Improving biological condition assessment accuracy by multimetric index approach with microalgae in streams and lakes.

Multimetric index (MMI) approach is a broadly used in ecological assessment because it can integrate information of various kinds of ecologically related metrics of freshwater ecosystems and provide an easily understandable score for purpose of further evaluation and managements. Accounting for natural variation and disentangling covariation between natural environmental factors and human disturbance factors are imperative for an accurate assessment. Lots of progress has been made recently on the aforementioned two aspects. Three approaches, a priori classification of sites by regions or typologies, site-specific modeling of expected reference condition and varying metrics in site groups, have been tested in lakes and streams to improve assessment accuracy. All existed studies support that site-specific modeling can efficiently account for natural variation and generate a MMI with good performance. However, until now, no strong evidence has shown that diatom/blue-algae typologies are better than regionalization frameworks on accounting for natural variation either in lakes or in streams. To separate the natural variation explained by site specific modeling from that of varying metrics is necessary for a thorough and accurate evaluation on the valuableness of site-grouping by typologies. Different performance of varying metrics among site groups of streams and lakes was most probably caused by the lack of representativeness of diatom metrics on biological condition rather than the complex multi-stressor gradients in streams and rivers. A recent study showed that blue-green algae enhanced performance of diatom-based MMI on defining lake condition under high level of human disturbance. On the other hand, with more and more extensive and intensive use of statistics techniques in developing MMI, we also discussed some statistical challenges faced by scientists in field of ecological assessment, especially on setting significance level of a statistical test and multiple comparison issue in MMI performance comparison.