Kalama subcounty, situated in Machakos County (Figure 2) spans approximately 200 km² between latitudes 1°37′S and 1°45′S and longitudes 37°15′E and 37°23′E. The area has experienced extensive environmental transformation, largely driven by human activities. According to Ellenkamp (2004), forest encroachment for settlement expansion, shifting landuse patterns, and widespread charcoal production have resulted in the progressive loss of vegetation cover and accelerated land degradation. The subcounty was selected as a case study because it encapsulates the core challenges and dynamics of food security in Kenya’s Arid and Semi-Arid Lands (ASALs). Administratively, Kalama subcounty consists of four locations and eight sub-locations; Kakayuni being the largest and most ecologically fragile. Its steep hills experience the highest levels of soil erosion, a condition exacerbated by deforestation linked to expanding settlements and cultivation on hilltops. The area is underlain by metamorphic basement rocks, with mountainous zones overlain by shallow, well-drained stony sandy clay loams, while adjacent plains and uplands are dominated by poorly drained black cracking clays and, in dissected uplands, well-drained dark reddish-brown clay and sandy clay soils with limited agricultural potential (Siderius, 1987). Hydrologically, the region is drained by two seasonal rivers—Thwake and Kaiti—which play a critical role in local water supply and agricultural practices. The climate is semi-arid, with a mean annual rainfall of about 602 mm, falling mainly during the long rains (March–May) and the short rains (October–December), with a distinct dry season separating the two (Ellenkamp, 2004). Rainfall is typically more prolonged on the southern and eastern mountain slopes. Temperatures vary considerably, with monthly maximums ranging between 22.2 °C and 27.3 °C and minimums between 11.1 °C and 15.2 °C. Mixed cropping is the principal farming system, featuring maize, pigeon peas, beans, and various fruit trees as the main crops (Onduru et al., 2002), while livestock—especially cattle and goats—are widely kept for milk production and manure to support soil fertility.

The study approach involved a mixed methods design involving both secondary and primary data sets. Quantitative climate and land-use analyses were conducted first to identify key environmental trends and changes in Kalama. These results informed the design and focus of the qualitative interviews and FGDs, which were used to interpret how these changes are experienced locally and translated into food security outcomes through market and institutional processes. Secondary datasets were mainly used for quantitative data analyses on precipitation and temperature trends as well as land use and cover changes from the remotely sensed satellite data. Descriptive analysis of population trends and market prices of selected crops is also done using data acquired from the Kenya National Bureau of Statistics (KNBS) and the Kenya Agricultural Market Information System (KAMIS) of the Ministry of Agriculture. The secondary data acquisition and analysis was followed by the primary data collection involving mainly qualitative data on the drivers of food insecurity in Machakos County using Kalama subcounty as the case study. The impacts of flooding, drought, and other shocks on the availability, stability, utilization, and access to food supplies were investigated through Focus Group Discussions (FGDs) and Key Informant Interviews (KIIs) in November 2024. Observation and interviews with sellers were also done during a visit to one of the main local markets in Kalama. Interview and FGD excerpts are presented to illustrate recurring themes; they are selected for clarity and representativeness rather than verbatim coverage of all responses.

A comprehensive assessment of rainfall and temperature trends was undertaken to characterize climatic variability and detect long-term changes relevant to food security and agricultural resilience in the semi-arid environment of Kalama subcounty.

Monthly time series data on rainfall and maximum surface air temperature for 1981–2024 were obtained from the NASA Prediction of Worldwide Energy Resources (POWER) project. This dataset was selected over ground station data due to its comprehensive spatial coverage, temporal consistency, and minimal gaps critical advantages in semi-arid regions like Kalama where meteorological stations are sparse and historical records are often incomplete or discontinuous. NASA POWER provides quality-controlled, satellite-derived climate data that has been validated for trend analysis and is widely used in data-scarce regions (Stackhouse et al., 2018). This dataset offers consistent and quality-controlled satellite-based climate data, widely used in data-scarce regions. The extracted data were geographically referenced to Kalama subcounty and aggregated into Kenya’s four primary climatological seasons: MAM (March–May, long rains) JJA (June–August, dry season), OND (October–December, short rains), DJF (December–February, crossover season), Data preprocessing included temporal aggregation, quality checks, and gap handling.

To complement the climate trend analysis, satellite imagery, LULC mapping was performed for the years 1990 and 2023 to assess changes in forest, shrubland, and agricultural land, and their implications on environmental sustainability. Landsat imagery corresponding to the dry season (to reduce cloud interference) was used for both years. The images were geometrically and atmospherically corrected, and classification was performed using a supervised Maximum Likelihood Classifier (MLC). Training samples for each land-cover class (forest, shrubland, agricultural land, and others) were derived through image photointerpretation supported by high-resolution historical basemaps (Google Earth Pro) and local expert knowledge. Following established recommendations for supervised classification (Jensen, 2015), approximately 50–60 homogeneous training pixels per class were selected for the 1990 image and 70–80 pixels per class for the 2023 image, with the larger sample size reflecting increased landscape heterogeneity associated with agricultural expansion. To reduce classification bias and spectral confusion, training sites were spatially well distributed across the study area, located within homogeneous patches away from class boundaries, and cross-validated using ancillary data and local knowledge.

The accuracy of the land-use/land-cover classification was evaluated using standard accuracy assessment procedures, yielding an overall accuracy of 0.88 and a Kappa coefficient of 0.81. These values exceed commonly accepted thresholds for environmental and land-cover mapping (≥85% overall accuracy; Olofsson et al., 2014), indicating that the classification results are sufficiently robust for subsequent change-detection analysis and for informing land-use planning and climate adaptation applications.

In order to understand the trends in quantitative data analyses, a guide to key informant interviews and focus group discussion was developed with open ended questions on climate change effects, food access and adaptation strategies. The number of key informant interviews were eight and involved four local farmers, two local administrators, one county government official and one official from an NGO working in Machakos County on food security. This selection ensured representation of key stakeholder groups relevant to climate change, land use, and food security. Data collection continued until thematic saturation was reached, as no substantively new themes emerged in the later interviews. The resulting data were adequate to address the study’s research questions. The information gathered through KIIs was validated through two Focus Group Discussions, one with farmers of different age groups and another one with the youth also known as the Gen Zs. To supplement the secondary data on market prices for foods, an observation sheet was developed to gather primary data during market day.

Information gathered through KIIs and FGDs was organized in the word document application and analysed thematically through coding. The coding was done inductively so as to gather the perspectives of the residents of Kalama subcounty.

This study involved human participants through key informant interviews (KIIs) and focus group discussions (FGDs). Ethical approval was obtained from the National Commission for Science, Technology and Innovation (NACOSTI), under license number NACOSTI/P/25/414302, issued on 16th February 2025. All participants provided informed consent prior to participation. Participation was voluntary, and respondents were assured of confidentiality and anonymity; no personal identifiers are reported in this manuscript.

Long-term trend analysis of annual rainfall in Kalama subcounty shows a statistically significant decline over the 44-year study period (Figure 3), with rainfall decreasing at an average rate of approximately 3.83 mm per year. The coefficient of determination (R² = 0.157) indicates that 15.7% of the variability in annual rainfall is explained by temporal progression. While this R² value is relatively low, reflecting the high interannual variability characteristic of semi-arid climates; the statistically significant p-value (p = 0.007) confirms a non-random, long-term declining trend. This consistent downward trajectory suggests a systematic reduction in rainfall over time, albeit superimposed on considerable year-to-year fluctuations. The declining rainfall trend aligns with broader environmental changes observed in Kalama, where extensive land degradation, deforestation of hilltops, vegetation loss, and charcoal burning have progressively altered the local microclimate (Ellenkamp, 2004). Reduced vegetative cover limits evapotranspiration and weakens local moisture recycling, diminishing the likelihood of convective cloud formation, particularly in the uplands and plains. Additionally, the increasing prevalence of bare rocky surfaces and poorly structured soils has reduced soil moisture retention, further lowering humidity levels that typically support localized rainfall. Beyond local factors, regional climate variability associated with the Indian Ocean Dipole (IOD) and El Niño–Southern Oscillation (ENSO) has contributed to more erratic rainfall patterns across eastern Kenya, often resulting in shorter rainy seasons and prolonged dry spells.

The seasonal rainfall trend analysis (Figure 4) for 1981–2024 in Kalama Ward shows marked differences across the four climatological seasons, driven by both local conditions and large-scale climate systems. The MAM long rains exhibit strong interannual variability, with wet years such as 1998 and 2018 linked to El Niño–enhanced moisture influx, while deficits in 2000 and 2015 correspond to neutral or negative IOD phases that suppress convection. The OND short rains are more consistent but show notable peaks in 1997, 2006, and 2023, reflecting the strong influence of positive IOD and El Niño events, which warm the western Indian Ocean and intensify moisture transport into eastern Kenya. The JJA season remains the driest period, as the ITCZ shifts far north and southeasterly winds bring cool, dry air that limits rainfall. Meanwhile, the DJF crossover season provides moderate but occasionally significant rainfall contributions, with wetter years such as 1998, 2001, and 2006 often following strong OND rainfall associated with El Niño.

Rainfall variability was further assessed using the Rainfall Anomaly Index (RAI) (Figure 5), confirmed alternating wet and dry years as a persistent feature of the regional climate. Positive RAI values (RAI > 0), indicating above-average rainfall, were observed in years such as 1982, 1997, 2006, and 2023, whereas negative values (RAI < 0), representing drought years, were recorded in 1983, 2000, 2009, and 2015.

Seasonal trend analysis (Table 1) using the Mann–Kendall test and Sen’s slope estimation indicated a statistically significant declining trend in MAM rainfall, with a Sen’s slope of −1.32 mm per season per year (p = 0.042). In contrast, OND rainfall showed a weak positive slope of +0.68 mm per season per year, which was not statistically significant (p = 0.15). These results suggest a weakening of the long rains while the short rains remain relatively stable.

The analysis of surface air temperature trends over Kalama Ward from 1981 to 2024 reveals a significant long-term warming signal in maximum temperature (Figure 6). The annual maximum temperature increased at an average rate of +0.23 °C per decade (p < 0.001), equating to approximately 1.0 °C warming over the 44-year period. Seasonal disaggregation indicates particularly strong warming during the MAM season, which showed an increase of +0.25 °C per decade (p = 0.003). April exhibited the most pronounced monthly warming, at +0.32 °C per decade. The OND season also demonstrated a statistically significant warming trend of +0.19 °C per decade (p = 0.012). These warming patterns coincide temporally with the declining trend in MAM rainfall, suggesting a potential shift in regional hydroclimatic conditions.

Concurrently, annual maximum temperatures have risen ~1.0 °C since 1981 (+0.23 °C/decade; p < 0.001), with amplified warming during MAM (+0.25 °C/decade). This aligns with IPCC (2021) projections of accelerated warming in East Africa and suggests emerging aridification, as higher temperatures compound soil moisture deficits during the critical growing season (Cook et al., 2020). For the primary crops of Machakos, this implies heightened vulnerability for moisture-sensitive maize, while underscoring the adaptive potential of drought-resilient pigeon peas. Such trends may intensify crop water stress, reflecting regional observations of compound heat-drought extremes (Zscheischler et al., 2020).

Adaptation strategies require localized innovations, including drought-resistant crop varieties (Wainwright et al., 2019), expanded water harvesting (Oiganji et al., 2025), and integration of seasonal forecasts (Abidoye et al., 2025). Institutional capacity-building at county levels, as recommended by Kenya’s Climate Change Act (2016), will be critical to mainstream climate resilience in rural development.

The analysis of land use and land cover (LULC) change between 1990 and 2023 reveals a significant transformation in the Kalama landscape.

GIS classification results show a dramatic increase in agricultural land from 11.33 km² (8.96%) in 1987 (Figure 7) to 53.72 km² (42.41%) in 2023 (Figure 8), representing an approximate 374% expansion. This surge indicates widespread conversion of shrubland and forest areas into farmland, likely driven by population pressure, rising food demand, and the need for livelihood diversification in response to climate variability (Lambin and Meyfroidt, 2011). The 374% agricultural expansion represents a logical, locally-driven adaptation to population pressure and climate variability, seeking to secure immediate food production. However, this comes at a significant environmental cost: the large-scale conversion of shrubland and forest, which declined by 35 and 44.7%, respectively undermining key ecosystem services. The loss of these natural covers reduces soil organic matter, diminishes water infiltration and retention, increases erosion vulnerability, and degrades the microclimatic buffers that moderate local temperatures and moisture recycling. In the context of climate change, this unregulated land conversion can create a negative feedback loop: it addresses immediate food needs but contributes to increased landscape aridity and systemic land degradation (Wainwright et al., 2019), which may ultimately threaten the productivity gains it initially sought. Therefore, integrated land-use planning and the promotion of sustainable farming approaches—such as agroforestry, water harvesting, and conservation agriculture—are urgently needed to balance immediate food security with the preservation of the ecological foundations for long-term resilience (Rani et al., 2025).

Figures 7, 8: changes from 11.33 km² (8.96%) in 1987 (Figure 7) to 53.72 km² (42.41%) in 2023 (Fi8); visually confirms that agricultural land in 1987 was confined to small, scattered patches, while shrubland dominated the landscape.

Figure 8 reveals the extensive conversion to agriculture (green areas) by 2023, particularly in former shrubland regions, visually representing the 374% expansion quantified above.

Simultaneously, shrubland cover declined by 35%, from 94.14 km² (74.40%) in 1990 to 61.30 km² (48.39%) in 2023. This reduction is associated with increased clearance for cultivation and fuelwood collection, a trend commonly observed in semi-arid Kenyan rangelands (Kiage, 2013). Shrublands play a critical role in maintaining soil structure, regulating surface runoff, and supporting biodiversity. Their loss increases the risk of land degradation and further reduces the resilience of local farming systems to climate shocks.

Additionally, forest cover decreased by 44.7%, falling from 21.07 km² (16.65%) to 11.66 km² (9.20%) (Figure 9). Forests are essential for carbon sequestration, microclimate regulation, and maintaining rainfall patterns. The observed deforestation is likely due to fuelwood demand, encroachment for agriculture, and weak enforcement of forest protection policies (Munyao and Munyao, 2016). This poses serious implications for climate adaptation, particularly in regions already experiencing declining rainfall and rising temperatures.

The kappa accuracy assessment of the classified maps confirmed the robustness of the results, with overall classification accuracy improving from 0.82 in 1990 (Kappa = 0.70) to 0.88 in 2023 (Kappa = 0.81) which validates the reliability of the findings for planning and policy analysis.

These LULC dynamics highlight an interesting relationship between agricultural expansion and shrubland decrease, it may address short-term food production needs, but undermine long-term environmental sustainability. In the context of climate change, unregulated land conversion contributes to increased aridity and land degradation (Wainwright et al., 2019). Integrated land-use planning and sustainable farming approaches—such as agroforestry, water harvesting, and conservation agriculture—are urgently needed to balance food security with environmental protection Rani et al., 2025.

Figure 9 provides an immediate visual confirmation of the 374% agricultural expansion and the concurrent, significant contraction of both forest and shrubland cover between 1987 and 2023.

The study integrates key drivers of food security—population growth, crop productivity, and market prices—by explicitly linking them to the FAO food security dimensions (FAO, 2009). Crop productivity primarily affects availability, as it determines the quantity of food produced locally. Population growth and fluctuations in market prices influence access, shaping household demand, purchasing power, and the affordability of staple foods. Qualitative data from interviews and focus group discussions highlighted that households often experienced insufficient food, particularly during dry seasons, illustrating how these drivers translate into real-world food insecurity. By connecting each driver to specific dimensions of food security, the analysis provides a clearer understanding of how environmental and socio-economic changes impact food security outcomes in Kalama.

Machakos County is considered one of the food insecure counties in Kenya. The county being located in the arid and semi arid zones experiences erratic rains that hinder sufficient food yields to support the growing population.

The population for Machakos County has been rising steadily over the years as shown in the Figure 10 above. There is a need to match food productivity with population growth.

The two main food crops grown in Machakos County are Maize and pigeon peas. Figure 11 shows maize and pigeon production between 2019 and 2023.

Crop productivity in Machakos has been fluctuating with time (Figure 11). Despite the wide range of factors influencing yields per hectare, climatic conditions—especially temperature and precipitation—are among the most significant drivers of productivity. Whereas maize requires high moisture content for optimal productivity, pigeon peas can still thrive in low moisture conditions.

Access to food is not only influenced by food productivity but also by the purchasing power by households. The level of disposable income and food prices are key determinants of households’ ability to meet their dietary requirements. The Figure 12 indicates fluctuations in market prices for Maize and Pigeon peas between 2021 and 2024.

The average price of food crops has been on the rise generally in the Kenyan markets. The rise has been associated with not only the food supply dynamics (e.g., transportation and manufacturing costs) but also the rising cost of production inputs such as fertilizers. With 80% of the residents of Machakos County depending on farming as their source of livelihoods (CIDP, 2023–2027), any disturbance in the agricultural sector is bound to affect their welbeing. Global events such as climate change, the COVID-19 pandemic, and more recently the Russia–Ukraine war have significantly affected market supply chains, leading to fluctuations in food prices. For instance, food production in Kenya was widely reported to have declined during the COVID-19 period. Similarly, the Russia–Ukraine conflict has contributed to increases in global oil prices. Together, these disruptions have created cascading effects on food production, supply systems, and overall market prices. A field visit to the local market revealed that most of the food crops were being supplied from the neigbouring markets and that the prices were fluctuating with time as shown in Table 2.

The market survey indicates that most crops and livestock have experienced price decreases over the past few months, reflecting reduced demand and economic pressures on local consumers. Notable exceptions include spinach, cowpeas, and sisal ropes, which have seen slight price increases. The data also highlight reliance on both local and external sources, with products such as cabbages and sweet potatoes mainly coming from Narok County, and watermelons. These trends suggest ongoing challenges related to household incomes, food availability, and market accessibility in the region. The interplay between the environmental and human factors has led to Machakos County being food insecure. In this paper, the environmental factors are considered as those that impact on food production such as rainfall and temperature variability while the human factors involve decisions and processes that occur at global, regional and local levels that have an effect on the production of crops, prices of food commodities as well as the purchasing power of households. Consequently, the interaction between the two broad factors render the marginalized communities vulnerable to food insecurity as well as other challenges such as gender based violence as explained by one local leader below:

Interview excerpt with the assistant chief in her office at Kyanzasu, Makaveti in Kalama subcounty.

As evidenced in the excerpt above, gendered perceptions of food insecurity were evident in the qualitative data. Respondents described how high unemployment among men, women’s labour migration, and shifting household responsibilities affect food availability, childcare, and nutrition outcomes. Women were often burdened with securing water and food under conditions of rainfall variability and rising water prices, while economic stress was linked to increased household conflict and gender-based violence. These findings reflect how climate variability and livelihood constraints interact with existing gender inequalities to shape household food security, consistent with evidence that gender roles and access to resources are central to food insecurity outcomes in Kenya (FAO, 2021; Njuki et al., 2021).

The food access challenges have affected other aspects of the community for example school attendance and performance by the children as reported during a focus group discussion below:

Interview excerpt during an FGD at St. Stephens Mwanyani Church, Kitumbutu village, Mwanyani sub-location, Kiima Kimwe location, Kiima Kimwe Ward on 16th November 2024.

As a way of assisting communities to adapt to the effects of climate change, the government introduced initiatives such as relief food and cash transfers to the elderly and other vulnerable groups in Kenya. Conversations with the community members have revealed that these initiatives are inadequately addressing climate change vulnerability:

Interview excerpt with the Kalama ward agricultural officer, in Kalama subcounty on 15th November 2024.

Machakos County has long been recognized as a leading example of environmental management, particularly through its soil conservation measures using terracing systems. In the context of climate change and its impacts on local communities, the county continues to play a pioneering role in adaptation efforts. Various initiatives, both community-driven and externally supported, have emerged to help farmers respond to changing climatic conditions.

With the collapse of government support to farmers following the 1980’s world bank conditioned structural adjustment programs (SAPs), the extension service system has been weakened. This was reported by the local administrator as follows:

Interview excerpt with the senior chief in his office at Nziuni in Kalama subcounty on 14^th November 2024.

Therefore, agricultural technical and financial support for farmers has been made by the external partners. One notable externally initiated project is the Circular Regenerative Agriculture Program, implemented by the Hand in Hand NGO in Machakos town. The program aims to enhance productivity and address climate-related challenges by building on existing community innovations. Demonstrations compare traditional and modern production methods, enabling farmers to make informed decisions, while training from extension officers and other partners supports skill development. The program also provides guidance on preparing organic manure and the careful use of herbicides, helping farmers reduce production costs while promoting sustainable practices. It focuses also on women empowerment as majority of the beneficiaries are women.

Interview excerpt with Hand in Hand Business Development Officer on 11th November 2024 at Machakos town.

Results from the field observations and interviews with local farmers reveal a wide array of technologies used to conserve water and the soil. Some of the agricultural innovations in Kalama subcounty included drip irrigation, sand dams, drought resistant crops among others. An interview with one farmer revealed the use of indigenous knowledge and practices in improving farm productivity as a climate resilience strategy. This is reported below:

Interview excerpt with a farmer in Kamweleni Village, in Kalama sub county on 18th November 2024.

The study results also indicate that farmers utilize information communications technology in their agricultural practice. It was reported that weather forecasts are sent directly to the mobile phones of the locals. Also, to access extension services from private providers, one could simply request from their mobile phones as demonstrated by one local farmer;

Interview excerpt with a farmer on her farm at Kalama on 19th November 2024.

The study results revealed other measures by the households have been adapted as a way of increasing resilience to climate change shocks. Examples include self help group initiatives known as merry -go- round and table banking. In a merry go round, farmers provide labour on farms collectively from one household to the next until all the members’ farms are covered. In table banking community members contribute their savings and, from that collective pool, members can immediately borrow short-term or long-term loans. The loans are essential in financing farm inputs for improved crop productivity. Through these efforts, Machakos County exemplifies how communities can combine local knowledge with external support to adapt effectively to climate change and improve agricultural resilience.