What is influencing patchy space-use by wildebeest in western-central Kruger?
K. Yoganand and N. Owen-Smith
Center for African EcologySchool of Animal Plant and Environmental SciencesUniversity of the Witwatersrand
Background
Over 1.4 million wildebeest in Serengeti
Kruger, comparable in area (20,000 vs. 25,000 km2 of Serengeti), has fewer than 12,000 Wildebeest
WB population in central Kruger declined sharply after western-boundary fencing in 1960s & 70s (Whyte 1985)
This sub-population has not recovered in abundance nor resumed migration after fence removal in the 90s
The themes we examine here are:
The habitat and landscape features that govern space-use and thus restrict the distribution of wildebeest
The factors include:The forage conditions (quantity and quality features)
Vegetation cover and visibility features
Landscape-scale features
We placed GPS collars on (breeding) herds of wildebeest, zebra and buffalo in western-central region of Kruger (Orpen gate)
Obtained hourly locations
Measured forage and habitat features at locations/patches of ‘use’and in the surrounding ‘unused’ landscape
Spatial analyses of data
We placed GPS collars on (breeding) herds of wildebeest, zebra and buffalo in western-central region of Kruger (Orpen gate)
Obtained hourly locations
Measured forage and habitat features at locations/patches of ‘use’and in the surrounding ‘unused’ landscape
Spatial analyses of data
•Based on 40,000+ locations obtained at hourly intervals over the last year
•From 10 herds of wildebeest
•Data ranges from 3 to 11 months for each herd
•The collars are still working and we are continuing data collection
• Example of four herds
•Early wet season
• Late wet season
• Dry season
Movement paths covering the convex polygon justifying use of it as a ‘known’ area
0
20
40
60
80
100
120
Early-Wet Late-Wet Dry Annual
Area
in k
m2
Area-knownPatch-used
% of area ‘used’out of area ‘known’: 12.29 12.78 16.67 5.8
Summary of space-use for all (4) wildebeest herds
How did the wildebeest use these patches?
Rate of movement of herds was much lower within patches (indicating foraging) than between patchesMovements between patches were seldom done during nightThese are ‘real’ functional patches and not some artefacts of (auto-correlated) location data or a fractal pattern
69.35 (4.9)
561.31 (47.1)
0
100
200
300
400
500
600
700
movment rate (m/h) within-patch movment rate (m/h) betwn-patches
0
0.1
0.2
0.3
0.4
0.5
0.6
night : day movment within patch night : day movment betwn patches
Example from one herd
Are other large herbivores similar in space-use?
The patchy space-use of wildebeests is in contrast with zebra and buffalo herdsZebra and buffalo herds used a large proportion of the area ‘known’ to them
Factors that may influence patchy space-use by wildebeest at different scales
GeologySoilRainfallGrass biomass, structure & qualityPredation risk
Main landscape classes:
Mixed Combretum spp. / Terminaliawoodland on granite (light brown)
Thornveld on Gabbro (darker reddish brown)
Patchy use of grazing lawns in the wet season
Patchy use of seeplines in the dry season
Large-scale factors
The wet season patches were in grazing lawns and sodic sites, and the dry season patches were mainly along seeplines
The processes driving these patterns
What are these patches and surrounding landscape made of?
The two main hypotheses:Forage abundance (grass with favourable characteristics) and nutrient qualityPredation risk
Field sampling
Forage and cover characteristics In class intervals (3, 6, or 7-point scales)Most prevalent class in each plotSummarized by frequency distributions and modes
Sampling in ‘used’ locations and ‘unused’patches
‘Unused’ patches and the transects used to sample those patches
The main ‘forage’ and ‘cover’ variables
Grass species compositionLeaf-table heightAerial and basal cover
Tree coverShrub coverGrass height Distance to hidden space (which could offer cover for predators)
Forage and cover features compared between ‘unused’ patches and ‘used’ locations
Grass species prevalence
0%
5%
10%
15%
20%
25%
30%
35%
40%
Uro.
mos
Dig.
eri
Pan.
max
Dac.
aus
Chl.v
ir
Era.
sp
Pan.
col
The.
tri
Set.in
c
Het.c
on
Cym
.plu
% o
ccur
renc
e as
1st
or
2nd
mos
t ab
unda
nt s
p.
'Used' locations'Unused' patches
Grass leaf-table height
0
10
20
30
4050
60
70
80
90
1-10cm 11-20 21-30 31-40 41-50 51-60 >60cm
Rel
ativ
e fr
eque
ncy
(%)
'unused' patches'used' locations
Grass cover
0
10
20
30
40
50
60
11-25% 26-50% 51-75% 76-90%
Rel
ativ
e fre
quen
cy (%
)
'Unused'patches
'Used'locations
Tree canopy cover
0
10
20
30
40
50
60
0% 1-10% 11-25% 26-50% 51-75%
Rel
. fre
quen
cy (%
)
'Unused' patches'Used' locations
Grass height
0
10
20
30
40
50
60
1-10cm 11-20 21-40 41-60 >60cm
Rel
ativ
e fre
quen
cy (%
)
'Unused' patches'Used' locations
Distance to hidden area / cover for predators
0
10
20
30
40
50
60
0-10m 11-20m 21-30m 31-50m 51-100m
Rel
. fre
quen
cy (%
)
'Unused'patches'Used'locations
Features of ‘Used’ locations compared between wet and dry seasons
Stark contrasts between early-wet and late-dry seasons in variables such as;
Prevalent grass species; leaf greenness; distance to hidden area; tree cover; shrub cover; etc.
Remarkably, ‘grass height’ was similar between the two seasons
The key inference: Wildebeest herds are seeking and finding patches on granite in dry season that offer forage comparable in height/structure to the wet season grazing lawns
Grass height in 'used' locations
0%
10%
20%
30%
40%
50%
60%
70%
1-10cm 11-20 21-40 41-60
Rel
ativ
e fre
quen
cy (%
)
Early-wetLate-dry
Test of the main hypothesesThe statistical models we built with our data (logistic regression models in model-selection framework) show that:
Abiotic, forage (‘bottom-up’ processes) and predation-risk (‘top-down’ process) factors simultaneously influence space-use by wildebeest herds
The ‘forage’ factors alone explain a majority of the variation in the data, but ‘predation risk’ has substantial influence
Plans for further spatial analyses to separate the influence of the two factors
ConclusionsWildebeest herds use space patchily in all seasons
The patches used formed only a fraction of areaavailable to the herds
Short-grass seems to be the most important habitat feature sought after by wildebeest herds
Model results suggest influence of both forage and predation risk factors on wildebeest space-use
However, pending further spatial analysis, we are delaying our key conclusions
Our next step is to determine what factors restrict wildebeest distribution in the landscape
AcknowledgementsAugusto Mabunda – for field assistance and ‘protection’Sanparks Scientific Services
GIS data and support (Sandra MacFadyen and Izak Smit) Thembi Khoza, Patricia Khoza and Adolf Manganyi – for liaisoning
Sanparks Veterinary Services and Helicopter team for collaring and ‘re-collaring’ operations (Peter Buss, Markus Hofmeyr, Johan, Grant Knight, Charlie Thompson)
Sanparks Protection Services – Richard Sowry and KFI field rangers Manyaleti Nature Reserve – Jimmy Thanyani, Mark Bourn and
Dr. Ferreira (Mpumalanga Veterinarian)Timbavati Private Nature ReserveAfrica Wildlife Telemetry – for the collars and data downloadsNRF and University of the Witwatersrand for funding supportAPES, Wits University staff
for administrative supportBarend Erasmus – for discussions on GIS issues