Reach-scale models show heterogeneity of stream benthic invertebrate responses to eutrophication stress

Statistical stressor-response models are a common approach to derive biologically relevant water quality criteria for the management of waterbody health. These types of models are typically derived at state-wide or ecoregional scales and consequently incorporate a large amount of natural and disturbance-related variability that may obscure the relationship that one is interested in quantifying. We demonstrate an alternative approach termed "reach-specific modeling" to identify potential biological response thresholds to eutrophication in the Santa Margarita River watershed (California, U.S.A.). Individual models of benthic invertebrate response to eutrophication stress were created for both bioassessment sampling sites and NHD + stream-segments in the watershed (46 sites, 832 segments). Each model was built using only data from environmentally similar sites from a state-wide dataset to minimize variation from natural environmental gradients, while allowing eutro-phication stress to vary. Thresholds of potential biological impact were extracted from each stressor-response model. Across the whole watershed thresholds varied from location-to-location: total nitrogen (1.14-1.26 mg L-1 TN), total phosphorus (0.12-0.15 mg L-1 TP), benthic algal biomass (29-39 mg m(2) benthic chl-a), and benthic ash-free dry mass (2.5-mg cm(-2) AFDM). Notably, nearly all of the thresholds derived from these reach-specific models were similar to 10-90 % higher than those from a similar state-wide model. Furthermore, there were a number of spatial groupings of thresholds for each eutrophication indicator across the watershed, suggesting reach-scale natural gradients in hydrogeomorphology and natural land cover type may mediate the stressor-biology interaction. Reach-scale models tended to have better fits than their state-wide counterparts, but had equivalent or slightly worse accuracy. The reach-specific approach to threshold development illustrates that the biological response to stress is likely not uniform within a single system, much less between systems. As a consequence, this approach can allow managers to identify systems that are more sensitive or resistant to a given stressor across diverse landscapes and make better informed decisions on their management accordingly.