3. How the systems of shallow lakes function

Understanding of the differences between shallow and deep lakes has come from the failure of methods, trialled on deep lakes, to restore shallow lakes successfully from eutrophication. In the UK, the Norfolk Broads in particular [106] and on mainland Europe, a huge range of lakes in the Netherlands [152] and Denmark [87] were showing symptoms of eutrophication by the 1970s. They had lost their clear waters and extensive beds of submerged plants (the term 'macrophytes' is often used to describe the large macroscopic plants, as distinct from the microscopic plants, the algae, but the simpler terms 'plants' and 'algae' will be used here). With loss of the plants, an associated diverse fauna disappeared and the lakes had become turbid with either suspended silt, or phytoplankton algal populations or a mixture of both (Fig 5). Secondary symptoms included erosion of reed fringes and banks, fish kills and bird deaths from botulism. The conservation and amenity values, including the availability of large fish for angling, were dwindling and concern was being expressed

Fig 5 The systems of clear and turbid shallow lakes are very different. An urban lake in an English public park and Hudsons Bay in the Norfolk Broadland.

It seemed likely, from comparison with deeper lakes, that increasing nutrient loading had allowed increased algal growth either in the phytoplankton or in the attached (periphyton) communities coating the plants so that the algae were competing successfully with the plants for light or carbon dioxide. Because of the increase in phytoplankton, it was assumed that 'straightforward' eutrophication was involved and that the solution was therefore a reduction in phosphorus loading. This had been very successful in some deep lakes [18,19,121]; phosphorus precipitation from waste water effluents, or the diversion of such effluents away from the lake, had markedly improved Lake Washington in the USA [49,50], the St Lawrence Great Lakes [6] and the European alpine lakes. Models based on the Vollenweider equation (see above) were being widely used [45] to design restoration programmes based on phosphorus reduction

In the early nineteen-eighties, a programme of phosphorus removal from effluents discharging into Barton Broad, in Norfolk, UK, was completed and similar schemes were underway for many lakes in the Netherlands. All proved disappointing, however. Phosphorus control sometimes led to a reduction in algal crops and total P concentrations, though it was usually impossible, for lack of long term data, to say whether this was within the limits of 'normal' variation, but crops and concentrations remained high, waters did not clear and plant communities did not return (Fig 6). Worse, in a series of experimental ponds containing well-developed plant communities, subjected to a series of high nutrient loadings, the plants did not disappear, but thrived (Fig 7) [5,82]. It became clear that the simple linear model of eutrophication (Fig 8) in shallow lakes would have to be abandoned

Fig 6 Removal of phosphorus from waste water effluent entering Barton Broad in Norfolk had no major effect on the concentrations in the lake because of release from the sediment and hence on phytoplankton chlorophyll a. From Moss et al [112].

Fig 7 In experimental ponds in which the plants were left intact, addition of high nitrogen loads plus a graded series of phosphorus loadings did not displace the plants nor increase the concentrations of phosphorus or chlorophyll in the water. When the plants were previously removed, however, the loadings had a considerable effect. From Irvine et al [82].

Fig 8 Early ideas about the restoration of shallow lakes were summarised in a linear model which suggested that phosphorus increases alone had caused the changes and that phosphorus control alone would reverse them. This proved to be a gross oversimplifi-cation, but was a reasonable view to take at the time because this situation pertained in deep lakes.

In the failure of nutrients to displace the plants from the experimental ponds, there was, however, a clue to a better model. In some ponds, the added nutrients did increase total P and chlorophyll concentrations provided the plants were removed first. This began the development of an alternative states model for the changes in communities of shallow lakes [82,112,125,126,141] and a more complex approach to their restoration from eutrophication. The key feature of this model is that although nutrients, both phosphorus and nitrogen, are involved, they play a role in conjunction with other mechanisms not necessarily linked with nutrients