Introduction

Climate-smart agriculture (CSA), is an approach that helps guide actions to transform agri-food systems towards green and climate-resilient practices and is widely acknowledged, as an approach that presents opportunities for climate change adaptation (resilience building), enhancement of food security needs, and achievement of the Sustainable Development Goals (SDGs) (Azadi et al. 2021; Tong et al. 2019). CSA presents a wide range of practices and technologies including; weather-smart (weather forecasts, climate-informed agro-advisories, weather insurance and climate analogues for planning), water-smart (aquifer recharge, rainwater harvesting, community management of water, laser-land levelling, micro-irrigation, raised-bed planting, solar pumps), carbon/nutrient-smart practices (agroforestry, minimum tillage, organic materials input, integrated nutrient management, biofuels, and limited/no synthetic chemicals use), and institutional/market smart activities (learning platforms or farmer to farmer learning and capacity development, financial services, market information, gender-equitable approaches, and off-farm risk management strategies) (Aggarwal et al. 2018).

Under CSA, carbon/nutrient-smart practices have been indicated to have greenhouse gas (GHG) mitigation potential (Barasa et al. 2021; Branca et al. 2021). Whereas there are numerous anecdotal narratives on the GHG mitigation potential of CSA, quantitative empirical evidence is limited especially in developing countries where CSA is broadly implemented. Assessment of the mitigation potential of various low-emission agriculture practices strengthens local, national and international level reporting, GHG stock-taking and mitigation strategies (Streimikis et al. 2021). For instance, the adoption of CSA practices including organic materials input and agroforestry increased soil organic carbon (SOC) by 42‒196% at the depth of 0–15 cm in agricultural soils in Eastern Africa (Tanzania, Kanye & Uganda) (Ambaw et al. 2019). Reduced synthetic fertilizer use in Sub-Saharan Africa (SSA) agriculture soils showed a change in carbon (C) content from 12.2 g C soil kg-1 to 13.3 g C soil kg-1, thereby reducing GHG emissions significantly (Robroeck et al. 2015).

To bridge the science-policy knowledge gap, the Consultative Group for International Agricultural Research (CGIAR) program on Climate Change, Agriculture and Food Security (CCAFS) introduced the Climate-Smart Village (CSV) concept in South-East Asia, Africa, Latin America, and the Caribbean as a bottom-up approach for CSA assimilation (Aggarwal et al. 2019). The CSV concept seeks to understand CSA priorities, synergies, and trade-offs in the local community context (Campbell et al. 2016). The study examines the GHG mitigation potential of selected carbon/nutrient CSA practices (residue incorporation, organic fertilizer input and limited/no synthetic fertilizers/pesticides) using data from CCAFS CSVs in Ghana. The study further assesses local-level participation towards the acceleration of bottom-up CSA implementation.

Materials and methods

Study areas and the concept of CSVs

In Ghana, CCAFS has been working with smallholder farmers toward food systems transformation in the Upper West Region (UWR) through the CSVs. Located in northwest Ghana, the UWR is characterized by a semi-arid environment and possesses savanna agroecological characteristics (Fig. 1). The two experimental communities used in the study are Doggoh (latitude 10°32′N, longitude 2°43′W) in the Jirapa district and Bompari (latitude 10°37′N, longitude 2°54′W) in the Lawra district (Buah et al. 2017).

Fig. 1
figure 1

Showing the location of the climate-smart villages in Ghana

Full size image

The UWR has a single rainy season spanning between April and October, providing most of its annual rainfall (Totin et al. 2018). Data from the Ghana Meteorological Agency (GMeT) showed that the region has high rainfall variability. Rainfall variability ranged between 523.7 mm (1986) and 1036 mm (1963) with a long-term average of 1036 mm (1953–2011) (Nyantakyi-Frimpong and Bezner-Kerr 2015). Time series data concerning the region for twenty years (20) point to planting rains (defined as 20–30 mm followed by not more than ten dry days ) shifting from mid-March to April-May, extending the drought period. Subsequently, mean temperatures have increased, especially during the dry season (Nyantakyi-Frimpong and Bezner-Kerr 2015). The two communities covered in this study are high-risk climate-vulnerable areas (Fig. 1).

Data and sources

The implementation of the CSA practices in the CSVs was done from 2016 (baseline/before implementation of practices) to 2020 (after implementation of practices/mitigation). Farm-level data used for this study were collected within the framework of an integrated monitoring process of the CSVs in Ghana. The overall aim of the monitoring exercise was to evaluate the efficacy of CSA options and better understand their constraints and barriers to adoption and benefits. Within the evaluation, parameters were established to determine the GHG mitigation benefits of different CSA options. A questionnaire was developed to collect data on the mitigation technical options offered and implemented in CSVs (Table 1). The questionnaire was administered to 214 farmers who have been involved in the implementation of CSA practices in the CSVs (farm clusters shown in Fig. 2). Thirty-five (35) interviews were also conducted; 25 one-on-one interviews were conducted with farmers (15 females and 10 males), 5 AEOs (Agriculture Extension Officers), and 5 project officers (both national and international). Four (4) Focus Group Discussions (FGDs) were conducted (two in each CSV). The interviews and FDGs sought to identify farmers’ choices and willingness to adopt CSA practices. Local-level participation in CSA was measured by farmers’ reasons and willingness to adopt CSA practices through the qualitative interviews. All agricultural activities in the CSVs were rain-fed hence no data on irrigated crops were included.

Table 1 CSA practices included in the questionnaire and their codes. The percentage of acceptability indicates the general acceptance of the practices by farmers
Full size table
Fig. 2
figure 2

The specific farms from the climate-smart villages included in the study

Full size image

The data was collected using GeoFarmer, a flexible application (APP) used for data collection and efficient monitoring and evaluation of agricultural technologies and practices at the farm level (Eitzinger et al., 2019). In the current study, the GeoFarmer APP was uploaded on either a personal computer or smartphone. The Farm Calculator, a module in the GeoFarmer APP, enabled us to collect data and generate a database of soil properties, technologies, management practices and outputs from different crop and livestock systems. For instance, field-specific information on tillage and crop establishment, crop management, water use, fertilizer, residue management, biomass yield, feed characteristics and manure management were collected from individual farmers. In addition, we recorded data on livestock types, growth stages and management practices (i.e., feed consumption and manure management system). As the study involved human participants, ethical consideration was sought from the University of Ghana Research Ethics Committee. Respondents were asked to participate voluntarily and were assured of their anonymity since names and other personal details were not associated with specific responses given.

Data organization with cool farm tool (CFT)

Collected data were used to run the Cool Farm Tool (CFT) (Hillier et al., 2011, CFT: https://www.coolfarmtool.org/). The CFT is a GHG emission calculator that allows users to estimate annual GHG emissions associated with the production of crops and livestock products (Hillier, 2011). It comprises a generic set of empirical models that are used to estimate full farm-gate product emissions constituting a mix of Tier 1 (lower level estimation), Tier 2 and Tier 3 approaches (advance or robust estimations). The tool considers context-specific factors that influence GHG emissions such as pedo-climatic characteristics, production inputs and other management practices at the field and farm levels. The tool provides total GHG emission per unit area and per product, allowing users to estimate the performance and efficiency per unit of product (Pragya et al. 2017; Seebauer 2014).

The CFT has seven main sections (general, crop, livestock, sequestration, energy use, processing and transport) that allow for the calculations of GHG emissions. The study made use of mainly the General, Crop and Carbon Sequestration sections, since these were relevant to our study. The General section collects information on the product type, production area, climate, and unit of analysis. The Crop Section requires information on crop type, soil (soil texture, soil organic matter, soil pH, soil drainage and soil moisture), fertilizer use, pesticide application, and crop residue management. In the case of rice, we incorporated the flooding system (rain-fed) and the number of days under cultivation. The Carbon Sequestration section takes in data on land-use changes, management changes (tillage changes, cover cropping, compost, manure additions, and residue incorporation), and annual biomass for trees in the cropping system (CFT: https://www.coolfarmtool.org/). Information from literature was used to present the soil and climatic characteristics of the CSVs (presented in Table 2).

Table 2 Soil and climatic information about the CSVs. The information was extracted from existing literature on the CSVs
Full size table

Statistical analysis

The study adopted a novel quasi-experimental design: the Difference in Difference estimator (DID) in linear regression for the impact analysis. The estimation technique corrected bias and minimized extreme variable selection. The estimation was done for GHG emission per unit area and unit product. Subsequently, the significant coefficients from the regression results were used to generate coefficient plots (Figs. 3, 4 and 5) through STATA visualization. Also, ggplots and graphs were used to present the results of the study. GHG emissions up to farm-gate were calculated. Emissions from processing, marketing, and consumption (post-farm-gate) were unaccounted for.

Fig. 3
figure 3

Describes the standardized coefficients for the baseline and mitigation scenarios. The mitigation scenario shows that all the practices contributed to GHG abatement (kgCO2e/ha− 1). Below the zero line indicates GHG abatement. Organic fertilizer was significant at (P < 0.05), residue incorporation (P < 0.05), no-inorganic fertilizer (P < 0.001) and no-pesticide input (P < 0.01). Below the zero line indicates GHG removal

Full size image
Fig. 4
figure 4

Describes the effect of CSA adoption on the GHG account of crop types at the Bompari CSVs. Below the zero line indicates GHG mitigation potential for the various crops. CropCowpea:time, cropGroundnut:time, cropMaize:time, cropMillet:time, cropRice:time, cropSorghum:time and cropYam:time shows abatement potential after CSA adoption

Full size image
Fig. 5
figure 5

Describes the effect of CSA adoption on the GHG account of crop types at the Doggoh CSVs. Below the zero line indicates GHG mitigation potential for the various crops. CropCowpea:time, cropGroundnut:time, cropMaize:time, cropMillet:time, cropRice:time, cropSorghum:time and cropYam:time indicates abatement potential after CSA adoption

Full size image

We present the implementation of the estimator below:

$${y}_{ist}= {\beta }_{s}+{\beta D}_{st}+{e}_{ist}$$

Where, y captures the outcome variable GHG emission at time t for each crop (s) and farmer (i). D takes the dummy variable of the GHG emission-specific effect such as crop residue incorporation, organic fertilizer input, inorganic fertilizer and pesticides. The mitigation scenario “after” is represented by “time”: cropCowpea:time, cropGroundnut:time, cropSorghum:time, cropMillet:time, cropMaize:time, cropRice:time and cropYam:time.

Results and discussion

Inorganic fertilizer (125 kg ha− 1) and pesticide input (65 kg ha− 1) were higher at the baseline, whereas residue incorporation (1300 kg ha− 1) and organic fertilizer input (1125 kg ha− 1) improved after the implementation of the practices (Shown in Table 3). Organic fertilizer and residue incorporation are nature-based crop nutrient inputs that enhance the soil ecosystem. For instance, organic materials from residue incorporation and organic fertilizer can increase soil ability in carbon capture (Sapkota et al. 2019). Nonetheless, the implementation of residue incorporation and organic fertilizer will require mechanization due to the high volumes (Table 3), of which smallholder farmers cannot absorb the associated financial cost.

Table 3 Shows a change in the implementation of practices for the baseline and mitigation. The average rate of implementation was combined for both CSVs
Full size table

The associated GHG effect indicates that all the practices were significant in GHG emission reduction after implementation of practices (P < 0.0001) (Fig. 3). The estimation design was built under the common trend assumption that without the interventions, the treatment (after implementation) and control (baseline) will observe the same trend over time. However, different effects of the diverse interventions promoted were observed after implementation. Residue incorporation (40 ± 15 kg GHG ha-1, P < 0.05) and organic fertilizer input (49 ± 14 kg GHG ha-1, P < 0.05) resulted in GHG removal per unit area after implementation. Limiting inorganic fertilizer and pesticide input also reduced emissions by about 50.1 ± 15 kg GHG ha-1 area (P < 0.001) and 37 ± 12 kg GHG ha-1 area (P < 0.01) respectively. Synthetic fertilizers and pesticides are highly potent in generating emissions due to their artificial chemical composition (Giampieri et al., 2022). Chemical substances affect soil quality and reduce soil organic carbon (SOC) content, thereby generating emissions and reducing soil sequestration ability (Awasthi 2022; Ouédraogo et al. 2018). A study by Bordoloi et al. (2019) indicated that a reduction of synthetic fertilizer rate by 25% (thus, a reduction from 60 to 45 kg N ha-1) reduced emissions considerably from fertilized agricultural soils. Also, Robroeck et al. (2015) established that reduced synthetic fertilizer use in Sub-Saharan Africa (SSA) agriculture soils showed a change in carbon (C) content from 12.2 g C soil kg-1 to 13.3 g C soil kg-1 reducing GHG emissions significantly.

At the baseline, farmers were mostly engaged in savanna burning and residue removal from cropland. The burning of savanna, crop residue and biomass generates emissions (CH4 and N2O), especially in organic soils. Aside from the air pollution and emissions from the burning of savanna and biomass, it also results in the wastage of valuable organic resources (Balehegn et al., 2022). Residue addition can contribute significant amounts of N (nitrogen) to the soil lowering emissions (Buss et al. 2022). For example, the introduction of organic manure and residue addition increased SOC stocks by 42‒196% at the depth of 0–15 cm, and 19‒110% at cumulative one-meter depth soil profile compared to BAU (Business-as-Usual) in Eastern Africa (Tanzania, Kanye and Uganda) (Ambaw et al. 2019).

The effect of CSA adoption was analysed for the crop types in the CSVs, with ‘time’ representing the mitigation scenario. The adoption of the practices at the specific CSVs presented different GHG abatement opportunities for the various crop types. For instance, rice cultivation (298.2 kg CO2e GHG ha− 1 area, P < 0.01) recorded the highest mitigation potential at Bompari. Cowpea (147.2 ± 25 kg CO2e GHG ha− 1 area, P < 0.01), groundnut (92.1 kg CO2e GHG ha− 1 area, P < 0.05), millet (160.3 kg CO2e GHG ha− 1 area, P < 0.001) and yam (115.1 kg CO2e GHG ha− 1 area, P < 0.05) cultivation also significantly reduced GHG emissions (Fig. 4). At the Doggoh CSV, GHG removal came from yam, cowpea, rice and groundnut cultivation (Shown in Fig. 5). Cowpea and groundnut were consistent in GHG mitigation for both CSVs because they are leguminous crops with high nitrogen fixation ability, helping in formation and maintenance of soil structure and degradation of agrochemicals and pollutants thereby supporting the abatement of GHG emissions (Herridge et al. 2022). An experimental study by Mairura et al., (2022) in Central Kenya concluded that legume cultivation has a positive effect on agriculture by adding and recycling biologically fixed nitrogen, enhancing nutrient uptake, breaking down the crop pest cycle and reducing GHG emissions.

Rice paddies/fields are a major concern to science and policy communities because they produce threatening and long-lasting emissions (CH4 and N2O) (Cheng et al., 2022). Literature (Belenguer-Manzanedo et al. 2022; Gupta et al. 2021) suggests that 30% CH4 and 11% N2O of global agricultural emissions are from rice fields. Nonetheless, rice cultivation can present large GHG emission removal with simple changes and appropriate management systems (Islam et al. 2022; Saha et al. 2022). Rice contributed the largest abatement potential in this study, partly because rice was cultivated under rain-fed conditions. Rice cultivated under continuous flooding releases larger amounts of emissions due to anaerobic conditions in the soil, which can greatly increase under long-term inundation (Saha et al. 2022). Less continuous flooding of rice paddies was observed at the CSVs due to limited irrigation facilities. Available irrigation facilities were preferred for vegetables (tomato, pepper and garden eggs) farming to rice. Few scenarios of irrigated rice cultivation were observed especially during the dry season, but in all the instances, farmers practised intermittent flooding with less number of flooding days during cultivation.

Maize cultivation did not present any abatement opportunity at both CSVs (maize was not significant in GHG emission reduction under CSA interventions). Not a discouragement for maize production in the smallholder context but signifies the need for crop-specific mitigation measures. Farmers showed readiness to cultivate legumes such as cowpea and groundnut without synthetic fertilizers compared to maize. The nutrient requirement of maize was higher, limiting the sufficiency of organic inputs for maize production. Koglo et al. (2022) reported low SOC stock of maize cultivation under different management typologies in Togo, indicating the complexity of maize production and the need for targeted abatement measures. GHG emission per unit product of the different crops presented in the mean graphs (Figs. 6 and 7) shows higher emission intensities and lower crop yields during the baseline. For instance, Sorghum (Mean = 229 ± 23 kg CO2e GHG kg− 1 sorghum) and groundnut (Mean = 214 ± 25 kg CO2e GHG kg− 1 groundnut) recorded larger GHG emissions per unit product (Fig. 6). The mitigation scenario showed an increase in crop yields with lesser GHG emissions per unit product (Fig. 7). Larger abatement intensity was recorded for sorghum (93.2 kg CO2e GHG kg− 1 sorghum), rice (79.2 kg CO2e GHG kg− 1 rice) and groundnut (69.7 kg CO2e GHG kg− 1 groundnut). GHG emission intensities of crops were generally higher than emissions per unit area affirming the conclusion reported by earlier studies (Kim et al. 2021; Seebauer 2014) that smallholder agriculture tends to have lower absolute GHG emissions but higher emissions per unit product. Global crop-wise distribution of emission intensity has been uncertain with higher intensity reported for rice paddy management; peatland draining and nitrogen (N) fertilizer with different cropping practices contributing disproportionately to emissions (Carlson et al., 2017).

Fig. 6
figure 6

Mean graph that shows emission intensity and crop yield at the baseline. Crop yields and emission intensity are measured in kg per unit product.

Full size image
Fig. 7
figure 7

Mean graph that shows emission intensity and crop yield after implementation. Crop yields and emission intensity are measured in kg per unit product.

Full size image

Local-level participation in CSA implementation

A summary results of the determinants of farmers’ participation in CSA implementation is presented in Fig. 8. From the results, farmer participation in CSA is influenced by farmer eligibility/institutional arrangements (19%), willingness (45%) and farmer ability to participate/assets and capital (36%).

Fig. 8
figure 8

Indicates farmer participation in CSA in implementation

Full size image

Farmer eligibility in CSA

Organizations that promote agriculture sector initiatives and policies have the absolute power to decide where and when projects are implemented, beneficiaries and outcomes of projects. Farmer eligibility is dependent on whether a country, region or community is selected for an agriculture project. Interviews with the project implementers (CCAFS team) revealed that the institutional arrangement provided opportunities for farmer participation; “Each climate-smart village has its theory of change based on location and context” (Representative, CCAFS). The theory of change is linked to national priorities to ensure consistency of initiatives with local actions and landscape arrangement. Farmer eligibility is decided by vulnerability mapping, countries’ consistency in seeking support for climate change adaptation and mitigation, media alerts and calls from the scientific community about a specific area and needs (Representative, CCAFS). Nonetheless, we established that farmer participation in CSA decision-making was only visible during the period of implementation. Identification, planning, design and selection of practices were done without prior consultation with farmers; “No consultation was made with the farmers during the selection of the practices. Selected participants were only asked if they are interested to join the implementation” (F08, Doggoh CSV). Local-level participation is a catalyst for inclusiveness, socio-cultural understanding, enhancement of community development and a tool for sustainable development (Ningrum et al. 2022; Barasa and Jelagat 2013). Less involvement of locals in CSA implementation has the propensity to limit sustainability, and reduce utilization and ownership (Chiles et al., 2021). As found by Anugwa et al. (2022), local-level participation in planning, designing and implementing low-carbon agriculture programmes facilitated effective adoption and carbon benefits in Nigeria.

Farmers’ willingness to participate in CSA

At the local level, farmers need to cultivate a strong motivation to adopt CSA. Willingness to participate can be an internal or external factor that influences a farmer’s participation (Lee 2017). The study found that farmers’ willingness to participate in CSA was due to climate crisis, increase yield/productivity, technical knowledge and understanding of practices, risk aversion and fear and gender disparity.

Farmers indicated that prolonged drought and declined precipitation have severely affected their productivity over the years; “Fewer rains and long months of drought have affected farm output” (F02, Bompari CSV). Aside from the adverse on-farm climatic impacts, interviewees indicated that the climate crisis also affects their properties (houses), causes psychological worries (anxiety, depression and stress) and migration (people seeking better livelihoods elsewhere); “Floods and storms destroy houses and other properties. Community dwellers worry about the situation; insufficient food, destruction of properties and inability to meet family demands. Severe depression and stress can be associated with climate change in the community” (F07, Bompari CSV). Farmers, already frustrated about the adverse impacts of climate change were willing to adopt CSA to reduce the consequences of climate change on their livelihoods; “Experiences with adverse climate change impacts are clear, adoption of the CSA practices can help reduce the climate change impacts” (F04, Doggoh CSV).

Gender played a crucial role in farmers’ willingness to participate in CSA implementation. Male farmers spent less time on farms and had other sources of livelihood (buying and selling livestock) thereby less concern about CSA. The gender of farmers also played a role because male farmers felt they have more energy/strength to work on their farms and do not need any support; “There is energy to work - why depend on projects?” (F04, Doggoh CSV). On the contrary, female farmers were more willing to adopt CSA because they spend more time on the farms. Female farmers were also noticed to have multiple responsibilities within the household sphere including, child care, care for the elderly, cooking for the entire family and walking long distances to access water and fuel. These multiple household and farm-level responsibilities pushed female farmers to consider which options are more efficient, less labour intensive and are able to address their challenges. The practices were supporting more women to reach their production goals and household food needs without comprising time for other household responsibilities; “As a woman, the CSA practices have been supportive. The practices have reduced the number of hours spent on the farm and productivity is high. More women have joined with more expressing interest because of the observed benefits” (F20, Doggoh CSV). Our findings align with earlier researches (Musafiri et al., 2022; Anugwa et al. 2022) who established that women are more likely to adopt CSA practices and engage in good farming practices to enhance food security.

Farmer’s ability to participate in CSA

Farmer eligibility and willingness are important for CSA implementation; however, farmer ability (assets and capital) is needed to make CSA adoption possible (Lee 2017). Farmer ability includes assets and capital available to farmers to support their production activities (Branca et al. 2021; Clay and Zimmerer 2020). The study found that time and labour, physical and financial capital and land ownership influenced farmers’ ability to adopt CSA practices. At the local level, labour and time are of the essence to farmers. Interviewees indicated inadequate labour and time to engage in CSA activities; “The training and demonstrations are time-consuming. Combining the training and visiting demonstration fields with farm activities is tedious. Participants are burdened with these additional activities” (F20, Doggoh CSV).

Other interviewees indicated inadequate availability of farm tools to support implementation; “Wheelbarrows/motorized tricycles are needed to transfer household waste and organic manure unto the farms. Wheelbarrows were not provided making the process of organic fertilizer application difficult” (F03, Bompari CSV). The Agriculture Extension Officers (AEOs) indicated that the farmers needed some specific equipment to support implementation but were not provided by the project;“Wheelbarrows/ motorized tricycles were not provided. Farmers relied on their mediocre tools to organize and transfer manure unto farms” (AEO, Bompari CSV).

Ownership of land had a strong association with social and economic power. Land access and secure rights increased the bargaining power of vulnerable groups including women and migrants. The connection between land ownership and CSA adoption was established in the study. Farmers with absolute land ownership were more likely to engage in CSA. Farmers without absolute land control (probably using rented land, borrowed from a friend or dedicated by a husband) were hesitant to make huge investments on the land including adopting CSA; “If investments such as time and money are made in preparing and applying organic fertilizer on the farm and the land becomes fertile, landowners collect the farm” (F22, Bompari CSV). Land ownership is an issue of contention in the UWR of Ghana (Akugre et al. 2022). Encroachment, land grabbing and commercial farms have generated additional tension on land, especially for smallholder farmers’ use (Kansanga et al., 2022; Akugre et al. 2022). Farmers with absolute land rights were more willing to adopt CSA which is also reported in previous studies (Waaswa et al. 2022; Ogunyiola et al. 2022).

Conclusion

The study revealed that the introduction of CSA practices including organic fertilizer, residue incorporation, no-inorganic fertilizer and no-pesticide input reduced GHG emissions from cropland. The study also revealed that lower GHG emissions per unit product and higher crop yield can be achieved under CSA interventions. However, smallholder farmers in the CSVs need to be attentive to specific crop nutrient requirements to maximize the GHG mitigation potential of CSA interventions. For instance, maize cultivation did not present an abatement opportunity even though maize is a staple food in the communities, and produced in larger quantities. Targeted maize interventions could increase soil carbon stock in maize cultivation and reduce GHG emissions. As the study is aligned with the Sustainable Development Goals (SDGs), more specifically SDG 13 (climate action) and SDG 2 (zero hunger), it is imperative for CSA promoters such as the CGIAR, FAO and the World Bank to embrace participatory processes such as farmer schools to increase CSA uptake for food security and mitigation goals.