Regional watershed evaluation of influences on stream biogeochemistry in southeastern Alaska

Characterization of individual watershed types at regional scales has been a goal of many organizations. Broad regional approaches have attempted to use watershed information to guide environmental analysis or prediction, but regional assessments rely on extensive historical data. Southeast Alaska, which is mostly under the management of the Tongass National Forest, has extensive vegetation and soil information, but no comprehensive information on stream discharge, nutrient chemistry or biogeochemical measurements across the landscape. The Tongass Forest is larger than West Virginia, contains thousands of anadramous salmonid reaches, spans 6 degrees of latitude and is undergoing significant climate changes. Anticipating climate change effects or responses to timber harvest or other management requires a much better understanding of existing natural variability and what watershed characteristics drive biogeochemical cycles. We report results of a regional watershed evaluation using a statistically based synoptic sampling designed to determine what physical and biological features influence stream chemistry over the entire Tongass. Sixty-one watersheds across the Tongass were randomly chosen in two broad classes: wetland dominated watersheds (>50% wetland by area), and non-wetland watersheds (<50% wetland). The main drainage in each watershed was sampled in triplicate and water analyzed for dissolved organic carbon (DOC) and total dissolved nitrogen (TDN). Key attributes of each watershed were quantified using GIS coverages of wetland extent, harvest history, and ecoregional classification. DOC concentrations in stream water were significantly related to the wetland extent in the watershed, which explained nearly 25% of the variability among sampled watersheds in the region. This confirms the hypothesis that wetland soil processes exert an important influence on stream chemistry in the temperate rainforest of southeast Alaska independent of all other variables known to influence DOC concentration. When latitude, harvest history, and watershed size were included in the analysis the model explained about half of the variation in DOC concentration among watersheds. Analysis of variance with DOC concentration versus ecological subsection was not significant. Percent wetland cover did not significantly predict total dissolved nitrogen, but an ANOVA using ecological subsections explained 54% of TDN variation, suggesting that stream nitrogen is controlled by larger scaled landscape variables. Although based on one sample period, this regional sampling approach provides a possible tool for discriminating watershed characteristics over a diverse and large geographic area for management planning and prioritizing restoration efforts.