The Elephant Dilemma: A Long-term Perspective on the Management of African Savannas - Biodiversity Conservation and Environmental Change - page 22

fungal spores) that enabled the reconstruction of fire, climate, nutrient availability, and her-
bivory over millennial time-scales (Gillson and Ekblom 2009a, b, Ekblom and Gillson 2010a, c).
In order to be useful to Park managers, the pollen data had to be translated into a form that
was directly relevant to management goals, specifically the TPC for woody cover. The aim was
to compare long-term records of changing tree cover from pollen data in order to test whether
the TPC was in line with the normal range of variability of tree cover at the scale of hundreds to
thousands of years (Gillson and Duffin 2007).
A complicating factor in the use of fossil pollen data is that savanna trees are poorly repre-
sented in the pollen record, because they produce little pollen compared with grasses, for
example, which produce copious amounts. Therefore, in order to accurately translate arbor-
eal pollen abundance into an estimate of past tree cover, the relationship between pollen
abundance and tree abundance had to be calibrated using present-day measurements of tree
cover, and modern pollen rain (Duffin and Bunting 2008). Once the relationship between
pollen abundance and tree abundance was quantified (Figure 2.3), the fossil pollen data
could be converted to estimates of woody vegetation cover (Gillson and Duffin 2007).
The pollen data showed that the TPC (80% of highest ever value) showed no sign of being
exceeded in the past 5,000 years (Figure 2.4). However, longer palaeoecological records might

100

80

60

40

20

0

20 40 60

Simulated arboreal pollen (%)

80 100

Figure 2.3 Calibration of pollen data using simulated woody vegetation cover and arboreal pollen per-
centages, based on modern pollen and vegetation data from the Kruger National Park. The curve is
described by a quadratic regression equation ( y = 0.0085 x 2 + 1.7048 x + 8.1163) with 95% confidence limits
indicated. Examples of landscape scenarios created with MOSAIC are shown in the left hand column,
with woody vegetation of 20, 40, 60, and 80% shown as black patches on grey (grass) matrix (Gillson and
D u i n 200 7).