Jump to main content or area navigation.

Contact Us

Water: Monitoring & Assessment

1.0 Introduction

Bioindicators for Assessing Ecological Integrity of Prairie Wetlands
Report # EPA/ 600/ R-96/ 082
September 1995


1.1 Need for This Document
1.2 Document Organization
1.3 Cumulative Effects of Stressors
1.4 Glossary, Abbreviations, and Place Names
1.5 Statistical Analyses: Objectives and Methods

1. Introduction

1.1 Need for This Document

Under the Clean Water Act (CWA), wetlands are legally considered "waters of the state." Thus, states are required to adopt narrative standards and criteria for protecting quality of wetlands (USEPA 1987, 1989), just as states have developed standards and criteria for other surface waters.

The U.S. Environmental Protection Agency (USEPA) is responsible for developing regulations, policies, and guidance to help states implement a water quality standards program. USEPA policy requires that the states adopt biological criteria as part of their water quality standards for wetlands. USEPA recommends that the states use biological criteria to supplement the chemical and physical water quality standards they have used traditionally. USEPA has taken this approach because biological criteria measure the actual time-integrated response of the resident aquatic community to all environmental stresses, rather than inferring biological impairment from comparison of values derived from laboratory bioassays with instantaneous field measurements of the same or similar contaminants (USEPA 1990).

To satisfy USEPA requirements for biocriteria development, state agencies need technical information. Specifically, they need to know which biological resources to monitor in wetlands, how to monitor them, how to analyze and interpret the data, and what it costs. To adequately monitor wetlands, develop sound criteria and standards, and evaluate ecological risks, states also need information on what levels of various contaminants (or other regulated stresses) will impair the integrity of various kinds of wetland communities. USEPA is mandated to provide states with such technical guidance and information, drawn from comprehensive synthesis of literature, research, and expert knowledge.

This document is intended as a contribution to the effort to establish biocriteria in one region of the North America: the prairie region. The document compiles biological information on a single wetland type in this region: prairie potholes. These are wetlands that during most years are unconnected by surface water to lakes, rivers, or streams. The acreage of prairie pothole habitat has declined dramatically over the years as a result of human activities (Dahl 1990, Dahl et al. 1991), underlining the importance of monitoring and maintaining the quality of what habitat remains.

Processes for developing biocriteria may include:

  • developing and testing consistent and biologically meaningful classifications of ecoregions and wetland types
  • designing and conducting biosurveys, e.g., to establish and characterize reference sites and conditions
  • developing and calibrating sample metrics
  • evaluating data to assess environmental effects
  • analyzing collected data to devise biocriteria

This document assumes that the reader is generally familiar with these processes, and thus it does not discuss all of them in depth. Rather, the document emphasizes just one aspect of biocriteria development: the selection (targeting) of assemblages of biological indicators (bioindicators) and metrics, as a basis for designing and conducting biosurveys. Before meaningful biocriteria can be developed and implemented, appropriate bioindicators must be identified and tested. Bioindicators are species, species assemblages, or communities whose presence, abundance, and condition is indicative of a particular set of environmental conditions. Bioindicators can include assemblages of species that are important because they play pivotal roles in wetland ecosystems, or assemblages that show outstanding sensitivity to, or strong correlation with, particular anthropogenic or natural factors (stressors). Bioindicators can be used both to assess wetland condition and to help measure and diagnose the actual causes of impairment. Bioindicators that are cost-effective and sensitive to particular stressors are useful for measuring attainment of other wetland management objectives as well, such as criteria for successful restoration of wetlands. The most useful bioindicators are likely to be ones that can distinguish between natural variation (e.g., phenological changes, annual wet-dry cycles) and anthropogenic stresses, because the latter often mimic (and are overlaid upon) natural stresses, varying only in terms of relative magnitude.

Once bioindicator data have been collected, efforts are often made to integrate the data into metrics (e.g., density estimates, species counts) and ultimately to combine multiple metrics into indices of ecosystem condition. However, in the case of prairie wetlands, our knowledge of the relative performance of various metrics is severely limited, and no attempts have been made yet to devise and test indices of wetland integrity. For ecosystems generally, several excellent texts describe methods for developing and interpreting metrics and indices of condition (e.g., Green and Vascotto 1978, Gauch 1982, Pielou 1984, Isom 1986, Jongman et al. 1987, Ludwig and Reynolds 1988, Magurran 1988).

When developing biocriteria, it is seldom practical to address the environmental needs of all species within a particular assemblage of organisms (Landres 1992). Thus, many past efforts have focused on identifying functionally similar assemblages (or "guilds") of species and life stages. "Functionally similar" generally means similar with regard to reproductive strategy, food habits, and/or habitat preference. Examples of groupings specific to wetland or aquatic species include Dean-Ross and Mills 1989 (bacterial communities), Boutin and Keddy 1993 and Hills et al. 1994 (plants), Merritt and Cummins 1978 (macroinvertebrates), and Short 1989 (birds). A limitation of the functional ing approach is that much of the basic natural history information needed to validate the appropriateness of the ings and classifications for prairie wetlands is currently lacking.

1.2 Document Organization

Biological monitoring (biomonitoring) generally focuses on one or more broad assemblages of related organisms. For this reason, this document uses the commonly-used taxonomic assemblages of organisms (algae and microbes, vascular plants, invertebrates, amphibians, birds) as the main section headings. Fish and mammals were not discussed in this document because of their relatively low diversity in prairie wetlands and paucity of information. No attempt was made to give equal coverage to all topics in this document, because availability of data varies greatly.

The document provides information on each of the major assemblages of organisms in separate subsections:

Ecological Significance and Suitability as an Indicator: This describes why the particular assemblage is important to a wetland's functioning. That is, the subsection provides a rationale for using the assemblage as an assessment endpoint. The subsection also summarizes advantages and disadvantages of using the assemblage as an indicator of the ecological integrity of wetlands, and it defines ings that are conventionally applied within the larger taxonomic assemblage.

Potential Indicator Metrics: This subsection lists the metrics (measurable aspects that summarize biological structure or function, e.g., species richness) that show promise as indicators of the ecological integrity of wetlands when applied to the taxonomic assemblage. Metrics were included if they had been used previously in prairie wetlands, and/or were judged by the author to show promise (due to sensitivity, cost, and other factors) for reflecting wetland integrity. These lists of metrics are by no means definitive. Readers should not assume, because a metric is listed, that it has been "proven" by research, and should be aware that the listing is not all-inclusive. Many indicators deserve considerably more investigation and fine-tuning before they are used routinely. Whenever possible, users should obtain assistance from local wetland scientists when using the indicators to interpret wetland condition.

Previous and Ongoing Monitoring in the Region: From a review of over 400 publications, this subsection summarizes studies in the region that have addressed the taxonomic assemblage, and the most common themes among these studies. Some ongoing research (circa 1994) is also noted, but listings are not necessarily comprehensive. This subsection is provided to help readers understand the relative extent of knowledge about different taxa and stresses. Understanding the extent of the knowledge base allows users to exercise proper caution in interpreting statements in this documnet. Such an understanding also can help focus future research on important topics that previously have been understudied.

Response to Stressors: This is the largest of the subsections, and for each major taxonomic assemblage (primary headings), it compiles and organizes all available prairie wetland literature according to various stressors (secondary headings) and metrics (tertiary headings). Information on both anthropogenic and natural stressors is presented together because few prairie studies have reliably distinguished any differences in the responses of biological communities to the affects of these. Stressors are not necessarily "bad" for maintaining wetland resources and functions of interest to humans. Indeed, some degree of disturbance or stress, whether natural or anthropogenic, is vital to the evolution and sustainability of prairie wetlands, whereas excessive levels (too much or too little) of a stressor can spell the eventual loss of wetland function. Anthropogenic stressors are of special interest not only because they affect the structure and function of biological communities, but also an ecosystem's ability to rebound from natural stresses. They are also, by definition, more amenable to human control.

Information in the stressor subsections describes how each metric (e.g., biomass, richness) responds to each stress, with the caveat that much of this information is derived only from single, perhaps unrepresentative, studies. When information is sufficient, this subsection gives physiological thresholds for impacts occurring at broad taxonomic levels, e.g., the level of salinity at which most wetland plants are incapable of reproducing.

The stressors discussed in this subsection are factors that are most likely to impair the ecological integrity of prairie wetlands when present at levels (or times) that differ greatly from their usual natural occurrence, and belong to these categories:

Hydrologic Stressors: Changes in levels of surface water or water tables, i.e., drought or flood conditions, whether natural or aided by anthropogenic factors, e.g., drainage, groundwater withdrawal, global climate change.

Vegetative Cover Conditions: Changes in aerial cover and density of vascular plants, whether natural (e.g., from muskrat consumption) or aided by anthropogenic factors, e.g., grazing, burning, mowing, herbicide application.

Salinity: Changes in total dissolved solids in the water column, soils, or sediments, whether natural or aided by anthropogenic factors.

Sedimentation and Turbidity: Physical changes a wetland's benthic (bottom) substrate and/or changes in light penetration caused by introduction or resuspension of living or (especially) non-living matter, as aided by either natural or anthropogenic factors, e.g., tillage, erosion.

Excessive Nutrient Loads and Anoxia: Occurrence of available phosphorus and nitrogen at greater-than-natural-background levels, usually due to introduction of animal fecal material or application of fertilizers, and resultant spread of anoxic conditions (i.e., lack of dissolved oxygen) throughout sediments and the water column.

Pesticide and Heavy Metal Contamination: Occurrence of insecticides, herbicides, fungicides, heavy metals (e.g., mercury), and selenium at greater-than-natural-background levels, usually due to intentional application to crops or leaching from drained, irrigated, or mined soils.

The relative ecological risks of these stressors to all wetland functions (not just biota) were assessed in an earlier comprehensive review of the prairie pothole literature (Adamus 1992). Also, it is important to recognize that few stressors act alone; cumulative interactions among stressors are usually important, and are summarized in Section 1.3.

Variability and Reference Points: This subsection summarizes what is known about the spatial and temporal variability of each metric, e.g., the degree to which species richness changes within a wetland, among wetlands, within a year, and among years.

Also, this subsection summarizes published maximum values or ranges of values for several metrics, e.g., densities of macroinvertebrates, in order to provide crude reference points that are useful for planning a monitoring program, calibrating wetland models, conducting realistic simulations, interpreting monitoring data, evaluating the success of restoration projects, and (perhaps) defining "optimum" conditions. However, these reference values are not necessarily representative of the prairie wetland population generally because the studies from which they are drawn were not located according to a probability-based sampling scheme. The values reflect only the wetland that was studied, the season and year it was studied, and the equipment and techniques used to study it. Typically there was insufficient detail in descriptions of study areas and methods to permit meaningful comparisons among values or to distinguish natural variation from anthropogenic effects. Moreover, species richness values are notoriously difficult to compare because of additional biases introduced by variation in sample sizes, sampling frequency, and the levels of resolution in identification.

Monitoring Techniques: This subsection describes techniques, equipment, and general considerations for sampling the particular assemblage of organisms. This is intended to be a general description rather than a prescriptive manual or standard operating procedure. Information is intended to be sufficient to allow users to make choices among various types of equipment and protocols.

Collection of Ancillary Data: This subsection describes key variables that affect each assemblage, because biomonitoring data are easiest to interpret when collected simultaneously with data on other (mostly abiotic) variables.

Sampling Design and Required Level of Sampling Effort: The level of effort and costs of sampling depend directly on the number and layout of samples within or among wetlands, as well as the sampling frequency and duration (i.e., the sampling design). The sampling design that is most appropriate for a particular objective depends on the desired precision and accuracy. This subsection summarizes some of the sampling designs used previously when monitoring the taxonomic assemblage in prairie wetlands.

Another useful feature of this document is the series of appendices at the end, which list vascular plants (Appendices A and I), invertebrates (Appendix B), birds (Appendix C), and algae (Appendix H) that occur in prairie wetlands. These lists are not comprehensive; they primarily include species that were identified in the literature as being numerically or functionally dominant in at least one prairie wetland during at least one sampling period.

The appendices were prepared in similar format so they can be linked and cross-referenced using commercially available data base software, if users so desire. They have several uses. First, they can be a source of ideas for candidate species for laboratory toxicity testing; thus, bioassays that are used to help establish water quality criteria would be realistic because they would be run on species indigenous to the wetland type and region. Second, vegetation information in the appendices could be used as an aid in classifying wetlands during the development of state water quality criteria. Third, the information can be used to help develop site-specific criteria, e.g., as an information source for the "recalcitration" or "resident species" procedures described in USEPA's water quality program guidance (USEPA 1991). Fourth, the lists can serve as an aid in linking species composition with wetland condition. For example, users can compare the species they find in a particular wetland (e.g., a reference wetland or other wetland for which a state is developing a "wetland profile") with species listed in the appendices. Then, by noting in the appendices the conditions usually associated with those species, users can make inferences about the ecological integrity of the wetland and the possible identity of its stressors. As a result, the appendices help fulfill the recommendation of Smith (1991) for developing "natural community databases ... for evaluation of changes in plant community structure to determine the biotic integrity of specific habitat types," and serve as a foundation for a "habitat requirements approach" to biocriteria development as demonstrated in the Chesapeake Bay by Dennison et al. (1993).

1.3 Cumulative Effects of Stressors

When symptoms of change are noted in prairie wetlands, it is not always possible to attribute the symptoms to a single stressor (i.e., an agent of stress) because many stressors in prairie wetlands act in concert or manifest themselves similarly (Larson 1994). Thus, if species composition, richness, density, biomass, or other metrics are to be interpreted unambiguously and used as a basis for biocriteria, it is important to understand which stressors are most likely influence each other or exert a similar influence on a particular metric. The following examples are intended to further the understanding of the most frequently encountered interactions:

  • Hydrologic stressors can aggravate or mitigate the effects of several other stressors. Specifically, drought (or water level drawdown) aggravates the effects of salinity, turbidity, excessive nutrient enrichment, and contamination (chemicals within a wetland become concentrated and bottom sediments are more likely to be disturbed by wind mixing). However, floods (or water level increases) also can decrease salinity in wetlands (Neill 1993). Floods can increase turbidity and nutrient enrichment in wetlands if they deliver chemicals and sediments to the wetland via runoff and groundwater input.
  • Changes in vegetative cover are almost always the result of changes in hydrology, salinity, sedimentation/turbidity, nutrient enrichment, and/or contaminants. Droughts can decrease cover by allowing greater access to the center of usually-flooded wetlands by vehicles, livestock, and fire, or can increase cover in the long term by increasing the dominance of "drawdown" species (plants whose germination depends on periodic absence of water or shallow conditions). Floods usually decrease cover by drowning rooted wetland plants.
  • Turbidity can increase the toxicity of some herbicides (Hartman and Martin 1984, 1985) but can reduce the availability of others. Excessive sedimentation can increase the frequency of drought experienced by a wetland by decreasing wetland depth and isolating the wetland substrate from the water table. Flooding can increase, however, in downslope wetlands as water is displaced to these wetlands.

Thus, users who desire to fully know, for example, the biological effects of hydrologic alteration (or natural hydrologic cycles) will remember to look not only in the hydrologic stressor subsection of this document, but also in subsections on salinity, excessive nutrient enrichment, sedimentation and turbidity, and pesticide and heavy metal contamination. Figures 1-4 also illustrate some of these relationships, and a more complete analysis could be derived from the qualitative models of prairie wetlands detailed by Adamus (1992).

[not included on Web page; see published report]

Figure 1. Paths of effects that can result from increases in water levels above the long-term annual norm in a prairie wetland.

+ = increase in the variable, - = decrease in the variable, 0 = no change in the variable.

This diagram simplifies the processes involved. The extent and actual probability of these effects occurring may depend partly on the wetland type (e.g., semipermanent vs. temporary), initial condition (e.g., the point in a long-term wet-dry cycle the wetland is currently in), seasonal timing, presence/absence of fish, and characteristics of the specific water level process that triggers the effects (e.g., the type, frequency, duration, intensity, and timing of water level changes). See text for citations of supporting literature.

[not included on Web page; see published report ]

Figure 2. Paths of effects that can results from decreases in the density and percent cover of emergent vegetation in prairie wetlands.

+ = increase in the variable, - = decrease in the variable.

This diagram simplifies the processes involved. The extent and actual probability of these effects occurring may depend partly on the wetland type (e.g., semipermanent vs. temporary), the initial condition (e.g., the point in a long-term wet-dry cycle the wetland is currently in), seasonal timing, presence/absence of fish, and characteristics of the specific vegetation removal process that triggers the effects (e.g., the type, frequency, duration, intensity, and timing of herbicide application, grazing, fire, water level increase, etc.). See

text for citations of supporting literature.

[not included on Web page; see published report]

Figure 3. Paths of effect that can result from increases in sediment deposition in a prairie wetland.

+ = increase in the variable, - = decrease in the variable, 0 = no change in the variable.

This diagram simplifies the processes involved. The extent and actual probability of these effects occurring may depend partly on the wetland type (e.g., semipermanent vs. temporary), the initial condition (e.g., the point in a long-term wet-dry cycle the wetland is currently in), seasonal timing, and characteristics of the specific sediment deposition processes that trigger the effects (e.g., the type, frequency, duration, intensity, and timing of deposition). See text for citations of supporting literature.

[not included on Web page; see published report]

Figure 4. Paths of effect that can result from increases in nutrient loading to a prairie wetland.

+ = increase in the variable, - = decrease in the variable, 0 = no change in the variable.

This diagram simplifies the processes involved. The extent and actual probability of these effects occurring may depend partly on the wetland type (e.g., semipermanent vs. temporary), overall water chemistry, initial condition (e.g., the point in a long-term wet-dry cycle the wetland is currently in), seasonal timing, and characteristics of the specific nutrient loading process that triggers the effects, such as the frequency, duration, intensity, and timing of increased inputs (e.g., more fertilizer or greater seasonal runoff) or increased mobilization of nutrients previously immobilized in sediments. See text for citations of supporting literature.

1.4 Glossary, Abbreviations, and Place Names

For the sake of maintaining brevity, this document uses broadly certain terms that conventionally have a more narrow definition:

Basin. A topographic depression in the prairie landscape, which normally lacks a permanent natural connection to larger rivers and lakes, and which contains a wetland.

Biological Criteria (biocriteria). Criteria that use the condition of an organism or assemblage of organisms to describe the ecological integrity of unimpacted, least-impacted, or representative ("reference") areas. Biocriteria may be expressed in numeric or narrative terms.

Biomarker. A measurement, generally of biological tissue or physiological byproducts, that indicates previous or ongoing organism response and/or exposure to general or specific environmental stresses (Huggett et al. 1992).

Cover Ratio. The percent open water in a wetland, where "open water" is any part of the wetland that lacks a canopy of emergent vegetation and contains water during at least part of the growing season.

Density. The number of individuals per unit area or volume.

Ecological (or Biological) Integrity. The condition or "health" of an area, as defined by comparison of community structure and functions to those of unimpacted, least-impacted, or representative ("reference") areas.

Macrophytes. Plants generally visible to the unaided eye, including vascular plants and some of the larger algae.

Permanent Basins. Prairie pothole depressions that retain surface water throughout the year, as classified by Stewart and Kantrud (1971), and that contain wetland vegetation and soils. Used synonymously with "permanent (or permanently flooded) wetland."

Seasonal Basins. Prairie pothole depressions that retain surface water for much of the growing season (e.g., sometimes into July), as classified by Stewart and Kantrud (1971), and that contain wetland vegetation and soils. Used synonymously with "seasonal (or seasonally flooded) wetland."

Semipermanent Basins. Prairie pothole depressions that retain surface water throughout most of the growing season, as classified by Stewart and Kantrud (1971), and that contain wetland vegetation and soils. Used synonymously with "semipermanent (or semipermanently flooded) wetland."

Temporary Basins. Prairie pothole depressions that retain surface water only during the first weeks of the growing season, as classified by Stewart and Kantrud (1971), and that contain wetland vegetation and soils. Used synonymously with "temporary (or temporarily flooded) wetland."

Species Composition. The identity and relative abundance of species in a biological community. Used synonymously with "community composition."

Species Richness. The number of species (or any other taxonomic denomination) per sample, per wetland, per number of individuals. Used synonymously with "taxa richness" and "family richness" because many data sets combine a variety of levels of taxonomic resolution.

Throughout this document several place names and abbreviations are used without elaboration to maintain brevity. These are defined as follows:

Cottonwood Lakes. The Cottonwood Lakes Long-term Environmental Monitoring site, a large and varied complex of prairie pothole wetlands located in Stutsman County, North Dakota, in which data on waterfowl, climate, and vegetation dynamics have been collected for decades.

Delta Marsh. A large wetland complex in south-central Manitoba, Canada, a portion of which has been used by the Marsh Ecology Research Program (MERP) of Ducks Unlimited to conduct over 80 multi-year experiments using ten, 5-ha marsh cells, each with independent water level control.

EMAP. USEPA's Environmental Monitoring and Assessment Program, a long-term program intended to regularly monitor the ecological condition of ecosystems (including wetlands) throughout the nation using a probability-based sample design, and generate estimates of status and trends in ecosystem condition by region and ecosystem (e.g., wetland) type.

NPSC. The Northern Prairie Science Center, the federal research facility in Jamestown, North Dakota, that has investigated wetland ecology of the prairies for decades, run by the National Biological Service (formerly by the US Fish and Wildlife Service, USFWS).

1.5 Statistical Analyses: Objectives and Methods

One objective of this project was to estimate the probable number of samples needed to satisfy various purposes. To achieve these estimates, eight existing data sets were obtained from investigators in the region and were analyzed statistically. There was no particular reason for selecting these data sets, other than their availability. Two of the data sets pertained to wetland plants, four to macroinvertebrates, and two to birds. Detailed descriptions of the data sets are found in Appendix L. Data were analyzed to address two questions of practical relevance to sampling prairie wetlands:

1. How many samples need to be collected to find 50, 75, 90, 95, and 99% of the taxa found in the full suite of samples collected by a particular study? (Asymptotic Richness)

2. Given a particular number of samples containing information on biomass, number of individuals, or number of taxa, what size difference between two means (e.g., from different wetlands or different dates) can be detected at usual levels of statistical significance? (Power of Detection, or "precision")

Information of this type is essential to estimating costs and levels of effort required for monitoring programs. The results are presented in Sections 3.8, 4.8, and 6.8.

Information on the first objective (Asymptotic Richness) is needed to help determine if a population has been oversampled or undersampled with regard to detecting most taxa that are present. To address this objective, we used a bootstrap subsampling technique to quantify species accumulation rates (Szaro and King 1990). This reflects a basic principle of diminishing returns: as one samples a population, the number of taxa in samples at first rises rapidly, but then levels off as additional samples add only a few new taxa.

A computer program was written in SAS to estimate species accumulation rates. The program first tallied the number of taxa in the entire data set. Samples that had been collected then were selected randomly without replacement until the number of taxa they cumulatively contained reached one of the specified points (50, 90, 95, 99% of species total from all samples). However, the number of samples needed to reach a particular point depends on the order in which the samples are combined. Thus, the random selection process was repeated 100 times, and the median, mean, and standard deviation of samples sizes estimated from the 100 runs were used to represent requisite sample size. Two assumptions were made when implementing the statistical analysis: (1) 100 runs are sufficient to stabilize the estimates of requisite sample size, and (2) the number of samples originally collected was sufficient to capture nearly all taxa in the target population.

Addressing the second objective (Power of Detection) at first seemed straightforward, inasmuch as many papers in the published literature have defined power of detection through use of elementary equations and simplifying assumptions (Downing 1979, Schwenneker and Hellenthal 1984, Canton and Chadwick 1988, Riddle 1989, Downing 1989, Niemi et al. 1993). Although coefficients of variation calculated for all data sets (Appendix N) might have been used in such equations, the use of simplified approaches limits the generality of the results. Thus, a more involved approach (Components of Variance) was used. Variance component estimates for random factors were calculated using the SAS MIXED procedure for fitting mixed linear models (experimental designs having both fixed and random effects). The estimated variance components were used as estimates of the population variances in the equations below. Estimated variances were obtained by incorporating the variance component into the expected mean square for the random effect of interest. We made statistical comparisons only within data sets, not between them, e.g., to determine which taxon is least variable, or which metric varies the most seasonally. Further, we did not transform any values or test assumptions that routinely underlie the analysis of variance (ANOVA). The analyzed data represent samples that are subject to uncontrolled influences such as weather. In the future, it might be informative to use data collected over several years as independent experiments to address the issue of the effect of temporal variation on the estimates. We approximated the degrees of freedom (df1) for obtaining F values by using Satterthwaite's effective df (Steel and Torrie 1980):


We used two methods to calculate precision (i.e., detectable difference) over a range of sample sizes. The first equation provides an "optimistic" estimate for the difference between two means that is detectable at a given sample size.

[not included on Web page; see published report]


The second equation is more conservative and takes into account the assurance that the study has the desired precision. Thus, Fß provides greater assurance that the difference between means in future experiments will be no greater than the estimated ability to detect the specified difference in the means.

[not included on Web page; see published report]


Where: P1n = optimistic precision estimate for the nth sample

P2n = conservative precision estimate for the nth sample

t = value from t-table

df1 = effective degrees of freedom for the error term

s2 = estimated Variance

r = replicate

F = value from F-table

We present the results of applying the equations over a range of n values. Specifically, P1 and P2 were calculated by varying the replicates and subsequent degrees of freedom over a range of values. Curves of the calculated values of P1 and P2 were plotted on the same graph for each random effect variable from each of six datasets. Figure 5 is an example. The upper curve on each plot represents the conservative precision estimates and the lower curve the more optimistic estimates. These curves make it possible to assess relationships between the number of replicates and the precision estimate. Because of the large number of curves generated, results have been summarized tabularly (Appendix M).

[not included on Web page; see published report]

Figure 5. The Difference Between Two Means (of the number of Conchostraca in sweep nets) That Can Be Detected By Various Sample Sizes

The upper curve is based on an equation that estimates precision conservatively, whereas the lower one estimates precision optimistically. See explanation on page ?. Results from curves for all major taxa and metrics are compiled tabularly in Appendix M.

Understand that the values presented in this document, while generally realistic, are not intended as exact estimates of requisite sample sizes. The requisite number of samples or resultant levels of precision could differ if sampling is done according to a design or under conditions (e.g., weather, season, wetland type, equipment) that differ from those upon which these estimates were based.

Jump to main content.