Using watershed characteristics to inform cost-effective stream temperature monitoring

Water temperature is a key driver of aquatic processes. Monitoring stream water temperature is key to understanding current species distributions and future climate change impacts on freshwater ecosystems. However, a very small fraction of streams are continuously monitored for water temperature throughout North America, due to prohibitive logistical costs. We develop a framework that aids in developing cost-effective stream temperature monitoring by using stream habitat features to inform strategic site selection of temperature monitoring sites. We test this framework using sockeye salmon spawning streams as a model, which included 19 streams in the northern-most watershed of the Fraser River Basin, British Columbia, Canada. The objective of this framework is to evaluate the trade-off between cost (i.e., the number of streams monitored) and the effectiveness of monitoring scenarios at meeting different monitoring objectives. We compared monitoring scenarios that were informed by well-established relationships between variables and that are commonly collected or available as part of other monitoring activities (stream length, magnitude, order, gradient, wetted width, and spot temperatures) and water temperature metrics (maximum, mean, and variance during August) derived from continuously monitored streams to monitoring scenarios where streams were randomly selected. Informed scenarios included streams that were selected in order of watershed level and stream habitat characteristics (e.g., longest to shortest); ordering was based on the relationship between each habitat variable and temperature metrics. Informed monitoring scenarios were then compared to random selection of monitoring sites with regard to how well monitoring scenarios met two management objectives during the critical salmon spawning period: (1) identifying streams that exceed a temperature threshold and (2) identifying streams that represent the temperature regime of a complex of streams (e.g., mean and variance of streams within an aggregate of streams). Management objectives were met by monitoring fewer streams using the informed monitoring scenarios rather than the average of the random scenarios. This highlights how common inexpensive watershed level variables that relate to stream temperature can inform the strategic selection of sites and lead to more cost-effective stream temperature monitoring.