2003 Progress Report: Protocols for Selection of Classification System and Reference Conditions: A Comparison of Methods

EPA Grant Number: R828777
Title: Protocols for Selection of Classification System and Reference Conditions: A Comparison of Methods
Investigators: Johnson, Lucinda , Richards, Carl , Schuldt, Jeffrey A. , Host, George E. , Brady, Valerie J , Holland, Jeffrey L. , Hollenhorst, Thomas P. , Breneman, Dan , Ciborowski, Jan , Gathman, Joseph
Institution: University of Minnesota - Duluth
EPA Project Officer: Packard, Benjamin H
Project Period: March 26, 2001 through March 25, 2004 (Extended to March 25, 2005)
Project Period Covered by this Report: March 26, 2003 through March 25,2004
Project Amount: $707,404
RFA: Development of National Aquatic Ecosystem Classifications and Reference Conditions (2000) RFA Text |  Recipients Lists
Research Category: Aquatic Ecosystems , Water , Ecosystems


The objectives of this research project are to:  (1) apply an a priori classification system to Great Lakes coastal ecosystems; (2) quantitatively identify least impacted sites within classification units using widely available spatial databases; (3) define biological conditions that best characterize reference classification units; (4) use biological data to test the efficacy of reference classifications; and (5) determine whether biological conditions in degraded classification units differ from conditions in reference areas at a subset of classification units.


We will combine an a priori classification system, pressure indicators quantifying the degree of anthropogenic impact, and biotic communities in an iterative process to identify the most useful level of classification and reference conditions within the classification system. To initiate this iterative process we will apply a highly specific classification system to Great Lakes coastal ecosystems. The degree of anthropogenic influence on individual ecosystems will be assessed extrinsically from remote sensed data and spatial data bases. Within classification units, reference ecosystems will be identified as those sites with the least amount of anthropogenic influence. We will then define the biotic (macrobenthos and fish) conditions of these reference ecosystems. If we cannot identify unique reference biotic conditions at the most specific level of classification, we will refine our classification system using agglomerative approaches. Because of the potential interdependence of the classification system and the reference conditions identified within the system, this process is not completely subject to quantitative testing. We have, however, introduced quantitative decision points in the process. We will also test whether the biological conditions at degraded sites differ from reference conditions for a subset of our classification units.

Progress Summary:

A Priori Reference Area Selection Procedure
We first sought to develop a quantitative a priori process for selecting reference sites that eliminated the subjectivity associated with “best professional judgement” reference site selection.  Although there are several ways to define reference condition, for this project we defined reference condition as the relative absence of anthropogenic stresses (or “least impacted”).
Classification/Hydrogeomorphic Classification of Shorelines
Two types of classifications were employed to stratify sites across the Great Lakes coastal region: an ecoregion-based classification (Omernik 1987) and a geomorphic classification of shoreline units.  An inventory of the geomorphic types in each of the Great Lakes ecological sections was completed in Year 1 of the project.
Criteria for Identifying Reference Condition Locations
Measures of residential/commercial land use, agricultural land use, human population density, and distance to point sources were summarized for each wetland and its receiving area.  We calculated an anthropogenic pressure (AP) score for high-energy zones and wetlands across the basin according to the methods described in Host, et al. (2004).  Sites falling into the lowest 20th percentile AP scores within an ecosection were considered as reference sites.
Evaluating Reference Conditions Using Fish and Invertebrates
We have elected to evaluate our reference condition approach by characterizing fish, macroinvertebrates, and physical habitat attributes.  The underlying assumption of the reference condition approach is that the biological integrity is closely tied to environmental condition. Thus, biota of nonreference sites is expected to differ from that of reference sites.  This information will allow us to establish a benchmark (both a value and some measure of variability) that reflects the best attainable abiotic and biotic conditions and evaluate the appropriate spatial scale for developing reference areas for coastal regions.
Preliminary Results
Classification Responses.  Preliminary analyses of the fish data from all sites show that high- energy shorelines have markedly different fish communities than those found in Great Lakes’ wetlands and embayments (multiresponse permutation parameter p < 0.0001).  Analyses using nonmetric multidimensional scaling analysis and multiresponse permutation procedure showed a clear separation between sites in the northern and southern ecoprovinces (p < 0.0001) for wetland and high-energy sites.  Consequently, the northern and southern sites were analyzed separately (Brady, et al., 2004; Bhagat, et al., 2004).  In both ecoprovinces, fish community structure reflected the degree of hydrodynamic exposure to which sites were subjected, with the more open types (e.g., fringing coastal wetlands) grouping somewhat distinctly from more protected site types (protected and riverine coastal wetlands).  There also were some distinct fish community characteristics within each of the five Great Lakes (p = 0.05).

Table 1. Fish Community Metrics for Reference (U.S. Environmental Protection Agency) and Non reference (Great Lakes Environmental Indicator) Sites Across the Northern (212) and Southern (222) Ecoprovinces of the Great Lakes. Numbers represent means (and standard deviations).














% Piscivore Indiv

0.09 (0.06)

0.07 (0.07)

0.2 (0.09)

0.05 (0.05)

% Exotic Indiv

0.04 (0.08)

0 (0)

0.01 (0.01)

0.02 (0.04)

% Exotic Sp

0.1 (0.04)

0.01 (0.03)

0.11 (0.08)

0.08 (0.09)

% Piscivore Sp

0.35 (0.15)

0.16 0.1)

0.5 (0.1)

0.1 (0.06)

Exotic Richness

1.67 (0.82)

0.2 (0.42)

1.6 (0.89)

1.15 (1.68)

Reference Community Identification.  To validate the approach for selecting reference areas, we compared the observed fish community to the community that would be expected under “pristine” conditions.  The expected community composition was derived from four sources, including:  Scott and Crossman (1973), the Michigan Department of Natural Resources Fish Atlas, the Ontario Freshwater Fishes Life History Database, and the Fishbase Web site (www.fishbase.org/search.cfm Exit ).  We used range data to create a site-by-species matrix defining whether or not each site was in each species’ range and expected habitat.  We then created a distance matrix (1-Jaccard) between reference sites based on expected fish assemblages.  The two distance matrices were compared using a Mantel test.  There was a weak, but significant, relation between the expected assemblages and the actual assemblages (r = 0.15, p < 0.0012), suggesting that the reference condition sites are acting as such, according to the fish species caught.  We hope to refine these analyses based on higher-resolution fish distribution data.

Expected Results:

The proposed research will identify scientifically defensible protocols for selecting an appropriate classification system and defining reference conditions. These protocols will rely on a combination of readily available, as well as newly collected field data. The methods will be appropriate to large geographic regions with a number of distinct ecosystem types.

Future Activities:

We will complete the benthic sample processing and invertebrate identifications.  Ongoing analyses are taking place to:  (1) determine the spatial scale of fish species responses to environmental stressors; (2) quantify fish community structure and variability in reference and nonreference sites; (3) quantify habitat structure and variability in reference and nonreference sites; (4) assess classifications of Great Lakes coastal wetlands; and (5) reference conditions and axes of environmental stress, including developing, integrating, and evaluating indicators of environmental conditions at Great Lakes coastal margins.


Bhagat Y, Ciborowski JJH, Hrabik TR, Johnson LB, Richards C, Schuldt J.  The use of fish communities to assess water quality along the U.S. Great Lakes coastline:  a comparison of multimetric and multivariate approaches.  Presented to the International Association of Great Lakes Research, Waterloo, Ontario, Canada, June 2004.
Brady V, Schuldt J, Johnson L, Ciborowski JJH, Host GE, Hollenhorst T, Richards C, Breneman D, Gathman J.  Identification of reference Great Lakes coastal wetlands and comparison of fish communities between reference and nonreference systems.  Presented at the Environmental Monitoring and Assessment Program Symposium, Newport, RI, May 2004.
Omernik JM.  Ecoregions of the conterminous United States.  Annals of the Association of American Geographers 1987;77:118-125.
Scott WB, Crossman EJ.  Freshwater Fishes of Canada.  Fisheries Research Board of Canada - Bulletin 184. 1973.

Journal Articles on this Report : 2 Displayed | Download in RIS Format

Other project views: All 10 publications 2 publications in selected types All 2 journal articles
Type Citation Project Document Sources
Journal Article Grigorovich IA, Mills EL, Richards CB, Breneman D, Ciborowski JJH. European valve snail Valvata piscinalis (Müller) in the Laurentian Great Lakes basin. Journal of Great Lakes Research 2005;31(2):135-143. R828777 (2003)
R828675 (Final)
R828675C003 (2004)
R828675C003 (Final)
  • Full-text: ScienceDirect-PDF
  • Abstract: ScienceDirect-Abstract
  • Other: IAGLR-Abstract
  • Journal Article Host GE, Schuldt J, Ciborowski JJH, Johnson LB, Hollenhorst T, Richards C. Use of GIS and remotely sensed data for a priori identification of reference areas for Great Lakes coastal ecosystems. International Journal of Remote Sensing 2005;26(23):5325-5342. R828777 (2003)
    R828675 (Final)
    R828675C003 (Final)
  • Full-text: GLEI-PDF
  • Abstract: Taylor & Francis Online-Abstract
  • Supplemental Keywords:

    watersheds, water, ecological effects, aquatic, habitat, integrated assessment, biology, ecology, modeling, surveys, measurement methods, satellite, landsat, remote sensing, Great Lakes, Midwest, pressure indicator, classification system, environmental reference condition, biotic reference condition, Great Lakes, pressure indicator, classification system,, RFA, Scientific Discipline, Geographic Area, Ecosystem Protection/Environmental Exposure & Risk, Hydrology, Aquatic Ecosystems & Estuarine Research, Aquatic Ecosystem, Ecology and Ecosystems, Great Lakes, reference condition, coastal environments, anthropogenic impact, comparison of methods, environmental indicators, biological indicators, ecological classification, remotely sensed data

    Progress and Final Reports:

    Original Abstract
  • 2001 Progress Report
  • 2002 Progress Report
  • Final