Assessing the Contribution of Farm Forestry Farmer Field Schools to Climate Resilience in a Mixed Crop–Livestock System in Dryland Kenya

Abstract

This study examines the role of farm forestry Farmer Field Schools (FFSs) in strengthening climate resilience in mixed crop–livestock systems in dryland Kenya. Based on interviews and focus group discussions in Embu and Taita Taveta, this study finds that FFS participation enhanced tree cultivation, market monitoring, and group-based learning, with greater involvement of women in decision-making. While FFS households showed stronger motivation for continued learning and experimentation, it has not consistently translated into statistically significant improvements in climate resilience outcomes as measured by recent drought and disturbance impacts. Limited water access emerged as a major barrier. The findings suggest that while FFSs foster adaptive learning and farm-level innovation, their contribution to climate resilience requires integration with cross-sectoral strategies, especially water management and institutional support.

1. Introduction

Rural livelihoods in Kenya’s drylands are highly vulnerable to climate change due to increasing droughts, floods, and erratic rainfall [1,2,3]. There is increased national and international recognition of the need for initiatives that enhance rural communities’ resilience to climate disturbances and promote climate change adaptation in these regions [4]. The Farmer Field School (FFS) approach is promoted as a means of building resilience by equipping farmers with practical knowledge and adaptive learning skills [5,6]. Farm forestry FFS programs extend this by integrating tree cultivation into farming systems, offering potential benefits such as improved soil fertility, diversified production, and enhanced drought tolerance [7]. While there are already various guidance and training documents for incorporating a tree growing component in FFSs [8], limited research exists on how such programs specifically affect resilience in mixed crop–livestock systems. This study addresses this gap by examining the contributions and limitations of farm forestry FFS in Kenya, with particular attention to how the farm forestry FFS approach contributes to enhancing climate resilience at the local level, and the limitations in coping with and recovering from climate disturbances.

2. Framing Climate Resilience in Kenya’s Drylands

This paper applies the general definition of climate resilience, understood as the capacity of a rural farmer to cope with and recover from climate disturbances [9,10].

We specifically focus on two complementary perspectives related to climate resilience: the system-based perspective, which emphasizes understanding and managing ecological and agricultural systems under stress, and the learning-based perspective, which builds farmers’ adaptive capacity through experiential, participatory learning. Together, these approaches enhance resilience by enabling communities to analyze climate risks, test context-specific solutions, and sustainably adapt their livelihoods [5]. The system-based perspective addresses the capacity of a subject to maintain key structures and functions of a livelihood system in the face of climate disturbances [10,11,12,13,14].

In the context of dryland Kenya, livelihoods are largely sustained by a mixed crop–livestock farming system. Livestock provides draft power and manure for crop cultivation, while crops supply residues that feed livestock. Products from both components are used for household consumption and income generation, embedding the system in both subsistence and local market economies. As shown in Figure 1, this system is structured around the co-existence of crops and livestock at the household level, with both linked to local markets. It functions through interconnected flows of materials and energy—such as manure, crop residues, biomass, and nutrients—facilitating a reciprocal relationship: livestock support crop production, and crops, in turn, support livestock.

The system-based perspective has several important implications for enhancing the climate resilience of mixed crop–livestock systems in dryland Kenya. For example, farmers’ access to drought-resistant seeds is frequently proposed in preparation for climate disturbances [16]. From the system-based perspective, the establishment of stable production and distribution mechanisms for seeds is recommended to ensure that these are acquired by those who need them, including poorer households, through the creation of enabling conditions [17]. Securing access to water is also a critical aspect of enhancing climate resilience [18]. Digging deep wells within a community is a quick solution; however, from the system-based perspective, the establishment of well-monitoring mechanisms for regular maintenance and urgent action in the event of water shortages should also accompany these efforts.

The learning-based perspective emphasizes the capacity to modify or transform an existing system or to create mechanisms to adapt to fluctuating situations caused by climate change and future uncertainties. The learning-based perspective needs to be assessed at both individual and collective levels [6]. In addition to experiential and participatory learning, the concept of transformative learning—whereby individuals shift their perspectives through critical reflection and engagement with broader structural issues [19,20]—has been proposed as essential for fostering resilience in complex systems. Because the vulnerability to climate disturbances is context-dependent [21], which ultimately varies from household to household and even from person to person, enhancing individual learning capacity is a way to address such context-dependent vulnerability at the household and individual levels. Collective learning is a way to transform an existing system or to develop a new mechanism that benefits a wider community, as in the cases of seed distribution or deep well monitoring described above. Such systems may or may not be fully implemented within a community. In the latter case, in particular, collective learning that involves relevant external stakeholders is imperative to support system change or development.

Based on the framework described above, this paper addresses two research questions. The first is how farm forestry FFSs contribute to improving a mixed crop–livestock system to cope with and recover from climate disturbances. The second is what new actions emerge through learning in farm forestry FFSs, which may contribute to modifying or transforming the existing mixed crop–livestock system. Addressing these questions allows us to examine the contributions and limitations of farm forestry FFSs in enhancing climate resilience at the local level.

3. Farm Forestry FFSs Within a Mixed Crop–Livestock System

3.1. Role of Trees in Mixed Farming Systems

The brief description of the mixed crop–livestock system in Section 2 does not explicitly mention the role of trees and forests; however, these can be integrated as vital components of the system. Incorporating trees through approaches such as agroforestry can enhance ecological functions, diversify production, and strengthen climate resilience. In general, the addition of trees to a mixed crop–livestock system can provide a broader set of options for securing firewood, timber, fodder, food, income, and other environmental services such as soil fertility improvement, erosion control, and water table stabilization [22]. Fodder trees, in particular, play an important role for livestock during drought periods, when the nutritional value of grasses is low [23]. As such, the addition of trees plays a significant role in securing livelihoods in dryland Kenya.

When trees are viewed from the structural and functional viewpoint of a mixed crop–livestock system, however, the interaction of trees with crops and livestock tends to be one-way—typically from trees to crops and livestock. Functions such as soil fertility improvement, erosion control, and water table stabilization are performed by trees and forests that benefit cropping, and the production of fodder for livestock [24]. These are the key characteristics of trees in the system. In other words, trees play a more supportive role in a mixed crop–livestock system rather than forming its core structure. In this paper, we refer to this augmented system as an expanded mixed crop–livestock system—a farming system in which the traditional crop–livestock interactions are enhanced by integrating trees that provide environmental, economic, and feed support, thereby improving resilience. Figure 2 presents a diagram of the expanded system, illustrating the integration of trees and forests into the mixed crop–livestock system.

3.2. Farm Forestry FFSs and Climate Resilience

Farm forestry FFSs integrate multiple enterprises across both agriculture and forestry. Agricultural components typically include production activities of cereals, vegetables, and livestock, complemented by conservation practices such as contour planting, composting, and agroforestry. The forestry component involves activities like nursery establishment, tree planting, and management. The field-based learning approach mirrors that of conventional agricultural FFSs, emphasizing experiential learning and critical reflection grounded in adult learning theory. This participatory, hands-on method fosters deeper understanding and allows skill development across integrated farm systems.

As we argued earlier, trees play a more supportive role in a mixed crop–livestock system rather than forming its core structure. This supportive role includes functions that help strengthen the capacity of a mixed crop–livestock system to cope with and recover from climate disturbances. Trees are often more drought resistant than crops and grasses [23]. For example, fodder trees can provide feed to livestock when grasses have died; trees that improve soil fertility, control erosion, or stabilize the water table help maintain the drought resistance capacity of the system; fruit trees are sources of nutrition, food, and income; and fuelwood and timber can become key income sources [23,25,26].

The supporting role of trees in a mixed crop–livestock system is contingent on the selection and management of effective, context-appropriate tree species. What is particularly distinctive about farm forestry FFSs is that, beyond introducing beneficial tree species, they build participants’ learning capacity. Graduates of FFSs gain the skills to explore the root causes of local challenges and collaboratively experiment with potential solutions. This capacity for problem-solving and adaptive learning enables them to respond more effectively to future climate disturbances alongside their communities.

4. Method and Data

This study employs a case study approach to address the research questions in the context of dryland Kenya. The case is the Capacity Development Project for Sustainable Forest Management (CADEP-SFM), which implemented farm forestry FFSs in Embu and Taita Taveta counties of Kenya from 2016 to 2021 [27]. The implementation in Taita Taveta was limited to lowland areas. Land and climate characteristics of these counties are summarized in

Table 1. Land and climate of Embu and Taita Taveta counties.

County	Embu	Taita Taveta
Land area	2818 km2	17,084 km2
Average altitude	1221 m	695 m
Annual average mean temperature	21 °C	23 °C
Average annual precipitation	640–1495 mm	450–500 mm

Remark: Average annual precipitation for Taita Taveta reflects conditions in the lowland areas. Source: [28,29,30].

. Embu and lowland Taita Taveta have similar temperature ranges, but the latter experiences significantly lower rainfall, indicating a drier environment.

Between 2017 and 2020, three cycles of FFSs were conducted, producing 47 groups (31 in Embu and 16 in Taita Taveta), with a total of 851 graduates—70% of whom were women [31]. As described in [27], invitations were extended to community groups likely to complete the training, assessed via 14 criteria including activeness and experience in forestry-related activities. The FFS training focused on four main enterprises: woodlots, crops, fruit trees, and tree nurseries. Crop selection was tailored to local agroecological conditions. Embu emphasized green grams, cowpeas, maize, beans, sorghum, and dolichos beans (Njahe), while Taita Taveta concentrated on green grams, cowpeas, and maize.

We conducted key informant interviews (KIIs) and focus group discussions (FGDs) in November 2022, followed by a household survey in May–June 2023. KIIs were conducted for Forest Department staff who were responsible for the implementation of the FFS program (two staff at the Headquarters, two at the Embu office, and two at the Taita Taveta office). A total of ten FGDs were conducted—six in Embu and four in Taita Taveta. Each session lasted approximately two hours and took place either at a member’s field or a community center. The discussions were facilitated by a member of the research team who was fluent in Kiswahili, while Forest Department staff assisted with translation into local languages where needed. Detailed notes were taken during the discussions to ensure accurate documentation of participants’ views and experiences. To complement the FGDs, we conducted field observations at the farms of selected FFS members to visually assess reported improvements and practices discussed during the group sessions.

The household survey included 153 FFS graduates and 191 non-FFS control households (

Table 2. Covariates used in this study.

Covariate	Description	Min	Max	Mean
Gender	Gender of the household head (female = 0, male = 1)	0	1	0.72
Age	Age of the household head (year)	20	88	53.2
Edu	Education level of the household head (informal = 1, primary = 2, secondary = 3, tertiary = 4, university = 5)	1	5	2.39
Parcel	Number of parcels used by the household for agriculture, forestry, or livestock production (count)	1	5	1.33
Land	Total land area of parcels used by the household for agriculture, forestry, or livestock production (acre)	0.1	25	3.69

Remarks: Units of measurement or scoring scales are shown in parentheses. The summary statistics are those of all 344 households surveyed. Source: [27].

). Semi-structured interviews facilitated systematic data collection addressing our research objectives. Survey participants from FFS groups were randomly selected from lists provided by project staff. Non-FFS households were chosen using multistage random sampling, with support from Kenya Forest Service officials and FFS facilitators. Indicators for the interviews included the following:

(i)

Structural and functional elements of a mixed crop–livestock system:

-

Asset holding: land, livestock, facilities, and equipment;

-

Knowledge and skills in agricultural and forestry practices;

-

Awareness of climate, environment, market, and institution (e.g., bank, insurance);

-

Production and income.

(ii)

Participants’ learning capacity:

-

Motivation to acquire new knowledge or skills and engage in experimentation;

-

Motivation to participate in group activities;

-

Motivation to invest.

Because the CADEP-SFM selected FFS participants from community groups with a high potential to benefit from farm forestry, households with FFS graduates may have had greater capacity than non-FFS households from the outset. This introduces potential selection bias. While it is not possible to eliminate this bias entirely, we sought to reduce it by applying a propensity score matching (PSM) approach to data collected through semi-structured interviews. Rather than directly comparing the means of treated (households with FFS graduates) and control (non-FFS households) groups using an independent t-test, we matched treated and control observations based on several key observed covariates known to influence agricultural practices and learning and applied a paired t-test.

In selecting covariates for the PSM procedure, we included gender, age, education level, and landholding size—variables that are well established in the literature as key determinants of agricultural decision-making, innovation adoption, and learning capacity. Gender influences access to resources, information, and decision-making power, with female farmers often facing structural constraints that affect their ability to participate in training programs and adopt new technologies [32,33]. Age reflects experience and responsiveness to new information, as older farmers may be more risk-averse while younger ones tend to be more open to innovation [33,34] (Kassie et al., 2015; Loison, 2015). Education level enhances a farmer’s ability to understand, evaluate, and apply new knowledge, making it a strong predictor of engagement with extension services and sustainable intensification practices [32,34]. Landholding size serves as a proxy for wealth and productive capacity, shaping both the incentive and ability to implement improved agricultural practices [32,33]. Including these covariates thus strengthens the internal validity of the analysis by reducing overt selection bias related to FFS participation and ensuring more credible comparisons between treated and control groups.

Covariates were analyzed separately for each county. Propensity scores were estimated using a logistic regression model, and matching was conducted using one-to-one nearest neighbor matching with a caliper width of 0.2 without replacement [27]. The analysis was implemented using the “MatchIt” package in the R statistical software. Table 2 provides an overview of the covariates used to estimate the propensity scores.

Table 3. Sample size before and after the matching.

	FFS Households		Non-FFS Households		Total
County	Embu	Taita Taveta	Embu	Taita Taveta	Embu	Taita Taveta
Before matching	109	44	151	40	260	84
	153		191		344
After matching	91	32	91	32	182	64
	123		123		246
Unmatched	18	12	60	8	78	20
	30		68		98

summarizes the sample sizes before and after matching. The results presented in the next section are based on the matched dataset.

It is important to note that this study analyzes multiple outcome variables—each of which could be influenced by distinct sets of confounders. From a strict statistical standpoint, this implies that using a single set of covariates for matching may not fully account for all potential sources of bias across outcomes [35]. However, given the exploratory and partly qualitative nature of this study, the intent of using PSM is not to achieve perfect causal inference for each individual outcome, but rather to strengthen the robustness of the comparisons by reducing overt selection bias. Although the shared covariates may not capture all relevant confounders for each outcome and thus potential endogeneity—i.e., the influence of unobserved factors—remains a concern, the estimated differences should be interpreted as statistical associations rather than definitive causal effects, consistent with the view of [36] that observed associations do not, by themselves, establish causation. By doing so, the approach represents a meaningful improvement over simple mean comparisons, offering greater robustness while maintaining alignment with this study’s broader qualitative framework.

As described in the Introduction, this study explicitly situates its analysis within the specific socio-economic and institutional contexts in Embu and Taita Taveta. All evidence is drawn from socio-economic data—such as KIIs, FGDs, and a household survey—and no natural-science measurements (e.g., biophysical monitoring of soil, water, or vegetation) were undertaken. This boundary should be borne in mind when interpreting the findings, as they reflect reported practices and perceptions rather than direct environmental measurements.

5. Results

5.1. Characteristics of a Mixed Crop–Livestock System

All the households (i.e., 246 in total after matching) have cropland and are engaged in farming practices. Regarding livestock,

Table 4. Percentage of households that holds cattle and/or goat and average number of animals kept.

Animal Holdings	FFS	Percentage of Holdings				Average Number of Animals
		Embu		Taita Taveta		Embu		Taita Taveta
Cattle	FFS	65%	***	47%		1.4	***	0.8
	Non-FFS	37%		34%		0.7		0.7
Goat	FFS	82%		78%		4.6		3.6
	Non-FFS	76%		78%		3.5		4.8
No cattle, no goat	FFS	11%	*	16%
	Non-FFS	22%		19%

Remarks: ***, * indicate p < 0.001 and p < 0.05, respectively.

presents the ownership and number of cattle and goats. The overall rate of livestock ownership is high. Differences between households with FFS graduates and non-FFS households can be observed in two main aspects. First, the rate of cattle ownership is higher among FFS households. Second, the proportion of households holding neither cattle nor goats is lower among FFS households compared to non-FFS households, particularly in Embu.

One possible explanation for the differences in cattle ownership is that participation in farm forestry FFSs leads to improved farm productivity and diversification, which may result in higher or more stable household incomes. With increased financial capacity, households are better positioned to invest in livestock as a complementary asset, enhancing income security and livelihood resilience. Additionally, the integrated nature of the farming system promoted through FFSs—where crops, trees, and livestock interact—may further encourage livestock keeping. For example, improved crop yields provide more residues for animal feed; tree components can supply fodder and shade; and livestock contribute manure to enrich soils. This mutually reinforcing system creates synergies that make livestock rearing more feasible and attractive to FFS graduates.

Although data on the use of livestock manure as fertilizer were not systematically collected through the household survey, both key informant interviews (KIIs) and focus group discussions (FGDs) confirmed that the practice is common among farmers. This finding is supported by the 2019 CADEP-SFM study [37], which highlighted the widespread integration of crops and livestock in both counties. However, the quality of livestock manure varies: some households purchase it as fertilizer, while others apply it directly by transporting manure from livestock barns to their crop fields. Additionally, crop residues are commonly used as livestock feed, further demonstrating the interconnectedness of crop and livestock systems. As such, a mixed crop–livestock system is considered a major farming model that supports rural livelihoods in the target areas, and its adoption is higher among FFS graduate households.

5.2. Impact of Climate Disturbances on a Mixed Crop–Livestock System

In dryland Kenya, drought, floods, and crop and livestock pests and diseases are major climate disturbances that have affected mixed crop–livestock systems in various ways [38]. Studies on trends in average temperature and precipitation show that average temperatures in Taita Taveta are rising, and annual precipitation in Embu is highly variable [29,39]. Overall, conditions are challenging for farmers engaged in rain-fed agriculture as they are obliged to adapt to higher temperatures and increasingly unpredictable water availability. Such climate conditions directly affect the livelihoods of farm households, as confirmed by interviewees in Embu and Taita Taveta, who reported losses of crops and livestock during droughts (see

Table 5. Relationship between farm forestry FFSs and tree growing.

	Embu			Taita Taveta
	FFS	(3)	Non	FFS	(3)	Non
Engaged in tree growing	99%	**	91%	94%		97%
Number of grown trees	64.9		72.9	42.0		29.1
Survival rate of planted trees (1)	69%		66%	57%		65%
Primary purpose of tree growing
Self-consumption	59%		63%	38%		53%
For sales	20%	*	10%	13%	*	0%
Supplementary purpose
For fodder	54%		43%	47%		34%
For conservation	77%		80%	97%		94%
Growing trees on/around farms	82%		79%	69%		69%
Engaged in nursery management	78%	***	46%	81%	**	50%
Monitoring local wood price (2)	3.43	**	2.85	2.94		2.81
Attention to tree-soil relations (2)	4.68	*	4.48	4.75		4.69

Remarks: ⁽¹⁾ Trees planted in 2017–2020 and still growing at the time of interview in 2023; ⁽²⁾ Answer was on the Lickert scale 1–5; ⁽³⁾ ***, **, * indicate p < 0.001, p < 0.01, and p < 0.05, respectively.

).

The results of the KIIs and FGDs suggest that drought occurred in three consecutive years from 2020 to 2022 in Taita Taveta, causing extensive damage to both crops and livestock [40]. Farmers had difficulty accessing water, which resulted in many crops and trees going unharvested and widespread livestock deaths. Livelihoods in the area were sustained through external aid, as a mixed crop–livestock system failed to function effectively. A pest outbreak that damaged crop production was also reported in the region and is believed to be linked to changing climate patterns [41,42].

Figure 3 and Figure 4 indicate that there is no statistically significant difference in the impact of drought and other adverse events between FFS and non-FFS households; both groups experienced significant hardship. Notably, households in Taita Taveta were more severely affected than those in Embu, largely due to substantially lower annual precipitation. Moreover, the average household income in Taita Taveta (12,543 KES) is significantly lower than that in Embu (16,588 KES), with the difference being statistically significant at the 1% level. This suggests that adverse events exert a more pronounced negative effect on livelihoods in Taita Taveta. In certain cases—such as the impacts of drought and livestock losses in Embu—FFS households appear to have been more severely affected.

However, using the same dataset as this study, [27] reported that: (i) FFS households diversified their agricultural products for sale more than non-FFS households, (ii) households with more diversified agricultural products for sale were associated with lower losses from drought and crop pests and diseases, and (iii) households with more active participation in group activities experienced lower losses from crop pests and diseases. These quantitative findings were substantiated by qualitative evidence from FGDs, which highlighted practical examples of resilience-enhancing strategies adopted by FFS households. For instance, one female-headed household in Embu had, following her graduation from the FFS program, established a fast-growing tree plantation on half of her 2.5-hectare plot and installed a roof catchment system connected to a water storage tank. She also cultivated peas and fodder grasses beneath the planted trees and remained involved in group gatherings. At the time of the FGD, her garden and plantation were observed to be well managed and free from any damage.

As discussed below, active participation in group activities is one of the characteristics of FFS households. Therefore, although the survey results reveal some ambiguity regarding the impact of the FFS intervention, it may be the case that FFS households are better able to cope with climatic events under certain conditions such as the diversification of agricultural products and a learning-oriented mindset fostered through active participation in group activities.

5.3. Effect of Farm Forestry FFSs on a Mixed Crop–Livestock System

As farmers generally recognize the multiple benefits of tree growing, more than 90% of the interviewed households grow trees regardless of their participation in farm forestry FFSs. Table 5 indicates that farm forestry FFSs have a clear positive association with commercial orientation, encouraging farmers to grow trees for economic purposes rather than solely for domestic use. This is supported by the finding that FFS households monitor wood prices at local markets more frequently than non-FFS households. Although it is not statistically significant, FFS households in both counties are more likely to be connected to the wood market and to earn higher income from wood sales than non-FFS households (see

Table 6. Characteristics related to household income.

	Embu			Taita Taveta
	FFS	Non		FFS	Non
Households with wood sales income	29%	24%		19%	9%
Average per year (KES)	3298	2217		2436	1031
Overall income (KES)	100,219	147,219		165,219	125,906
Average % of wood sales income	3.3%	1.5%		1.5%	0.8%

Remarks: No statistically significant differences were observed between FFS and Non-FFS households.

). At this stage, timber revenues account for a small percentage of total income, as plantations have only recently been established. The revenue from wood sales is expected to increase in the years to come.

Table 5 also shows that both FFS and non-FFS households are generally concerned with conservation as one of their motivations for tree growing. Approximately 80% of households in Embu and 70% in Taita Taveta grow trees on or around farmland, either as part of agroforestry systems or as boundary planting. Ratings for “attention to tree-soil relations” are high in both counties, with FFS households showing a slightly higher figure compared to non-FFS households. This finding aligns with insights from KIIs and FGDs, that indicate that household interest in conservation is largely focused on preserving farmland for crop production. It is worth noting that there is no significant difference in tree survival rates between FFS and non-FFS households. While not statistically significant, FFS households are more likely to consider tree leaves as a source of fodder when planting trees—a crucial perspective within the context of a mixed crop–livestock system.

5.4. Experimenting and Learning as a Normal Practice

Typically, FFS graduates experience group experimentation and learning processes through which they formulate their own questions and assumptions, monitor progress independently, identify results, and analyze causes of failure. Then, they develop possible solutions. Through repeated engagement in such processes, experimentation and learning become part of their routine behavior.

Table 7. Relationship between farm forestry FFSs and behavioral characteristics.

	Embu			Taita Taveta
	FFS	(3)	Non	FFS	(3)	Non
Early adoption of new practices (1)	3.98	***	3.47	4.13		3.81
Involvement in group activities	100%	***	67%	94%	*	78%
Average number of groups (2)	1.84	***	0.99	1.75	***	1.16
On agriculture	83%	***	11%	88%	***	50%
On tree growing/nursery	11%	***	0%	9%	**	0%
Women’s group	32%	*	19%	28%		19%

Remarks: ⁽¹⁾ Answer was on the Lickert scale 1–5; ⁽²⁾ The number of groups that interviewees and their household members are involved; ⁽³⁾ ***, **, * indicate p < 0.001, p < 0.01, and p < 0.05, respectively.

illustrates a behavioral pattern among FFS households, particularly in their willingness to adopt new farming practices early. These households responded positively to questions related to early adoption. Notably, activities such as nursery management and monitoring local wood prices—which are among the practices showing statistically significant uptake in Table 5—can be considered examples of early adoption. These practices reflect the application of knowledge and skills acquired through the farm forestry FFS process.

FFS households exhibit a high level of participation in group activities, reflecting the inherently group-based nature of the FFS learning process. While group participation is generally common in rural Kenya [43], what stands out is the sustained engagement of FFS graduates—even years after completing the program. In Embu, all FFS households continue to participate in group activities, and in Taita Taveta, 94% do so. These activities often relate directly to FFSs, such as nursery management or collective marketing, but they also extend to broader community or livelihood initiatives, including table banking, farming support groups, and local welfare associations. In several villages in Taita Taveta, women’s groups have collaborated to protect planted trees from goats by wrapping mats and clothes around tree trunks. This suggests that collaborative learning and group-based action have become embedded in the everyday practices of these households, strengthening both social cohesion and adaptive capacity.

5.5. Perceptual and Behavioral Change on the Decision-Making in a Household

Gender dynamics play a central role in household decision-making, particularly in agricultural and forestry practices. FFSs are widely recognized for promoting the active participation of women throughout the training process, thereby contributing to more inclusive and balanced decision-making within households [44].

Table 8. Effects of farm forestry FFSs on the decision making ⁽¹⁾.

	Embu			Taita Taveta
	FFS	(3)	Non	FFS	(3)	Non
Joint decision-making on (2)
Crop for self-consumption	67% (90)	*	53% (91)	58% (31)		55% (31)
Cash crop farming	72% (58)	*	57% (54)	73% (15)		75% (12)
Tree nursery management	68% (71)		67% (42)	58% (26)		75% (16)
Tree growing	69% (85)	**	50% (68)	56% (32)		59% (27)
Livestock rearing	71% (82)	*	53% (77)	61% (26)		61% (28)

Remarks: ⁽¹⁾ Independent t-test was used to check if FFS households show a difference from non-FSS households in their decision-making style. Numbers in parentheses indicate the number of households that were engaged in the activity in question. The sample number is 91 each for FFS and non-FFS households in Embu and 32 each for Taita Taveta; ⁽²⁾ Answer was on yes (1) or no (0) for all the questions; ⁽³⁾ **, * indicate p < 0.01, and p < 0.05, respectively.

presents data on household decision-making arrangements related to agriculture and forestry. The results show that more than half of the households surveyed reported joint decision-making between husbands and wives. In Embu, the proportion of FFS households practicing joint decision-making is notably higher than among non-FFS households, suggesting that participation in FFSs may contribute to fostering more equitable household dynamics.

Interestingly, focus group discussions (FGDs) revealed that tree nursery activities are often managed by women. This is supported by household interview data in

Table 9. Situation of coping with water scarcity.

	Embu			Taita Taveta
	FFS (1)	Non		FFS (1)	Non
Recognition of water issues in tree growing (2)	3.76	3.96		4.09	3.75
Possession of roof catchment system	18%	13%		3%	0%
Access to borehole water	3%	8%		34% (3)	34% (3)

Remarks: ⁽¹⁾ No statistically significant differences were observed between FFS and Non-FFS households; ⁽²⁾ Answer was on the Lickert scale 1–5; ⁽³⁾ In Taita Taveta, approximately 95% of households with access to borehole water during the dry season were concentrated in a single ward. In this particular ward, the proportion of households with access to borehole water reached 72%. A binary logistic regression analysis conducted for this ward—using access to borehole water as the dependent variable and “FFS graduation,” “gender of household head,” “age of household head,” “education level,” “landholding size,” and “household income” as independent variables—revealed that none of the independent variables had a statistically significant association with borehole water access.

, where non-FFS households demonstrate a similarly high rate of joint decision-making as FFS households, suggesting that women are frequently involved in nursery management. Women’s active role in these activities can be partly attributed to the accessibility of nurseries, which are often located near the homestead and allow for flexible scheduling. This makes them compatible with women’s daily responsibilities observed in the study areas, such as childcare and household chores. In addition, many FFS programs are intentionally designed to promote female participation, and nursery activities offer an accessible entry point that requires relatively low capital investment and technical complexity. The implication is that women’s involvement in nurseries not only enhances household resilience—through improved tree cover and soil fertility—but also strengthens their role in decision-making and contributes to more gender-equitable outcomes in agricultural development.

5.6. Effect on Access to Water

Enterprises that were conducted under farm forestry FFSs of the CADEP-SFM did not specifically focus on water issues. However, water is obviously an indispensable resource for tree growing as well as for supporting a mixed crop–livestock system. Therefore, it is rational to assume that FFS households would pay attention to water issues through learning processes on crop and livestock farming and tree growing, and they might take action to improve access to water.

FGDs in Embu provided stories of how FFS households secured access to water for tree growing. Because average annual precipitation is 640–1495 mm in Embu, the installation of a water tank for collecting rain water is a cost-effective way and relatively popular among FFS households. In Taita Taveta, where the precipitation is 450–500 mm, we did not observe actual cases of FFS households which improved access to water although FGDs suggested that digging a deep well and installing a water pipe from a watershed source are ways to ensure access to water during the dry season.

Table 9 shows the result of semi-structured interviews on water-related issues. These results challenge the earlier assumption that FFS households would have greater awareness of water-related challenges and improved access to secure water sources through their exposure to integrated learning on tree growing and farming systems. In practice, however, FFS households do not show a significant difference in the awareness of the importance of water issues for tree growing and in access to a more secure water procurement system (roof catchment system in Embu and a borehole in Taita Taveta, which are relatively popular in each County). In Taita Taveta, access to borehole water helped households mitigate the losses caused by adverse climatic events, as shown in

Table 10. Effect of access to borehole water on the loss value caused by negative events ⁽¹⁾.

	Embu			Taita Taveta
	Access to borehole	(3)	No	Access to borehole	(3)	No
The number of households	6		254	20		64
Drought (KES) (2)	28,900		23,896	26,385	·	45,414
Crop disease and pest (KES) (2)	6167		6581	4700		3250
Death of livestock (KES) (2)	22,167		6685	35,950		44,477

Remarks: ⁽¹⁾ An independent t-test was used to determine if the access to borehole water significantly affects the impact of “Drought”, “Crop disease and pest”, and “Death of livestock”. The average loss value is from the past one year at the time of the interview; ⁽²⁾ KES: Kenyan Shilling; ⁽³⁾ · indicate p < 0.1.

. It is important to note that access to borehole water does not appear to be determined by household-specific characteristics, including economic conditions, as indicated in the Remark of Table 9. Although quantitative data are not available, discussions from FGDs suggest that boreholes are often installed with support from governments or NGOs, as well as through community level efforts.

6. Discussion

In this section, we discuss the findings of this study. Unlike our initial assumption that farm forestry FFSs could enhance the climate resilience of smallholder households [45], both Figure 3 and Figure 4 show that there is no clear difference in the impact of drought, crop disease and pest and death of livestock between FFS and non-FFS households during the past year. Both groups were severely impacted by events that appear to be climate-related.

The subsequent discussion explores the contributions and limitations of farm forestry FFSs in strengthening the resilience of mixed crop–livestock systems. The focus is on understanding why the anticipated differences between FFS and non-FFS households did not materialize. The discussion is structured around the two complementary perspectives on climate resilience that were introduced earlier in this study: the system-based perspective and the learning-based perspective.

6.1. Mixed Crop–Livestock System Through Farm Forestry FFSs

The findings suggest that farm forestry FFSs contribute to the strengthening and evolution of the mixed crop–livestock system. First, tree-growing activities among FFS households are systematic and purposeful, involving both self-consumption and commercial orientation. Rather than simply planting trees, these households tend to engage in a sequence of coordinated tasks—such as operating nurseries, monitoring market prices, and ultimately generating income. Second, FFS households in Embu demonstrate concern for the soil conditions of farm forestry sites, as trees often coexist with crops in agroforestry arrangements. This aligns with widespread practices observed among tree-growing farmers in other regions [25]. Although no significant difference was observed in Taita Taveta, both FFS and non-FFS households show a higher level of concern for the soil conditions, partly due to severe soil erosion that often occurred in Taita Taveta [46]. One issue that needs to be examined regarding the impact on the expanded mixed crop–livestock system is the result of the tree survival rate, which shows no significant difference between FFS and non-FFS households. Since the FFS program included forestry components such as nursery establishment, tree planting, and management, it might have been expected to result in higher tree survival rates. However, the absence of significant differences suggests that other factors—such as water availability—could have played a more dominant role [47]. One possible explanation is that the relatively low survival rate may stem from inadequate water access during dry seasons—a factor not directly addressed by the FFS program, which focused more on agronomic and forestry techniques than on water infrastructure.

In terms of resilience to climate disturbances, despite the contribution of farm forestry FFSs to strengthening the mixed crop–livestock system that sustains rural livelihood, no statistically significant difference was observed in the impact of drought, crop disease and pest and death of livestock between FFS and non-FFS households. We explore the reasons for this gap separately in Embu and Taita Taveta because the conditions are different. In Embu, although average annual precipitation is much higher than in Taita Taveta, the irregularity of rainfall seems to have affected rain-fed farming [48]. FGD results indicated that some FFS households were coping with the rainfall irregularity by installing a roof catchment system connected to a water storage tank or implementing a drip irrigation system on a farm, but others faced difficulties in water access when river water dried up, indicating a lack of preparation for rainfall irregularity. The FFS program involved conservation practices such as contour planting, composting, and agroforestry; however, addressing water access issues appears to have been insufficient. Similar finding is reported from Odisha state, India that the farmers who are linked to government extension services, including FFSs, are more likely to adopt crop diversification, crop rotation, and drought resistant seeds but the adoption of micro-irrigation is not statistically significant [49].

In Taita Taveta, the main reason is considered as continuous drought that severely affected the county. The lack of water damaged the expanded mixed crop–livestock system, leaving crops unharvested and livestock and trees dead. As the prolonged drought impaired the core interactions among crops, livestock, and trees, FFS households faced severe limitations in applying their acquired knowledge and skills effectively. In other words, the situation can be understood with the ecological threshold concept by which environmental stress leads to abrupt shifts in ecosystem structure and function [50,51]. Once such a threshold is crossed, the system may shift to an alternative stable state, and previously effective management practices become insufficient. In the Taita Taveta case, the prolonged drought likely pushed the mixed crop–livestock system beyond such a threshold, resulting in a systemic breakdown that constrained the applicability of FFS-acquired skills.

This understanding provides a critical insight into the farm forestry FFS approach. As discussed earlier, trees are generally considered more drought resistant compared to grasses and annual crops and play a supporting role in a mixed crop–livestock system. However, securing access to water during the dry season is considered as a precondition for trees to play a supporting role in a system. On the other hand, the survival rate of planted trees is more than 50% even after three consecutive drought years in Taita Taveta. This situation suggests that although trees may no longer support crops and livestock in their typical ecological functions under extreme drought, they continue to provide essential resources such as firewood, timber, and fruits for both subsistence and income generation [52]. This continuing role of trees is observed regardless of FFS participation, as the available data do not show statistically significant differences in survival rates between FFS and non-FFS households. Although timber revenues account for a small percentage of total income at this stage, they are expected to increase as trees grow, and it will still be possible to maintain this income source even if severe drought occurs that heavily damages crops and livestock. Trees may thus continue to play a role in coping with climate disturbances, or an expanded mixed crop–livestock system would still work as shown in the arrow (g) and (d) of Figure 2.

Although FFS households follow the expanded system more explicitly than non-FFS households, several issues remain to be addressed. First, the number of FFS households actively practicing this system is still relatively small. For example, only about 20–30% of the interviewed FFS households reported earning income from wood sales. Second, the quality of activities still needs improvement. No statistically significant difference in tree survival rates was observed between FFS and non-FFS households, suggesting a need for enhanced tree management and maintenance skills and/or a mechanism to improve access to water during the dry season. Third, value addition in tree-based enterprises remains limited. While FFS households are more likely to monitor local wood prices, their income levels from wood sales remain low. To realize greater benefits, more value-added opportunities beyond local markets should be explored and developed. Fourth, due to a lack of longitudinal and event-specific data, this study could not quantitatively assess whether the expanded system is more resilient than the traditional mixed crop–livestock system in coping with and recovering from climate disturbances [53]. Such a quantitative assessment requires data on the scale of disturbances and their potential impact on a rain-fed and mixed crop–livestock farming system, which the current study does not provide.

6.2. Learning Capacity for an Expanded Mixed Crop–Livestock System and Beyond

The study results demonstrate that FFS graduates exhibit a strong and persistent motivation to adopt new farm forestry practices and participate in group activities, even three or more years after graduation. More than 80% of FFS households continue to engage in agricultural group activities, and approximately 10% are actively experimenting with tree-growing and nursery management. This enduring behavioral pattern suggests that learning through FFSs are not merely a temporary intervention but have become embedded in the everyday practices of farm households [54].

Importantly, the continuous experimentation and learning observed among FFS households suggest that the expanded mixed crop–livestock system, as illustrated in Figure 2, should not be understood as static but rather as dynamically evolving. It is continuously being examined, refined, and improved through collective processes centered on agriculture and forestry. These group-based practices—rooted in local realities and reinforced by experiential learning—constitute a foundational element of resilience in the face of climate change [55]. While this understanding may apply in theory to FFS households in both Embu and Taita Taveta, empirical evidence suggests a divergence in outcomes.

As discussed in Section 6.1 and presented in Figure 3 and Figure 4, ongoing group-based learning has not translated into effective responses to climate disturbances. In our cross-sectional data, we do not find statistically significant differences between FFS and non-FFS households in water-related awareness or access, and households’ adaptation efforts have remained focused on agriculture and forestry, with no clear evidence that FFS households more consistently extend actions toward securing access to water—a critical factor in sustaining a functional mixed crop–livestock system during dry seasons. This suggests that water access outcomes are likely shaped by contextual and institutional factors beyond household learning alone, so any interpretation of “FFS not extending to water” should be framed cautiously as a hypothesis for further study. In this regard, current learning processes fostered through farm forestry FFSs remain inadequate for enhancing climate resilience. In contrast, there are FFS cases in Kenya, such as those among tea farmers, where learning has expanded to domains such as health, sanitation, and nutrition security [20]. In the present study sites, however, such horizontal spillovers into non-agricultural domains have yet to emerge, suggesting that while the conditions for transformative learning may be in place, they are largely underutilized.

Unlike farm-based agricultural and forestry practices, which can often be improved through internal household efforts, securing water access—such as through roof catchment systems in Embu or borehole drilling in Taita Taveta—requires capital investment. This includes costs for storage tanks, drilling equipment, and piping infrastructure, which may constitute financial and logistical barriers to adoption. Studies in Niger and Ethiopia found that households with more assets are more likely to adopt irrigation and water conservation techniques [56,57]. Nevertheless, as shown in Table 10, access to borehole water does not exhibit a statistically significant relationship with household-specific characteristics, including income. Instead, it appears to reflect external support mechanisms and the proactive initiative of a community. Although qualitative evidence from FGDs suggests the importance of external support for water infrastructure, further research is needed to systematically examine how access to water interacts with FFS learning processes and practice adoption. This observation highlights possible entry points for intervention through expanded and collaborative learning processes.

To enable these learning processes to transcend current sectoral boundaries and become transformative, the next critical step lies in building stronger linkages between FFS groups and external institutions, including local governments, NGOs, and private sector actors [58]. By fostering such connections, FFS groups can act as catalysts for broader institutional change. For example, coordinated initiatives aimed at improving access to water—in partnership with public agencies—could support infrastructure development, including deep well drilling or the construction of piped water systems from upland sources. Strengthening these linkages should therefore be integrated as a central theme in future FFS curricula, as this is likely to enhance both the scope and impact of learning processes [59,60].

7. Conclusions

Farm forestry FFSs contribute to strengthening mixed crop–livestock systems by promoting tree cultivation, market-oriented practices, and group-based learning. However, these improvements have not consistently translated into reduced impacts from droughts or other climate disturbances. A key limitation lies in the persistent challenge of water access, which remains largely unaddressed. To enhance their impact, future farm forestry FFS programs should integrate transformative learning that goes beyond farm-level practices, enabling farmers to engage with cross-sectoral issues such as water management and institutional linkages through experiential approaches. These broader perspectives and direct cross-sectoral experiences are essential for building sustainable climate resilience in dryland farming communities.