Development of Biological Indicators for New England Lakes Based on Benthic Macroinvertebrate and Zooplankton Assemblages
Yong Cao 1 , Charles P. Hawkins 1 and Alan T. Herlihy 2
1 Western Center for Monitoring and Assessment of Freshwater Ecosystems, Department of Watershed Sciences, and Ecology Center, Utah State University, Logan, Utah
2 Department of Fisheries and Wildlife, Oregon State University, Corvallis, Oregon
We used data collected from New England lakes (US EPA–EMAP) to explore the performance of two types of biological indicators based on benthic macroinvertebrate and zooplankton assemblages: O/E and multimetric indices.
We applied Random Forests (RF) predictive models to data collected from 24 (macroinvertebrates) and 74 (zooplankton) reference quality lakes to develop the O/E indices. These two O/E indices were reasonably precise (CV = 0.13 and 0.14 for macroinvertebrate and zooplankton indices, respectively) and responsive (86% and 77% of stressed sites rated as impaired, respectively). We also used RF to model the effects of natural environmental gradients (e.g., lake size, depth, and temperature) on values of 44 macroinvertebrate and 40 zooplankton candidate metrics, 9 and 15 of which varied with natural gradients.
Model residuals for these metrics, together with the other unadjusted metrics, were further screened for redundancy and discrimination efficiency. Four metrics were selected (two unadjusted and two adjusted metrics) for inclusion in both the macroinvertebrate and zooplankton multimetric indices. These two indices were less precise (CV=0.21 and 0.17 for benthos and zooplanton, respectively) than the O/E indices and less responsive (77% and 63% of degraded sites rated as impaired, respectively). Given that these indices performed as well as those used in streams, biological indicators based on either benthic or zooplankton assemblages should be useful in characterizing the ecological condition of lakes. However, our generally poor understanding of the tolerances of lake macroinvertebrate and zooplankton taxa to stressors may constrain the utility of multimetric indices in lake assessments.
Keywords: lake assessment, RIVPACS-type model, multimetric indices, Random Forest