Lester L. Yuan

Condition of stream ecosystems in the US: an overview of the first national assessment

Abstract The Wadeable Streams Assessment (WSA) provided the first statistically sound summary of the ecological condition of streams and small rivers in the US. Information provided in the assessment filled an important gap in meeting the requirements of the US Clean Water Act. The purpose of the WSA was to: 1) report on the ecological condition of all wadeable, perennial streams and rivers within the conterminous US, 2) describe the biological condition of these systems with direct measures of aquatic life, and 3) identify and rank the relative importance of chemical and physical stressors affecting stream and river condition. The assessment included perennial wadeable streams and rivers that accounted for 95% of the length of flowing waters in the US. The US Environmental Protection Agency, states, and tribes collected chemical, physical, and biological data at 1392 randomly selected sites. Nationally, 42% of the length of US streams was in poor condition compared to best available reference sites in their ecoregions, 25% was in fair condition, and 28% was in good condition. Results were reported for 3 major regions: Eastern Highlands, Plains and Lowlands, and West. In the West, 45% of the length of wadeable flowing waters was in good condition. In the Eastern Highlands, only 18% of the length of wadeable streams and rivers was in good condition and 52% was in poor condition. In the Plains and Lowlands, almost 30% of the length of wadeable streams and rivers was in good condition and 40% was in poor condition. The most widespread stressors observed nationally and in each of the 3 major regions were N, P, riparian disturbance, and streambed sediments. Excess nutrients and excess streambed sediments had the highest impact on biological condition; streams scoring poor for these stressors were at 2 to 3× higher risk of having poor biological condition than were streams that scored in the good range for the same stressors.

Striving for consistency in a national assessment: the challenges of applying a reference-condition approach at a continental scale

Abstract One of the biggest challenges when conducting a continental-scale assessment of streams is setting appropriate expectations for the assessed sites. The challenge occurs for 2 reasons: 1) tremendous natural environmental heterogeneity exists within a continental landscape and 2) reference sites vary in quality both across and within major regions of the continent. We describe the process used to set expectations for the multimetric index of biotic integrity (MIBI) and observed/expected (O/E) indices generated from predictive models used to assess stream condition for the US Wadeable Streams Assessment (WSA). The assessment was based on a reference-site approach, in which the least-disturbed sites in each region of the US were used to establish benchmarks for assessing the condition of macroinvertebrate assemblages at other sites. Reference sites were compiled by filtering WSA sample sites for disturbance using a series of abiotic variables. Additional reference sites were needed and were obtained from other state, university, and federal monitoring programs. This pool of potential reference sites was then assessed for uniformity in site quality and comparability of macroinvertebrate sample data. Ultimately, 1625 sites were used to set reference expectations for the WSA. Reference-site data were used to help define 9 large ecoregions that minimized the naturally occurring variation in macroinvertebrate assemblages associated with continental-wide differences in biogeography. These ecoregions were used as a basis for developing MIBI and O/E indices and for reporting results. A least-disturbed definition of reference condition was used nationally, but we suspect that the quality of the best extant sites in ecoregions, such as the Northern Plains and Temperate Plains, was lower than that of sites in other ecoregions. For the MIBI assessment, we used a simple modeling approach to adjust scores in ecoregions where gradients in reference-site quality could be demonstrated conclusively. The WSA provided an unparalleled opportunity to push the limits of our conceptual and technical understanding of how to best apply a reference-condition approach to a real-world need. Our hope is that we have learned enough from this exercise to improve the technical quality of the next round of national assessments.

Algae–P relationships, thresholds, and frequency distributions guide nutrient criterion development

Abstract We used complementary information collected using different conceptual approaches to develop recommendations for a stream nutrient criterion based on responses of algal assemblages to anthropogenic P enrichment. Benthic algal attributes, water chemistry, physical habitat, and human activities in watersheds were measured in streams of the Mid-Atlantic Highlands region as part of the Environmental Monitoring and Assessment Program of the US Environmental Protection Agency. Diatom species composition differed greatly between low- and high-pH reference streams; therefore, analyses for criterion development were limited to a subset of 149 well-buffered streams to control for natural variability among streams caused by pH. Regression models showed that TP concentrations were ∼10 μg/L in streams with low levels of human activities in watersheds and that TP increased with % agriculture and urban land uses in watersheds. The 75th percentile at reference sites was 12 μg TP/L. Chlorophyll a and ash-free dry mass increased and acid and alkaline phosphatase activities decreased with increasing TP concentration. The number of diatom taxa, evenness, proportion of expected native taxa, and number of high-P taxa increased with TP concentration in streams. In contrast, the number of low-P native taxa and % low-P individuals decreased with increasing TP. Lowess regression and regression tree analysis indicated nonlinear relationships for many diversity indices and attributes of taxonomic composition with respect to TP. Thresholds in these responses occurred between 10 and 20 μg/L and helped justify recommending a P criterion between 10 and 12 μg TP/L to protect high-quality biological conditions in streams of the Mid-Atlantic Highlands.

Comparing responses of macroinvertebrate metrics to increasing stress

Metrics characterizing the benthic macroinvertebrate assemblages in wadeable streams in the Mid-Atlantic region of the United States were analyzed to explore the relative responses of the metrics to different types of anthropogenic stress. The data used in our study were collected by the US Environmental Protection Agency Environmental Monitoring and Assessment Program from 1993 to 1996. Regression models were developed relating metric values at reference sites to natural sources of variability. These models were then used to predict reference values at test sites. Test site metric observations were scaled by subtracting the predicted reference value and dividing by the standard deviation of residuals at reference sites. Stressor–response relationships for each scaled metric were then estimated using generalized additive models. Metric responses to 4 groups of stressors (nutrient enrichment, habitat degradation, elevated metals concentrations, and elevated ion concentrations) were different. The proportional abundance of tolerant taxa was the most sensitive indicator of nutrient enrichment and habitat degradation, whereas Ephemeroptera richness was the most sensitive indicator of elevated metals or ion concentrations.

Estimating the effects of excess nutrients on stream invertebrates from observational data.

Increased nutrient concentrations in streams and rivers have altered biological structure and function. Manipulative studies have provided insights into different mechanisms by which changes in nutrient concentrations influence aquatic biota, but these studies are limited in spatial scope and in their quantification of nutrient effects on aggregate measures of the invertebrate assemblage. Observational data provide a complementary source of information to manipulative studies, but these data must be analyzed such that the potential effects of spurious correlations are minimized. Propensity scores, a technique developed to analyze human health observational data, are applied here to estimate the effects of increased nutrients on the total taxon richness of stream invertebrates in a large observational data set collected from the western United States. The analysis indicates that increases in nutrient concentration are strongly associated with and cause decreases in invertebrate richness in large, but wadeable, open-canopied streams. These decreases in invertebrate richness were not mediated by periphyton biomass, a commonly proposed mechanism by which nutrients influence invertebrates. In smaller, closed-canopied streams, increases in nutrients were associated with small increases in total richness that were not statistically significant. Using propensity scores can greatly improve the accuracy of insights drawn from observational data by minimizing the potential that factors other than the factor of interest may confound the results.

CADDIS: The Causal Analysis/Diagnosis Decision Information System

Biological monitoring and assessment methods have become indispensable tools for evaluating the condition of aquatic and terrestrial ecosystems. When an undesirable biological condition is observed (e.g., a depauperate fish assemblage), its cause (e.g., toxic substances, excess fine sediments, or nutrients) must be determined in order to design appropriate remedial management actions. Causal analysis challenges environmental scientists to bring together, analyze, and synthesize a broad variety of information from monitoring studies, models, and experiments to determine the probable cause of ecological effects. Decision-support systems can play an important role in improving the efficiency, quality and transparency of causal analyses.

Effects of regionalization decisions on an O/E index for the US national assessment

Abstract We examined the effects of different regionalization schemes on the performance of River InVertebrate Prediction and Classification System (RIVPACS)-type predictive models in assessing the biological conditions of streams of the US for the National Wadeable Streams Assessment (WSA). Three regionalization schemes were considered: a single national predictive model (MOD1), separate predictive models for each of the 9 WSA aggregated Omernik level III ecoregions (MOD9), and 3 predictive models roughly corresponding to the western US, the Appalachian Mountains, and the Central and Coastal Plains (MOD3). The goal of the WSA was to assess stream condition at the national scale and at the scale of WSA aggregated ecoregions, so we compared the performance of the ratio of the observed number of taxa to the expected number of taxa (O/E) index estimated using different regionalization schemes at both of these spatial scales. We assessed model performance with a randomized resampling procedure, in which we set aside 10% of the reference sites, calibrated the model with the remaining sites, and applied the model to the set-aside sites. Performance statistics for the set-aside reference sites were accumulated over 10 iterations. When summarized at the national scale, mean model predictions of O/E for set-aside reference sites from the 3 different regionalization schemes were all reasonably close to 1. When summarized by the 9 aggregated ecoregions, MOD1 and MOD3 predictions of O/E differed systematically from 1 in certain aggregated ecoregions. Over all 9 ecoregions, the magnitude of these differences was significantly greater than observed with MOD9 predictions. Results from our analysis suggest that O/E values at test sites should be interpreted with respect to mean and SD of O/E of reference sites from the same region to minimize the effects of systematic biases in the predictions. RIVPACS-type predictive models also should be calibrated at a spatial scale similar to the scale at which summary statistics are reported.

Community Response Patterns: Evaluating Benthic Invertebrate Composition In Metal-Polluted Streams

Human activities are modifying the condition and character of ecosystems at a rapid rate. Because of these rapid changes, questions concerning how ecosystems and their assemblages respond to anthropogenic stressors have been of general interest. Accurate prediction of assemblage composition in ecosystems with anthropogenic degradation requires that we assess both how assemblages respond to stressors and the generality of the responses. We ask whether assemblage composition among stream sites becomes more similar after exposure to a common stressor. Using data from biological monitoring programs in the southern Rocky Mountain ecoregion of Colorado and in West Virginia, we compare benthic invertebrate similarity and assemblage composition among sites having different levels (background, low, medium, and high) of heavy-metal pollution. Invertebrate assemblages were most similar within the background metal category, and similarity was progressively lower in low, medium, and high metal categories. An analysis of the frequency of occurrence of genera within metal categories reveals taxonomic shifts that conform to expectations based on metal tolerance of benthic invertebrates. However, different metal-tolerant genera were found at different metal-impacted sites, suggesting that local abiotic and biotic processes may influence the identity of the metal-tolerant genera that become established in polluted sites. Low community similarity in the medium and high-metal categories suggests that accurate prediction of assemblage composition at impacted sites may be challenging.

Using spatial interpolation to estimate stressor levels in unsampled streams.

Accurate estimates of stressor levels in unsampled streams would provide valuable information for managing these resources over large regions. Spatial interpolation of stream characteristics have rarely been attempted, partly because defining separation distances between distinct stream samples is not straightforward. That is, conventional Eulerian definitions of separation distance may not apply to stream networks where information flows along distinct paths. A two-stage model for estimating stressor levels in unsampled streams is presented. Mean characteristics within streams are predicted using a generalized additive model and residual variation is estimated using a conventional application of spatial statistics. The model is developed and tested using stream survey data collected in the state of Maryland, USA. Model efficiency is compared for three stream variables (nitrate concentration, sulfate concentration, and epifaunal substrate score) known to be associated with biological impairments in streams. Accounting for spatial autocorrelation in the residual variation improved model R2 from 0.71 to 0.81 for nitrate, from 0.29 to 0.63 for sulfate, and from 0.21 to 0.31 for epifaunal substrate score.

Classifying lakes to improve precision of nutrient–chlorophyll relationships

Abstract: Accurate and precise estimates of relationships between stressors and environmental responses can inform management decisions most usefully when models can be easily interpreted. Here, we describe an approach for classifying lakes and reservoirs that can improve estimates of the relationships between total P (TP) and chlorophyll a (chl a) concentration, while preserving a model that can be readily interpreted by environmental managers and stakeholders. We selected classification variables statistically with a classification and regression tree in which relationships between TP and chl a were the terminal nodes of the tree. We developed a set of classification trees from bootstrapped replicates of the calibration data to explore a broader range of possible trees. We chose a final tree based on its predictive performance with a validation data set. The total N:TP mass ratio was the classification variable selected most frequently from a broad array of biological, chemical, and physical candidate classification variables. Relationships between TP and chl a in the resulting lake classes provided predictions that were substantially more accurate than predictions computed using nutrient ecoregions based on aggregations of Omernik Level III ecoregions, but predictions from a random forest model that averaged an ensemble of trees were even more accurate. Thus, the classification approach presented here sacrifices a small amount of predictive accuracy to retain a tree structure that is readily interpretable.