Using structural equation models to evaluate long-term aquatic effectiveness monitoring efforts

Background/Question/Methods

Conceptual models outlining anthropogenic stressors and their hypothesized impacts to biological integrity are critical to the design of effectiveness monitoring programs. Benthic macroinvertebrates (BMI) are one of the primary tools for quantifying the cumulative impacts of land management activities on biological integrity of stream ecosystems. While studies have shown BMIs are sensitive indicators of change within lowland, urban, and some forested applications, a paucity of studies have empirically evaluated the hypothesized responses of BMIs to large-scale land management activities. The reliance and emphasis on collecting macroinvertebrate data can be cost prohibitive for long term monitoring, thus verifying their utility as bioindicators is imperative.

We use data collected in 2008 and 2009 from 507 stream reaches probabilistically selected within the Interior Columbia River and Upper Missouri River basins as part of the Pacific Anadromous Fish Strategy and Inland Fish Strategy Biological Opinions program. We verify whether there is empirical evidence for the a priori hypothesized causal relationships using Bayesian structural equation models. Specifically, we hypothesize that macroinvertebrate assemblages will be negatively impacted by grazing and roads indirectly through stream habitat degradation (more fine sediments), altered stream temperature regimes, and degraded riparian habitat.

Results/Conclusions

Our preliminary results support the direct positive effect of grazing on in stream fine sediments [credible interval 0.01 to 0.36].  We also found evidence for the hypothesized negative direct effect of fine sediments on O/E score [credible interval -0.05 to -0.02]. The total indirect effect of grazing on O/E score was estimated to be negative [-0.03 to -0.01], albeit the magnitude of the effect was quite small. We did find evidence of a negative direct effect of grazing on O/E scores [-0.8 to -0.2]. We explore the added ecological insights structural equation models provide compared to other more common analytical methods used for such data.

We discuss the potential advantages and disadvantages of reallocating monitoring resources based on our results. Incorporating empirical evidence into the evaluation of the conceptual model underpinning the long term monitoring program is a necessary step to provide a critical evaluation of whether the a priori set of measurable variables is appropriate for continued monitoring.