access to and interaction among five key forms of capital: human, natural, financial, social, and physical40. These capitals jointly determine a household’s ability to cope with and adapt to socio-environmental challenges. In rural Nepal, male out-migration interrupts this balance by reducing household labour (human capital) while augmenting financial capital through remittances47. Remittances may offer financial relief, but they often have unintended consequences. It raises managerial and physical burdens on women left behind, who often lack adequate institutional support and access to agricultural resources13,48. The feminisation of agriculture understood as a growing involvement of women in farm management in the absence of male household members1. It emerges not just as a demographic trend but as a deeper structural shift in rural livelihoods48. As women increasingly take charge of agricultural decisions and labour, their ability to adopt SAPs becomes pivotal at the household level. However, this capacity is shaped by different livelihood capitals and a set of demographic characteristics (age, education, and household size), farm attributes (farming systems, land tenure, and income levels), and institutional factors (access to credit or participation in farmer groups). Additionally, migration-specific variables such as the volume and stability of remittances, the gender composition of migrants, and household involvement in off-farm income-generating activities can either strengthen or constrain women’s role in agricultural decision-making13,48,49. By combining SLF with intersectionality, our study does not treat rural women as a homogenous category. Instead, it recognises that women’s capacity to adopt SAPs is shaped by the intersection of gender with other social identities such as caste, class, ethnicity and marital status. These overlapping identities influence their access to livelihood capitals and their position within intra-household and community-level power structures50. For instance, a female-headed household in a nuclear family may exercise greater decision-making autonomy than a woman in a multi-generational household where patriarchal norms, even if both receive remittances10,31. Moreover, looking at the intra-household perspectives, women are often responsible for domestic labour and caregiving, which limits their time, mobility, and participation in decision-making and income-generating activities. Such intersecting inequalities limit women’s bargaining power in agricultural markets and influence their ability to engage effectively in purchasing and selling agricultural goods. These dynamics affect women’s access to and control of key livelihood capitals (human, social, financial, natural, and physical), ultimately shaping their ability to adopt and sustain SAPs. Within this framework, adopting SAPs is not solely determined by access to resources or awareness, but is deeply embedded in gendered power relations and evolving livelihood strategies. By situating male out-migration within the broader context of the SLF and applying an intersectional lens, our study highlights how shifts in household composition reshape women’s roles in agriculture. Fig. 1Framework illustrating the intersection of male out-migration, gender dynamics, and the adoption of SAPs.Source: Authors own construction based on the SLF39,40 and concept of the feminisation of agriculture. Framework illustrating the intersection of male out-migration, gender dynamics, and the adoption of SAPs. Source: Authors own construction based on the SLF39,40 and concept of the feminisation of agriculture. Materials and methods Study zones and environmental features Our research was conducted in Nepal, a country in the broader Hindu Kush Himalayan (HKH) region, known for its sharp ecological gradients and diverse farming systems51. To capture this variation, we adopted a zonal framework based on elevation: the lowland Terai zone (67–610 m), the mid-hill zone (610–2,500 m), and the high mountain zone (2,500–8,848 m above sea level). Our data were collected from three districts representing each zone: Chitwan (Terai), Baglung (mid-hills), and Mustang (mountains) as shown in Fig. 2. These sites were selected to reflect variations in altitude, climate stress, migration patterns and farming practices. Each zone faces unique opportunities and constraints. The Terai offers fertile land and irrigation access, supporting cash crop production and livestock farming, but it is increasingly affected by floods, erratic rainfall, and labour shortages driven by migration52. Meanwhile, the mid-hills zone relies on terrace farming to manage sloped terrain, yet they suffer from landslides, limited access to credit, and declining labour availability5. The high mountain zone features short growing seasons, limited rainfall, and harsh winters, where households often depend on remittances, migration, and off-farm activities to maintain their livelihood5,52. Fig. 2Map of study zones.Source: Authors own construction using QGIS. Map of study zones. Source: Authors own construction using QGIS. Sampling technique and data collection Our study adopted a multistage, purposive sampling strategy across three agro-ecological zones: lowland (Terai), mid-hill, and high mountain regions. We selected one representative district from each zone – e.g. Chitwan (lowland), Baglung (mid-hill), and Mustang (high mountain). One rural municipality, Nepal’s smallest administrative unit was purposively chosen from each district based on farming communities. Within these municipalities, we selected specific villages as primary data collection sites. Lanku, Ratnanagar, and Sharadpur villages were selected from the Chitwan district, whereas Bhuskat, Hila, and Tara were chosen from the Baglung district. Meanwhile, we selected Thasang, Gharapojung, and Baragaun Muktikshetra from the Mustang district. A snowball sampling method was employed within purposively selected villages to identify smallholder farmers. This approach was appropriate for reaching dispersed farming households in remote areas where no comprehensive household lists were available. A total of 400 farmers were surveyed: 70 from Chitwan, 150 from Baglung, and 180 from Mustang districts. The sample distribution reflects the relative prevalence of smallholder farming across these regions. Because both site selection and farmer identification followed non-probability procedures, the sample should not be interpreted as statistically representative of all farming households in Nepal. Nonetheless, we conducted post-stratification checks using district-level data from the 2021 National Population and Housing Census of Nepal to evaluate the comparability of key demographic characteristics (Appendix 1, Table A1). For demographic variables such as age, household size, caste, and formal education, the sample broadly aligns with district averages. Data on farm income, remittances, credit access, and farmer group membership were unavailable at the district level. Perception-based variables (e.g., labor shortages, rainfall variability) are inherently context-specific and not directly comparable to census data. Data collection relied on a semi-structured questionnaire based on the study’s conceptual framework and existing literature1,22,53–55. Moreover, inputs from focus group discussions with local and local key informants helped to refine the questionnaires. We also conducted a pilot survey with 28 participants across the three zones: six in Terai, 10 in mid-hills, and 12 in mountainous to ratify the questionnaires. The survey was conducted during March and July 2021 by a team of 15 well-trained enumerators (five per zone). We conducted face-to-face interviews using printed questionnaires translated from English into Nepali language to ensure clear communication with rural farmers. To avoid any misconception, we obtained formal permissions from the municipality mayor or secretary before data collection. The initial entry points for the snowball sampling were identified through the secretary of the wards, local agricultural extension officers and farmer group leaders, who helped connect enumerators with the first respondents in each research site. In some scenarios, the initial contact with households was facilitated through village representatives. While this approach facilitated access to otherwise hard-to-reach households, it also posed challenges such as reliance on social networks, which may have limited diversity among referred participants. Our survey data covered various topics, including socio-demographics, farming, institutions, migration, and remittance information. According to national regulations in Nepal at the time of data collection, formal ethical approval was not required for social survey-based studies of this nature. Nevertheless, the study was conducted in line with internationally recognized ethical principles, including the Declaration of Helsinki. Prior to each interview, we informed participants about the objectives of the study, and their right to withdraw at any stage without consequence. Informed consent was obtained verbally, as many participants had limited literacy levels, making written consent less practical. All participants provided explicit verbal consent before participation. No minors were interviewed for this study. Data analysis Poisson regression model This study used Poisson regression to analyze the effect of migration, remittances and gender dynamics on the adoption of sustainable agricultural practices, since the dependent variable is counting data. Specifically, the outcome variable represents the number of sustainable agricultural practices adopted by each household make the Poisson regression most appropriate for the analysis56,57. Let \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Y}_{i}$$\end{document}​ represent the count of practices adopted by households \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:i$$\end{document}. The Poisson regression model assumes: 1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Y}_{i}\sim Poisson\:\left({\lambda\:}_{i}\right)$$\end{document}​2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:Pr\:\left({Y}_{i}=yi\right)={\varepsilon\:}^{\_{\lambda\:}_{i}\:}{\lambda\:}_{i}^{yi}/{yi}^{!}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{i}$$\end{document}​ is the expected count, expressed as a log-linear function of a vector of covariates \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{X}_{i}$$\end{document}:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:log\:\left({\lambda\:}_{i}\right)={\beta\:}_{0}+{\beta\:}_{1}{X}_{1i}+{\beta\:}_{2}{X}_{2i}+..\:.\:\:\:\:\:\:+{\beta\:}_{k}{X}_{ki}+\varepsilon\:$$\end{document}4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:{\lambda\:}_{i}=\mathrm{e}\mathrm{x}\mathrm{p}\left({\beta\:}_{0}+\sum\nolimits_{k=1}^{k}{\beta\:}_{k}{X}_{ki})\right)$$\end{document} The model was estimated using Maximum Likelihood Estimation (MLE), and the results are presented in terms of coefficients, incident rate ratios (IRRs), and marginal effects for robustness checks. Selection of variables Our study incorporates three categories of variables: outcome, variable of interest and control (see Table 1). The outcome variable is the adoption of SAPs, measured through a range of strategies such as crop diversification, changes in planting time or crop type, early maturing and drought-resistant varieties, irrigation, rainwater harvesting, reduced tillage, mulching, agroforestry, use of organic fertilizers, and integrated pest management system. We considered as SAP adoption if one of these strategies was implemented by households in the last 10–15 years. The variables of interest include female-managed farms, household migration status, remittances, and the number of female migrants to capture household labour dynamics and gendered decision-making. Control variables account for demographic, farming, environmental, and institutional factors. These include age, caste, education, marital status, household size, off-farm work, farm income, land size, farming type, labour availability, erratic rainfall, access to credit, and farmer groups membership. Variable selection is informed by the study’s conceptual framework and supported by prior research on migration, gender, and sustainable agricultural practices1,22,53–55. Table 1Definition of variables (n = 400).VariablesDefinitionsMeanStd. Dev. Dependent variable Sustainable agricultural practicesNumber of sustainable agricultural practices adopted by household (total SAP = 10)5.792.21 Independent variables Migration and remittances (variables of interest) Female managed farm1 if farm is managed by female, 0 otherwise0.280.45Migration1 if household has migrated members, 0 otherwise0.350.47Remittances1 if the household receives remittances from migrated members, 0 otherwise0.290.45Female migrantsTotal number of female migrants from the households1.072.01 Demographic characteristics AgeYears of household head50.3113.99Higher caste1 if household belongs to ethnic groups “so-called” higher caste, 0 otherwise0.240.42Formal education1 if household head has formal education, 0 otherwise0.650.47Household sizeTotal number household members5.802.88Marital status1 if household head is married, 0 otherwise0.850.35Off-farm activities1 if any members of the household are involved in the off-farm activities, 0 otherwise0.570.49 Farm characteristics Farming practices1 if household practices mix farming system, 0 otherwise0.880.31Cash cropsTotal percentage of cultivated land for cash crops such as potato production31.2722.24LivestockTotal number of livestock such as cattle of the household as major livestock1.402.15Farm incomeScale (1–5), total household earnings from farms, 1 being earning less than 50,000 NPR1 annually to 5 being earning more than 200,000 NPR11.760.99Cultivated landTotal size of cultivated land (Ropani2)9.8815.54Land sizeTotal land size owned by household in Ropani13.1617.86Lack of labor1 if households perceive lack of labor for agricultural work, 0 otherwise0.790.40Erratic rainfall1 if farmers perceived negative impact of erratic rainfall on their farm over the past 10–15 years, 0 otherwise0.860.34 Institutional characteristics Access to formal credit1 if household has access to formal credit sources such as bank, 0 otherwise0.030.19Farmer groups1 if household is member of farmer groups, 0 otherwise0.670.47Note: 1NPR= Nepalese rupee (Currency of Nepal), 1$ = 120 NPR as of May 2021. 2Ropani = A unit of the area measured in Nepal, 1 Ropani = 0.051 hectare. Table 1Definition of variables (n = 400).VariablesDefinitionsMeanStd. Dev. Dependent variable Sustainable agricultural practicesNumber of sustainable agricultural practices adopted by household (total SAP = 10)5.792.21 Independent variables Migration and remittances (variables of interest) Female managed farm1 if farm is managed by female, 0 otherwise0.280.45Migration1 if household has migrated members, 0 otherwise0.350.47Remittances1 if the household receives remittances from migrated members, 0 otherwise0.290.45Female migrantsTotal number of female migrants from the households1.072.01 Demographic characteristics AgeYears of household head50.3113.99Higher caste1 if household belongs to ethnic groups “so-called” higher caste, 0 otherwise0.240.42Formal education1 if household head has formal education, 0 otherwise0.650.47Household sizeTotal number household members5.802.88Marital status1 if household head is married, 0 otherwise0.850.35Off-farm activities1 if any members of the household are involved in the off-farm activities, 0 otherwise0.570.49 Farm characteristics Farming practices1 if household practices mix farming system, 0 otherwise0.880.31Cash cropsTotal percentage of cultivated land for cash crops such as potato production31.2722.24LivestockTotal number of livestock such as cattle of the household as major livestock1.402.15Farm incomeScale (1–5), total household earnings from farms, 1 being earning less than 50,000 NPR1 annually to 5 being earning more than 200,000 NPR11.760.99Cultivated landTotal size of cultivated land (Ropani2)9.8815.54Land sizeTotal land size owned by household in Ropani13.1617.86Lack of labor1 if households perceive lack of labor for agricultural work, 0 otherwise0.790.40Erratic rainfall1 if farmers perceived negative impact of erratic rainfall on their farm over the past 10–15 years, 0 otherwise0.860.34 Institutional characteristics Access to formal credit1 if household has access to formal credit sources such as bank, 0 otherwise0.030.19Farmer groups1 if household is member of farmer groups, 0 otherwise0.670.47Note: 1NPR= Nepalese rupee (Currency of Nepal), 1$ = 120 NPR as of May 2021. 2Ropani = A unit of the area measured in Nepal, 1 Ropani = 0.051 hectare. Definition of variables (n = 400). Note: 1NPR= Nepalese rupee (Currency of Nepal), 1$ = 120 NPR as of May 2021. 2Ropani = A unit of the area measured in Nepal, 1 Ropani = 0.051 hectare. Model robustness checks and diagnostics summary Several diagnostic tests were performed to validate the model and ensure the reliability of the coefficient estimates and other parameters. The overdispersion test (goodness-of-fit chi² = 252.75, p = 1.000) indicates no significant overdispersion issue, supporting the appropriateness of the Poisson model for the count data (Appendix 2, Table A2). Although a Negative Binomial regression was not strictly required since the overdispersion test indicated no significant overdispersion, we estimated it as an additional robustness check. The results were consistent with the Poisson model in terms of coefficient signs and significance, confirming the robustness of our main findings (Appendix 5 Table A5). The Variance Inflation Factor (VIF) was used to assess multicollinearity among the independent variables. The VIF values for all the covariates were below 10 (Appendix 3, Table A3), indicating that multicollinearity is not a concern in the model58,59. This confirms that the explanatory variables are not linearly dependent and that the regression coefficients are stable. To evaluate potential model misspecification due to omitted variables or incorrect functional form, the link test for model specification was employed (Appendix 4, Table A4). The test regresses the dependent variable on the predicted value (_hat) and its square (_hatsq). In a correctly specified model, _hat should be significant while _hatsq should not. Results showed that the _hatsq term was not statistically significant at p > 0.05. This indicates that the model is well-specified, with no strong evidence of misspecification or omitted variable bias60,61. Although the link test indicates correct model specification, we acknowledge that potential endogeneity due to unobserved factors beyond the observed controls such as farmer groups membership and access to formal credit may still be present and is recognized as a limitation of the study. Furthermore, a zero-inflated Poisson (ZIP) model was estimated as a robustness check. The ZIP model shows a slightly higher log pseudolikelihood (–833.09) than the standard Poisson model (–836.29), indicating only a marginal improvement in fit (Appendix 6, Table A6). However, given that the dataset contained only one zero count and the overdispersion test did not indicate significant dispersion issues. As so zero-inflated or hurdle models were deemed unnecessary. Sample description On average, the farmers adopt approximately six practices at a time, with a standard deviation of 2.22, this indicates moderate but varied uptake of SAPs across the sample (Table 1). Migration and remittances are central to this study, about 28% of farms are managed by women, often due to male out-migration. These gender-specific roles and constraints may influence the choice and extent of SAPs adoption. Additionally, 35% of households have at least one migrated member, which may impact labor supply and alter household decision-making regarding farm management and innovation. About one-third of the households receive remittances (29%), these financial inflows can support SAP adoption by easing liquidity constraints or substitute agricultural engagement entirely. Households report an average of one female migrant, with a standard deviation of 2.02. This reflects gendered patterns in migration and potential shifts in household farm labor dynamics. Regarding the demographic characteristics of the sample, the average age of household heads is 50 years, with a standard deviation of 14 years, indicating a mature farming population that may be experienced and risk-averse when adopting new practices. About one-fourth of respondents belong to higher caste groups, which could influence access to resources and networks that facilitate SAPs adoption. Two-thirds of the household head have formal education, a factor that demonstrates greater openness to and understanding of sustainable practices. Households have an average of six members, with a standard deviation of three, which could reflect labor availability for farm work. The majority (85%) of household heads are married, which may indicate responsibility in household decision-making. More than half (57.8%) of households have at least one member engaged in non-agricultural employment, which could influence farm investment and resource allocation. In terms of farm characteristics, a large proportion (88.5%) of households practice mixed farming systems, which could improve the integration of various SAPs. On average, 31% of cultivated land is used for cash crops such as potatoes, suggesting a market orientation that could affect input choices and land use decisions. Households keep an average of one livestock animal, which may contribute to nutrient cycling and influence practices such as the use of organic manure. On a scale of 1 (earning below 50,000 NPR annually) to 5 (earning above 250,000 NPR annually), the average farm income score is 1.76, suggesting relatively low-income levels from farming activities. The average cultivated land area is 9.89 Ropani (approximately 0.5 hectares), and the average total land ownership is 13.16 Ropani (approximately 0.67 hectares), indicating generally small-scale farming. Notably, 79% reported a lack of labor as a constraint in agricultural work, likely due to out-migration or youth disengagement, and 86% perceive that erratic rainfall has negatively affected their farms over the past 10–15 years. This indicates strong climate vulnerability and may motivate the adoption of resilient practices. Only 3.7% of households have access to formal credit sources, such as banks, revealing major financial constraints that could hinder SAPs adoption. Two-third (67%) of households belong to farmer groups, which may enhance information dissemination and peer learning, supporting broader SAPs implementation. Results We implemented several statistical measures such as Pseudo R2, chi2 statistics, the Akaike Information Criterion (AIC), the probability associated with the chi-square test, and the Bayesian Information Criterion (BIC) to evaluate the overall performance of the regression model (Table 2). The model’s chi2 statistic was 181.515, with an associated probability value (prob > chi2) of 0.000. This indicates that the model is statistically significant and confirms that the set of explanatory variables included in the model significantly improve fit compared to an intercept-only model. Whereas the Poisson regression results offer important insights into how migration dynamics, farm characteristics, and institutional factors influence the adoption of SAPs in Nepal. Effect of migration, remittances, and gender dynamics on the adoption of SAPs Table 2 display the Poisson regression result of effect of male out-migration on the adoption of SAPs, with regards to the variable of interest. Our results shows that female-managed farm has a positive and statistically significant effect on SAPs adoption. The marginal effect (0.425) implies that when a household is managed by a female, the number of SAPs adopted increases by approximately 0.43 practices, holding other variables constant. Migration status also positively affects SAPs adoption of the household, with a marginal effect of 0.609. Households with at least one migrant member adopt around 0.61 more practices compared to households without migrant members. Remittances exhibit a strong positive and significant effect with SAPs adoption. The marginal effect of 1.25 indicates that households receiving remittances adopt, on average, 1.26 more SAPs. To further explore the nature of remittance flows, Fig. 3 categorizes them into four types: money, farm inputs, household items, and food items. Respondents rated the frequency of receiving each type on a five-point Likert scale: never, rarely, sometimes, often, and very often. The results indicate that the majority of households often receive money, making it the most common form of remittance. The direct farm input, by contrast, were largely never received. Household items such as clothing and electronics were sometimes received, while food items were generally received rarely or not at all. Fig. 3Frequency of distribution of remittances by types. Frequency of distribution of remittances by types. To complement the analysis of remittance types, we further examined their perceived importance across various household expenditure categories. Figure 4 presents how respondents rated the importance of remittances for different uses, using a five-point Likert range ranging from not at all important to extremely important. The results indicate that house construction is considered the most important use of remittances, with the highest proportion of respondents rating it as extremely important. This is closely followed by migration costs and repayment of debts, both of which were also widely perceived as very to extremely important. Spending on health and education is also ranked high, with a significant share of respondents identifying them as very or extremely important. In contrast, agricultural investments such as seeds, pesticides, and agricultural tools were generally viewed as less important. A substantial proportion of respondents rated these categories as neutral or slightly important, indicating a relatively lower priority. Food fell in the mid-range, with more varied responses, reflecting mixed levels of perceived importance. Fig. 4Perceived importance of remittance across various household expenditure categories. Perceived importance of remittance across various household expenditure categories. Another important variable of interest in our study was number of female migrant members in the household. This is negatively and significantly associated with SAPs adoption. More specifically, a unit increase in the number of female migrants leads to a reduction of 0.40 SAPs adoption. Regarding demographic characteristics, household size positively and significantly affects SAPs adoption. Each additional member contributes to the adoption of 0.14 more practices. Off-farm activities are positive and significantly associated with adoption of SAPs. The marginal effect is 0.58, implying households involved in non-farm work adopt 0.58 more practices. Table 2Effect of migration and remittances on sustainable agricultural practices (n = 400).VariablesCoefficientStandard errorIncident rate ratioStandard errorMarginal effectStandard error Migration and remittances (variables of interest) Female managed farm0.0730.038*1.0760.041*0.4250.220*Migration0.1050.058*1.1110.06*0.6090.338*Remittances0.2170.085**1.2420.106**1.2550.496**Female migrants−0.0690.022***0.9330.02***−0.4010.128*** Demographic characteristics Age0.0000.0011.0000.0010.0010.019Higher caste0.0510.0371.0520.0390.2950.001Formal education0.0380.0411.0380.0430.2180.001Household size0.0240.011**1.0240.011**0.1360.019**Marital status0.0260.0441.0270.0790.1520.001Off-farm activities0.1000.037***1.1050.041***0.5790.211*** Farm characteristics Farming practices0.2230.063***1.2500.079***1.2910.365***Cash crops0.0030.001***1.0030.001***0.0170.004***Livestock−0.0270.009***0.9740.009***−0.1550.054***Farm income0.0400.018**1.0410.019**0.2340.104***Cultivated land0.0040.001***1.0040.001***0.0200.007***Land size0.0010.0011.0010.0010.0040.004Lack of labor−0.0210.0430.9790.042−0.1220.249Erratic rainfall0.1280.055**1.1360.062**0.7400.316** Institutional characteristics Access to formal credit0.0380.0691.0380.0720.2190.402Farmer groups0.0720.038*1.0740.04*0.4150.219*Constant0.9130.143***2.4910.357***--Pseudo R20.0520.052Chi-square181.515181.515Akaike criteria (AIC)1714.5761714.576Prob > chi20.0000.000Bayesian criteria (BIC)1798.3961798.396Note: ***, **, and * represent significance at 1%, 5%, and 10% probability levels, respectively. Table 2Effect of migration and remittances on sustainable agricultural practices (n = 400).VariablesCoefficientStandard errorIncident rate ratioStandard errorMarginal effectStandard error Migration and remittances (variables of interest) Female managed farm0.0730.038*1.0760.041*0.4250.220*Migration0.1050.058*1.1110.06*0.6090.338*Remittances0.2170.085**1.2420.106**1.2550.496**Female migrants−0.0690.022***0.9330.02***−0.4010.128*** Demographic characteristics Age0.0000.0011.0000.0010.0010.019Higher caste0.0510.0371.0520.0390.2950.001Formal education0.0380.0411.0380.0430.2180.001Household size0.0240.011**1.0240.011**0.1360.019**Marital status0.0260.0441.0270.0790.1520.001Off-farm activities0.1000.037***1.1050.041***0.5790.211*** Farm characteristics Farming practices0.2230.063***1.2500.079***1.2910.365***Cash crops0.0030.001***1.0030.001***0.0170.004***Livestock−0.0270.009***0.9740.009***−0.1550.054***Farm income0.0400.018**1.0410.019**0.2340.104***Cultivated land0.0040.001***1.0040.001***0.0200.007***Land size0.0010.0011.0010.0010.0040.004Lack of labor−0.0210.0430.9790.042−0.1220.249Erratic rainfall0.1280.055**1.1360.062**0.7400.316** Institutional characteristics Access to formal credit0.0380.0691.0380.0720.2190.402Farmer groups0.0720.038*1.0740.04*0.4150.219*Constant0.9130.143***2.4910.357***--Pseudo R20.0520.052Chi-square181.515181.515Akaike criteria (AIC)1714.5761714.576Prob > chi20.0000.000Bayesian criteria (BIC)1798.3961798.396Note: ***, **, and * represent significance at 1%, 5%, and 10% probability levels, respectively. Effect of migration and remittances on sustainable agricultural practices (n = 400). Note: ***, **, and * represent significance at 1%, 5%, and 10% probability levels, respectively. In terms of farm characteristics, mixed farming practices significantly enhance SAP adoption. More specifically there was a high marginal effect of 1.29. This indicates that households practicing mixed-farming adopted 1.29 more SAPs. Cash crop cultivation shows a small but significant positive impact, with each additional percentage of land under cash crops increasing SAPs by 0.02. The households with livestock holdings have a significant negative association, reducing SAPs adoption by 0.15 practices per unit increase in livestock such as cattle and buffalo. Farm income, cultivated land, and erratic rainfall perception are all positively and significantly associated with SAPs by 0.23, 0.02 and 0.74 respectively. Farmer groups membership has a positive and marginally significant effect, increasing SAP adoption by 0.41. Discussion Female-managed farms and adoption of SAPs Our study results indicate that female-managed farms are positively and significantly associated with the adoption of SAPs. Specifically, households where women are identified as the primary farm managers adopt more SAPs. This finding underscores the crucial role women play in promoting environmentally sustainable farming in rural Nepal. In the study area, female farmers, especially in smallholder contexts, often manage diverse and multifunctional farms. Their decision-making tends to prioritize household food security, soil conservation, and long-term productivity over short-term yield maximization. Several studies in South Asia and Sub-Saharan Africa have shown that women are more inclined than men to adopt sustainable practices such as intercropping, organic inputs, and crop diversification34,62–64. Additionally, female farm managers may be more likely to adopt SAPs out of necessity, particularly in households facing male out-migration or limited access to credit or mechanization65. The practices like composting, mulching, and crop rotation are often low-cost, labor-based strategies that align well with women-managed plots together with their household activates66. In this context, SAPs are not just environmentally sound they are also economically rational for women operating under resource constraints67. Another plausible reason is female farmers often participate actively in female originated and other informal community networks (e.g., mothers’ group, seed exchanges, local cooperatives, savings groups), which serve as key channels for disseminating sustainable agricultural practices68. Although formal extension services may underserve women, these peer-based systems often compensate for institutional gaps and encourage mutual learning69. Despite their positive impact, female-managed farms may still face systemic constraints. Access to land, extension services, training, and inputs tends to be lower for women compared to men, even when they are de facto or de jure household heads70,71. Thus, the observed positive association may be even stronger in contexts where these institutional inequalities are reduced. Moreover, in Nepal, the increasing feminization of agriculture due to male migration is often not accompanied by a matching shift in institutional recognition or support for female farmers. While many women are assuming managerial roles on the farm, their legal and social recognition as “farmers” remains weak72,73. This disconnect can limit their access to subsidies, training, or credit programs intended to support sustainable agriculture. Rural out-migration and adoption of SAPs In terms of rural out-migration, households with at least one migrant member tend to adopt more SAPs compared to their counterparts. This implies that the benefits of migration extend beyond economic contributions through remittances. Migration can facilitate behavioral change and strategic innovation within rural households. Specifically, migration can reshape household decision-making processes by increasing access to different knowledge systems, changing risk perceptions, and encouraging forward-looking investment behavior. Given that most Nepali migrants are employed in Gulf countries or any other countries, often within the agriculture or food sectors, there is potential for the transfer of agricultural knowledge, skills, or exposure to new practices to their households back home. This transnational flow of knowledge, though often informal, may influence the adoption of sustainable agricultural practices, especially when returning migrants or their family members integrate these ideas into local farming systems. Moreover, migrants are often exposed to new technologies, environmental management approaches, and socio-economic norms in their destination areas19,74,75. Through regular communication or return visits, migrants can share these new insights and aspirations with their communities of origin. This “social remittance” effect can influence how households view agriculture, risk, and long-term sustainability76,77. For example, exposure to more mechanized or conservation-oriented farming systems in urban or international contexts may cause migrants to adopt similar methods at home. Furthermore, migration can serve as an informal risk mitigation strategy. The knowledge that a household member can provide support during agricultural shocks, such as crop failure or market fluctuations, may lower the perceived risk of experimenting with new, potentially costly, sustainable practices78,79. This dynamic highlights how migration can indirectly foster adaptive capacity in the face of environmental and economic uncertainty. In Nepal, rural male out-migration reshapes intra-household roles and supports the adoption of SAPs. In Terai region such as Chitwan district, migrant-member households adopt labor-saving innovations like integrated crop–livestock systems, influenced by knowledge shared through migrant networks80. In mid hills region such as Baglung district, the absence of male labor has increased women’s decision-making authority, leading to the uptake of soil conservation and agroforestry practices81. In the mountain district of Mustang, households sustain apple and herb farming by reorganizing labor or hiring workers, demonstrating adaptive strategies under shifting gender and labor dynamics82,83. Remittances and adoption of SAPs The results further reveal that remittances exert a strong, positive, and statistically significant influence on the adoption of SAPs. Households that receive remittances tend to adopt more such practices, suggesting that these financial inflows play a pivotal role in supporting environmentally conscious farming decisions. Due to Nepal’s diverse geography from the fertile Terai plains to the Mid-Hills and mountainous regions farmers encounter varying agro-ecological constraints, making financial support through remittances vital for adopting region-specific sustainable farming methods that enhance productivity, resilience, and environmental conservation16. Remittance income may facilitate by funding improved irrigation systems, the adoption of drought-resistant crop varieties, diversification into higher-value crops, and the application of integrated pest management techniques. This finding aligns with a growing body of empirical research emphasizing the importance of remittances as a critical source of liquidity for rural households, particularly in regions where access to credit and financial services is limited84,85. Remittances can alleviate liquidity constraints that often prevent households from investing in sustainable but capital-intensive farming inputs, such as organic fertilizers, improved seeds, irrigation infrastructure, or conservation tools22,55. In many rural settings, these inputs are considered risky or unaffordable, especially for resource-constrained smallholders. However, with remittance income, households are better positioned to take on these investments, thereby facilitating the transition toward more resilient and sustainable agricultural systems86. Beyond merely increasing purchasing power, remittances may also influence risk preferences and planning horizons. Households with stable external income sources may be more willing to experiment with long-term sustainability measures that do not yield immediate returns but contribute to environmental and livelihood resilience over time87. Additionally, remittance-receiving households might allocate labor more flexibly, hiring additional help or reducing dependence on family labor, thus allowing time and resources to be directed toward the management of more complex, sustainable farming systems88. Women migrant-sending households and adoption of SAPs Our results show a negative association with female migrants and adoption of SAPs in the study areas. This implies that when women migrate, households may face labor or knowledge gaps, especially given the central role women play in subsistence agriculture in South Asia70. The negative association may be explained by several mechanisms. First, loss of skilled labor, for example women typically manage essential tasks related to input selection, weeding, harvesting, and seed management89. Their migration can result in a substantial loss of labor and knowledge. Dependency on less skilled labor, households may compensate for female migration by hiring external labor or relying on remaining members with less experience or motivation to implement sustainable practices. This finding aligns with studies that emphasize the gendered dimensions of migration and caution against assuming that all remittance-receiving households benefit equally90,91. The result suggests that when female migration occurs without adequate institutional or community support, it can undermine sustainability goals in agriculture. This finding resonates with studies that highlight the “feminization of agriculture” without adequate empowerment or resource access for remaining members90. While the analysis assumes that female migrants may send remittances, the study does not measure how these remittances are used; therefore, the mechanisms described should be viewed as plausible pathways rather than directly observed effects. Second, in rural Nepal, particularly in districts like Mustang, Baglung, and Chitwan, the lower adoption of SAPs in women migrant-sending households can be explained by how remittances are allocated. Money sent by female migrants is often not invested directly in farming; in unmarried or single women migrants, families frequently deposit the funds in bank accounts for the woman’s future use upon return, or spend it on high-value, non-productive assets such as gold jewellery. This economic diversion, which diverts away from farm inputs and infrastructure, reduces the likelihood of adopting improved irrigation, new crop varieties, crop diversification, or integrated pest management92. Third, female migrants move abroad for educational purposes, which requires households to send money to support tuition fees and living costs rather than investing in agricultural improvements, which could not be the case for male migrants. Combined with reduced on-farm labour availability and limited engagement with agricultural extension services, these financial choices contribute to the observed lower uptake of SAPs. Socio-demographic, farm, institutional characteristics and adoption of SAPs In terms of demographic factors, larger households are more likely to adopt SAPs, likely due to better labor availability, which aligns with findings from studies in Nigeria and Malawi93,94. Involvement in off-farm activities suggests that income diversification can facilitate sustainable farming, possibly by providing cash flow for inputs or equipment95. With regards to farm characteristics, mixed farming systems show a strong positive association, reflecting the complementary use of crop-livestock systems and integrated nutrient management, as promoted in conservation agriculture literature96. The negative relationship with livestock holdings may be due to competition for land, labor, or even preference for more traditional practices incompatible with SAPs97. The positive impact of cash cropping may reflect better market access or the need to maintain soil quality for commercial yields98. Additionally, perception of erratic rainfall is positively associated with SAPs adoption, implying that farmers use these practices as climate adaptation strategies34. The institutional support reveals that membership in farmer groups enhances adoption, consistent with evidence that social networks facilitate information sharing, risk reduction, and peer support99. However, the low significance of access to formal credit reflects a systemic barrier in rural Nepal, where informal borrowing is more prevalent and formal finance is underutilized63. Similar patterns were observed in studies conducted in Nigeria, Ethiopia and Pakistan, suggesting a possible universal trend of limited access to formal credit among women farmers95,100,101. Limitations of the study The study has some limitations to acknowledge. Firstly, the utilisation of cross-sectional data impedes our capacity to discern temporal changes and to ascertain causal relationships. Secondly, while the categorization of remittances as tangible types (e.g., money, goods, and agricultural inputs) was employed, this approach may overlook informal or non-monetary contributions, such as knowledge exchange, decision-making support, or labor assistance, that also shape farm-level practices. Future research endeavors may benefit from the adoption of mixed-method designs, which have been shown to enhance the capture of informal support systems and the nuanced gender dynamics that are frequently overlooked in quantitative analyses. Thirdly, despite the implementation of controls for salient factors, including membership in farmer groups and access to formal credit, the potential for endogeneity resulting from unobserved variables cannot be entirely excluded. Finally, while the present analysis accounts for male migration, the potential bi-directional relationship between male out-migration and female farm management may complicate causal interpretation. Further exploration of this relationship is recommended, ideally through the use of longitudinal or panel studies. Conclusion This study uses data from 400 households across three agro-ecological zones in Nepal, which provide empirical insights into how migration, remittances, and gender dynamics impact the adoption of sustainable agricultural practices. Moreover, it highlights women’s capability on farms to adopt more sustainable and resilient farming practices while facing systemic vulnerabilities within society. Our findings emphasise that financial flows from migration, particularly remittances, are key drivers enabling households to invest in sustainable agricultural practices. However, increased migration of women appears to constrain adoption, likely due to labour shortages and the diversion of funds toward savings, education, or household consumption rather than direct agricultural investment. Beyond gender, migration and remittance-related factors, farming systems, cash crop production, and cultivated land size significantly support implementing sustainable agricultural practices in the study area. Conversely, reliance on livestock limits such adoption, possibly reflecting competing demands on household resources. Moreover, engagement in local farmer groups increases adoption of sustainable agricultural practices. It emphasises the importance of social networks for knowledge exchange and community support in advancing sustainable agricultural transitions. Considering the positive link between female farm management and the adoption of sustainable agricultural practices, policymakers and development agencies should prioritise gender-responsive agricultural interventions. These programs must focus on empowering women, especially in migrant-sending communities where women increasingly manage farms. Empowerment can be achieved by gender-targeted agricultural training, easy access to credit, tailored extension services, and strengthening local women’s groups. Moreover, remittance utilisation strategies need better alignment with agricultural development goals. Since remittances were often spent on household consumption rather than farming, creating innovative schemes and providing subsidies such as matched savings schemes, agricultural input vouchers, and community investment funds can encourage households to invest remittances into sustainable technologies and practices. Finally, the adverse effects of the sequence of female household members’ migration on implementing SAPs highlight the necessity of addressing labour and knowledge deficits resulting from their absence. Community labour-sharing networks, public-private partnerships for mechanisation, and subsidised agricultural services can help mitigate these challenges. By integrating gender empowerment, strategic remittance use, and community-driven support systems, climate resilience can be enhanced, and sustainable agriculture can be effectively fostered. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1