Effort-based predictors of headwater stream conditions: comparing the proximity of land use pressures and instream stressors on macroinvertebrate assemblages

• Published in: Aquatic sciences Vol. 79; no. 3; pp. 765 - 781
• Main Authors: Pond, Gregory J., Krock, Kelly J. G., Cruz, Jonathan V., Ettema, Leah F.
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 01-07-2017; Springer Nature B.V
• Subjects: Algae; Autocorrelation; Biomedical and Life Sciences; Buffers (chemistry); Catchment area; Chemistry; Conductance; Conservation; Cost engineering; Density; Ecological monitoring; Ecology; Forest watersheds; Freshwater & Marine Ecology; Habitats; Headwaters; Health; Laboratories; Land use; Life Sciences; Macroinvertebrates; Marine & Freshwater Sciences; Oceanography; Ordination; Predictions; Proximity; Regression analysis; Remediation; Research Article; Resistance; Riparian forests; Riparian land; Rivers; Scales; Streams; Strength; Upstream; Urban areas; Water; Water chemistry; Zoobenthos; Buffer zone; Predictive models; Urbanization; Land use alteration; Bioassessment; Headwater
• ISSN: 1015-1621; 1420-9055
• DOI: 10.1007/s00027-017-0534-3

Environmental agencies are often faced with resource and time constraints in assessing waterbody health. We compared the strengths of varying levels of effort (field measures, laboratory chemistry, land use, and multiple combinations of these) to explain macroinvertebrate assemblage response along a gradient of urban land use intensity among 30 headwater streams in northern West Virginia. Because the spatial arrangement of human disturbance can govern biotic response, land use effects were analyzed at five spatial scales (whole catchment, and 100 m buffer zone at three fixed upstream distances and total stream network upstream of site); instream ecological measures included physical habitat, algal concentrations and water chemistry. Of the five spatial scales, we predicted that riparian land use nearest the site would explain the most variation but that instream measures would be the overall driver of the macroinvertebrate assemblages. Regression analysis evaluated the strength of single and multiple variables in explaining macroinvertebrate multimetric index (MMI) and ordination patterns, and revealed that assemblages were highly responsive to numerous stressors. In contrast to predictions, total upstream network riparian forest cover explained the most variation overall (83%) while specific conductance was the single best instream measure (64%). Stepwise regression models using combinations of field, laboratory, and land use variables all performed reasonably well but we found that a 3-variable model [% forest (catchment), road density, and specific conductance] that minimized colinearity and cost/effort explained 90% of the variation in the MMI. Validation and spatial autocorrelation results suggest that this model could potentially be used to forecast stream condition for prioritizing conservation and remediation efforts in headwaters within the ecoregion, and our general approach would be broadly applicable in other settings.