LAUSR

Abstract In tropical lakes, deep water-column mixing exerts great influence on nutrient dynamics and food-web structure, and on the probability for hazardous accumulation of bio- or geogenic gases in the hypolimnion. However, conventional criteria to infer a lake's mixing regime in temperate regions, such as the changing seasonal position of the thermocline or the depth to anoxia, often lead to faulty conclusions when applied to stratifying tropical lakes. The water column of these lakes tends to exhibit multiple thermoclines, and deep-water anoxia tends to spread quickly throughout the hypolimnion in all but the least productive lakes, complicating the interpretation of isolated temperature and dissolved-oxygen profiles. Here, multiple depth profiles of temperature, dissolved oxygen and conductivity from a diverse selection of 11 small tropical crater lakes in western Uganda are combined with time series of temperature-logger data to determine the occurrence and maximum depth of water-column mixing within a given time period. We propose three reliable indicators of past mixing: 1) presence of oxygen below the photic zone; 2) evidence of recent cooling at a given depth; and 3) disappearance of previously documented thermo- and chemoclines. The greatest depth of mixing can only be inferred with a fair degree of certainty from a chronosequence of at least two sets of profiles.