Climate Change: Defining Reference Conditions and Restoring Freshwater Ecosystems - Climate Change Impacts on Freshwater Ecosystems - page 214

associated with higher nutrient concentrations (e.g. fine-leaved Potamogeton spp.
including P. pectinatus L. and P. pusillus L., C allitriche truncata Guss. and
Nymphaeaceae).
This approach is especially valuable in shallow, lowland lakes as there are few
undisturbed examples of this lake type remaining in the current population of
European water bodies and, therefore, spatial approaches are often inappropriate.
The direct ecological information recorded by the sediments provides a sounder
basis on which to establish reference conditions and formulate management
decisions such as the definition of restoration targets. In the future it would be
interesting to quantify the uncertainties associated with different sediments -
macrofossil approaches to better understand how variability in establishing the
reference condition affects inferences of anthropogenic-induced change.
Reference conditions and uncertainty
Although palaeolimnological records offer an excellent method for establishing
reference conditions for lakes, as discussed above, the methods used in
reconstructing past chemical and biological conditions have inherent uncertainty -
as exemplified above in the studies comparing discrepancies between historical
plant records and use of macrofossils. Although there is now a growing interest
in using multiple methods, both direct (e.g. multiple groups of organisms) and
indirect (e.g. different forms of inference), few studies have compared and
contrasted the use of different methods to better understand their intrinsic
uncertainties. One notable exception is a recent study by Battarbee et al . (2005),
which assessed the relative accuracy of different approaches in reconstructing the
past pH of a Scottish loch.
Battarbee et al . (2005) used contemporary monitoring data (pH and diatoms
from 1991 to the present day) from the Round Loch of Glenhead, an acidified
Scottish loch (cf. Flower & Battarbee 1983), to evaluate the performance of
three pH inference models. Each of the pH inference models, based on different
calibration training sets, was used to reconstruct pH from the diatom assemblages
of three cores taken in different years, resulting in nine reconstructions of pH
(Fig. 9.5). Taking the year AD 1800 as the reference date, the pH values for this
date, derived using the transfer functions, were compared with the pH inferred
using a diatom-cladocera modern analogue technique (Simpson et al . 2005) and
with hindcast values using the MAGIC model (Model of Acidification of
Groundwaters in Catchments, Cosby et al . 1985).
Comparison of the different methods showed that the inferred reference pH
for the Scottish loch in AD 1800 varied by less than an order of magnitude (between
5.5 and 6.1). Relatively good agreement, with respect to both between-core and
between-training set differences, was found between the diatom-pH transfer
functions. Diatom-pH transfer functions indicated a reference pH of between 5.5
and 5.7, and the weighted average pH using the modern analogue method gave a
similar value (5.8). By contrast, the MAGIC-model hindcast estimate for 1850
was somewhat higher (6.1). Although there is no way to ascertain which estimate
is the most accurate, the discrepancy between the MAGIC hindcast and the
diatom-based values is the most striking. According to Battarbee et al . (2005),
one explanation is that the MAGIC model does not allow for DOC concentrations