Climate-mediated changes in small lakes inferred from midge assemblages: the influence of thermal regime and lake depth

Abstract

Subfossil midge (Chironomidae and Chaoboridae) assemblages were examined in the surficial sediments (0–1 cm) from small inland lakes in the Experimental Lakes Area (ELA) of northwestern Ontario, Canada. In these boreal lakes, maximum depth (Zmax), alkalinity, Secchi depth and chlorophyll-a concentrations explained significant amounts of variation in the subfossil assemblages. Constrained ordinations (redundancy analysis) indicated that the relationship between Zmax (as sqrt Zmax) and assemblage composition was strong enough to develop a paleolimnological inference model. Model statistics suggested that a robust model was generated (r 2 = 0.78, RMSEP = 0.533, max bias = 0.674); however, when the model was applied to a subfossil stratigraphy from an ELA lake sediment core, results suggested that the inference model had produced an unrealistically shallow Zmax inference. Further analyses indicated that thermal regime had a strong influence on assemblage composition; when the influence of thermal regime was partialled out, there was a much weaker relationship between Zmax and assemblage composition, particularly for stratified lakes. A thermal regime inference model was subsequently developed, which, when applied to the lake sediment core stratigraphy, indicated that the shallow Zmax inference may have been the result of a period of increased mixing or polymixis in this stratified lake. Water column mixing may increase due to hypolimnetic warming and increased water clarity resulting from declines in dissolved organic carbon. In a training set where there are strong correlations between lake depth and assemblage composition, this relationship is not necessarily a strict function of lake depth, but of some other highly correlated variable, likely thermal conditions.