Long-term climatic records are not available from instrumental data, and so proxy data are required to infer long-term climatic trends. The sedimentary records of closed-basin lakes in arid and semi-arid regions are particulary good archives of paleohydrological and paleoclimatic signals as lake levels are governed largely by the balance between precipitation and evaporation. In Canada, little is known of the enviromental and climatic histories of arid regions such as the Great Plains and the southern interior of British Columbia.

In order to provide a quantitive means of evaluating fluctuations in salinity, and potentially climate, a diatom-based salinity transfer function (r$\sp2$ = 0.87, bootstrap RMSE = 0.37) was developed to infer past salinities and climatic trends from fossil diatom assemblages preserved in lake sediment cores from closed-basin lakes in western Canada. The modern diatom assemblages and limnological data were collected from 208 fresh and saline lakes from the southern Interior of British Columbia and from 11 saline lakes from the northern Great Plains. This dataset is an expansion of an earlier dataset of 102 lakes. Evaluation of dataset size and species deletion criteria using this 219 lake dataset emphasizes the value of constructing calibration sets with a large number of lakes and many taxa to provide better estimates of species optima and tolerances to various enviromental variables, such as salinity, and to provide more and closer modern analogues for inferring salinity downcore.

Using the above salinity transfer function, Holocene salinity changes were reconstructed for Harris Lake, in the Cypress Hills of Saskatchewan, and from Clearwater Lake, Saskatchewan. The diatom paleosalinity record from Harris Lake indicated that the lake remained fresh throughout the Holocene, even when many closed-basin lakes in the surrounding plains experienced higher salinities during the warmer early Holocene. The diatom data contradict an earlier geochemical study that suggested episodes of hypersalinity in the early Holocene.

A high-resolution early Holocene diatom record from Clearwater Lake revealed that the lake had much higher salinities ($\rm {>}20 g\cdot L\sp{-1}$) between 9700-8840 years BP, compared to the present ($\rm {\sim}1 g\cdot L\sp{-1}$. It is unclear whether this high salinity period was the result of evaporative stress caused by increased summer insolation, as suggested by Milankovitch cycles, or was the result of changes in groundwater flow to the lake. The diatom-inferred salinities clearly showed that Clearwater Lake had become fresh by ca. 7300 BP. The more recent diatom record from Clearwater Lake shows that the lake is not particulary sensitive to short-term climatic fluctuations, although it appears to show signs of responding to at least some known droughts of the recent past.