USING THE PAST TO PREDICT THE FUTURE - LAKE-SEDIMENTS AND THE MODELING OF LIMNOLOGICAL DISTURBANCE

Most lakes have been disturbed to varying degrees but for an individua l lake the timescale of these disturbances is rarely known. Lake sedim ents, however, can be used as natural archives of perturbation histori es, e.g. acidification and eutrophication. At present the use of simpl e weighted averaging models permits the reconstruction of a variety of water chemical variables from diatom and other microfossils preserved in lake sediments (pH, total phosphorus, salinity and lakewater tempe rature). Sediment records can, therefore, provide lake-specific backgr ound data for lake management as well as information about their ecolo gical histories. The common models used in palaeolimnology (dating, tr ansfer-functions) are reviewed and their role in environmental monitor ing discussed. Predictions of future lake water quality following lake restoration methods tend to be made from dynamic mathematical models, but they are also used for hindcasting (e.g. the MAGIC model of catch ment acidification). A problem with using dynamic models is that they are often site-specific and require calibration for a given lake. Comb ined with reliable dating, chemical reconstructions from microfossil-b ased transfer functions offer the possibility of testing hindcast pred ictions derived from dynamic mathematical models, e.g. for salinity, T P and pH. In this way, sediment microfossil-based models can assist wi th the parameterization of more complex, dynamic models of contemporar y processes. In this review, comparisons between the two approaches (s ediment-based and dynamic models) are given and possible future intera ctions outlined. Validation of mathematical models by palaeolimnologic al data might enhance their predictive ability when used for forecasti ng lake recovery. There is clearly, however, a need for a more rigorou s approach to palaeolimnology, i.e. critical hypothesis generation. Mu ltidisciplinary studies of lake disturbance, that combine palaeolimnol ogy, dynamic modelling and contemporary process studies, would also be beneficial.