APPLICATION OF MULTIVARIATE-STATISTICS IN DETECTING TEMPORAL AND SPATIAL PATTERNS OF WATER CHEMISTRY IN LAKE-GEORGE, NEW-YORK

Cluster and component analyses were used to identify temporal and spat ial patterns of water chemistry in Lake George, a meso-oligotrophic la ke in northeastern New York, during 1981-1993. The lake includes two m ajor basins that have similar area and volume, but different biologica l community structure, plankton assemblages, watershed area, and water shed development. Analyses were based on total phosphorus, particulate phosphorus, dissolved organic phosphorus, dissolved inorganic phospho rus, nitrate, calcium, chlorophyll a, silica, chloride, and pH, indivi dually or in combinations. Total phosphorus, chlorophyll a, chloride, and particulate phosphorus were included in the first linear component indicating that these are probably the most important analytes in exp laining the total variance of the data. In spring or summer, three or four components explained 86 or 84% of the total variance, respectivel y. Cluster analysis based on the major components or on the original v ariables indicated that there are distinct differences in water chemis try between the two major basins of the lake. The only long-term tempo ral pattern that could be detected by cluster analysis was an increase in chloride concentrations. Cluster analysis is found to be a useful tool to detect both step (abrupt) and monotonic (gradual) changes in t ime and space.