Marc H. Weber

National and regional comparisons between Strahler order and stream size

Abstract Water-body size is one of the most important factors affecting the structure and function of aquatic ecosystems. The categorical variable, Strahler stream order, is commonly used as a surrogate for stream size, perhaps because stream size is a multidimensional attribute that defies simple definition. Some stream-size attributes, including continuous variables, such as catchment area, distance to source, and model estimates of discharge also are available as geographic information system (GIS)-derived or modeled variables. These GIS measures are commonly used by stream ecologists along with field-derived attributes, such as discharge, stream cross-sectional area, width, and depth, which are more direct measures of stream size as experienced by aquatic organisms. Our objective was to quantify how well some commonly used stream-size attributes are predicted from Strahler order in the US as a whole and within major ecoregions and hydrologic landscape regions. We based our analysis on field-channel-survey and digital-stream-trace data (NHD-Plus) from 2162 US stream sites, ranging from 1st to 8th order (at 1∶100,000 scale). Strahler order provided a surprisingly useful approximation of the ranges of catchment size, distance to source, modeled mean annual discharge, and field-based low-flow and bankfull channel dimensions for most streams within a given Strahler order. However, even within geoclimatically and ecologically similar regions, site-specific predictions of stream size from Strahler order can have large errors. Correlations between Strahler order and the size measures considered here varied widely (r  =  0.48–0.91). Within individual Strahler orders, the alternative size measures varied by 5 and 4 orders of magnitude at national and regional scales, respectively. The same size-measure value could occur in 1 to 7 different stream orders at the national scale and in some regions, with generally good agreement in mountains and poor agreement in plains. Therefore, we conclude that Strahler order is useful for relating information about stream size, but that researchers should base analyses on multiple, continuous measures of stream size and should communicate stream-size results or associations based on the size-related measurements. Two characteristics of Strahler order make it useful for selecting sites across the range of stream sizes encountered in regional and national surveys, as long as limitations are explicitly recognized. First, the number of Strahler orders is limited. Second, Strahler order is easy to extract from stream networks constructed from digital elevation data and from national hydrographic data sets.

The Stream‐Catchment (StreamCat) Dataset: A Database of Watershed Metrics for the Conterminous United States

We developed an extensive database of landscape metrics for ~2.65 million stream segments, and their associated catchments, within the conterminous United States (U.S.): The Stream-Catchment (StreamCat) Dataset. These data are publically available (http://www2.epa.gov/national-aquatic-resource-surveys/streamcat) and greatly reduce the specialized geospatial expertise needed by researchers and managers to acquire landscape information for both catchments (i.e., the nearby landscape flowing directly into streams) and full upstream watersheds of specific stream reaches. When combined with an existing geospatial framework of the Nations rivers and streams (National Hydrography Dataset Plus Version 2), the distribution of catchment and watershed characteristics can be visualized for the conterminous U.S. In this article, we document the development and main features of this dataset, including the suite of landscape features that were used to develop the data, scripts and algorithms used to accumulate and produce watershed summaries of landscape features, and the quality assurance procedures used to ensure data consistency. The StreamCat Dataset provides an important tool for stream researchers and managers to understand and characterize the Nations rivers and streams.

Survey design and extent estimates for the National Lakes Assessment

Abstract.  The US Environmental Protection Agency (USEPA) conducted a National Lakes Assessment (NLA) in the conterminous USA in 2007 as part of a national assessment of aquatic resources. The EPA used the National Hydrography Dataset (NHD) as the basis for the sample frame for the NLA. The target population was all lakes >4 ha, excluding the Laurentian Great Lakes and the Great Salt Lake. An unequal probability survey design was used to select 4472 candidate lakes for potential sampling. The unequal selection depended on 5 lake area classes and 9 aggregated Omernik level III ecoregions. In all, 2034 candidate lakes were evaluated for inclusion in the target population, and 1309 lakes (representing ∼68,000 lakes in the sample frame) met the criteria. A total of 1028 lakes (of 1309) were sampled and represented ∼50,000 lakes. The remaining lakes (231, representing ∼18,000 lakes) could not be sampled because of access denial or physical inaccessibility. The target population included natural (41 ± 2% [SE]) and man-made lakes (59 ± 2%). All target lakes in the Southern Appalachian region and >90% of the target population in the Southern Plains and Xeric regions were man-made. In the Upper Midwest region, 97 ± 1% of the target population were natural lakes. Small lakes (4–10 ha) made up 47 ± 2% of the target population, and lakes >50 ha made up ∼15% of the target population. The results raise 2 issues that have implications for current and future NLA projects: 1) the cost and effort required to identify lake features in the sample frame that do not meet the criteria for inclusion in the target population (∼50% in NLA 2007), and 2) the potential for biased estimates of the size and condition of the target population caused by lakes that cannot be sampled. Future NLA efforts involve refining the survey design to include smaller lakes and resampling lakes from previous NLAs. We offer approaches for addressing both issues, including use of a high-resolution version of NHD as the basis for developing the NLA sample frame. Developing a master sample frame of lakes would provide a consistent basis of lake numbers (or surface area) from which to estimate extent or assess ecological condition.

Modeling tribal exposures to methyl mercury from fish consumption.

Exposure assessment and risk management considerations for tribal fish consumption are different than for the general U.S. population because of higher fish intake from subsistence fishing and/or from unique cultural practices. This research summarizes analyses of available data and methodologies for estimating tribal fish consumption exposures to methyl mercury (MeHg). Large MeHg fish tissue data sets from the Environmental Protections Agencys (EPAs) Office of Water, USGSs EMMMA program, and other data sources, were integrated, analyzed, and combined with fish intake (consumption) data for exposure analyses using EPAs SHEDS-Dietary model. Results were mapped with GIS tools to depict spatial distributions of the MeHg in fish tissues and fish consumption exposure patterns. Contribution analyses indicates the major sources for those exposures, such as type and length of fish, geographical distribution (water bodies), and dietary exposure patterns. Sensitivity analyses identify the key variables and exposure pathways. Our results show that MeHg exposure of tribal populations from fish are about 3 to 10 times higher than the US general population and that exposure poses potential health risks. The estimated risks would be reduced as much as 50%, especially for high percentiles, just by avoiding consumption of fish species with higher MeHg concentrations such as walleye and bowfin, even without changing total fish intake. These exposure assessment methods and tools can help inform decisions regarding meal sizes and frequency, types of fish and water bodies to avoid, and other factors to minimize exposures and potential health risks from contaminated fish on tribal lands.

Mapping watershed integrity for the conterminous United States

Watershed integrity is the capacity of a watershed to support and maintain the full range of ecological processes and functions essential to sustainability. Using information from EPAs StreamCat dataset, we calculated and mapped an Index of Watershed Integrity (IWI) for 2.6 million watersheds in the conterminous US with first-order approximations of relationships between stressors and six watershed functions: hydrologic regulation, regulation of water chemistry, sediment regulation, hydrologic connectivity, temperature regulation, and habitat provision. Results show high integrity in the western US, intermediate integrity in the southern and eastern US, and the lowest integrity in the temperate plains and lower Mississippi Valley. Correlation between the six functional components was high (r = 0.85-0.98). A related Index of Catchment Integrity (ICI) was developed using local drainages of individual stream segments (i.e., excluding upstream information). We evaluated the ability of the IWI and ICI to predict six continuous site-level indicators with regression analyses - three biological indicators and principal components derived from water quality, habitat, and combined water quality and habitat variables - using data from EPAs National Rivers and Streams Assessment. Relationships were highly significant, but the IWI only accounted for 1-12% of the variation in the four biological and habitat variables. The IWI accounted for over 25% of the variation in the water quality and combined principal components nationally, and 32-39% in the Northern and Southern Appalachians. We also used multinomial logistic regression to compare the IWI with the categorical forms of the three biological indicators. Results were consistent: we found positive associations but modest results. We compared how the IWI and ICI predicted the water quality PC relative to agricultural and urban land use. The IWI or ICI are the best predictors of the water quality PC for the CONUS and six of the nine ecoregions, but they only perform marginally better than agriculture in most instances. However, results suggest that agriculture would not be appropriate in all parts of the country, and the index is meant to be responsive to all stressors. The IWI in its present form (available through the StreamCat website; https://www.epa.gov/national-aquatic-resource-surveys/streamcat) could be useful for management efforts at multiple scales, especially when combined with information on site condition. The IWI could be improved by incorporating empirical or literature-derived relationships between functional components and stressors. However, limitations concerning the absence of data for certain stressors should be considered.

Assessing the accuracy and stability of variable selection methods for random forest modeling in ecology

Random forest (RF) modeling has emerged as an important statistical learning method in ecology due to its exceptional predictive performance. However, for large and complex ecological data sets, there is limited guidance on variable selection methods for RF modeling. Typically, either a preselected set of predictor variables are used or stepwise procedures are employed which iteratively remove variables according to their importance measures. This paper investigates the application of variable selection methods to RF models for predicting probable biological stream condition. Our motivating data set consists of the good/poor condition of n = 1365 stream survey sites from the 2008/2009 National Rivers and Stream Assessment, and a large set (p = 212) of landscape features from the StreamCat data set as potential predictors. We compare two types of RF models: a full variable set model with all 212 predictors and a reduced variable set model selected using a backward elimination approach. We assess model accuracy using RF’s internal out-of-bag estimate, and a cross-validation procedure with validation folds external to the variable selection process. We also assess the stability of the spatial predictions generated by the RF models to changes in the number of predictors and argue that model selection needs to consider both accuracy and stability. The results suggest that RF modeling is robust to the inclusion of many variables of moderate to low importance. We found no substantial improvement in cross-validated accuracy as a result of variable reduction. Moreover, the backward elimination procedure tended to select too few variables and exhibited numerous issues such as upwardly biased out-of-bag accuracy estimates and instabilities in the spatial predictions. We use simulations to further support and generalize results from the analysis of real data. A main purpose of this work is to elucidate issues of model selection bias and instability to ecologists interested in using RF to develop predictive models with large environmental data sets.

How Misapplication of the Hydrologic Unit Framework Diminishes the Meaning of Watersheds

Hydrologic units provide a convenient but problematic nationwide set of geographic polygons based on subjectively determined subdivisions of land surface areas at several hierarchical levels. The problem is that it is impossible to map watersheds, basins, or catchments of relatively equal size and cover the whole country. The hydrologic unit framework is in fact composed mostly of watersheds and pieces of watersheds. The pieces include units that drain to segments of streams, remnant areas, noncontributing areas, and coastal or frontal units that can include multiple watersheds draining to an ocean or large lake. Hence, half or more of the hydrologic units are not watersheds as the name of the framework “Watershed Boundary Dataset” implies. Nonetheless, hydrologic units and watersheds are commonly treated as synonymous, and this misapplication and misunderstanding can have some serious scientific and management consequences. We discuss some of the strengths and limitations of watersheds and hydrologic units as spatial frameworks. Using examples from the Northwest and Southeast United States, we explain how the misapplication of the hydrologic unit framework has altered the meaning of watersheds and can impair understanding associations between spatial geographic characteristics and surface water conditions.

Predictive mapping of the biotic condition of conterminous U.S. rivers and streams

Understanding and mapping the spatial variation in stream biological condition could provide an important tool for conservation, assessment, and restoration of stream ecosystems. The USEPAs 2008-2009 National Rivers and Streams Assessment (NRSA) summarizes the percentage of stream lengths within the conterminous United States that are in good, fair, or poor biological condition based on a multimetric index of benthic invertebrate assemblages. However, condition is usually summarized at regional or national scales, and these assessments do not provide substantial insight into the spatial distribution of conditions at unsampled locations. We used random forests to model and predict the probable condition of several million kilometers of streams across the conterminous United States based on nearby and upstream landscape features, including human-related alterations to watersheds. To do so, we linked NRSA sample sites to the USEPAs StreamCat Dataset; a database of several hundred landscape metrics for all 1:100,000-scale streams and their associated watersheds within the conterminous United States. The StreamCat data provided geospatial indicators of nearby and upstream land use, land cover, climate, and other landscape features for modeling. Nationally, the model correctly predicted the biological condition class of 75% of NRSA sites. Although model evaluations suggested good discrimination among condition classes, we present maps as predicted probabilities of good condition, given upstream and nearby landscape settings. Inversely, the maps can be interpreted as the probability of a stream being in poor condition, given human-related watershed alterations. These predictions are available for download from the USEPAs StreamCat website. Finally, we illustrate how these predictions could be used to prioritize streams for conservation or restoration.

Regional patterns of total nitrogen concentrations in the National Rivers and Streams Assessment

Patterns of nitrogen (N) concentrations in streams sampled by the National Rivers and Streams Assessment (NRSA) were examined semiquantitatively to identify regional differences in stream N levels. The data were categorized and analyzed by watershed size classes to reveal patterns of the concentrations that are consistent with the spatial homogeneity in natural and anthropogenic characteristics associated with regional differences in N levels. Ecoregions and mapped information on human activities including agricultural practices were used to determine the resultant regions. Marked differences in N levels were found among the nine aggregations of ecoregions used to report the results of the NRSA. We identified distinct regional patterns of stream N concentrations within the reporting regions that are associated with the characteristics of specific Level III ecoregions, groups of Level III ecoregions, groups of Level IV ecoregions, certain geographic characteristics within ecoregions, and/or particular watershed size classes. We described each of these regions and illustrated their areal extent and median and range in N concentrations. Understanding the spatial variability of nutrient concentrations in flowing waters and the apparent contributions that human and nonhuman factors have on different sizes of streams and rivers is critical to the development of effective water quality assessment and management plans. This semi-quantitative analysis is also intended to identify areas within which more detailed quantitative work can be conducted to determine specific regional factors associated with variations in stream N concentrations.