NEWS
The findings from this study should
prompt increased attention to riparian
habitats and fine-scale, detailed work
aimed at further understanding the dy-
namics of these systems. This will go
some way in determining thresholds for
conservation concern in an attempt to
ensure persistence of these important
ecosystems, in not only KNP, but other
subtropical savannas.
Figure 2: Riparian areas were delineated within 300 m buffer polygons (white
boundary) of several rivers (blue line) in KNP. The riparian woodland extent was
manually drawn at each site (yellow boundary) and analysis was in these areas.
This image shows a bend in the Shingwedzi River in the 1950s.
tiny fraction of the land surface area.
Study determines trends in riparian
woodland cover
A recent MSc project completed by
Joshua Weiss under the supervision of
Prof Mike Cramer (University of Cape
Town) and Dr Dave Thompson (SAEON)
aimed to determine trends in riparian
woodland cover within KNP on a tem-
poral scale not done previously. Pos-
sible drivers of change were identified
by comparing long-term environmental
data to woodland extent over time.
Eighteen sites along 14 of the park’s riv-
ers (nine perennial, five non-perennial)
were selected for the study. A multi-
decadal time-series analysis of riparian
woodland extent was conducted using
aerial and satellite imagery, with the
oldest photograph used dating back
to 1936. Riparian areas were delineated
along a 10 km stretch of river at each
site and then the tree cover proportion
was estimated using random point plots
using GIS tools (Figure 2).
The time-series of tree cover at each
site was compared to river flow data
sourced from Department of Water and
Sanitation gauging stations (Figure 3)
located at the centre of each site. Rain-
fall data from the nearest South African
Weather Service station to each site was
also compared to the tree cover.
Particular attention was paid to cumula-
tive flow effects, as well as the frequency
and magnitude of large infrequent dis-
turbances (LIDs) such as droughts and
floods, which regulate the depth of the
water table and may cause physical dis-
turbance.
Tree cover typically fluctuated over
the time period, with the trajectory of
change being variable between sites.
Grassroots
Vol 19
No 4
Most (14 of 18) sites experienced de-
creasing overall tree cover between the
first and last observation, with these de-
creases being significant at six sites (red
and deep orange arrows, Figure 4).
Several sites were sparsely covered in
the 1940s to 1960s, before experienc-
ing increases in woody cover, which
has subsequently declined over the last
three decades. Tree cover increased
only at three sites (green arrows, Figure
4) over time, with these increases being
significant in one instance.
Increased riparian woodland cover
was associated with non-perennial riv-
ers which have transient surface water
only after large flood events. Further,
increased tree cover was only seen in
rivers where the catchments fall at least
90% within the protected area, high-
lighting a possible positive link to a lack
of anthropogenic disturbance.
Peak flow and maximum rainfall events,
however, were the strongest significant
association with decreases in riparian
tree cover, indicating that floods are
potentially the biggest drivers of tree
loss. Indeed, tree cover decreased sub-
stantially at several sites following the
mega-flood event of early 2000 (one of
the largest on record) and subsequent
large floods over the last decade, such
as those in 2012 and 2013.
Figure 3: One of the Department of
Water and Sanitation hydrological
gauging stations on the Olifants River,
KNP
A bleak future for these woodlands?
With predictions of increasing climate
variability and extreme climate events
into the future, and increasing an-
thropogenic disturbance, particularly
through water abstraction outside of
but impacting protected areas, the fu-
ture of the riparian woodlands of the
Lowveld, and of their associated biota,
looks bleak.
November 2019
Figure 4: Study sites with respective
arrows indicating riparian tree cover
losses/gains. Arrows pointing down in-
dicate a decrease in riparian tree cover
and vice versa, while the colour scale is
graded from red (greatest loss in tree
cover), through yellow (moderate gain/
loss in tree cover) to dark green (great-
est gain in tree cover). One site had a
net zero change.
24