Methods of pH calibration of sedimentary diatom remains for reconstructing history of pH in lakes

Abstract

The pH history of lakes can be inferred from diatom remains in dated sediment cores. To derive transfer functions for pH inference in acidic lakes, we counted diatoms in surface-sediment from 31 soft-water lakes in n. New England (NE) and 36 in Norway (N), covering pH 4.4–7.1. Cluster analysis of each data set indicates that pH 6 is an upper limit for a group of similar diatom assemblages. For each set, we developed multiple linear regressions to relate three versions of the diatom data to pH of surface-waters: (1) relative frequencies of selected diatom taxa, (2) the first principal component (1 PC) of these frequencies, and (3) the frequencies of Hustedt pH groups. Also, simple linear regressions were developed for two versions: (1) Index B and (2) Index Alpha, both based on pH groups. Regressions were run separately for lakes with pH ≤ 6; these are most relevant for pH inference in acidic lakes. The best regressions (N: taxa & 1 PC taxa) have r² 0.69–0.91 and S_e 0.24–0.31 pH units, the worst (NE: log alpha) have r² 0.27–0.57 and S_e 0.51. In all cases, errors for NE are greater than N, partly due to greater diversity of NE lakes. Regressions based on pH groups (directly & by indices) have smaller r² and larger S_e than those based on taxa and 1 taxa. The Index Alpha is least useful because its requirement for alkaline diatom units is unsatisfied at many acidic lakes. Regressions based on taxa may give erratic pH inferences due to sensitivity to unusual frequencies of individual taxa; this effect is reduced by using 1 PC taxa. Four regressions based on pH ≤ 6 lakes were used for inferring pH in a ²¹⁰Pb dated core from Nedre Målmesvatn, N (now pH 4.6). There is good agreement among three of the four (not for the regression based directly on taxa) that there has been a decrease of ca. 0.6 pH units starting in the late 1800's.