When are historical data sufficient for making watershed-level stream fish management and conservation decisions?

Addressing landscape-level threats to stream fishes such as habitat and hydrological alterations requires adequate watershed-level species inventories. Where watershed-level ichthyofaunal surveys are prohibitively expensive, existing (historical) data sources may provide an option for compiling species lists. However, it is critical that managers consider potential biases or limitations of species lists compiled from existing data. Here we assess the suitability of species lists compiled from existing data sources for making watershed-level fish management and conservation decisions. For nine Great Lakes watersheds, we developed existing species lists by compiling all available federal and state agency and museum fish survey data. We then compared the size and species composition of existing species lists to current species lists compiled from intensive field surveys, conducted in 2002, of the same watersheds. Species lists compiled from commonly available existing data sources, such as state and federal agency and museum data, missed many species detected during our 2002 field surveys. In most watersheds, more than 10 species were missed (range 5-21) on existing lists. Sampling over multiple years and seasons increased the size of both current (field) and existing species lists. Existing species lists compiled from surveys conducted over multiple years and seasons included an average of 15 species not captured during the 2002 field surveys. However, such multiyear existing datasets are rare and not available for many watersheds. In addition, species lists compiled from older existing surveys (e.g., before 1984) did not accurately represent current species composition of the watersheds and our results indicate several apparent misidentifications or errors on these lists. Lastly, while most game species were detected on existing lists, migratory species and recently introduced species were commonly missed on these lists. We conclude with recommendations for using existing data for watershed-level stream fish management and conservation decisions.