REGRESSION AND WEIGHTED AVERAGING MODELS RELATING SURFICIAL SEDIMENTARY DIATOM ASSEMBLAGES TO WATER DEPTH IN LAKE-ONTARIO

Diatom assemblages in surface sediment samples in depth profiles from Lake Ontario and from East Lake, a shoreline lake, were analyzed for t he purpose of describing the relationship between species distribution and water depth. At both sites species composition varied markedly wi th sample depth. In East Lake a multiple regression of four habitat gr oups: benthic, epiphytic, tychoplanktonic, and euplanktonic, against w ater depth (30 cm-8 m) produced a relationship with a standard error ( SE) of 1.5 m. An analysis of the ratio of euplanktonic diatoms to peri phytic diatoms over a transect from 3 m to 150 m in Lake Ontario produ ced a multiple regression with an SE of 11.8 m. In another approach, w ater depth optima for 91 diatom tara were developed using a weighted a veraging (WA) technique. A good correlation (r(2) > 0.9) was found bet ween measured and inferred water depth over the range 3 m to 30 m usin g a unimodal WA regression model. The transfer functions offer the pos sibility of inferring Holocene water level changes in Lake Ontario fro m fossil diatom assemblages in sediment cores.