Spatial cluster analysis for large herbivore distributions: Amboseli ecosystem, Kenya

doi:10.1016/j.ecoinf.2015.05.010

Ecological Informatics

Volume 30, November 2015, Pages 203-206

https://doi.org/10.1016/j.ecoinf.2015.05.010 Get rights and content

Highlights

•
We analyze large mammal species distribution in Amboseli, Kenya. We use spatial hierarchical clustering technique.
•
Six large mammal species were grouped into zones of most similar use based on a spatial hierarchical cluster analysis. The herbivores clustered into metabolic biomass size classes, suggesting variability in migration and habitat utilization patterns at different scales.
•
Results show significant range contraction of species pointing to increased human activities.

Abstract

We present an assessment of seasonal movements and species distributions in the Amboseli ecosystem, southern Kenya, using spatial cluster analysis for large herbivore populations over the period 1970 to 2010. Six large mammal species were grouped into zones of most similar use based on a spatial hierarchical cluster analysis. The herbivores clustered into metabolic biomass size classes, suggesting variability in migration and habitat utilization patterns at different scales. The analysis shows a steady loss of spatial spread and heterogeneity among six indicator species over the decades. The findings point to the spread of farming, loss of habitat, land subdivision, sedentarization and ivory poaching as key factors causing compression of species in and around Amboseli National Park. The national park and environs is the only cluster showing no loss of species heterogeneity. The analysis highlights large herbivores as more vulnerable to displacement and compression than small species. The hierarchical cluster analysis shows the use of spatial tools in detecting patterns of change across an ecosystem and identifying important areas for land conservation and restoration to sustain species migration and species diversity.

Introduction

African savannah ecosystems are characterized by extensive seasonal migrations of large herbivores (Sinclair et al., 2007). In recent decades, rapid human population and land use changes have caused a loss of populations and range among most large herbivores (Ogutu et al., 2011). Amboseli ecosystem in southern Kenya, which has been continuously studied since the 1960s, provides a rich dataset showing the impact of human activity on a savannah ecosystem and wildlife populations (Western and Nightingale, 2003). Despite the abundant documentation of population and habitat changes, little has been published on the spatial changes in wildlife movements and species richness patterns (Mose et al., 2013).

In this paper we present an assessment of species distributions and spatial utilization of selected species using spatial hierarchical cluster analysis of large herbivore populations from the 1970s to 2000s. The species include buffalo (BF), elephant (EL), Grant's gazelles (GG) Thomson's gazelles (TG), wildebeest (WL) and zebra (ZB). The analysis covers a range of body size to explore long term changes in spatial distribution and richness, the primary agents and the differential effects on species. We use spatial cluster analysis as a tool for detecting data patterns, the changes in pattern over time and the conservation implications.

Section snippets

Study area and methods

The 8500 km² (Fig. 1) Amboseli ecosystem straddling the Kenya–Tanzania border is defined by the seasonal movements of large herbivores and includes the 388 km² Amboseli National Park. Further details of the ecosystem are provided elsewhere (Western, 2006). Land subdivision, settlement, agriculture and loss of seasonal grazing (Okello, 2005) have led to range restriction and fragmentation of pastoral livestock and wildlife populations over the last four decades (Western, 2006, Worden, 2007).

Results

Species clustered into four groups in both the 1970s and 2000s. In the 1970s, species richness was relatively uniform within and across clusters (Fig. 3). Each of the six species fell into any of the four clusters identified fairly randomly. Clusters 1, 2, and 3 recorded the maximum species richness (Fig. 3). Cluster 4 had a species richness of 5. By the 2000s only cluster 2 had maximum species richness, indicating a compression of species into and around the national park (Fig. 2). Cluster 4

Discussion

In the 1970s species were widespread across the ecosystem, but by the 2000s showed a heavy contraction of range and concentration in the vicinity of Amboseli National Park (Fig. 2). The major causes of compression are an expansion of the agricultural areas (Fig. 5), draining of the swamps east of the national park for irrigation, increase in human settlement (Fig. 4), land subdivision and sedentarization (Western and Nightingale, 2003), leading to loss of rangeland productivity (Groom and

Acknowledgments

We wish to thank the many wardens and the staff of Kenya Wildlife Service for their support over the years, David Maitumo for being part of the data collection team and Lucy Waruingi for logistical support. Liz Claiborne Ortenberg Foundation: www.lcaof.org (Grant Number: LCAOF-ACP2014) funded the data collection.

References (18)

R.J. Groom et al.
Impact of land subdivision and sedentarization on wildlife in Kenya's southern rangelands
Rangel. Ecol. Manag.
(2013)
V.N. Mose et al.
Modelling herbivore population dynamics in the Amboseli National Park, Kenya: application of spatial aggregation of variables to derive a master model
Ecol. Complex.
(2012)
V.N. Mose et al.
Modelling the dynamics of migrations for large herbivore populations in the Amboseli National Park
Kenya Ecol. Model.
(2013)
D. Western et al.
The impact of subdivision and sedentarization of pastoral lands on wildlife in an African savanna ecosystem
Biol. Conserv.
(2009)
A.X.Y. Carvalho et al.
Spatial hierarchical clustering
Rev. Bras. Biom.
(2009)
R. Core et al.
R: A Language and Environment for Statistical Computing
(2014)
A.W. Illius et al.
Oecologia Modelling the Nutritional Ecology of Ungulate Herbivores: Evolution of Body Size and Competitive Interactions
(1992)
G.M. Jolly
Sampling methods for aerial censuses of wildlife populations
East Afr. Agric. J.
(1969)
Y. Lu
Spatial cluster analysis of point data: location quotients versus kernel density.
(2000)

There are more references available in the full text version of this article.

Cited by (5)

Application of open source tools for biodiversity conservation and natural resource management in East Africa
2018, Ecological Informatics
Citation Excerpt :
The open source tools require basic computer knowledge and can be executed by local community trainees using straight forward computer commands. The digital platforms allow biodiversity scenario building, model formulation such as the identification of migratory corridors, habitat utilization by animal species, and application to county spatial planning programs (Mose and Western, 2015). The Amboseli and Magadi ecosystems (Fig. 1) have a combined area of 9500 km2 and support a large population of livestock, wildlife and pastoralists (Groom and Western, 2013) freely moving over the rangelands extending into Northern Tanzania.
We use forty years of animal and plant data collected in the Amboseli and Magadi ecosystems in Kajiado county, southern Kenya to demonstrate the application of open source tools to spatial analysis and the visualization of results. These highly interactive tools developed in R contain a series of commands that perform various data functions including: examining the data structure, aggregation, exploratory data analysis, vegetation biomass, greenness and grazing pressure estimations, NDVI extraction, species population estimation and spatial mapping, among other tasks. We customized the tools to suit the local community information needs and level of expertise without losing the underlying data richness or integrity. The implementation requires basic computer knowledge and can be undertaken by local community trainees using straight forward commands.
The results show that with these automated tools, long-term data in large databases can be analyzed rapidly and presented in a way that encourages community participation, ownership and uptake locally and regionally. The tools allow biodiversity scenario building and other functions such as the identification of migratory corridors, habitat utilization by animal species and general land use and resource management.
Seasonal movements of wildlife and livestock in a heterogenous pastoral landscape: Implications for coexistence and community based conservation
2017, Global Ecology and Conservation
Citation Excerpt :
This area, the South Rift Valley, Kenya, is inhabited by indigenous Maasai pastoralists, who have a long history of coexistence with wildlife. Within many community and protected areas across East Africa seasonal wildlife movements have been drastically altered or stopped altogether (e.g Mose and Western, 2015), with this area as an exception to this rule. In this ecosystem, wildlife movements are not constrained by any physical barriers, such as settlement or fencing, and appear to respond to metabolic requirements, with wildlife grazers moving down a seasonal forage biomass gradient across the heterogeneous grazing areas (Russell et al., submitted).
Rangelands across the world are home to millions of pastoral people and vast wildlife populations, which create a complex landscape for conservation. Community based conservation has been used to promote human-wildlife coexistence on pastoral lands, protecting wildlife outside of official protected areas. With the spread of community based conservation within the rangelands there is a need for more information on successful management practices. This study provides an example of this in the South Rift, Kenya, where seasonal movements of pastoralists aid coexistence. We used Density Surface Modelling (DSM), a novel tool for conservation managers in the rangelands, to predict wildlife and livestock abundance across the landscape and seasons. Wildlife grazers, zebra (Equus burchelli) and wildebeest (Connochaetes taurinus), follow expected metabolic patterns, feeding on short grass outside the conservation area in the wet season, before returning to the taller-lower quality grazing in the conservation areas during the drought. Browsing wildlife, impala (Aepyceros melampus) and Grant’s gazelle (Nanger granti), move from open grassland and bushland areas into thicker, denser browse as the seasons progress towards the drought. Livestock, both shoats (Ovis aries, and Capra aegagrus hircus) and cattle (Bos indicus), are managed by community grazing committees, who enforce a grazing plan that creates spatial–temporal separation between wildlife and livestock. They exploit the high-quality grazing in the livestock area during the wet season while conserving pasture in the conservation area, which is utilized only as forage is depleted. This ensures that wildlife has access to a diverse resource base across all seasons and potentially reduces competition, allowing for a diverse and abundant wildlife community to coexist with livestock. This highlights the importance of the presence and maintenance of spatial and temporal heterogeneity of forage resources, through livestock management, for community based conservation. We encourage more community based conservation initiatives in pastoral landscapes to incorporate livestock management into planning.
Cascading effects of elephant–human interactions and the role of space and mobility in sustaining biodiversity
2023, Ecosphere
Conservation from the inside-out: Winning space and a place for wildlife in working landscapes
2020, People and Nature
Monitoring vegetation dynamics in southern Tunisia using SPOT-5 (Take5) data: A case study of the Tozeur oases
2018, Journal of Applied Remote Sensing

View full text

Spatial cluster analysis for large herbivore distributions: Amboseli ecosystem, Kenya

Highlights

Abstract

Introduction

Section snippets

Study area and methods

Results

Discussion

Acknowledgments

Rangel. Ecol. Manag.

Ecol. Complex.

Kenya Ecol. Model.

Biol. Conserv.

Spatial hierarchical clustering

Rev. Bras. Biom.

R: A Language and Environment for Statistical Computing

Oecologia Modelling the Nutritional Ecology of Ungulate Herbivores: Evolution of Body Size and Competitive Interactions

Sampling methods for aerial censuses of wildlife populations

East Afr. Agric. J.

Spatial cluster analysis of point data: location quotients versus kernel density.