ALBERT

Notes: 1. Autumn circulation in lakes is currently conceived to occur very rapidly, being controlled mainly by wind-power dynamics, decreasing irradiance and heat flux. In addition, autumn mixing is usually related to nutrient redistribution in the vertical column, resulting in its overall increase. To test these assumptions, mixing and nutrient dynamics in a Spanish small, wind-sheltered, mesotrophic, seepage lake were studied daily during autumn circulation.2. The seasonal erosion of the pycnocline in Las Madres Lake was the outcome of vertical and horizontal exchanges of heat and matter. The overall mixing of the water column lasted 3 months, which was an unexpected period for a rather shallow lake. Two periods of mixing could be envisaged until full circulation was attained. First, a slightly faster period of pycnocline deepening than that predicted by the heat flux and wind stress model of Fisher et al. (1979) occurred for 41 days, mixing most of the water column down to within two meters of the bottom. Then a much slower process took place promoting frequent instability of the bottom layer and resulting in entire mixing in a further 52 days.3. Vertically, the whole mixing process was a response to weak surface cooling, resulting from the mild air temperatures of the semiarid climate of the area, and weak wind stress, because of low wind fetch and high shelter. Horizontally, a gravity current transporting cold, denser water from western shallower areas of the lake and materials produced by the decomposition of organic matter of littoral origin may produce a bottom layer of increased density, thus impinging on vertical stability. Seepage inputs of water of roughly constant temperature might also have increased bottom density. Bottom density enhancement resulted in a double diffusion process.4. Only in-lake nitrogen content increased until full circulation was attained, whereas carbon showed no trend and phosphorus declined. External processes, such as seepage exchange and atmospheric deposition, coupled to internal processes, such as nitrification, oxidised phosphorus precipitation and complexation with organic carbon, might have been responsible for the areal nutrient patterns observed.5. Our study demonstrates that current models of water column mixing and nutrient redistribution in lakes during autumn circulation must be improved to encompass the effects of external inputs, including horizontal heat and matter exchange.