Adapting agriculture to climate change: institutional determinants of adoption of climate-smart agriculture among smallholder farmers in Kenya

Abstract

Climate-smart agriculture (CSA) adoption rates have remained low in Kenya despite being promoted by the Kenyan government and its development partners. Analyzing institutional factors could help inform efforts to mitigate potential climate adaptation mal-actions in Kenya and other contexts. This study examined the relationship between institutional factors and CSA adoption among smallholder potato farmers in Gilgil Sub County of Nakuru County, Kenya. The institutional factors selected for this study included access to credit, training on CSA, non-governmental organization (NGO) support, and farmers' group membership. A binary logistic regression analysis unveiled that smallholder potato farmers' adoption of CSA was positively related to institutional factors, which was statistically significant at a 5% significance level (χ² = 10.219, df = 4, p < 0.05). However, only access to credit was positive and statistically significant at a 5% significance level (Wald χ² = 4.524, df = 1, p < 0.05) among the four explanatory variables included in the binary logistic regression model. Therefore, warranting access to credit is deemed to produce favorable requisites for adopting farming methods befitted to adapting agriculture to the effects of climate change. One way of warranting access to credit could be mobilizing farmers to join groups that serve as Savings and Credit Cooperatives (SACCOs) and Village Savings and Lending Associations (VSLAs) because farmers that join SACCOs or VSLAs have access to loans that may otherwise be challenging to obtain from conventional financial institutions.

Keywords:

• Climate change
• climate change adaptation
• climate-smart agriculture
• institutional factors
• adoption
• potato production
• smallholder farmers

Reviewing Editor:

• Manuel Tejada, Universidad de Sevilla, Spain

Subjects:

• Agriculture & Environmental Sciences; Botany; Plant & Animal Ecology; Soil Sciences; Environmental Studies; Environmental Management; Conservation - Environment Studies; Biodiversity & Conservation; Ecology - Environment Studies; Hazards & Disasters

1. Introduction

The effects of Corona Virus Disease 2019 (COVID-19) pandemic are focal in promoting and accepting diverse sustainable and climate-resilient strategies (Ingutia, 2021). This is because the pandemic worsened the existing effects of climate change. Yet, climate-resilient and sustainable agricultural practices are vital in lessening the combined effects of climate change and the global pandemic (Gates, 2020; Ingutia, 2021). In recent years, there has been increasing acknowledgment among the development and scientific community that nature, poverty, and climate change are intertwined (Waaswa et al., 2021a; Waaswa & Satognon, 2020). These are usually called triple jeopardy, triple emergency, or triple crisis. Kenya has been in the lead in developing a climate change¹ adaptation strategy and launched the National Climate Change Action Plan (NCCAP 2013–2017) in 2013 following the successful launch of the National Climate Change Response Strategy (NCCRS) in 2010 (Ambrosino et al., 2020).

Agriculture fundamentally depends on climatic circumstances; therefore, it is one of the most exposed sectors to climate change hazards. Climate change's adverse effects on agriculture may be magnified by the interplays between the elements of the climate system, resource exhaustion, and other aspects of global climate change, such as changes in land uses, biosphere integrity, freshwater use, and nutrient sequences (Aguilera et al., 2020; Satognon, Owido, et al., 2021; Steffen et al., 2015). These relationships between global changes indicate the necessity for combined approaches to establish non-resource exhaustive agricultural production systems beyond climate resilience to global change resilience (Cramer et al., 2018; Malek & Verburg, 2018; Waaswa & Morgan, 2023). Climate change interacts with resource exhaustion and other global changes to imperil livestock and crop production. Kenya's food production has been a victim of these changes (MoALF, 2016; Waaswa et al., 2021b). The interaction between the ever-changing temperature and unreliable/irregular precipitation is a significant factor responsible for the low food productivity (see Figure 1).

Figure 1. Average annual rainfall and temperatures (1991–2020) for Kenya.

Source: Authors' computation of temperature and rainfall data from World Bank Climate Change Knowledge Portal (2021).

Moreover, if the food production industry actors do not take appropriate measures to cushion climate change, cereal and other crop yields are predicted to continue declining (Bird et al., 2016). Additionally, climate change influences the most significant critical factors in food production across crop and livestock production because it defines the existence of feeds and pastures that directly affect animal health. This manifests in favoring pests and pathogens multiplication and infestation, checking on reproduction and growth, and reducing water availability yet increasing its requirements (Aguilera et al., 2020; Rojas-Downing et al., 2017). These circumstances have been experienced in Kenya over time, partly explaining the country's persistent average number of malnourished people (see Figure 2).

Figure 2. Number of undernourished people (2000–2019) in Kenya.

Source: Authors' computation of FAOSTAT data (2018).

This can also be explained by the uneven intensification of sustainable agricultural practices, and it appears to have left food insecurity and high poverty levels unchecked, hence nutrition insecurity prevalence like in many other Sub-Saharan Countries (Clay & Zimmerer, 2020; Rasmussen et al., 2018). Kenya's environment and agriculture have been threatened by climate change. The farmers have encroached on the environment to open more land for agriculture. However, due to climate change effects, the yields of major food crops like cereals and potatoes continue to fall (see Figure 3).

Figure 3. Land use and trend in crop yields (2012–2017) in Kenya.

Source: Authors' computation of FAOSTAT data (2018).

Severe climatic incidences intimidate food production in Kenya and impose pressure on the country's economy, livelihoods, food and nutrition security, and the environment (Bryan et al., 2013; Ogola et al., 2023; Waaswa et al., 2021c). These instances are most likely to intensify if the remedies proposed by various climate change adaptation activists are not taken up, and farmers will be significantly affected. This is because the future temperature and precipitation forecasts for 2020 to 2039 show an undulation that may not favor food production (see Figure 4).

Figure 4. Average projected monthly precipitation and temperature (2020–2039) for Kenya.

Source: Authors' computation of temperature and rainfall data from World Bank Climate Change Knowledge Portal (2021).

As a result, Kenya's government is obliged to intensify its agriculture sector's resilience, sustainability, and productivity (MoALF, 2016; Satognon et al., 2021b). In response to the effects of climate change on agricultural production, the Kenyan government launched the Kenya Climate-Smart Agriculture Strategy (KCSAS) in 2017. This was a result of collaboration between the Environment and Agriculture ministries. KCSAS guides agricultural systems change by employing integrated agriculture, development, environment, food security, and climate change adaptation strategies. This strategy details CSA as an 'excellent opportunity for transformation by uniting agriculture, development, and climate change under a common agenda' (Government of Kenya (GoK), 2017, p. 8).

CSA is promoted as an innovative, transformative tool to realize the required integration and adjustments, but its adoption among farmers is still low. Sustainable development advocates should study and document cases of prosperous CSA uptake with scrutiny to draw lessons to inform future CSA development and scaling efforts (Chandra et al., 2018). If lessons to other contexts are to be drawn from Kenya, follow-up research would be invaluable to inform the development of sectoral policies and strategic plans (Faling, 2020). This study sought to address this recommendation. This paper is part of the study by Waaswa (2021) conducted in Gilgil Sub County, Kenya. Gilgil Sub County was selected because of its location in a climate change hotspot area (Rampa & Knaepen, 2019). In addition, farmers in the area are active potato producers, and it is their second staple food crop after maize. Due to the growing water deprivation in Gilgil Sub County for crop growth and watering of animals (MoALF, 2016), several development agencies have diffused adaptation management strategies in the area. The various CSA practices being diffused in the area include using cover crops, moisture management practices, soil erosion and water management practices, and high-yielding and drought-resistant crop varieties. This could be attributed to the high susceptibility of potatoes to the effects of climate change since it also forms part of the major cash crops in the area and is a second major staple crop. Therefore, efforts to help farmers maintain and or improve their livelihood options have shown interest in extending CSA as an adaptation strategy to this area (Kibe et al., 2019; Satognon, Lelei, et al., 2021b; Waaswa et al., 2021a).

The CSA also involves actions within a more all-inclusive system and policy frameworks, such as statutes and incentive provisions that sustain the on-ground climate-smart agricultural practices [CSAPs] (Lipper, 2017). Such support systems and incentives include farmers' access to credit, non-governmental organizations (NGOs), farmer groups, and training on CSA, among others. According to studies conducted elsewhere, support systems help accelerate the adoption of CSAPs (Agrimonti et al., 2021; Bryan et al., 2013; Wekesa et al., 2018). For example, Zakaria et al. (2020) found a positive and significant relationship between attending training and the adoption of CSA, where trained farmers were 49.8% more inclined to adopt CSA than untrained fellows. Similarly, Bryan et al. (2013) and Schmitt (2012) found that agrarian and poor farmers adapted to climate change effects following the support received from several institutions, including NGOs, and claimed these were very influential players. Also, Baudoin (2014) and Comoé and Siegrist (2015) applauded NGOs for effectively enhancing farmers' adaptive decisions. Numerous studies have discovered a strong and significant connection between FGM and the adoption of CSA (Awotide et al., 2016; Danso-Abbeam et al., 2017; Mango et al., 2017). These claim that group membership intensifies informative networking that builds trust among group members and facilitates experience sharing about CSA, creating courage among the persuaded farmers to adopt CSA fully.

Further, in some localities, farmers' access to extension services and training on how technologies are applied, access to credit to fund the technology adoption process, and Non-government support services may hasten technology adoption process if available and accessible by all (Ikendi et al., 2023a; Kalungu & Leal Filho, 2018). However, the reverse may hold if these factors are unavailable or inaccessible. Understanding how these factors affect technology uptake, especially CSA, is necessary and can be helpful in the context of policy, planning, and development. Cautiously derived generalizations about their relationship with the adoption of CSA practices among smallholder farmers can shed light on appropriate ways for developing future planned CSA practices and other climate change adaptation strategies in a way that accounts for the past failures and successes experienced by local communities and development projects (Faling, 2020). Without an adequate study of the influence of critical determinants² of CSA adoption, there is a danger of recurrent failures of externally developed CSA programs in Kenya and related contexts. Because like several African countries, CSAPs have been promoted in Kenya even before the inception of KCSAS (Waaswa et al., 2021a). However, the adoption rates have remained low due to several factors, notwithstanding support from the Kenyan government and its development partners (MoALF, 2016), which include Kenya's nationally determined contribution (INDC), which sets out a greenhouse gas emissions reduction target of 30% by 2030 compared to a business-as-usual (BAU) scenario of 143 megatons of carbon dioxide equivalent [MtCO2-eq] (International Institute for Sustainable Development (IISD), 2015). An analysis of institutional factors³ could help inform efforts to mitigate potential climate adaptation mal-actions in Kenya and other ­developing countries by improving risk and stakeholder-resource consciousness linked to development preferences in smallholder farming societies (Forsyth, 2018). Awareness of what can foster CSA adoption at the farm level and reasonably classifying CSA practices according to the available support can promote such consciousness. Such clarity can assist in knowledge building by relating research on CSA uptake with innovation diffusion and farmer decision-making. It can also aid in evaluating the contribution of externally developed CSA mediations in encouraging on-farm uptake of CSAPs (Sova et al., 2018). The fact that Gilgil Sub-County has not been studied on the institutional factors influencing the adoption of CSA practices despite their relevance in other contexts and CSA diffusion efforts on ground informed the hypothesis that this research sought to test – that there is no statistically significant relationship between institutional factors and the adoption of CSA among the smallholder potato farmers in Gilgil sub-County, Kenya.

2. Theoretical framework

This research was based on Rogers M. Everett's theory of diffusion and adoption of technologies. Unlike other theories of technology adoption, like rural household behavior under market failure by de Janvry and Sadoulet (2006) and induced innovation theory by Binswanger et al. (1978) that acknowledge innovation as being part of the economic system, Rogers (2003) explains how several factors interact to influence technology adoption. Rogers (2003) points out socioeconomic factors and information dissemination pathways and identifies the key features exacerbating the adoption rate. The features are combined in the perceived relative advantage of the technology, compatibility, ease of use/trial-ability, complexity, and how it is communicated. However, this study acknowledged the difference in the contexts. Since communities are discrete, the effect of these factors may vary from one community to another by seeking to establish the relationship between selected factors and the adoption of CSA. Rogers (2003) states that technology adoption decision follows five steps (Figure 5), which chronologically follow one another in a time-ordered way.

Figure 5. Stages in the technology adoption process.

Source: Rogers (2003).

Individuals who learn first about the technology before others are found to have exposure to mass media and more channels of interpersonal communication, have more education, are of high social status, have more access to change agents, and exercise high levels of social participation in their communities (Rogers, 2003). Therefore, the inception of this study was premised on the ideas put forward by this theory to understand how CSA has been diffused and adopted by the smallholder potato farmers in Gilgil Sub-County, Nakuru County, Kenya.

3. Materials and methods

3.1. Research design

A cross-sectional survey methodology was used for data collection from the study population. Such a design enables data collection on the population at a particular time. While using this design, the respondents can describe existing phenomena without intervention. This design allows the comparison of literate/illiterate, youth/aged, and male/female, among other categories, without manipulating the independent variable (Mugenda & Mugenda, 2003). Adoption of CSA was the dependent variable in the proposed study, and the independent variables were the institutional factors.

3.2. Study area and protocol approval

The National Commission for Science, Technology, and Innovation (NACOSTI) of Kenya approved the protocols used in this study under license No. NACOSTI/P/21/9627. The study was conducted in Gilgil Sub-County of Nakuru County, Kenya. Gilgil is subdivided into five wards, namely, Gilgil, Elementaita, Malewa West, Mbaruk/Eburu, and Morendat wards (Figure 6).

Figure 6. Map showing the location of the study area.

Gilgil Sub-County covers an area of 1,348.43 square kilometers, with a total population of 171,839 (Rampa & Knaepen, 2019). The study area is located at coordinates 36°10^′0^ʺE 0°40^′0^ʺS, and it is known for its annual rainfall of between 500 and 870 mm, with maize, beans, and potatoes as the significant crops covering 86.4% of the arable land area (Rampa & Knaepen, 2019).

3.2.1. Target population

The study targeted smallholder potato farmers in Gilgil Sub-County. According to the 2019 agricultural census, 15,359 smallholder farmers⁴ actively engage in potato production in Gilgil Sub-County (MoALF, 2019). These formed this study's target population. The accessible population consisted of all the 10,889 potato farmers found in Morendat ward (4,287) and Mbaruk/Eburu ward [6,602] (Gilgil Sub-County, 2019).

3.3. Sampling procedure and sample size

Gilgil Sub-County was purposively considered for this study because of its susceptibility to the effects of climate change (MoALF, 2016). This has attracted several interventions; for example, the Climate and Water Smart Agriculture Centre (CaWSA-C) project by the Netherlands Development Organisation (SNV) and the Kenyan government through the Sub County and Ward extension officers to foster CSA adoption to cushion the area against the shock. The Kenyan government implements the CSA under its CSA implementation framework in the study area. The SNV has encouraged the adoption of CSA among the smallholder potato farmers in the study area. Besides, farmers in the study area are actively engaged in potato production.

Of the five wards in Gilgil Sub-County, Mbaruk/Eburu and Morendat wards were purposively selected because they comprise the most significant number of potato farmers in the Sub-County. Additionally, these two form the major farming communities in the Sub-county, unlike other wards like the Gilgil ward, a town with rocky soils that result in low farming activities (Rampa & Knaepen, 2019). The sample size was calculated based on the coefficient of variation formula suggested by Nassiuma (2000) following the acquisition of the list of smallholder potato farmers from the Gigil Smallholder Farmers' Association. For this study, a 21% coefficient of variation and 0.02 standard error were used to compute the sample size using Nassiuma (2000) equation (see equation 1(1) $$\mathrm{n}=\frac{N{C}^2}{C^2+(N-1)e^2}$$(1) ). These parameters were chosen, assuming the lower coefficient of variation and standard error, to minimize variability and error in the sample. Besides, considering that the maximum coefficient of variation is 30% above which, it is not justified. A low coefficient of variation below the one used in this study leads to a small sample size, which may not be suitable for the survey research (Kollipara et al., 2011). (1) $$\mathrm{n}=\frac{N{C}^2}{C^2+(N-1)e^2}$$(1) where:

n = sample

N = population

C = Coefficient of variatione = standard error $$\mathrm{n}=\frac{10889\times {\left(21\mathrm{\%}\right)}^2}{{\left(21\mathrm{\%}\right)}^2+(10889-1){0.02}^2}$$ $$\mathrm{n}=109$$

The n value, which is above the suggested sample size of 100 for survey studies, is appropriate to provide the necessary level of accuracy (Kathuri & Pals, 1993).

To cater for non-responses, attrition, and the purposes of a representative sample, the researchers revised the sample size to 120 by adding 10% of 109. The wards, Sub-County extension officers, and Nakuru smallholder farmers' association chairperson helped identify all the potato smallholder farmers in the study area. Proportionate random sampling was utilized to determine the number of respondents for the purposefully sampled wards (Table 1). The actual responders from the wards were then drawn using a simple random sampling method.

Table 1. Proportion of sample size per ward.

Ward	Number of Potato Farmers	Proportion	Sample size
Mbaruk / Eburu	6,602	60.63	73
Morendat	4,287	39.37	47
Total	10,889	100	120

Source: Gilgil Sub-County (2019).

3.4. Data collection procedure and analysis

A structured questionnaire administered by the researchers was used to gather the primary data. Visits to the sampled potato farmers were made, and one of the visits gave the researchers an opportunity to seek informed consent from the participants for their voluntary participation, assuring them of protection from harm and keeping their responses anonymous. The researchers collected data with a translator's assistance to counteract the language barrier challenge. A translator helped some farmers who knew the CSAPs investigated by different local names fully understand what was being asked. The collected data was organized and keyed into the Statistical Package for Social Sciences (SPSS) version 25. Descriptive analysis was used to compute the percentages and frequencies for some variables against the adoption of CSAPs. A binary logistic regression model was used to analyze the relationship between institutional factors and the adoption of CSAPs. Logistic regression allows predicting a distinct outcome from a set of variables that may be discrete, dichotomous, continuous, or a combination.

4. Results and discussions

4.1. Relationship between institutional factors and adoption of climate-smart agriculture

Table 2 shows the selected institutional factors whose relationship with the adoption of CSA was investigated among smallholder potato farmers in Gilgil Sub County, Kenya.

Table 2. Descriptive statistics for institutional factors among smallholder potato farmers.

Institutional factors	Frequency	Percentage
Access to credit
No	68	56.7
Yes	52	43.3
Sources of credit
SACCOs	36	69.2
Private money lenders	3	5.8
Relatives and friends	7	13.5
Banks	6	11.5
Amount of credit accessed annually
< KES 10,000	16	30.8
KES 10,001- KES 25,000	23	44.2
KES 25,001- KES 35,000	5	9.6
>KES 35,000	8	15.4
Trained on CSA
No	82	68.3
Yes	38	31.7
Time when first training on CSA was attended
Last season	22	57.9
One year ago	13	34.2
Two years ago	3	7.9
Three years ago and above
Number of trainings on CSA attended
One	13	34.2
Two	13	34.2
Three	6	15.8
Four and above	6	15.8
Source of training on CSA
Ministry of Agriculture	23	60.5
Private companies	6	15.8
NGOs	9	23.7
Universities
Access to NGO support on CSA
No	108	90
Yes	12	10
Kind of support received from NGOs on CSA
Extension services	7	58.3
Credit	4	33.3
Inputs	1	8.3
Farmers' group membership
No	67	55.8
Yes	53	44.2
Receive support from farmers' group on CSA
No	9	17
Yes	44	83
Kind of support received from farmers' group on CSA
Eases access to extension services	38	86.4
Eases access to credit	2	4.5
Eases access to inputs	4	9.1

Source: Own computation of survey data, 2021.

Around 56.7% of the interviewed smallholder potato farmers had no access to credit. Of the 43.3% who had access to credit, 69.2% accessed it from SACCOs, 5.8% (the least) from private moneylenders, 13.5% from relatives and friends, and 11.5% from banks. Majority (44.2%) accessed KES 10,001- KES 25,000, followed by 30.8% (less than KES 10,000), 15.4% (above KES 35,000), and the least (9.6%) who accessed KES 25,001- KES 35,000.

A relatively low number of farmers with access to credit, moreover the most considerable percentage with a low amount and locally sourced may mean that farmers in the study area find difficulties accessing and adopting CSA. Because most CSAPs require financial muscle to execute them as recommended. This assertion is consistent with previous findings (Gebrehiwot & van der Veen, 2013; Meijer et al., 2015; Mihiretu et al., 2019) where access to credit was appreciated to aid in climate change adaptation in agriculture by facilitating access to CSAPs and other adaptation strategies. Further affirmation is that farmers without access to credit might be prevented from implementing appropriate adaptive strategies because access to the required information is dictated by available resources, especially in monetary terms (Arimi, 2014).

Most (68.3%) farmers reported no CSA training, and only 31.7% had been trained. Over 57.9% of those who received training had been trained within the previous season, 34.2% within one year, 7.9% had received training two years ago, and none had been exposed to training within the last three years. An estimated 34.2% received training once and twice, respectively, and the rest (15.8%) were trained three to four times and above. Of these, 60.5% received training from the Ministry of Agriculture, 15.8% from private companies, 23.7% from NGOs, and non-from universities besides their involvement in diffusing CSA in the study area.

The low percentage of trained farmers may result in a low adoption rate of CSA among the farmers. Because training helps to expose farmers to institutions that support agricultural development and sources of inputs and credit that farmers require to execute CSA (Nkuba et al., 2020). For example, agricultural input dealers are always invited during training to enable the farmers to implement what they have learned from the training by locating quality inputs. During the training, farmers access information on the weather forecast, which drives them to seek adaptation strategies like CSA (Arimi, 2014), especially when these trainings are crop specific, they further benefit farmers through interactions with the trainers which influence adoption of CSAs (Ikendi et al., 2023b).

Having few farmers trained more than three times may also mean that farmers are left ill-equipped with the knowledge required to put CSA into practice⁵ because farmers, being adult learners, may need to be trained repeatedly to grasp the concept. This contention agrees with Danso-Abbeam et al. (2017), who indicated that training is key to farmers' decision to adopt improved maize varieties. On the other hand, many farmers getting training from the Ministry of Agriculture followed by NGOs may mean that new practices can easily be communicated to the farmers. Because the ministry may have a hand in generating the innovations done by sister institutions like the research institutions (Rogers, 2003).

Similarly, most farmers (90%) never received NGO support, and of the 10% who received NGO support, the most considerable percentage (58.3%) were supported with extension services, 33.3% with credit, and 8.3% with farm inputs. Likewise, over 55.8% of the potato farmers who participated in this study did not belong to any farmer's group. Of the 44.2% members of different farmers' groups, the majority (83%) received support to adopt one or more of the CSAPs from their farmers' groups. Nevertheless, 86.4% were supported by easing their access to extension services, 4.5% through facilitating their access to credit, and 9.1% found it easy to access farm inputs through their respective farmers' groups.

The most considerable percentage of farmers not receiving NGO support may mean that farmers are not exposed to enough external support that may bolster the execution of CSA. This is because NGOs tend to expose farmers to new adaptive strategies, support them with the required information, and, in some cases, facilitate farmer-to-farmer learning (Ikendi et al., 2023a; Meijer et al., 2015). Therefore, having a high percentage of farmers receiving extension services as NGO support among those supported could mean the increased adoption of CSA among such a category (Zhang et al., 2012). On the contrary, low NGO support may create room for the farmers to look for alternatives by themselves, resulting in the increased adoption of CSA. This is because NGOs tend to generate an overdependence attitude among farmers, which may make them reluctant to take action against climate change (Elbehri & Lee, 2011).

In addition, most farmers not belonging to farmer groups may imply the low adoption of CSA, especially those requiring collective action for their execution. For example, establishing agroforestry may burden and shun individual farmers mainly in sourcing the seedlings, but this can be done when farmers pool resources and do collective sourcing of the required planting materials. This contention is supported by previous studies (Arimi, 2014; Hunecke et al., 2017; Tazeze et al., 2012), which found farmers with farmer group membership (FGM) easily accessed information on various CSAPs and other adaptive measures, collectively worked together in adopting CSA. A typical example is the case of the lower Nyando soil and water conservation initiative, where a successful farmer was a group member (Gbegbelegbe et al., 2018). However, having a few farmers belonging to farmer groups may be advantageous for CSA scaling efforts. Because as noted by Waaswa et al. (2021a), ties of trust and collaboration among group members can result in withdrawal behavior, making farmers less inclined to embrace and try new agricultural discoveries. Withdrawal behavior happens when a few group members shift their responsibilities to active members. Due to ties of trust, the burdened members may not question the non-participation of their colleagues.

4.1.1. Access to credit and adoption of climate-smart agriculture

Table 3 shows that almost all CSAPs were highly adopted by over 69.2% of the farmers with access to credit. However, mulching scored low by being adopted by 59.6%, followed by irrigation (28.8%) and the two (potato seedlings and minitubers) that were least adopted by only 15.4% and 3.80% of farmers, respectively.

Table 3. Access to credit and adoption of climate-smart agriculture.

Climate-smart agricultural practices	Smallholder potato farmers (N = 120)
	No Access (n = 68)		With Access (n = 52)
	F	%	F	%
Rainwater harvesting and storage	54	79.4	46	88.5
Irrigation	23	33.8	15	28.8
Mulching	46	67.6	31	59.6
Minimal tillage	45	66.2	42	80.8
Improved crop varieties	36	52.9	35	67.3
Terracing	48	70.6	42	80.8
Drainage management	46	67.6	39	75.0
Intercropping	58	85.3	49	94.2
Agroforestry	55	80.9	47	90.4
Synthetic fertilizers	64	94.1	50	96.2
Composting	46	67.6	45	86.5
Furrow/ ridge planting	48	70.6	41	78.8
Crop rotation	52	76.5	48	92.3
Apical rooted cuttings (potato seedlings)	1	1.5	8	15.4
Mini-tubers	0	0.0	2	3.8

Source: Own computation of survey data, 2021.

On the other hand, smallholder potato farmers with no access to credit also adopted most of the CSAPs, though not at the same rate as their counterparts. Over 61.0% of the farmers in this category adopted CSA, with improved crop varieties being adopted by 52.9%, followed by irrigation (33.8%), potato seedlings (1.5%), and minitubers (0.0%). For some CSAPs, access to credit may affect them less or not at all. For example, irrigation practices are high among farmers without access to credit. These farmers may use free options for watering their crops, like irrigating by flooding, which only requires labor and determination. Other investigators encountered similar scenarios (Pattanayak et al., 2003; Silvestri et al., 2012) where credit didn't seem to influence the adoption of some CSAPs.

Contrarily, the general and low percentage of adoption of CSA among farmers with no access to credit compared to those with access to credit can prove the contention by Teklewold et al. (2013) that limited access to credit stifled farmers' ability to adopt and intensify the use of some CSA. This also explains why farmers with access to credit show a high percentage of CSA adoption. Conversely, the low adoption of mulching by farmers with access to credit could be due to differences in access to credit. Because in some areas, men have more access to credit, yet women commonly adopt some practices, making them unable to access the necessary resources to adequately put CSA into adoption (Gumucio et al., 2020).

4.1.2. Training on and adoption of climate-smart agriculture

Surprisingly, an average of 69.0% of the farmers who had been trained adopted CSA (Table 4). This is nearly similar to the farmers' response with access to credit. Furthermore, an average of 62% of the farmers who were not trained also adopted CSA.

Table 4. Training on and adoption of climate-smart agriculture.

Climate-smart agricultural practices	Smallholder potato farmers (N = 120)
	Not Trained (n = 82)		Trained (n = 38)
	F	%	F	%
Rainwater harvesting and storage	68	82.9	32	84.2
Irrigation	17	20.7	21	55.3
Mulching	47	57.3	30	78.9
Minimal tillage	57	69.5	30	78.9
Improved crop varieties	42	51.2	29	76.3
Terracing	60	73.2	30	78.9
Drainage management	55	67.1	30	78.9
Intercropping	74	90.2	33	86.8
Agroforestry	72	87.8	30	78.9
Synthetic fertilizers	79	96.3	35	92.1
Composting	61	74.4	30	78.9
Furrow/ ridge planting	63	76.8	26	68.4
Crop rotation	65	79.3	35	92.1
Apical rooted cuttings (potato seedlings)	7	8.5	2	5.3
Mini-tubers	1	1.2	1	2.6

Source: Own computation of survey data, 2021.

Besides, the percentage of farmers with no training adopting CSA, generally, is not far from that exhibited by their counterparts. Because this category highly adopted some CSAPs. For example, intercropping (90.2%), agroforestry (87.8%), synthetic fertilizers (96.3%), furrow/ridge planting (76.8%), and potato seedlings (8.5%), which were adopted by 86.8%, 78.9%, 92.1%, 68.4%, and 5.3% respectively of the farmers who had been trained on CSA. However, a noticeable difference was observed in adopting the rest of the CSAPs, where trained farmers scored high, especially in irrigation, mulching, and improved crop varieties, and a minimal difference in adopting mini-tubers.

Though small, a discrepancy existed in the adoption of CSA between the two categories of farmers. The reason could be that trained farmers can see how some practices are adopted or even witness the results of a given CSAP. This argument aligns with previous findings (Moges & Taye, 2017; Nkuba et al., 2020), where trained farmers adapted by adopting improved crop varieties and other adaptive strategies. Ikendi et al. (2023c) also found that farmers participation in agronomy training increased their adoption of land sparing techniques of production including sack and keyhole gardens which are climate change mitigation and food and nutrition security strategies. Danso-Abbeam et al. (2017) added that farmers participating in farm demonstrations had increased chances of allocating more land for improved maize varieties. However, a relatively higher percentage of adoption of CSA by untrained farmers could mean that the training is not a panacea, but other factors like social relationships in the communities come into play (Waaswa et al., 2021a) and also their indigenous knowledge of practicing sustainable agriculture (Ikendi, 2023).

4.1.3. Non-government organization support and adoption of climate-smart agriculture

On average, a minimal difference existed between the adoption of CSA by the smallholder potato farmers who received NGO support (66%) and those who did not receive [64%] (Table 5). Some CSAPs were adopted at the same rate by farmers receiving NGO support and those that did not. These included terracing and compositing, adopted by 75% of both categories, with a slight difference in the adoption of compositing (adopted by 75.9% of the farmers who did not receive NGO support).

Table 5. Non-government organization support and adoption of climate-smart agriculture.

Climate-smart agricultural practices	Smallholder potato farmers (N = 120)
	Not Supported (n = 108)		Supported (n = 12)
	F	%	F	%
Rainwater harvesting and storage	89	82.4	11	91.7
Irrigation	31	28.7	7	58.3
Mulching	68	63.0	9	75.0
Minimal tillage	78	72.2	9	75.0
Improved crop varieties	64	59.3	7	58.3
Terracing	81	75.0	9	75.0
Drainage management	76	70.4	9	75.0
Intercropping	97	89.8	10	83.3
Agroforestry	94	87.0	8	66.7
Synthetic fertilizers	104	96.3	10	83.3
Composting	82	75.9	9	75.0
Furrow/ ridge planting	82	75.9	7	58.3
Crop rotation	89	82.4	11	91.7
Apical rooted cuttings (potato seedlings)	7	6.5	2	16.7
Mini-tubers	1	0.9	1	8.3

Source: Own computation of survey data, 2021.

Amazingly, some CSAPs were adopted by a higher percentage of farmers with no NGO support than their counterparts. Among these included improved crop varieties (59.3%), intercropping (89.8%), agroforestry (87.0%), synthetic fertilizers (96.3%), and furrow/ridge planting (75.9%) compared to 58.3%, 83.3%, 66.7%, 83.3%, and 58.3% respectively adopted by the farmers receiving NGO support.

Results also show that other than the latter, a high percentage of adoption of the rest of the CSAPs was recorded among farmers receiving NGO support. A prominent difference existed in the adoption of irrigation, where only 28.7% of the farmers with no NGO support adopted it compared to 58.3% of the farmers with NGO support. However, with a difference in adoption between the two categories, both farmer categories adopted potato seedlings and minitubers below average, with the lowest percentages among farmers without NGO support.

A possible explanation for the higher adoption percentage of some CSAPs by non-NGO-supported farmers could be that NGOs may introduce non-agricultural projects, especially for women. For example, World Vision (Pouw & Elbers, 2012) and the Center for Sustainable Rural Livelihoods (Ikendi et al., 2023a) supports women in Uganda with sewing machines that tend to occupy them, increasing their financial livelihoods strategies though it compromising the time devoted to agriculture. This translates into low execution of some CSAPs besides receiving support from NGOs. Some CSAPs, such as intercropping, are primarily adopted by women (Waaswa et al., 2021a). However, a slightly high percentage of adoption of some CSAPs among NGO-supported farmers could be explained by the reality that these farmers are exposed to various pieces of training, and conferences and are sometimes directly supplied with farm inputs that aid them in adopting CSAPs (Danso-Abbeam et al., 2017; Gbegbelegbe et al., 2018; Meijer et al., 2015).

4.1.4. Farmer group membership (FGM) and adoption of climate-smart agriculture

Table 6 unveils that a considerable percentage (92.5%) of the farmers with no FGM adopted rainwater harvesting and storage compared to the 71.7% that adopted the same and have FGM. Though a lower percentage for both categories was embraced, a slightly lower percentage (29.9%) of farmers with no FGM adopted irrigation than those with FGM (34.0%). Except for potato seedlings, minitubers, and the latter, over 71.7% of the farmers with FGM adopted the rest of the CSAPs. However, this is not the case with farmers with no FGM, where some CSAPs, like improved crop varieties, were adopted below average (46.3%). In addition, farmers with FGM adopted minitubers (1.5%) and potato seedlings (4.5%) below average. Additionally, other than improved crop varieties, minitubers, and mulching adopted by 58.2% of the farmers with no FGM, the rest of the CSAPs were adopted by over 67.2% within this category.

Table 6. Farmer group membership and adoption of climate-smart agriculture.

Climate-smart agricultural practices	Smallholder potato farmers (120)
	No FGM (n = 67)		Has FGM (n = 53)
	F	%	F	%
Rainwater harvesting and storage	62	92.5	38	71.7
Irrigation	20	29.9	18	34.0
Mulching	39	58.2	38	71.7
Minimal tillage	46	68.7	41	77.4
Improved crop varieties	31	46.3	40	75.5
Terracing	50	74.6	40	75.5
Drainage management	45	67.2	40	75.5
Intercropping	58	86.6	49	92.5
Agroforestry	53	79.1	49	92.5
Synthetic fertilizers	62	92.5	52	98.1
Composting	45	67.2	46	86.8
Furrow/ ridge planting	47	70.1	42	79.2
Crop rotation	54	80.6	46	86.8
Apical rooted cuttings (potato seedlings)	3	4.50	6	11.3
Mini-tubers	1	1.50	1	1.9

Source: Own computation of survey data, 2021.

A possible explanation for the high percentage of adoption of some practices among smallholder potato farmers with no FGM could be the free-riding tendency that diminishes active participation among some members of the groups. According to Waaswa et al. (2021a), this problem poses a burden on some active members of the groups. It stifles initiatives like a high percentage in the aggregate adoption of some CSAPs that would be realized in its absence. Yet some practices like rainwater harvesting are labor-intensive, and collective efforts would be the best solution for income-constrained smallholder farmers. Instead, farmers with no FGM realize their incapability, which raises their motives to work hard to equalize with their counterparts significantly, hence ending up doing even better than compromised group members (Hackman & Katz, 2010).

On the other hand, a few farmers with no FGM adopted some CSAPs, especially irrigation, mulching, improved crop varieties, potato seedlings, and mini-tubers. This could be because of limited access to information, lack of motivation, and social support required in executing these CSAPs. This is premised on the fact that farmers with FGM are supported by each other and easily access information on new technologies and adaptive measures. This assertion is in line with previous studies (Ikendi et al., 2023a; Tambo & Abdoulaye, 2012) that found members of farmer groups to have considerable access to credit, new knowledge, inputs, and more labor as a result of collective action. Changes might follow this in on-farm-management practices due to collaborative learning and information spillover, thus increasing the possibility of adopting CSA. For example, group participation might allow farmers to consolidate labor (Teklewold et al., 2013). This can facilitate the creation of terraces on a farmer's garden before proceeding to the subsequent farmer (Shikuku et al., 2017), while credit pooled from the group members may increase the possibility of adopting CSA whose access requires money, for example, synthetic fertilizers (Ajayi et al., 2003). This also explains the overall average of CSA adoption among farmers with FGM being slightly higher than that seen among their counterparts.

4.2. Extent of CSA Adoption

On average, most farmers moderately (47.2%) adopted CSA, 38.5% indicated they adopted but to a lower extent, and only 14.3% highly adopted CSA practices (Figure 7).

Figure 7. Extent of CSA adoption.

At least every CSA practice had farmers reporting its adoption at a low, moderate, and high rate except for minitubers, which were only adopted at a low and moderate extent. The high extent adoption rate rates of most CSA practices by some farmers could be because they had access to information, which raises possibilities of adoption. This contention agrees with García de Jalón et al. (2015) and Ikendi et al. (2023b), who asserted that higher knowledge may further support farmers' decision to use a given climate change adaptation strategy. This is because information helps the farmers fully understand a given technology, and therefore, they may be willing to try it out. On the contrary, CSAPs, like irrigation, were adopted to a lower extent, and this could be because they may require heavy capital investment, which is not always readily available or accessed by some farmers.

Similarly, cultural bias against some CSA practices may curtail their widespread adoption. This assertion is commensurate with Nyasimi et al. (2017), who revealed cultural constraints as bottlenecks to adopting CSA. Yet, facilitating the adoption of such CSA may deter crop losses and increase agricultural output per area by enabling dry season production and eventually reducing over-reliance on rain-fed agriculture (Gebrehiwot & van der Veen, 2013; Ikendi et al., 2023b).

On the contrary, some CSA practices like potato seedlings and minitubers recorded low adoption rates because these CSA practices seem to be new in the area, and according to literature (Rogers, 2003), farmers are always skeptical about adopting new technologies. Another cause of the low adoption of traditionally dressed farming practices in CSA, like mulching and improved crop varieties, might be the information sources used to communicate these practices. For example, televisions, telephones, and radios (media) may not fully orient the farmers to get the full knowledge required to execute CSA. This claim is in line with Arimi (2014), who found farmers 'half-baked' regarding adopting adaptation strategies like the growth of biennial root and tuber crops, yet they received information from the media. This is because media gives general information (Comoé & Siegrist, 2015), not sufficient for the farmers to make complete adoption decisions, requiring a need to adopt narrative communication strategies between scientists with non-scientific audiences that have elements of persuasion and also their structure relays the cause-effect relation to influence adoption rates (Dahlstrom, 2014; Ikendi, 2023).

4.3. Statistical Test Results

Binary logistic regression was used in testing the hypothesis that there is no significant relationship between institutional factors and the adoption of CSA among the smallholder potato farmers in Gilgil sub-County, Kenya, and the analysis of institutional factors as independent variables relating to the adoption of CSA was statistically significant. Table 7 shows that the relationship between institutional factors and adoption of CSA was statistically significant at a 5% significance level (χ2 = 10.219, df = 4, p < 0.05).

Table 7. Omnibus tests of model coefficients for institutional factors.

		Chi-square	df	Sig
Step 1	Step	10.219	4	0.037
	Block	10.219	4	0.037
	Model	10.219	4	0.037

Results from Table 7 signify that a relationship existed between institutional factors and the adoption of CSA. Hence, the null hypothesis is rejected. This inference is consistent with findings from previous studies (Aryal et al., 2018; Diallo et al., 2019; Tran et al., 2020), where institutional factors were accountable for the adoption of CSA. Besides, between 12.1% (Cox & Snell R Square) and 8.2% (Nagelkerke R Square) of the variance in the adoption of CSA is explained by the institutional factors (Table 8).

Table 8. Institutional factors' model summary.

Step	−2 Log likelihood	Cox & Snell R Square	Nagelkerke R Square
1	124.741^a	0.082	0.121

^a Estimation terminated at iteration number 5 because parameter estimates changed by less than .001.

Like the latter, such a variance shows the relevance of institutional factors to the adoption of CSA and how they relate. In addition, a percentage accuracy classification (PAC) of 75.0% was yielded by the institutional factors' BLRM (Table 9).

Table 9. Percentage accuracy classification table^a for institutional factors' model.

Observed			Predicted
			Adoption of CSA		Percentage Correct
			.00	1.00
Step 1	Adoption of CSA	.00	0	30	0.0
		1.00	0	90	100.0
	Overall Percentage				75.0

^a The cut value is .500.

Results from Table 9 imply that the explanatory variables in the institutional factors' model accurately predict the adoption of CSA by the smallholder potato farmers by 75.0%, inferring that 75.0% of the time, we predict smallholder potato farmers to adopt CSA is correct. That said, a goodness of fit test results declared that the model used was fit for institutional factors. This is confirmed by the insignificant values (χ2 = 10.458, df = 6, p > 0.05), which support the model (Table 10). This, therefore, presents insufficient evidence to challenge the model that it does not adequately fit the data.

Table 10. Hosmer and lemeshow test for institutional factors' BLRM.

Step	Chi-square	Df	Sig
1	10.458	6	.107

Table 10 indicates that the institutional factor's BLRM is liable for the invisible traits across smallholder potato farmers' decisions to adopt CSA. The model's findings indicated a relationship between the explanatory variables and adoption of CSA amongst smallholder potato farmers differs considerably (Table 11).

Table 11. Institutional variables in the binary logistic regression equation.

Institutional variables		B	S.E	Wald	df	Sig	Exp(B)
Step 1^a	Access to credit	1.072	0.504	4.524	1	0.033	2.922
	Trained on CSA	0.777	0.611	1.619	1	0.203	2.176
	Receive NGO support on CSA	−0.768	0.827	0.863	1	0.353	0.464
	Farmers' group membership	0.406	0.496	0.670	1	0.413	1.501
	Constant	0.423	0.301	1.970	1	0.160	1.526

^a Variable(s) entered on step 1: Access to credit, Trained on CSA, Receive NGO support on CSA, Farmers' group membership.

The results also show that within the four hypothesized explanatory institutional variables included in the model, only one was found to have a significant relationship with the adoption of CSA. This was access to credit, which is described below. First, it is worth noting that an odds ratio of less than one symbolizes a negative relationship.

4.3.1. Access to credit

There is a positive relationship between access to credit⁶ and the adoption of CSA. This is statistically significant at a 5% level of significance (Wald χ2 = 4.524, df = 1, p < 0.05). The results show that farmers with access to credit are 2.922 times more likely to practice CSA than those without access to credit. This could be because access to credit empowers the farmers to meet labor costs, transport costs, and the rest of the costs related to production. Additionally, farmers with access to credit can adopt more than one CSAPs because it enables them to utilize existing information and change farm management methods in response to climate change (Kandlikar & Risbey, 2000). This finding is consistent with several past studies (Ali & Erenstein, 2017; Awotide et al., 2016; Danso-Abbeam et al., 2017; Mulwa et al., 2017) that found a positive and significant relationship between access to credit and adoption of CSA and other adaptation strategies. The latter noted that access to credit is required to fund the uptake of agricultural innovations and is regularly quoted as a factor influencing varied adoption rates. Contrastingly, some studies (Aryal, Jat, et al., 2018; Silvestri et al., 2012) found no significant relationship between access to credit and adoption of CSA.

5. Conclusions and policy implications

This study found a positive and significant relationship between institution factors, credit access, particularly, and adoption of CSA. Access to credit is therefore thought to result in favorable conditions for implementing farming practices appropriate for adjusting agriculture to the effects of climate change. One way to ensure access to credit should be by mobilizing farmers to join groups that serve as SACCOs and Village Savings and Lending Associations (VSLAs). This is because farmers who join SACCOs or VSLAs have access to loans that may otherwise be challenging to obtain from conventional financial institutions. These organizations enable members to pool their savings and lend money to one another depending on the need and ability to repay, and farmers can lessen their reliance on moneylenders, who frequently demand exorbitant interest rates. SACCOs and VSLAs would grant farmers access to farming resources like tools, equipment, and knowledge and provide financial help through cooperative sharing, which may increase adaptation potential and agricultural production. This study calls for increased efforts to facilitate smallholder farmers' access to credit since it is critical to fostering CSA adoption.

Acknowledgments

The authors acknowledged the support provided by the MasterCard Foundation through the Regional Universities Forum for Capacity Building in Agriculture (RUFORUM). The authors also acknowledged the family of Mr. Bernard Mwenja Ngigi of Gilgil, Kenya, for hosting the researchers and ensuring that the sampled smallholder potato farmers were successfully reached to participate in the study.

Disclosure statement

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

Data availability statement

The data that support the findings of this study are available from the corresponding author, AW, upon reasonable request.

Notes

1 Climate change refers to any change in climatic factors (rainfall, wind, sunshine, solar radiation, temperature) over time, which may be due to natural variability or as a consequence of human activities (Intergovernmental Panel on Climate Change (IPCC), 2012). This definition was adopted for this study.

2 Determinants: These are factors or causal elements that lead a given phenomenon to happen or directly influence a given decision (Merriam-Webster, 2020). In this study, determinants meant selected institutional factors that affect the adoption of CSA.

3 Institutional factors are related to the control measures developed by communities to moderate/govern the people. Some of these control measures are formal e.g. legal standards while the rest are informal (Warsaw, 2013). In this study, institutional factors constituted access to credit, access to training on CSA and NGO support.

4 Smallholder farmers: These are farmers who farm on less than 2 hectares (5 acres) of land (FAO, 2015). This definition was adopted for this study.

5 Practice/adoption is when one decides to fully use a technology that is perceived as the best option available. It entails the farmer learning the new knowledge and putting it into use. It includes purchasing of the tools and equipment necessary to execute the technology (Rogers, 2003). This definition was adopted for this study. Therefore, Practice/adoption of CSA is the use of a given CSA technology. In this study, it was measured by the percentage of farmers implementing CSA on their farms.

6 Farmers' access to credit is the ability of agricultural producers to get financial resources (credit) to support their farming operations and other activities (Chen, 2021). They can use this credit to pay for various farming-related expenses, including hiring personnel, managing day-to-day operating costs, and buying seeds, fertilizer, equipment, and machinery. This definition was adopted for this study and the sources of credit referred to included Savings and Credit Co-operatives (SACCOs), banks, borrowing from money lenders, neighbors and relatives.