The diatom-inferred pH reconstructions for a naturally neutralized pit lake in south-west Poland using the Mining and the Combined pH training sets

- The Mining pH training set contains the modern chemical and diatom data from 69 pit lakes located in SW Poland.
- The lakes included in the database have extremely acidic to slightly alkaline water.
- The reconstruction of DI-pH in pit lake TR-33 was more reliable using the Mining pH training set.
- The Mining pH training set can be used to reconstruct water pH changes in other pit lakes.

Many artificial lakes in various parts of the world were formed as a result of the exploitation of lignite and other mineral deposits. Usually, lakes created after the end of lignite mine activity have extremely acidic or very acidic water, with pH < 3.0, caused by the oxidation of pyrite and marcasite and other sulfide compounds. After the closure of mines, the lakes undergo natural neutralization. However, this process is long-term and depends on many different factors, such as the amount of carbonates, the geological structure of bedrock, hydrology and hydrobiology, climate and the method of lignite exploitation. In some types of anthropogenic lakes (i.e., clay pits), water pH is almost neutral or alkaline due to the adsorption properties of clay minerals. Along the Polish-German border in the Łuk Mużakowa Geopark, over 100 post-mining lakes were formed as a result of lignite and clay exploitation. The Mining pH training set of modern diatom samples and water chemistry was created from 69 lakes to quantify the relationship between the diatom assemblages living in these water bodies and the lakes' chemical composition. Then, this database was applied to the reconstruction of diatom-inferred pH in the sediments of a post-mining lake TR-33. The results of reconstruction was compared to similar analysis based on the Combined pH training set containing more variety of diatom taxa from a higher number of lakes. Although the second dataset was larger and gave a smaller root mean square error, the Mining pH training set has a longer gradient of pH including lakes with extremely acidic water and gives more reliable results. The reconstruction of DI-pH in TR-33 which had very acidic water in the past and currently neutralized from acidification was more precise and reliable using the database sensu stricto, which consists of only the post-mining lakes.

344