Dryland Pastoralism Climate Landscape and Food Security in the Suam River Basin of Kenya

River basins in the drylands of Sub-Saharan Africa have traditionally been utilized for pastoral livelihoods under communal land tenure. Communities in West Pokot in Kenya have continued to experience increased precipitation and temperature as a result of climate variability and change. This study aimed at assessing the impact of climate variability and change at micro-basin level in order to address research and policy gaps on climate change and food security as policy arena shifts from centralized to decentralized governance in Kenya. Primary quantitative data was collected from 387 households’ perceptions of climate variability and change and its implications on food security were measured. Food security index score was calculated. The annual rainfall trend over Suam river basin for the period (1981-2020), was characterized by a linearly increasing annual rainfall trend. Mann Kendall test Z-statistics and Tau were at 2.3578 and 0.0720 respectively. The basin experienced the highest rainfall variability during the first decade (1981-1990) with the highest coefficient of rainfall variation noted at 11.5%. The highest temperature was recorded in the third decade (2001-2010) and fourth decade (2011-2020) at 27.0 and 28.2 degrees Celsius respectively. However, the overall index score for food security was 55.78 with food availability scoring the highest index, mean (SD) of 63.41 (36.52). This was attributed to households’ practice of both nomadic pastoralism and agro-pastoralism activities. Climate variability and change, have resulted in increased amount of rainfall received providing for opportunity investment in rain water harvesting to support both pastoralism and agro-pastoralism production to enhance food security.


Introduction
Climate variability and change have for decades been considered the main risk to agricultural activities, the main means of subsistence and livelihoods of rural poor among smallholder farmers, pastoralists and agro-pastoralists in the drylands of Sub-Saharan Africa. While these climate change risks remain well documented, at international level discussions on climate change mitigation and adaptation remain politically motivated without overarching agreements (Vrålstad, 2010). At national and local levels, policies are created without implementation, however, greenhouse gases emission continues while pastoralists and smallholder farmers are left on their own to struggle with adaptation. FAO (Food and Agriculture Organization of the United Nations (FAO), 2016) indicated that climate change through its impacts on agricultural production, whether smallholder farming, pastoral and or agro-pastoral, will have negative effects on food security in all dimensions. To make it worse, drylands of Sub-Saharan Africa are constrained with limited production capacities since its main natural resources (land and water) are either degraded, scarce and overstretched by the demands of the growing human and animal populations. One way through which climate affects food security, is through its impacts on natural resources such as water and land which are essential in agricultural production (Čadro, Cherni-Čadro, & Žurovec, 2019;Bilali, Bassole, Dambo, & Berjan, 2020). Increased temperatures shift precipitation patterns contributing to unpredicted droughts and floods thus affecting land productivity in different seasons. Studies suggest that without appropriate interventions, climate variability and change will affect agricultural yields, food security and add to the presently unacceptable levels of poverty in Sub-Saharan Africa (Zougmoré, Partey, Oué-draogo, Torquebiau, & Campbell, 2018), dryland river basins included. Further, studies have warned that changes in the mean climatic characteristics will not only affect the hydrological cycle and crop production but also accelerate land degradation and its associated human suffering (Easterling, 2007;Sivakumar & Ndiang'ui, 2007).
Previous studies which utilized food insecurity and climate change vulnerability index have revealed that today the highest levels of vulnerability to climaterelated food insecurity are in Sub-Saharan Africa (Programme, 2017). Climate change will always have far reaching impacts on the agricultural sector, and will actually affect smallholder farmers whose livelihoods are precisely dependent on rain fed agriculture and have a low capacity to adapt (Mashizha, Monga, & Dzvimbo, 2017). Climate change has the potential to transform food production, especially the patterns and productivity of crop, livestock and fishery systems, and to reconfigure food distribution, markets and access. Future impacts of climate change and land cover changes on livestock production, which is the main source of livelihood among the pastoralists and agro-pastoralists, are likely to be both direct through productivity losses owing to temperature increases and indirect through changes in the availability, quality and prices of inputs such as fodder, energy, disease management, housing and water (Thornton, 2010).

N. D. Naburi et al. American Journal of Climate Change
Schilling and others (Schilling, Akuno, Scheffran, & Weinzierl, 2014) observed that in East Africa, debates about climate change mainly focus on increasing temperatures and higher rainfall variability, with a growing likelihood of more frequent and extended droughts. Usually, when droughts occur pastoralists and agro-pastoralists ability to access food and water among the livelihood necessities are threatened (ILO, 2019). Although recently, there have been many discussions on the potentials of drylands resources for economic development in Kenya, climate change has been identified as the main threat to economic development in the Vision 2030 as well as the dryland development blueprint (Mutimba & Wanyoike, 2013). Climate variability and change are projected to increase drought episodes, food insecurity, irreversible decline in herd sizes and deepening poverty among the pastoralists and agro-pastoralists communities in the drylands (Tawane & Wakhungu, 2018). Recent studies on climate change in North Western Kenya found that the average temperature in the county is on the rise. Rainfall patterns have shifted and become more erratic, long rains have shortened and become drier while the short rains have become hotter and wetter, with low levels of annual rainfall (Schilling, Akuno, Scheffran, & Weinzierl, 2014).
Although studies have constantly emphasized that affected communities strive to adapt to climate variability in the dryland of Sub-Saharan Africa (Opiyo et al., 2012), the importance of dryland river basin in balancing the effects of climate variability and food security has received little attention among researchers in Africa. Dryland counties in Kenya, have continued to be warned that the region has recorded increased precipitation and temperature as a result of climate variability and change (MoALF, 2016;Koei, 2013;Shongwe et al., 2011;Kogo, Kuma, & Koech, 2021), these have been done at a macro-level and generalized the effect of climate variability and change on households' food security. Therefore, the need to assess the impact of climate variability and change at micro-basin level was seen as a gap that necessitated this study for purposes of addressing research and policy gaps on climate change, food security and environmental management in the dryland river basins as policy arena shifts from centralized to decentralized governance in Kenya. This is after realization that studies have concentrated on the negative impacts of climate change without emphasizing the opportunities that come with climate change and food security such as increased precipitation in the drylands. Just the same way anthropogenic global warming theorists put it, human activities are the main contributors of emission of Green House Gasses leading to global warming thus changes in climate (Gore, 2006;Intergovernmental-Panel-on-Climate-Change, 2007) in this study we find it necessary also to put forward that same human are in a better position to explore opportunities that come with those changes to ensure water and food security in the dry land river basins.
Widespread changes in climate as well as the environment are not only noted and observed by pastoralists but also agro-pastoralists communities in the dryl-

Study Area
This study was carried out in the drylands of North Western Kenya, Suam River

Study Design
To answer the main research question, the study examined the characteristics of climate variability using two main parameters: rainfall and temperature patterns, and food security in the Suam River Basin. Interdisciplinary approaches involving an exploratory sequential mixed methods design were used. Data on household food security was collected in the period between November, 2019 and November, 2021, a period of two years. Climatic data was obtained through  Monthly gridded Maximum and Minimum Era5 temperature data at a spatial resolution of 0.25˚ by 0.25˚ and spanning the period 1981-2020 were accessed from Copernicus Climate data store, ECMWF reanalysis from global climate.
This was used to reconstruct monthly average temperature dataset for spatial analysis.

1) Mann-Kendall trend analysis
Temporal analysis employed Mann-Kendall trend test which was used to detect the presence of monotonic trends in rainfall within Suam river basin and to determine whether the trend was statistically significant or not Sen Slope estimator (Helsel & Hirsch, 2002). The median of those slopes is the Sen Slope Estimator (Sen, 1968). Since there are chances of outliers to be present in the dataset, the non-parametric MK test is useful because its statistic is based on the (+  (Mann, 1945;Kendall, 1975) and (Yue, Pilon, & Cavadias, 2002) as shown in Equation (1).
2) Spatial Trend Analysis The seasonal and annual spatial rainfall trends were computed in Climate Data Operator (CDO) using the command operator. Analysed spatial rainfall trends (seasonal and annual) were mapped in ArcMap (Schulzweida, Kornblueh, & Quast, 2010). Classified symbology embedded within ArcMap interface was used to reclassify spatial rainfall coverage into different classes based on the rainfall range and variation.

3) Variability Analysis
Coefficient of Variation was used to examine the variability of rainfall at annual scales Hare (Hare, 2003). A high value of CV is an indicator of high variability in rainfall, while, low value of CV is an indicator of low rainfall variability.
Rainfall variability was computed using coefficient of variation (CV). The CV is as shown in Equation (2).
where CV is the coefficient of variation; S is sample standard deviation of the rainfall and X is the sample mean of rainfall. CV was computed by dividing rainfall sample standard deviation by rainfall sample mean and expressing as a percentage.

4) Household survey
Purposive sampling was used to select the two sub-counties within which Suam river traverses Pokot North and West Pokot in West Pokot County, thus forming a common hydrological basin. Quota sampling was used to select respondents who constituted focus group discussion teams. Primary quantitative data was basically drawn at the individual household level. A two-level multistage sampling was conducted to select a representative number of households.
In the first level, simple random sampling technique was used to select at least 10% of the locations hence two locations from each of the sub-counties whereas in the second level, two sub-locations from each selected location were identified using simple random sampling technique.
Proportionate sampling was used to distribute the samples in the sub-locations based on their population in the sample frame. Finally, a simple random technique was used to select the households that formed the unit of analysis while the household heads formed the unit of observation during data collection process. A sample size of 387 was obtained using Yamane's formula for small  (Yamane, 1967). A structured questionnaire with both close and open-ended question was use to collect data from the households on their perceptions on climate variability, change and food security.

5) Food security index score
The 17 constructs of dependent variables (food security) were re-coded into binary (0 "no" and 1 "yes") outcome where "1" indicated presence of food security and "0" showed food insecurity in the households. All the 17 variables for the food security dependent variable were included in calculation of index score of food security since the reliability tests showed tight coherence with a Cronbach's alpha of 0.917 showing high quality data. In order to generate coherent results, we singled out "1" as an interested figure and summed up to generate sum scores from 0 to 17 using the formula in Equation (3).
where; FSSUM = Food security sum score; F 1 =variable 1 which is "HHs willingness to change food production"…, and F 17 is last variable which is "HHs skills and knowledge to ensure good nutrition, food safety and sanitation".
The results (sum scores) were linearized by subjecting to percentage i.e. on a scale of 0 -100 interval) for each household and a new variable called "food security index score" was generated by applying formula in Equation (4).
where FSI = Food security index score The above formulae were used to generate the index scores for each category of food security for instance availability of food index score. Thereafter, mean and standard deviation were computed to measure central tendency and dispersion of the data in addition to overall index score for food security. The overall index for food security score was then categorised into two namely: Food insecure HHs (0) and Food secure HHs (1) implying those households scored FSI less than 50% and 50% and more, respectively.       (Table 3) The increase in rainfall trend as further explained by Sens's slope (2.4760 and 2.3413) for JJA and SON respectively is attributed to climate variability and climate change which have impacts on systems controlling climate over the region like Congo air mass, topography and inter tropical convergence zone. The spatial-temporal trends and variability in rainfall which have been observed in the region are important for planning of pastoral and agro-pastoral sectors in the study area (Muthoni et al., 2019) It is important to note that although MAM depicts a linear trend of seasonal rainfall, the amount of rainfall received remains higher compared to JJA, SON and DJF respectively. This could be attributed to the global systems especially the ITCZ that traverses the East African region setting the beginning of the long rains season. These seasonal rainfall patterns are similar to those that sweep over the Lake Victoria Basin N. D. Naburi et al. from Kisii station in the Southern part and Kapcherop, Elgeyo Marakwet Kapenguria towards the northern western part covering the dryland sub counties (Mugalavai et al., 2008;Kipkorir, Raes, Bargerei, & Mugalavai, 2007) During the focus group discussions, agro-pastoralist indicated that their farming activities were based on traditional rain patterns MAM, this finding however, reveals the need to shift to JJA in order to maximize on use of rains received within this period. It was further observed that best practices require integration of both community indigenous knowledge (IK) and the conventional methods used by the Kenya Meteorological Department (KMD).

Rainfall Variability
As depicted from the decadal coefficient of rainfall variation results, Suam river

Spatial Temperature Patterns
As illustrated in Figure 4 above, historically the basin has been under high temperatures which are attributed to its location in the desert climate; hot and dry.
The highest temperatures noted were in the third decad (

Households Level of Food Security
According to the World Food Summit (1996), globally the multidimensional nature of food security includes an analysis of: food access, food availability, food use and stability (FAO, 1996). Despite the increasing threats caused by climate variability and change in the dryland river basin of Suam, the results in Table 4 show a contrary perception from the households on their level of food security. To measure households' food availability as an aspect of food security, five indicators were tested: change in food production practices; access to productive technologies and practices; access to resources, labour, finance, agricultural inputs; secure and timely access to fertile land, water and ecosystem services; and knowledge and skills to improve food production. Results show recorded households' food availability mean index of 63.41 with a standard deviation of 36.52. This finding implied that food availability in the basin was above average.
In the focus group discussion, participants attributed this to the practice of nomadic pastoralism which was more resilient during droughts and agro-pastoralist activities which are mainly practiced during the rainy season along the river, and trade between the borders with Uganda and Kenya. El Bilali and others (Bilali, Bassole, Dambo, & Berjan, 2020) noted that in Sub-Saharan Africa where most American Journal of Climate Change  agreed with the earlier conclusion that rural livelihoods and as a result rural populations will suffer from the increase of food prices as well as the negative impacts of climate change on their sources of income and livelihood strategies relating to agriculture (Bilali, Bassole, Dambo, & Berjan, 2020). According to El Bilali and others (Bilali, Bassole, Dambo, & Berjan, 2020) climate variability and the increasingly frequent and intensive extreme climate events could affect the stability of food availability, access and use. In this study, households' food stability was measured using six variables: growing of climate adapted crops/breeds; households are energy efficient; land restoration including soil and water conservation and management; having and implementing preparedness plans to protect lives and assets; having coping strategies; and existence of resources and income which can be mobilized by households. The results show household stability of food mean index was 37.11 with a standard deviation of 51.94. From this finding, it is imperative to note that efficient responses to climate change require an understanding of the full spectrum of potential climate impacts on food utilization, access and availability, as well as on the underlying natural, built and governance systems in the dryland river basin (Keller et al., 2018).
The overall index score for food security in the basin was 55.78 implying that majority of the households were food secure based on the seventeen variables that were used to measure food security (Table 4). Therefore, the results show  (Kogo, Kuma, & Koech, 2021;Demombynes & Kiringai, 2011;Rao, Ndegwa, Kizito, & Oyoo, 2011).
Interestingly, results in Table 5 show that the type of ecological zone where the household is located did not show statistical significance with food security.
Households main source of food was strongly related with food security at 1 % significance level (p-value = 0.003). Own farm activities were seen as main source of food followed by the buying. Further, the results show that the household main occupation and main source of water were found to be statistically significant to the household level of food security. Interestingly, household average monthly income did not show statistical significance with household food security status.
The results in Table 6 show the association between climatic variables (rainfall, temperature and humidity), climatic hazards (droughts, floods, landslides and soil erosion), and water variability with food security. There was a weak, positive correlation between food security variables and climatic variables including water. For instance, climate variables and food security index scores were positively related though slightly weak at r = 0.23 and statistically significant at 1% significance level at p-value of 0.000***. Overall, all the four categories; climate variables, climatic hazards/risks and water variability showed statistically significant relationships with the overall food security index score at p-value of 0.000*** although the correlation was weak i.e. 0.1 > r < 0.4. The overall food security index score showed strong positive correlation with overall climatic variability index score at 1% significance level. Although there was positive relationship between the two variables, the strength was weak (r = 0.27).
Therefore, the variables were significant in determining the status of dryland pastoralist and agro-pastoralist households' food security along the Suam river N. D. Naburi et al.
99% confidence level, compared to temperature at p-value of 0.046 (5% level of significance). Climate hazards like floods and landslides on the other hand showed strong relationship(s) with food security at p-value of 0.000, 99% confidence interval compared to droughts at p-value of 0.056, 10% level of significance; and pests and livestock diseases, and gullies at p-value 0.025, 5% level of significance.

Conclusions and Recommendations
From this study, rainfall and temperature patterns show increasing seasonal trends though with a lot of variations, making it unpredictable and unreliable for traditional pastoral and agro-pastoral planning and utilization of river basin resources. Although agro-pastoralists largely rely on traditional methods (IK) in predicting climatic events it was observed that best practices require integration of both these methods and the conventional methods used by the Kenya Meteo-American Journal of Climate Change rological Department (KMD).
The level of food security remains to be above average as households continue to diversify pastoralism and agro-pastoralism activities. Agro-pastoralist farming activities were based on traditional rain patterns of MAM however, there is a need to shift towards JJA and SON with significant increasing trends in order to maximize the rains received within these seasons. Increase in rainfall however, is an opportunity that has not been exploited for intensification of agro-pastoralism production that comes as result of climate variability and change in the dryland of Suam River basin. Furthermore, it is also important to note from the findings that to ensure food security given the current threats of climate variability and change in the dryland river basin of Suam, it is necessary for the policy makers and implementers to ensure a balance in policy interventions that will address the needs of both the pastoralist and agro-pastoralist communities.
The study further established that efficient responses to climate change require an understanding of the full spectrum of potential climate impacts on food utilization, access and availability, as well as on the underlying natural, built and governance systems. Targeting only one mode of livelihood is likely to increase vulnerabilities to the impacts of climate change. It is therefore necessary to promote both pastoral and agro-pastoral policies and interventions in the dryland river basins to enhance food security. This could be achieved through mainstreaming pastoralism and agro-pastoral activities in devolved policies on climate change, environment and agriculture as a supportive system.
Tapping into existing institutionalized modes of climate change governance provided under decentralized governance systems including the county government system, regional development authorities and devolving financing mechanisms could provide more opportunities to utilize the increased rainfall for agricultural production for food security.