Chrysophyte cysts were identified from the surface sediment of 105 mountain lakes in the Pyrenees (NE Spain), and their statistical relationship to water chemistry was examined using canonical correspondence analysis (CCA). The chemical parameters that explained significant and independent amounts of variability were alkalinity, pH, potassium, nitrate and magnesium. In a CCA using these parameters, the first canonical axis was related to a gradient of alkalinity and pH, which reflected the varying nature of the watershed bedrock in the Pyrenees, while the second axis was correlated with potassium (negatively) and nitrate (positively). The potential for environmental reconstructions of the five chemical parameters was further studied by: (i) analyzing the distribution of optima and tolerances calculated by weighted-averaging (WA); (ii) carrying out detrended canonical correspondence analysis (DCCA) with a single environmental variable; and (iii) examining the performance of WA-PLS transfer functions. Acceptable transfer functions were obtained for alkalinity, pH and nitrate. However, for potassium and magnesium the tolerance of cysts was too broad and the distribution of optima too skewed, respectively. The possibility of reconstructing nitrogen-related issues using chrysophyte cysts is particularly interesting because of the lack of direct chemical records of nitrogen compounds in sediments. Nitrate reconstructions using transfer functions may be complemented by a holistic reconstruction using partial CCA, where, after subtracting the effects of other chemicals, samples are ordered on a plain defined by potassium and nitrate. This ordination could show down-core trends in lake productivity and renewal time.

This study was supported by the CICYT of the Spanish Government (contract AMB93-0814-CO2-01), the European Commission (contract ENV4-CT97-0642, CHILL-10,000 project), the Comissionat per a Universitats i Recerca of the Catalan Government (grant 1999SGR00029) and a Ph. D. fellowship to SP from the Spanish Government.

14 páginas, 5 tablas, 7 figuras.

Peer reviewed