Research

INTRODUCTION

The United Nations’ Sustainable Development Goals (SDGs) are a country-specific development agenda that foregrounds the role of functioning ecosystems in achieving sustainable improvements in human well-being (Griggs et al. 2013). A defining feature of the SDGs is the interaction of the 17 constituent goals (Smith et al. 2018). These SDG interactions are modulated by the specific socioeconomic context of each country, with interactions manifesting over diverse time horizons (Scherer et al. 2018, Nilsson et al. 2018). Among the 17 goals, the second goal of achieving food security (SDG 2) has been recognized as a nexus issue: i.e., across contexts, it is characterized by repeated interactions with multiple other goals (Bleischwitz et al. 2018). For example, pathways to achieving SDG 2 result in direct impacts on land use and water, which can in turn impede progress toward achieving provisioning of clean water (SDG 6) and terrestrial biodiversity conservation (SDG 15; Pham-Truffert et al. 2020). Identifying effective policy interventions to achieve food security therefore depends on understanding SDG 2 interactions and balancing the resultant synergies and trade-offs with other sustainable development imperatives (Nilsson et al. 2016).

The nexus character of SDG 2 is exemplified by conditions prevailing in the arid and semi-arid lands (ASALs) of Kenya (Stavi et al. 2021). The ASALs cover approximately 80% of Kenya’s land area and are home to a third of its ethnically diverse population, which experiences chronic food insecurity (Oba 2001). Rural populations in the ASALs are predominantly engaged in traditional pastoralism centered on the raising of cattle and other livestock on rangelands (Ng’ang’a et al. 2016). Seasonally moving their households and herds to access widely dispersed critical resources, i.e., strategic mobility (Krätli et al. 2013), is a defining practice of traditional pastoralism and underpins the ability of pastoral households to withstand climatic shocks (Galvin 2009, McPeak and Little 2017). Over the past decades, forces operating at diverse scales within and beyond Kenyan pastoral systems have hindered this strategic mobility and modified pastoralist strategies, with cascading impacts on the rates of poverty (SDG 1) and food security of pastoral people, as well as changes in carbon sequestration and cycling (SDG 13; Reid et al. 2014). These forces include population growth, sedentarization and adoption of agriculture by pastoralists, livestock disease, and changes in land use driven by economic development and conservation interventions. Pastoral mobility also mediates the complex relationship between food security and patterns of ethnic conflict in the ASALs (Anyango et al. 2017). For example, movements of pastoralists into agricultural areas, or into pastures across ethnic and national boundaries during periods of extreme drought, may precipitate new conflicts or inflame existing tensions (Berger 2003, Kuznar 2005).

Maintaining food security in Kenya’s ASALs without undermining the other SDGs associated with this diverse region is imperative. The policy coherence necessary to achieve this goal requires a deeper consideration of the spatiotemporal heterogeneity in SDG 2 interactions and their relationship with existing livelihood strategies practiced by pastoral households. Of relevance today is the question of how the Kenyan government’s response to global calls for intensifying biodiversity conservation efforts (SDG 15) will interact with goals to improve food security in the ASALs. For example, the 30 by 30 initiative envisages, through multiple, national-level commitments, setting aside 30% of the planet for biodiversity conservation (Target 3, CBD 2022). This 30 by 30 goal is one among several such global targets that have emerged in response to documented global declines in biodiversity (Butchart et al. 2010). The achievement of many of these targets involves the strategic deployment of land sparing– and land sharing–based conservation measures. Whereas the language associated with these policies increasingly reflects a strong commitment toward equity-based and participatory conservation, they do not fully contend with the wider political, social, and economic ripples that protected area (PA) creation and management can set in motion (Brockington and Wilkie 2015, Gurney et al. 2023).

Several iconic PAs exist within Kenya’s ASALs and along its borders within nations such a Tanzania, Ethiopia, and Uganda. Whereas these PAs are known for their exceptional biodiversity, they also often enclose key wetlands and grazing pastures that are of critical importance to pastoral groups, particularly for surviving periods of drought. In conjunction with other drivers of change, constrained access to critical resources via PA establishment has led to changes in pastoral diets and lifestyles, and has introduced new resource conflicts (Reid et al. 2004, Kieti et al. 2020). Numerous studies have explored pathways by which access to key resources impacts pastoral household well-being, as well as how PAs can serve both to impoverish and enrich pastoral populations in their vicinity (Boone et al. 2011, Brockington and Wilkie 2015, Mojo et al. 2020). Yet, given the large variation in the size of PAs in the region, the availability of pastures within and around PAs, and the pastoralism practices of households in their vicinity, PA access is likely to be of varying importance to food security across Kenya. Similarly, the role of PA access in contributing to conflicts among pastoral groups is also poorly understood (Berger 2003, Greiner 2012). Consequently, ensuring that efforts expended toward achieving SDG 2 and SDG 15 in the ASALs are effective and sustainable requires engaging with the enmeshed character of these goals. To this end, we endeavored to understand how resources within existing PAs in Kenya’s ASALs contribute to pastoral household well-being, particularly household food security and exposure to inter-ethnic conflicts.

We explore this question using our newly developed agent-based model called SPIRALL. We begin first with a detailed description of SPIRALL, including a short review of its antecedents. We then demonstrate, using a baseline simulation, SPIRALL’s validity as a tool to explore our questions of interest. We then conduct a scenario analysis to explore how PA access modulates food security, poverty, and inter-ethnic conflicts across Kenya’s ASALs.

MODEL DESCRIPTION

Overview

The need for heuristic tools to examine SDG interactions has been reiterated (Nilsson et al. 2016, Breuer et al. 2019). Yet, no modeling tools exist to explore the social-ecological drivers and interactions of SDG 2 across Kenya. Scenario analyses using discrete-event simulations have been used to explore and explain the complexity of pastoral social-ecological systems in eastern Africa and more specifically in Kenya (Galvin et al. 2006, Boone and Lesorogol 2016). For example, the agent-based model RiftLand (Kennedy et al. 2014) simulates household decision-making over a large swath of eastern Africa. The model represents household responses to their environment with high granularity; however, the impacts of these household decisions on the environment are not considered. Household decision-making models such as Pastoral Household Economics and Welfare Simulator (PHEWS; Thornton et al. 2003) and Decisions under Conditions of Uncertainty by Modeled Agents (DECUMA; Boone et al. 2011), are distinguished by their explicit linkage with models that simulate ecosystem services. These models simulate the reciprocal interactions between pastoral and agro-pastoral households and the environment, albeit at smaller spatial scales such as counties.

Building on these previous efforts we developed Simulating Prosperity in Rural and Land-based Livelihood communities (SPIRALL), an agent-based (Bonabeau 2002), pastoral-household decision-making model tailored for the ASALs of Kenya. SPIRALL is written in NetLogo Version 6.1.1, a multi-agent modeling environment (Wilensky 1999). We linked SPIRALL to L-Range, a localized version of the global rangelands model G-Range (Boone et al. 2018, Sircely et al. 2019), and used L-Range to simulate ecosystem function across Kenyan rangelands. The resulting coupled social-ecological systems model pivots on the diverse pastoral mobility strategies (non-sedentary, sedentary, agropastoral) and includes an explicit rendering of the conversion of ecosystem production to household calories and income. We used the Overview, Design concepts, and Details protocol (ODD) to describe SPIRALL (Grimm et al. 2020). We summarize the ODD here and include a detailed description in Appendix 1.

The purpose of SPIRALL is to enable the exploration of SDG 2 outcomes and interactions across Kenya. At its broadest scale, Kenyan food security is the ultimate outcome of a complex interplay between national-level policies, environmental conditions, international markets, as well as global political and environmental shocks (FEWS NET 2017). However, SPIRALL primarily serves to describe the set of intrinsic social and ecological factors that may predispose different parts of the ASALs to food insecurity (Edmonds et al. 2019). Specifically, the model allows us to learn, via scenarios, how the environment, and policies that can modify pastoral household interactions with their environment, can modulate food security and other indicators of well-being in Kenya’s ASALs. SPIRALL favors a granular description of household-level behaviors, over a comprehensive integration of cross-scale drivers of food security. Consequently, it serves as a tool to understand social-ecological drivers of vulnerability to food insecurity, but not as a forecasting tool. Patterns in food security across Kenya, seasonal movements, and calorie consumption by households serve as model evaluation criteria.

SPIRALL is composed of two types of entities: patches and pastoral households (agents) distributed on these patches (Fig. 1). Each patch is 10 x 10 km and together the patches represent 1.94 million km² of eastern Africa. A set of variables define patch and household attributes (Table 1). Patch attributes together determine the social-ecological character of each patch. Household attributes determine the composition of the household, its socioeconomic characteristics, and pastoral practices, i.e., non-sedentary, sedentary, or agropastoralist. A set of parameters that are common to all households are also defined (Appendix 2). These include parameters that modulate movement decisions of households or livestock herds, livestock herd dynamics, livestock diet composition, economic transactions, and social interactions. Simulations in SPIRALL typically span multiple years, such as 20 years in our scenario exploration, with households making decisions at monthly time steps. Simulation outcomes can be summarized at national and sub-national spatial scales and at temporal intervals of a single month or longer. Processes occurring within patches, e.g., distribution of agents or foraging by livestock, are spatially implicit.

At every time step the ecosystem model L-Range reads in monthly weather data, i.e., precipitation and minimum and maximum temperature, and updates the availability of biomass across eight biomass pools for all patches within SPIRALL. This biomass availability is used to determine maximum stocking density of different livestock species on each patch. Households use the information on stocking density to move the entire household or their herds to appropriate patches. These movements are governed by rules specific to the type of pastoralism they practice, i.e., non-sedentary, sedentary, or agropastoral (Table 1). Households try to graze herds within county boundaries when possible. When a household or its herds crosses the county boundary, they are labeled as at risk for conflicts and may lose their herd at a fixed probability. Once all households have located themselves on appropriate patches, livestock species consume and deplete specific biomass pools. Livestock change weight based on the total energy gained and lost in accessing food. During reproductive months, i.e., April for all livestock, surviving individuals of livestock reproduce probabilistically and commence lactation. During harvest months, i.e., July, agropastoral households harvest maize. Each month, households attempt to meet calorie requirements using household sources such as milk, meat from dead animals, and purchased or harvested maize. Each month households attempt to meet expenses using available cash such as earnings from labor, business, and sale of livestock and crops, or make purchases when available cash exceeds expenses. Households who have failed to meet their calorie needs may receive milk from neighboring households who have some to spare. Similarly, households who have lost all their livestock may receive cattle as gifts from other wealthier households in their social network. Households that fail to meet calorie needs each month are labeled as food insecure and those that fail to meet monthly expenses are labeled as in debt. At the end of each month, L-Range reads from SPIRALL the fraction of biomass from each pool lost to grazing on each patch and uses that to continue simulation of ecosystem dynamics.

A design concept that underlies SPIRALL is that pastoralists have knowledge of the availability of forage in a subset of patches. Additionally, resource access is constrained by movement rules associated with the type of pastoralism practiced by the household. Each month households attempt to increase their herd size while minimizing the potential for conflicts with other ethnic groups. Finally, the reciprocal interactions between households and the environment hinge on explicitly tracking the conversion of ecosystem production to livestock and agricultural production in SPIRALL and tracking the ecosystem-wide impacts of grazing in L-Range.

Initialization

Households were initialized by distributing them across 14 ASAL counties in Kenya where pastoralism is practiced by more than 10% of the population (Table 2). A total of 10,844 households were simulated, representing approximately 5% of the pastoral population across these counties in the year 2000. Household distribution followed census data (SEDAC 2016) and the estimated pastoral population within each county (Krätli and Swift 2014). Household members were assigned such that each household had at least one adult male and female member with a mean household size of eight. Livestock holdings and external income sources of households were set to scale positively with the number of members in the household. We summarized livestock herds owned by a household using tropical livestock units (TLU), where one TLU represents a 250-kg animal. We also standardized the number of members in a household in terms of adult equivalents (AE; Appendix 2). We assumed that agropastoral households conduct rainfed agriculture and only grow maize with a maximum possible maize harvest at 606 kg / ha (Thornton et al. 2006) and annual harvests varying in proportion to the green herb biomass simulated by L-Range for the patch. In the simulation, pastoral households can know the quality of, and access patches within a 100-km radius around their home patch (Table 1; McCabe 2011). Finally, we assumed that there is no human population growth. To stabilize household behaviors, a five-year spin-up simulation for all households was conducted using randomized weather data for the period between 1980 and 2019. The state of all households at the end of this simulation was stored in a file and used to set initial conditions for all subsequent simulations.

SIMULATION EXPERIMENTS

Baseline simulation

We explored household behaviors across Kenya under a baseline scenario that allowed pastoralists minimally constrained access to livestock-grazing pastures, including those within PAs occurring within their movement orbits. Even though these PAs are intended as inviolate spaces, the use of these areas by pastoralists is accommodated to an extent (Butt 2011). The SPIRALL L-Range coupled model was set up to represent climatic, environmental, and demographic conditions that were extant between the years 2000 and 2019. We conducted 20 runs of the baseline simulation, and calculated means and standard errors for variables of interest. An exploratory analysis revealed that 20 simulations yielded narrow standard errors (< 2% of the mean) around variable means.

Evaluating a model such as SPIRALL that spans large spatial scales and a diversity of pastoral practices and land tenures represents challenges resulting from the uncertainties associated with parameters and the necessary simplification of complex social practices. We therefore relied on the principles of pattern-oriented modeling (Grimm et al. 2005, Gallagher et al. 2021) to assess our model’s ability to capture spatial and temporal trends relevant to exploring to our research question. Pattern-oriented modeling entails ensuring that simulation results match observed patterns in the study system across diverse scales. We assessed spatial and seasonal trends in rates of food insecurity across the ASAL counties and calorie consumption patterns. We compared estimates of food security from SPIRALL with those reported by the Famine Early Warning Systems Network (FEWS NET; FEWS NET 2017). For the period between 2010 and 2015, FEWS NET provides quarterly reports of food security outlooks for each county in Kenya. For this period, Kenyan counties are classified based on the food insecurity severity scale into five classes: (1) no acute food insecurity; (2) moderately food insecure; (3) highly food insecure; (4) extremely food insecure; and (5) famine. We created a numeric classification scale by assigning values from zero to four to these five classes. For each county, for the period between 2010 and 2015, we created an index of food security by summing together the quarterly food security scores; counties with higher scores could be interpreted to be more chronically food insecure. We then ranked these counties based on their food security scores. Similarly, we calculated a seasonal food security index for each county by separately summing the food security scores for the months of January, April, July, and October, months for which FEWS NET provides immediate or current food security projections.

Alternative Scenario: No PA access

We explored the impacts of PA access on measures of household well-being by comparing our baseline simulations against a scenario where households were denied access to PAs within their movement orbits. PAs (Fig. 1) cover approximately 9% of the study region. Households were denied access to both strict PAs (IUCN category I–VI; UNEP-IUCN 2018) and other protected patches within their movement orbits (Fig. 1). Our intention was to emulate a scenario where PAs are maintained as inviolate spaces as envisaged in land-sparing conservation initiatives (Phalan et al. 2011), where landscapes are composed of protected patches and areas of intensified agriculture. We conducted 20 independent runs of this alternative scenario and report mean changes in rates of hunger, debt, and conflicts within different Kenyan counties and household types relative to the baseline simulations.

RESULTS

Baseline simulation

Overall, the baseline simulations were stable, evidenced by low variability in mean annual estimates of per-capita livestock holdings for counties across repeated model runs (Fig. 2). Climate was a strong driver of SPIRALL dynamics. Fluctuations in livestock holdings were consistent with precipitation trends between 2000 and 2019. For example, counties such as Turkana and Laikipia, which experienced steady increases in precipitation, were also characterized by small increases or stable livestock populations. Similarly, in counties where precipitation levels were low and fluctuated substantially, e.g., Wajir (Fig. 2), livestock holdings were characterized by declining trends.

Seasonal trends

We summarized results for each year over quarterly intervals. These intervals approximately coincide with the four seasons in the ASALs (Table 3; Little et al. 1999). The fraction of food-insecure households was higher in the short-wet and late-dry seasons than in the wet season. These seasonal trends in vulnerability to food insecurity mirror long-term trends in food security predictions for the region (FEWS NET 2017). In the wet season, on average, households were able to meet up to 57% of their calorie needs from milk and meat derived from their own livestock herds (Little et al. 1999, Thornton et al. 2003).

Household incomes peaked in the wet season with a large increase in the fractional contribution from the sale of milk. Income from livestock sales were highest in the short-wet and late-dry seasons but did not exceed 10% of the overall income earned. The average distance traveled by households or their herds each month was lowest in the short-wet season. The mean distance traveled in the wet season does not represent households moving to find pastures because in SPIRALL, non-sedentary household agents are programmed to return to their home patches and stay there for the duration of the wet season. The fraction of households crossing county boundaries was highest in the late-dry season. We interpret these county crossings as an index for potential inter-ethnic conflicts. Households accessed protected areas in all seasons.

Spatial trends

We calculated a county-specific index of vulnerability to food insecurity by summing together the fraction of food-insecure households each month in the period between 2010 and 2015. We then ranked the counties based on this index, from most food secure, i.e., low vulnerability, to least food secure, i.e., high vulnerability (Table 4). We compared these ranks with county rankings qualitatively, based on FEWS NET food security predictions and using a Spearman’s Rank correlation coefficient. County-wise food security, i.e., vulnerability, predictions from SPIRALL and FEWS NET showed moderate positive correlation (ρ = 0.46). Relative to FEWS NET predictions, SPIRALL simulated higher rates of food insecurity, i.e., vulnerability, in Baringo and Narok Counties and lower rates in Tana River, Marsabit, and Isiolo Counties. Trends in the fractions of households failing to meet their monthly expenses in each county scaled positively with food security trends.

The mean distance traveled per move to access pastures varied considerably across counties. Households in northern counties such as Turkana, Mandera, Marsabit, Samburu, and West Pokot moved greater distances to access pastures relative to southern counties (Table 4). The average household moved 2.6 times each year. Use of protected areas by households for livestock grazing was higher in counties such as Narok, Laikipia, Kajiado, and West Pokot that border or contain large PAs. Nearly 62% of households in Narok County accessed the adjoining Masai Mara National Reserve and Serengeti National Park located in Tanzania. The number of households crossing county boundaries to graze their livestock, an index for conflict potential, was higher in northern counties such as Turkana, Baringo, Mandera, and Wajir. In counties such as Turkana and Mandera that border other nations, a large fraction of trans-boundary movements were into Uganda and Ethiopia respectively.

No PA access

Loss of PA access caused a decline in household livestock holdings. Agropastoral and non-sedentary households experienced the steepest declines (A, Fig. 3). The loss of PA access increased the annual fraction of non-sedentary and agropastoral households experiencing hunger. However, in both scenarios, agropastoral households experienced lower rates of hunger than non-sedentary and sedentary households (B, Fig. 3). In the baseline scenario, all pastoralist household types experienced increases in their large livestock, i.e., camel and cattle, holdings over a 20-year period, with the largest percent increases occurring among agropastoral households followed by non-sedentary households. However, when PA access is lost, agropastoral households experienced the largest percent declines in their large livestock holdings (Fig. 4). Small livestock numbers declined for all households in the baseline simulation with the largest declines occurring among non-sedentary households. Agropastoral households on the other hand experienced an increase in small livestock numbers when PA access was lost. Households across all counties experienced changes in hunger, debt, and trans-boundary movements. Increases in the incidence of hunger and debt were highest in Kajiado, Laikipia, Narok, and West Pokot where, as per the baseline simulation, a large fraction of households was dependent on PAs (Table 1, Fig. 5). Counties such as Garissa, Wajir, Tana River, and Isiolo experienced the smallest increases in rates of hunger and debt. Losing access to PAs also resulted in changes in the rates of trans-boundary movements with the sharpest increases occurring in Narok, West Pokot, Laikipia, and Kajiado (Fig. 3; Fig. A3.1, Appendix 3). County-specific impacts on other measures of household behaviors such as effects on income earned from the sale of livestock are included in Appendix 3 (Tables A3.1 and A3.2, Appendix 3).

DISCUSSION

Loss of PA access

Our scenario analysis reiterates the nexus character of SDG 2 across Kenya’s ASALs. Within this region, household food security is sensitive to PA management and in turn can affect household exposure to poverty and inter-ethnic conflicts. Regardless of the pastoralism type practiced, PA access was linked to household hunger and debt, suggesting that PAs across Kenya’s ASALs harbor resources critical to ensuring pastoral well-being (Boone et al. 2011). Declines in well-being stemmed primarily from declines in herd size and changes in herd composition. In our baseline simulations, small stock made up the larger fraction of household herds. However, when households lose access to PAs, sedentary and agropastoral households experience further skews in herd composition in favor of small stock, mirroring trends observed among sedentarizing households (Österle 2008). Such changes in livestock composition in the vicinity of PAs can drive changes in the ratio of woody and herbaceous vegetation with consequences for wild browsers, as well as for rates of carbon cycling (Österle 2008, Veldhuis et al. 2019).

Impacts of losing PA access were spatially heterogeneous and were most pronounced in small counties such as Narok and West Pokot, which share extensive boundaries with PAs. Households in SPIRALL can access patches within a 100-km radius of their home patch. Therefore, a larger fraction of households within small counties have movement orbits that can overlap adjoining PAs, partially explaining our results. Across eastern Africa, pastoralists are increasingly sedentarizing or adopting agropastoral lifestyles in response to changes in climate, land tenure, market, and services access (McCabe et al. 2010, Galvin 2021). In our simulations, despite their adaptive advantages, agropastoral and non-sedentary households experience the largest increases in hunger and herd declines when access to PAs is lost. Simultaneously, under both scenarios, non-sedentary and agropastoral households experience the highest and lowest rates of hunger respectively. These trends arise both from the specific ways by which these households translate ecosystem production into calories and sources of income, and because of their geographic location. For example, agropastoral households are typically located in areas that receive higher precipitation and rely on both livestock and agricultural production to meet calorie needs.

Agropastoralism is dominant in the productive southern parts of Kenya, where the largest PAs in the region are located. Increasing adoption of agropastoralism around these PAs has been a leading cause of rangeland fragmentation, disrupting both livestock and wildlife movements (Thornton et al. 2003, Reid et al. 2004, Veldhuis et al. 2019). For example, in Narok County, which adjoins the Masai Mara National Reserve, a large fraction, nearly 60% in SPIRALL, of the modeled households are agropastoral. Crop cultivation helps reduce overall food insecurity among agropastoral households. However, these households maintain large cattle herds by strategically moving them to suitable pastures and depend on livestock sources for a significant fraction of their calorie needs. When PA access is lost, cattle herds experience declines owing to competition for limited pastures. On the contrary, non-sedentary pastoralism dominates in the northern counties, i.e., Turkana, Mandera, and West Pokot, characterized by low rainfall. Within these regions the effects of loss of PA access potentially exacerbate the patchiness in the availability of grazing resources. Whereas in reality, non-sedentary pastoralists can adjust their movement orbits to accommodate this increased patchiness (McCabe 2011), in SPIRALL, specified maximum movement radii constrain the set of patches that can be accessed during simulations by these households.

The establishment of PAs and the rules of access associated with them has the potential to alter existing social dynamics, thereby introducing novel resource conflicts (West and Brockington 2006, Greiner 2012). Inter-ethnic conflicts are common across the ASALs, but their nature, underlying causes, and intensity vary (Van Weezel 2019). We focus only on conflicts that may arise when households graze livestock beyond their own ethnic group boundaries. In our simulations, loss of PA access resulted in an increase in trans-boundary movements in several counties, including movements beyond national boundaries (Table A3.2, Appendix 3). These transnational movements, which often result in conflicts, have been well documented (Leff et al. 2009). Access to PAs that lie just beyond Kenyan national boundaries may have a modulating effect on these conflicts. This emphasizes the need for greater cooperation in PA management among nations in this region, such as through the establishment of trans-frontier conservation areas in a manner that integrates the needs and aspirations of local communities (Duffy 2005, Hanks 2008, Bourgeois 2023).

Accelerating declines in biodiversity have promoted ambitious calls for the expansion of the global PA network, and undergird ambitions related to SDG 15 focused on terrestrial biodiversity conservation. The Nature Needs Half and Half-Earth movements call for the setting aside of 50% of the planet expressly for conservation purposes (Dinerstein et al. 2017). Similarly, the Convention on Biodiversity’s 30 by 30 target (2022), to which Kenya is a signatory, aims to protect 30% of terrestrial and aquatic areas for biodiversity conservation by 2030. Mehrabi et al. (2018) estimate that protecting half the planet would entail significant losses in food calories, particularly in Africa and Asia, resulting from the return of agriculture and pastureland back to nature. These authors and subsequent studies (Ellis and Mehrabi 2019), also posit that these losses may be mitigated through conservation actions centered on shared multifunctional landscapes. Our results suggest that PAs in and around Kenya harbor key resources that are critical to ensuring food security (SDG 2), reducing poverty (SDG 1), and increasing peace and justice (SDG 16) within several counties. Because of the geographic context within which they are practiced, existing pastoral strategies fail to alleviate the declines in household well-being that result from a loss of access to these resources.

Baseline simulation

Results from the baseline simulation demonstrate that SPIRALL coupled with L-Range can emulate key aspects of Kenyan pastoral household behaviors, specifically those that determine household-level food security. It captures the seasonal contribution of livestock-based calories to pastoral diets as well as the spatial variation in pastoral movement patterns. The baseline simulations reveal climate as a driver of change in per-capita livestock holdings (TLU / AE), which is a determinant of pastoral household well-being. The centrality of climate in determining food security in this region has been elucidated by previous studies (Galvin et al. 2001, Shukla et al. 2021). The wet season was characterized by improved access to calories and overall increases in household income driven by the productivity of livestock herds. In the driest months on the other hand, households experienced increases in food insecurity and trans-boundary movements, which we interpret as an index for inter-ethnic conflict risks. Climate change–driven increases in drying conditions in the region can increase the number of months over which households are exposed to these intersecting risks (Kogo et al. 2020). These seasonal trends may in part be driven by rules governing household movements. As is common in Kenya’s pastoral practices, non-sedentary households in SPIRALL return to their home patch during the wet season, thereby reducing the risk of conflicts (Boone et al. 2011). Conversely, this movement rule also means that many households are forced to return to poor quality patches and thereby fail to fully accrue the benefits of increased productivity in the wet season.

In Kenya, precipitation increases along a north-south gradient (Ayugi et al. 2016). This precipitation gradient underlies the spatial heterogeneity in rangeland productivity and pastoral strategies seen in the ASALs (Ellis and Galvin 1994). The baseline simulation adequately captured the spatial heterogeneity in household behaviors and their exposure to risks. The highly variable climatic conditions prevailing in the northern counties such as Turkana, Mandera, Marsabit, and West Pokot strongly limit the growth of livestock herds. Because of smaller per-capita livestock holdings, households in these counties experienced higher rates of hunger and debt. In addition, these households moved longer distances to access pastures that consequently exposed them to greater conflict risks.

Model uses and future applications

Disparities in county-level food-security scores based on SPIRALL and FEWS NET underscores a key design aspect of SPIRAL: the model is not intended as a forecasting tool. FEWS NET food-security forecasts are based on a near-term scenario analysis that depends on predicting livestock population responses to climate, analyzing agricultural output, and assessing national and global market trends. Spatial trends in SPIRALL emerge because of the reciprocal interactions between households and their immediate environment, without any external influences such as markets, or temporal changes such as increases in the pastoral population over time. SPIRALL does not accommodate local or regional markets even though fluctuations in maize and livestock prices are known to exert influences on pastoral household economic decisions and diets (Little et al. 2014). Similarly, household sources of income are inferred based on available survey data, which have spotty spatial and temporal coverage. Such simplifications are an inevitable aspect of modeling efforts that seek to simulate complex phenomena underpinned by cross-scale processes. These simplifications are also part of large-scale, spatiotemporally explicit, and coupled social-ecological models where scale constraints imposed by one or both models dictate which datasets can be leveraged to simulate processes of interest. On the contrary, the granular representation of household decisions pertaining to livestock-rearing and livestock-environment interactions is the strongest aspect of SPIRALL. This is because of the wealth of anthropological information on Kenyan pastoralists, the availability of ecological studies on pastoral livestock dynamics, and the precedents established by other pastoral household–decision models such as DECUMA (Boone at al. 2011) and PHEWS (Thornton et al. 2003).

Taken together, SPIRALL may be best viewed as a model that describes how pastoral households across Kenyan ASALs translate primary production into food calories. The outcomes illustrate spatiotemporal vulnerabilities with regard to food insecurity, poverty, and conflicts for traditional pastoral households across the ASALs. In addition, SPIRALL coupled with L-Range explicates the links between climate and pastoral-household well-being in the absence of attenuating factors such as local, national, and international institutions and markets. Consequently, it is an effective tool to explore the challenges posed to SDG 2 achievement across the ASALs by climate and environmental change. Owing to its detailed representation of livestock population dynamics, SPIRALL can also be used to explore how livestock populations and composition are likely to change across Kenya under changing climate and land-use scenarios and their consequent impacts on plant production and carbon cycling. The coupled model can also be extended, via linkage with other simulation models, to explore SDG 2 interactions under future climate and land-use scenarios.

LITERATURE CITED

Anyango, D. Y., W. N. Muhingi, M. T. Okuku, T. Mutavi, and S. Mwalw. 2017. Socioeconomic effects of cattle-rustling in Borabu Sub-County, Nyamira County, Kenya. IOSR Journal of Humanities and Social Sciences 22(10):70-90.

Ayugi, B. O., W. Wang, and D. Chepkemoi. 2016. Analysis of spatial and temporal patterns of rainfall variations over Kenya. Journal of Environment and Earth Science 6(11):69-83.

Berger, R. 2003. Conflict over natural resources among pastoralists in northern Kenya: a look at recent initiatives in conflict resolution. Journal of International Development 15(2):245-257. https://doi.org/10.1002/jid.985

Bleischwitz, R., C. Spataru, S. D. VanDeveer, M. Obersteiner, E. van der Voet, C. Johnson, P. Andrews-Speed, T. Boersma, H. Hoff, and D. P. van Vuuren. 2018. Resource nexus perspectives towards the United Nations Sustainable Development Goals. Nature Sustainability 1(12):737-743. https://doi.org/10.1038/s41893-018-0173-2

Bonabeau, E. 2002. Agent-based modeling: methods and techniques for simulating human systems. Proceedings of the National Academy of Sciences 99(3):7280-7287. https://doi.org/10.1073/pnas.082080899

Boone, R. B., R. T. Conant, J. Sircely, P. K. Thornton, and M. Herrero. 2018. Climate change impacts on selected global rangeland ecosystem services. Global Change Biology 24:1382-1393. https://doi.org/10.1111/gcb.13995

Boone, R. B., K. A. Galvin, S. B. BurnSilver, P. K. Thornton, D. S. Ojima, and J. R. Jawson. 2011. Using coupled simulation models to link pastoral decision making and ecosystem services. Ecology and Society 16(2):6. https://doi.org/10.5751/ES-04035-160206

Boone, R. B., and C. K. Lesorogol. 2016. Modeling coupled human-natural systems of pastoralism in east Africa. Pages 251-280 in S. Dong, K. A. S. Kassam, J. F. Tourrand, and R. B. Boone, editors. Building resilience of human-natural systems of pastoralism in the developing world: interdisciplinary perspectives. Springer, New York, New York, USA. https://doi.org/10.1007/978-3-319-30732-9_7

Bourgeois, R., C. Guerbois, N. Giva, P. Mugabe, B. Mukamuri, R. Fynn, W. Dare, M. Motsholapheko, L. Nare, E. Delay, et al. 2023. Using anticipation to unveil drivers of local livelihoods in transfrontier conservation areas: a call for more environmental justice. People and Nature 5(2):726-741. https://doi.org/10.1002/pan3.10446

Breuer, A., H. Janetschek, and D. Malerba. 2019. Translating sustainable development goal (SDG) interdependencies into policy advice. Sustainability 11(7):2092. https://doi.org/10.3390/su11072092

Brockington, D., and D. Wilkie. 2015. Protected areas and poverty. Philosophical Transactions of the Royal Society B: Biological Sciences 370(1681):247-249. https://doi.org/10.1098/rstb.2014.0271

Butchart, S. H., M. Walpole, B. Collen, A. Van Strien, J. P. Scharlemann, R. E. Almond, J. E. Baillie, B. Bomhard, C. Brown, J. Bruno, et al. 2010. Global biodiversity: indicators of recent declines. Science 328(5982):1164-1168. https://doi.org/10.1126/science.1187512

Butt, B. 2011. Coping with uncertainty and variability: the influence of protected areas on pastoral herding strategies in East Africa. Human Ecology 39:289-307. https://doi.org/10.1007/s10745-011-9399-6

Convention on Biological Diversity (CBD). 2022. Kunming-Montreal Global Biodiversity Framework. United Nations Environment Programme, Montréal, Québec, Canada.

Dinerstein, E., D. Olson, A. Joshi, C. Vynne, N. D. Burgess, E. Wikramanayake, N. Hahn, S. Palminteri, P. Hedao, R. Noss, et al. 2017. An Ecoregion-based approach to protecting half the terrestrial realm. BioScience 67(6):534-545. https://doi.org/10.1093/biosci/bix014

Duffy, R. 2005. The politics of global environmental governance: the powers and limitations of transfrontier conservation areas in Central America. Review of International Studies 31(2):307-323 https://doi.org/10.1017/S0260210505006467

Edmonds, B., C. L. Page, M. Bithell, E. Chattoe-Brown, V. Grimm, R. Meyer, C. Montanola-Sales, P. Ormerod, H. Root, and F. Squazzoni. 2019. Different modelling purposes. Journal of Artificial Societies and Social Simulation 22(3):6 https://doi.org/10.1007/978-3-319-66948-9_4

Ellis, E. C., and Z. Mehrabi. 2019. Half earth: promises, pitfalls, and prospects of dedicating half of earth’s land to conservation. Current Opinion in Environmental Sustainability 38:22-30. https://doi.org/10.1016/j.cosust.2019.04.008

Ellis, J., and K. A. Galvin. 1994. Climate patterns and land-use practices in the dry zones of Africa. BioScience 44(5):340-349. https://doi.org/10.2307/1312384

European Space Agency (ESA). 2017. Land cover CCI product use guide. Version 2.0. UCL-Geomatics, Louvain-la-Neuve, Belgium. http://maps.elie.ucl.ac.be/CCI/viewer/download/ESACCI-LC-Ph2-PUGv2_2.0.pdf

Famine Early Warning Systems Network (FEWS NET). 2017. Kenya acute food insecurity mapping. United States Agency for International Development, Washington, D.C., USA. https://fews.net/east-africa/kenya

Gallagher, C. A., M. Chudzinska, A. Larsen-Gray, C. J. Pollock, S. N. Sells, P. J. C. White, and U. Berger. 2021. From theory to practice in pattern-oriented modelling: identifying and using empirical patterns in predictive models. Biological Reviews 96:1868-1888. https://doi.org/10.1111/brv.12729

Galvin, K. A. 2009. Transitions: pastoralists living with change. Annual Review of Anthropology 38:185-198. https://doi.org/10.1146/annurev-anthro-091908-164442

Galvin. K. A. 2021. Transformational adaptation in drylands. Current Opinion in Environmental Sustainability 50:64-71. https://doi.org/10.1016/j.cosust.2021.03.003

Galvin, K. A., R. B. Boone, N. M. Smith, and S. J. Lynn. 2001. Impacts of climate variability on East African pastoralists: linking social science and remote sensing. Climate Research 19:161-172. https://doi.org/10.3354/cr019161

Galvin, K. A., P. K. Thornton, J. R. de Pinho, J. Sunderland, and R. B. Boone. 2006. Integrated modeling and its potential for resolving conflicts between conservation and people in the rangelands of East Africa. Human Ecology 34(2):155-183. https://doi.org/10.1007/s10745-006-9012-6

Greiner, C. 2012. Unexpected consequences: wildlife conservation and territorial conflict in Northern Kenya. Human Ecology 40:415-425. https://doi.org/10.1007/s10745-012-9491-6

Griggs, D., M. Stafford-Smith, O. Gaffney, J. Rockström, M. C. Öhman, P. Shyamsundar, W. Steffen, G. Glaser, N. Kanie, and I. Noble. 2013. Policy: sustainable development goals for people and planet. Nature 495:305-307. https://doi.org/10.1038/495305a

Grimm, V., S. F. Railsback, E. Christian, U. B. Vincenot, C. Gallagher, D. L. DeAngelis, B. Edmonds, J. Ge, J Giske, J. Groeneveld, et al. 2020. The ODD protocol for describing agent-based and other simulation models: a second update to improve clarity, replication, and structural Realism. Journal of Artificial Societies and Social Simulation 23(2):7. https://doi.org/10.18564/jasss.4259

Grimm, V., E. Revilla, U. Berger, F. Jeltsch, W. M. Mooij, S. F. Railsback, H.-H. Thulke, J. Weiner, T. Wiegand, and D. L. DeAngelis. 2005. Pattern-oriented modeling of agent-based complex systems: lessons from ecology. Science 310:987-991. https://doi.org/10.1126/science.1116681

Gurney, G. G., V. M. Adams, J. G. Alvarez-Romero, and J. Claudet. 2023. Area-based conservation: taking stock and looking ahead. One Earth 6(2):98-104 https://doi.org/10.1016/j.oneear.2023.01.012

Hanks, J. 2008. Transfrontier conservation areas (TFCAs) in Southern Africa: their role in conserving biodiversity, socioeconomic development, and promoting a culture of peace. 2003. Journal of Sustainable Forestry 17(1-2):127-148 https://doi.org/10.1300/J091v17n01_08

Kennedy, W., C. C. Reddy, T. Gulden, M. Coletti., and C. Cioffi-Revilla. 2014. Towards validating a model of households and societies in East Africa. Pages 315-328 in S. H. Chen, I. Terano, H. Yamamoto, and C.C. Tai, editors. Advances in Computation Social Science. Agent-Based Social Systems, Volume 11. Springer, Tokyo, Japan. https://doi.org/10.1007/978-4-431-54847-8_20

Kieti, D., R. Nthiga, J. Plimo, P. Sambajee, A. Ndiuini, E. Kiage, P. Mutinda, and T. Baum. 2020. An African dilemma: pastoralists, conservationists, and tourists-reconciling conflicting issues in Kenya. Development Southern Africa 37(5):758-772. https://doi.org/10.1080/0376835X.2020.1747988

Koech, G., G. O. Makokha, and C. N. Mundia. 2019. Climate change vulnerability assessment using a GIS modelling approach in ASAL ecosystem: a case study of Upper Ewaso Nyiro basin, Kenya. Modeling Earth Systems and Environment 6(1):479-498. https://doi.org/10.1007/s40808-019-00695-8

Kogo, B. K., L. Kumar, and R. Koech. 2020. Climate change and variability in Kenya: a review of impacts on agriculture and food security. Environment, Development, and Sustainability 23(1):23-43. https://doi.org/10.1007/s10668-020-00589-1

Krätli, S., C. Huelsebusch, S. Brooks, and B. Kaufmann. 2013. Pastoralism: a critical asset for food security under global climate change. Animal Frontiers 3(1):42-50. https://doi.org/10.2527/af.2013-0007

Krätli, S., and J. Swift. 2014. “Counting pastoralists” in Kenya. Dryland Learning and Capacity Building Initiative / Regional Learning and Advocacy Program, Nairobi, Kenya.

Kuznar, L. A. 2005. Collective violence in Darfur: an agent-based model of pastoral nomad / sedentary peasant interaction. Mathematical Anthropology and Cultural Theory 1:1-22.

Leff, J., U. States, W. Bergmann, and R. D. Crutchfield. 2009. Pastoralists at war: violence and security in the Kenya-Sudan-Uganda border region. International Journal of Conflict and Violence 3(2):188-203.

Little, M. A., R. Dyson-Hudson, N. Dyson-Hudson, and N. L. Winterbauer. 1999. Environmental variations in the south Turkana ecosystem. Pages 317-319 in M. A. Little and P. W. Leslie, eds. Turkana herders of the dry Savanna. Oxford University Press, Oxford, UK. https://doi.org/10.1093/oso/9780198549215.003.0016

Little, P. D., D. N. Debsu, and W. Tiki. 2014. How pastoralists perceive and respond to market opportunities: the case of the Horn of Africa. Food Policy 49:389-397. https://doi.org/10.1016/j.foodpol.2014.10.004

McCabe, J. T. 2011. Cattle bring us to our enemies. Fifth edition. University of Michigan Press, Ann Arbor, Michigan, USA. https://doi.org/10.3998/mpub.23477

McCabe, J. T., P. W. Leslie, and L. DeLuca. 2010. Adopting cultivation to remain pastoralists: the diversification of Maasai livelihoods in northern Tanzania. Human Ecology 38:321-334. https://doi.org/10.1007/s10745-010-9312-8

McPeak, J. G., and P. D. Little. 2017. Applying the concept of resilience to pastoralist household data. Pastoralism 7(1):1-18. https://doi.org/10.1186/s13570-017-0082-4

Mehrabi, Z., E. C. Ellis, and N. Ramankutty. 2018. The challenge of feeding the world while conserving half the planet. Nature Sustainability 1(8):409-412. https://doi.org/10.1038/s41893-018-0119-8

Mojo, D., A. M. O. Oduor, C. Fu, Y. Bai, H. Long, G. Wang, and L. Zhang. 2020. Effects of protected areas on welfare of local households: the case of Maasai Mara national reserve in Kenya. People and Nature 2(3):856-867. https://doi.org/10.1002/pan3.10123

Ng’ang’a, S. K., E. H. Bulte, K. E. Giller, N. N. Ndiwa, S. C. Kifugo, J. M. McIntire, M. Herrero, and M. C. Rufino. 2016. Livestock wealth and social capital as insurance against climate risk: a case study of Samburu County in Kenya. Agricultural Systems 146:44-54. https://doi.org/10.1016/j.agsy.2016.04.004

Nilsson, M., E. Chisholm, D. Griggs, P. Howden-Chapman, D. McCollum, P. Messerli, B. Neumann, A. S. Stevance, M. Visbeck, and M. Stafford-Smith. 2018. Mapping interactions between the sustainable development goals: lessons learned and ways forward. Sustainability Science 13(6):1489-1503. https://doi.org/10.1007/s11625-018-0604-z

Nilsson, M., D. Griggs, and M. Visbeck. 2016. Policy: map the interactions between sustainable development goals. Nature 534:320-322. https://doi.org/10.1038/534320a

Nyabira, B. C., and Z. A. Ayele. 2016. The state of political inclusion of ethnic communities under Kenya’s devolved system. Law, Democracy, & Development 20:131-153. https://doi.org/10.4314/ldd.v20i1.7

Oba, G. 2001. The importance of pastoralists’ indigenous coping strategies for planning drought management in the arid zone of Kenya. Nomadic Peoples 5:89-119. https://doi.org/10.3167/082279401782310998

Österle, M. 2008. From cattle to goats: the transformation of East Pokot pastoralism in Kenya. Nomadic Peoples 12(1):81-91. https://doi.org/10.3167/np.2008.120105

Phalan, B., M. Onial, A. Balmford, and R. E. Green. 2011. Reconciling food production and biodiversity conservation: land sharing and land sparing compared. Science 333:1289-1291. https://doi.org/10.1126/science.1208742

Pham-Truffert, M., F. Metz, M. Fischer, H. Rueff, and P. Messerli. 2020. Interactions among sustainable development goals: knowledge for identifying multipliers and virtuous cycles. Sustainable Development 28(5):1236-1250. https://doi.org/10.1002/sd.2073

Reid, R. S., M. E. Fernández-Giménez, and K. A. Galvin. 2014. Dynamics and resilience of rangelands and pastoral peoples around the globe. Annual Review of Environment and Resources 39:217-242. https://doi.org/10.1146/annurev-environ-020713-163329

Reid, R. S., P. K. Thornton, and R. L. Kruska. 2004. Loss and fragmentation of habitat for pastoral people and wildlife in east Africa: concepts and issues. African Journal of Range and Forage Science 21(3):171-181. https://doi.org/10.2989/10220110409485849

Scherer, L., P. Behrens, B. Sprecher, A. De Koning, R. Heijungs, and A. Tukker. 2018. Trade-offs between social and environmental sustainable development goals. Environmental Science and Policy 90:65-72 https://doi.org/10.1016/j.envsci.2018.10.002

Shukla, S., G. Husak, W. Turner, F. Davenport, C. Funk, L. Harrison, and N. Krell. 2021. A slow rainy season onset is a reliable harbinger of drought in most food insecure regions in Sub-Saharan Africa. PLoS ONE 16(1):e0242883. https://doi.org/10.1371/journal.pone.0242883

Sircely, J., R. T. Conant, and R. B. Boone. 2019. Simulating rangeland ecosystems with G-Range: model description and evaluation at global and site scales. Rangeland Ecology & Management 72:846-857. https://doi.org/10.1016/j.rama.2019.03.002

Smith, M. S., C. Cook, Y. Sokona, T. Elmqvist, K. Fukushi, W. Broadgate, and M. P. Jarzebski. 2018. Advancing sustainability science for the SDGs. Sustainability Science 13(6):1483-1487. https://doi.org/10.1007/s11625-018-0645-3

Socioeconomic Data and Applications Center (SEDAC). 2016. Gridded population of the world (GPW), version 4. SEDAC, New York, New York, USA.

Stavi, I., J. R. de Pinho, A. K. Paschalidou, S. B. Adamo, K. Galvin, A. de Sherbinin, T. Even, C. Heaviside, and K. van der Geest. 2021. Food security among dryland pastoralists and agropastoralists: the climate, land-use change, and population dynamics nexus. Anthropocene Review 9(3):299-323. https://doi.org/10.1177/20530196211007512

Thornton, P. K., S. B. BurnSilver, R. B. Boone, and K. A. Galvin. 2006. Modelling the impacts of group ranch subdivision on agro-pastoral households in Kajiado, Kenya. Agricultural Systems 87(3):331-356. https://doi.org/10.1016/j.agsy.2005.03.001

Thornton, P. K., K. A. Galvin, and R. B. Boone. 2003. An agro-pastoral household model for the rangelands of East Africa. Agricultural Systems 76(2):601-622. https://doi.org/10.1016/S0308-521X(01)00094-4

United Nations Environment Programme and the International Union for Conservation of Nature (UNEP-IUCN). 2018. The world database on protected areas (WDPA). UNEP-IUCN, Cambridge, UK.

Veldhuis, M. P., M. E. Ritchie, J. O. Ogutu, T. A. Morrison, C. M. Beale, A. B. Estes, W. Mwakilema, G. O. Ojwang, C. L. Parr, J. Probert, et al. 2019. Cross-boundary human impacts compromise the Serengeti-Mara ecosystem. Science 363(6434):1424-1428. https://doi.org/10.1126/science.aav0564

Van Weezel, S. 2019. On climate and conflict: precipitation decline and communal conflict in Ethiopia and Kenya. Journal of Peace Research 56(4):514-528. https://doi.org/10.1177/0022343319826409

West, P., and D. Brockington. 2006. An anthropological perspective on some unexpected consequences of protected areas. Conservation Biology 20(3):609-616. https://doi.org/10.1111/j.1523-1739.2006.00432.x

Wilensky, U. 1999. NetLogo. Center for Connected Learning and Computer-based Modeling, Northwestern University, Evanston, Illinois, USA.

Copyright © 2024 by the author(s). Published here under license by The Resilience Alliance. This article is under a Creative Commons Attribution 4.0 International License. You may share and adapt the work provided the original author and source are credited, you indicate whether any changes were made, and you include a link to the license.