Climate Change: Defining Reference Conditions and Restoring Freshwater Ecosystems - Climate Change Impacts on Freshwater Ecosystems - page 207

for lakes and 34% for watercourses), followed by expert judgement (30% lakes,
27% watercourses), historical data (18% lakes, 21% watercourses), predictive
models (9% lakes, 15% watercourses) and palaeo-reconstruction (12% lakes, 3%
watercourses). The selection of methods seemed to reflect the difficulty associated
with establishing reference conditions for the various quality elements and habitat
types. For example, spatial approaches were commonly used to establish reference
conditions for phytobenthos in watercourses (54%), whereas expert judgement
was more commonly used for macrophytes (43%). Not surprisingly, the use of
palaeo-reconstruction was almost twice as common in lakes (e.g. 30% for
phytobenthos) compared with watercourses (14% for phytoplankton). Although
a number of methods are currently being used for establishing reference
conditions, surprisingly little is known of their inherent error, in particular how
different levels of uncertainty may affect determinations of disturbance and
recovery.
Comparison of different approaches
Typology- and model-based approaches
Ecologists have for some time recognized the importance of biogeographical
drivers for species distribution patterns, and the use of spatial typologies for
partitioning natural variability (e.g. Hawkins et al . 2000; Johnson et al . 2007).
Establishing reference conditions and ecological targets for restoration is a direct
application of this knowledge. For example, cognizant of the importance of
regional drivers for aquatic biodiversity, the European Water Framework Directive
(European Commission 2000) specifies two alternative spatial approaches for
partitioning biological variability and assessing ecological quality. System-A
consists of four categories (e.g. ecoregion, altitude, catchment area and geology
for streams), whereas system-B consists of a mixture of obligatory as well as
optional factors. The value of typology-based approaches for explaining and
partitioning biological variability has, however, been much debated and findings
are hitherto equivocal. For instance, a number of studies have shown that aquatic
assemblages are related to large-scale patterns in vegetation and climate (e.g.
Feminella 2000; Rabeni & Doisy 2000; Verdonschot & Nijboer 2004), whilst
others have questioned the efficacy of classifications based solely on landscape
patterns in vegetation and climate, generally arguing that regional classifications
need to be augmented with other factors such as altitude, size and catchment
characteristics to discriminate communities effectively (e.g. Sandin & Johnson
2000a; Van Sickle & Hughes 2000).
Although many studies have focused on the importance of regional and local
scale variables for predicting the composition of aquatic habitats, few have
tested the efficacy of a priori classification systems (like system-A variables) for
predicting biotic assemblages and fewer still have quantified the uncertainties
associated with these predictions. In agreement with findings from earlier
studies, R.K. Johnson (unpublished) found that latitude, altitude and catchment
size were important predictors of invertebrate assemblages in boreal streams
(e.g. Hawkins et al . 2000; Johnson et al . 2004). However, use of system-A
variables alone resulted in many sites being misclassified and model uncertainty