Jump to main content or area navigation.

Contact Us

Water: Monitoring & Assessment

4. Invertebrates as Indicators of Prairie Wetland Integrity

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

4.1 Ecological Significance and Suitability as an Indicator
4.2 Potential Indicator Metrics
4.3 Previous and Ongoing Monitoring in the Region
4.4 Response to Stressors

4.4.1 Invertebrates as Indicators of Hydrologic Factors
4.4.2 Invertebrates as Indicators of Changes in Vegetative Cover
4.4.3 Invertebrates as Indicators of Wetland Salinity
4.4.4 Invertebrates as Indicators of Sedimentation and Turbidity
4.4.5 Invertebrates as Indicators of Excessive Nutrient Loads and Anoxia
4.4.6 Invertebrates as Indicators of Pesticide and Heavy Metal

4.5 Monitoring Techniques

4.5.1 General Considerations
4.5.2 Sampling Equipment
4.5.3 Time-Integrating Methods
4.5.4 Bioassay Methods
4.5.5 Bioaccumulation

4.6 Variability and Reference Points

4.6.1 Spatial Variability
4.6.2 Temporal Variability
4.6.3 Spatial vs. Temporal Variability

4.7 Collection of Ancillary Data
4.8 Sampling Design and Required Level of Sampling Effort

4.8.1 General Considerations
4.8.2 Asymptotic Richness: Results of Analysis
4.8.3 Power of Detection: Results of Analysis

4.9 Summary


4.1 Ecological Significance and Suitability as an Indicator

Invertebrates include aquatic insects, freshwater crustaceans (e.g., amphipods, crayfish), aquatic annelids (worms), zooplankton, and immature stages of certain terrestrial insects (e.g., Lepidoptera) that occur mainly in wetlands. The term "macroinvertebrate" or "macrofauna" refers to the larger organisms clearly visible to the unaided eye, as opposed to microinvertebrates, which include most smaller zooplankton, such as rotifers.

Although invertebrates occur in wetlands everywhere, prairie wetlands support notably great numbers. This is because prairie wetlands have especially rich soils, slow water turnover times, and seasonally fluctuating water tables, all of which support the high levels of algal production and spatially complex vegetative stands that are important to invertebrates.

Invertebrates are the vital link that makes algal production and emergent plant material available as an energy source for waterbirds and other animals. Invertebrates do this by consuming algal production and decaying plant material, and then being consumed by higher order animals (Driver et al. 1974). Invertebrates represent grazing, filtering, detritivore, and predator trophic pathways of energy flow, and thus should reflect the status of these fundamental processes in a wetland. Planktonic invertebrates (e.g., cladocerans) are potentially able to consume more than an entire day's production of algae (Porter 1977). In doing so, they considerably improve and maintain light penetration of the water column during the growing season. This in turn gives submersed aquatic plants a chance to flourish (Hanson and Butler 1990), and these macrophytes serve as a substrate that supports an even greater density of invertebrates, as well as a food source for many organisms.

In some cases, waterbirds appear to select wetlands having the greatest densities of invertebrates (Talent et al. 1982). Even where they do not, waterbirds spend more time foraging in wetlands that have greater abundance of macroinvertebrates (Kaminski and Prince 1981a,b, 1984). Whereas larvae are eaten mainly by ducks, emerging insects are consumed by many songbird species. The nutritional requirements of growing ducklings and breeding hens can be fulfilled only by an invertebrate diet (Swanson et al. 1974, 1977). However, the degree to which food supply -- as opposed to vegetative cover and predation -- truly limits the breeding and reproductive success of waterfowl populations at a regional scale is unknown. The variety (species richness) of invertebrates might be at least as important as the quantity, because waterfowl require or use different invertebrate species, from different parts of the wetland at different seasons (Swanson and Meyer 1973, Kaminski and Prince 1981a,b, 1984). Invertebrate richness supports elevated waterbird richness because different waterbirds use different invertebrate assemblages. If changes in hydrologic regime or turbidity cause changes in a key habitat component of invertebrates (e.g., submersed plants), the invertebrate species associated with that habitat could be reduced or eliminated from the wetland, even if the wetland remains well-vegetated with other types of plants. If the affected invertebrates are critical to waterbirds, waterbird productivity could suffer.

Soil macroinvertebrates (especially earthworms and certain midge larvae) also dominate the diet of several shorebird species that stop to feed in prairie wetlands during migration. Yet, soil invertebrates have seldom been monitored in temporary and seasonal wetlands, especially during portions of the growing season when surface water is lacking. Nematodes are one abundant, diverse, and sensitive invertebrate assemblage that has been found by many European studies to be a useful indicator of soil condition (Bongers 1990, Freckman and Ettema 1993), and might find similar application here. Quantitative sampling methods are relatively well-developed (Schouten and Arp 1991), and additional research could document the relative importance of nematodes to ecosystem processes.

Invertebrates are also important because they influence the amount of contaminants that are available to other components of the food web, and the rate of contaminant cycling across several ecotones (e.g., sediment-water column, wetland-upland). In the sediment, burrowing invertebrates can make more generally available the nutrients (or contaminants) contained in decaying plant roots. Nutrients released to the water column by invertebrates help sustain algal productivity.

Several characteristics of invertebrates make them favorable for use in monitoring ecosystem integrity (Adamus and Brandt 1990):

  • Characteristic responses to all major wetland stressors (hydroperiod, sediment, nutrients, contaminants) have been documented; many "indicator taxa" have been identified.
  • Larval lifespans are varied, ranging from short (e.g., cladocerans) to long (e.g., dragonflies), thus allowing invertebrates to be used as indicators of both chronic and acute disturbances.
  • Noninsect invertebrates reflect the quality of the wetland itself because they usually complete their entire life cycle within a single wetland.
  • Invertebrates can be confined for whole-effluent bioassays or in situ assessments.
  • Decay-resistant remains (e.g., shells) provide a means for establishing historical reference conditions in a wetland.
  • Sampling equipment is generally inexpensive and USEPA sampling protocols are available.

    Characteristics that are usually considered to be disadvantageous to the objective of monitoring wetland integrity include:

  • Excessive amount of time required to adequately isolate organisms from debris in many types of samples.
  • Laborious identification beyond the family level.
  • The occurrence of a particular species in an individual prairie pothole wetland is difficult to interpret because it is unknown whether occurrence is related to conditions within the wetland, proximity to sources of effective colonizers, or ephemeral conditions (e.g., favorable winds) that facilitated colonization (this is true mostly of insects, and is less true of invertebrates that do not emerge from the wetland as adults).
  • Difficulty in measuring precisely the true densities in dense stands of vegetation.
  • Many techniques and sampling tools are required to sample all important invertebrate assemblages present in a wetland.

4.2 Potential Indicator Metrics

The following measurements and metrics deserve consideration, as applied to invertebrate communities, for use in characterizing conditions in reference wetlands, identifying the relative degree of past disturbance of a prairie wetland, or assessing the current inhibition of key processes:

  • Richness of species and functional s (per unit of area or volume, or per a specified number of randomly-chosen individuals).
  • Number and biomass of individuals per unit of area or volume.
  • Relative dominance and richness of species reputedly tolerant to a named stressor.
  • Interannual variability in richness, density, and/or biomass.
  • Homogeneity of size or biomass classes within a species population.
  • Levels of tissue contaminants (biaccumulation).
  • Density of dormant but viable life stages.

The specific ways some of these metrics have been or could be interpreted as an indication of stressed conditions are described in Section 4.4. However, apparently no studies in prairie wetlands have systematically examined correlations among these metrics, or their merits relative to one another. A few studies of this type that have been completed in streams (Barbour et al. 1992, Kerans et al. 1992, Resh and Jackson 1993, Niemi et al. 1993) might be used as a model.

4.3 Previous and Ongoing Monitoring in the Region

Aquatic invertebrates have been the focus of at least 35 published studies, covering over 200 prairie wetlands (Appendix J). All of these studies measured numbers of individuals (or density) and at least four measured biomass. Apparently only Duffy and Birkelo (1993) have attempted to measure annual production. In most studies the invertebrates were identified only to family. Seven studies have sampled a wetland for more than two years, and only one-third of the studies involved sampling of more than five wetlands.

Montana's water quality monitoring agency is currently using benthic (bottom-dwelling) invertebrates on a trial basis as an indicator of the condition of five prairie wetlands. USEPA's EMAP has not yet undertaken monitoring of prairie wetland invertebrates at a regional scale, but has investigated various sampling methods in dozens of North Dakota wetlands. Variables that are being estimated include species richness and relative abundance. At a localized level, invertebrates have been used as possible indicators of the success of wetland restoration efforts in Iowa (Hemesath 1993) and Minnesota (Sewell 1989, Sewell and Higgins 1991). Research on ecological relationships of invertebrates, especially as affecting waterfowl, continues to be conducted by scientists at the NPSC, by Minnesota Department of Natural Resources, and by universities.

To draw conclusions about wetland integrity from samples of invertebrates, it is essential to have species-specific information on their tolerances and life histories. Appendix B indexes invertebrate taxa according to water regime, and more detailed data bases of this type have been assembled by Euliss (personal communication, NPSC, Jamestown, ND). Also, a data base that classifies prairie wetland invertebrates by feeding type and waterfowl food importance is maintained at North Dakota State University (Overland et al. 1993). The recent book by Rosenberg and Resh (1993) categorizes over 200 invertebrate species, many of them prairie wetland species, according to their tolerances to acidic conditions and organic pollution.

4.4 Response to Stressors

4.4.1 Invertebrates as Indicators of Hydrologic Stressors

Species Composition

The usefulness of species composition for inferring hydrologic conditions of prairie wetlands has been demonstrated with midges (Driver 1977, Euliss et al. 1993), water beetles (Hanson and Swanson 1989), and macroinvertebrates generally (Neckles et al. 1990, Bataille and Baldassarre 1993). Species composition can indicate how long and in what seasons a wetland has contained surface water. This requires that each species first be classified as to its hydrological requirements, a relatively simple procedure using life history categories such as defined by Hartland-Rowe (1966); McLachlan (1970, 1975, 1985); Wiggins et al. (1980); Jeffries (1989); and Eyre et al. (1991). Appendix B contains such information for dominant prairie wetland invertebrates.

In general, prairie wetlands can be cautiously deduced to be of greater hydrologic permanence when they contain a higher density and richness of longer-lived and/or relatively immobile species (e.g., snails, mollusks, amphipods, worms, leeches, crayfish), as compared with the density and richness of short-lived species (e.g., anostracans, conchostracans), species that survive the winter as drought-resistant eggs (e.g., Daphnia), and/or species that are relatively mobile (e.g., chironomid midges). This is probably due to the likelihood that drought and drawdown renders the less mobile species more vulnerable to predation, as well as causing their direct loss due to desiccation and salinity toxicity. From season-long, weekly activity-trap sampling of three pothole wetlands near the Delta Marsh, Bataille and Baldassarre (1993) found that a permanent wetland was dominated by cladocerans, a semipermanent wetland by ostracods, and a seasonal wetland by copepods. Considering just the emerging aquatic insect component, the permanent wetland was dominated by midges; the semipermanent wetland by water beetles (early season) and midges and other fly species (mid- and late-season); and the seasonal wetland by midges (mid-season) and other fly species (late season).

Some evidence (Neckles et al. 1990), however, suggests that wetland water regime in particular situations has little affect on the dominance of several major taxa that characteristically overwinter as adults or larvae (species of Dytiscidae, Corixidae, Ceratopogonidae, Ephydridae, and some Chironomidae). Caution is required in interpreting data because anecdotal evidence suggests that some species with supposedly minimal dispersal abilities are frequently carried passively into new areas by mobile waterbirds (Swanson 1984).

A shift from herbivorous to detrivorous species of macroinvertebrates, and in the ratio of open-water forms (e.g., zooplankton, water striders) to forms that characteristically dwell in vegetation (e.g., some mayflies), can suggest that a prairie wetland has recently undergone inundation (Murkin and Kadlec 1986, Murkin et al. 1991). In particular, densities of non-predatory midges (Chironomidae) increase greatly during the first year after flooding, and within this family, species characterized by the greatest tolerance for low oxygen levels increase the most (Murkin and Kadlec 1986b). Densities of swimming (nektonic) and bottom-dwelling (benthic) predatory invertebrates do not increase with flooding as much as do numbers of nektonic and benthic herbivores and detritivores. Predatory species can even decrease after flooding (Murkin et al. 1991), and they often increase as drought or drawdown progresses (Figure 6).

Long-term changes in wetland hydrology might be inferred by collecting decay-resistant remains of invertebrates from sediment cores or settling traps (see Section 4.6.3), and determining if the species present are ones that occur mostly in semipermanent, seasonal, or temporary wetlands.

Species Richness

Data from North Dakota indicate that even the wetlands that are flooded only temporarily have many more species than non-wetland areas (Euliss et al. 1993). Within wetlands, flooding can increase invertebrate richness somewhat, but perhaps only during the initial year of flooding (Figure 7). For example, flooding of Manitoba marshes containing cat-tail, hardstem bulrush, and common reed to a level 1 m above normal increased the variety of both nektonic and benthic invertebrates in vegetation, but not in open water (Murkin et al. 1991). The increase in benthic taxa persisted for only a short period after flooding (Murkin and Kadlec 1986b). A similar pattern was noted for midge diversity in other Manitoba wetlands by Driver (1977). When inundation persists for years with little fluctuation in water level, sediments often become anoxic and light deficits can reduce the amount and variety of aquatic plants available as invertebrate habitats, thus reducing invertebrate richness (Neckles et al. 1990). In drained wetlands whose water regime is restored, richness increases during the first few years following restoration (Nilsson and Danell 1981, Hemesath 1991).

Figure 6. Response of nektonic and benthic invertebrate herbivores/detritivores and predators to water level manipulations in a prairie wetland.

Habitat Condition Invertebrate Component and Its Response
    Nekton

(swimming invertebrates)

Benthos

(bottom dwelling invertebrates)

Open water Unflooded Herbivores/detritivores surpass predators Herbivore/detritivore numbers are about equal to those of predators
Open water Flooded Herbivores/detritivores greater than in unflooded habitat Predators and herbivores/detritivores are about equal, and both are about equal to their numbers in unflooded habitat
Emergent vegetation Unflooded Predators surpass or equal herbivores/detritivores Herbivore/detritivore numbers are about equal to those of predators
Emergent vegetation Flooded Predator and

herbivores/detritivore numbers are about equal to their numbers in unflooded habitat

Numbers of herbivores/detritivores are greater than in unflooded habitat

The unflooded condition describes the normal water level of the experimental wetland, and the flooded condition describes the water level in an otherwise similar wetland that had been raised about 1 m above normal at the beginning of the growing season and maintained at that level throughout the season. The invertebrate response is the response as measured at the beginning of the growing season one year after the water level had been raised in the flooded wetland. These conclusions were drawn from studies of the Delta Marsh (Murkin and Kadlec 1986b, Neckles et al. 1990, Murkin et al. 1991, Murkin et al. 1992).

Figure 7. Response of taxonomic richness of nektonic and benthic invertebrates to water level manipulations in a prairie wetland.

Condition Invertebrate Component and Its Response
  Nekton: Taxonomic Richness Benthos: Taxonomic Richness
Unflooded About equal in open water habitats to levels in emergent vegetation About equal in open water habitats to levels in emergent vegetation
Flooded About equal in flooded open water habitat to levels in unflooded open water, but in flooded emergent vegetation is greater than in unflooded emergent vegetation About equal in flooded open water to levels in unflooded open water, but in flooded emergent vegetation is greater than in unflooded emergent vegetation (except for cat-tail stands of emergents, where richness is unchanged from unflooded condition)

The unflooded condition describes the normal water level of the experimental wetland, and the flooded condition describes the water level in an otherwise similar wetland that had been raised about 1 m above normal at the beginning of the growing season and maintained at that level throughout the season. The invertebrate response is the response as measured at the beginning of the growing season one year after the water level had been raised in the flooded wetland. These conclusions were drawn from studies of the Delta Marsh (Murkin and Kadlec 1986b, Neckles et al. 1990, Murkin et al. 1991, Murkin et al. 1992).

Species richness of midges tends to be greater in wetlands having longer durations of standing water during the growing season (Driver 1977, Nelson and Butler 1987). This is partly because wetlands with longer hydroperiods generally are deeper and more likely to contain submersed and floating-leaved plants that diversify the range of habitats available to this assemblage of invertebrates. Also, wetlands with longer durations of flooding are less likely to experience deep freezing of sediments and types of human activities (e.g., soil compaction, cultivation) that reduce habitat quality for invertebrates (Swanson et al. 1974). However, in a of five wetlands in the Cottonwood Lakes area, 47 of water beetle taxa were found in seasonal wetlands whereas 38 were found in semipermanent wetlands (Hanson and Swanson 1989). The seasonal wetlands had 18 exclusive species whereas the semipermanent wetlands had only 11.

Density and Biomass

Flooding generally increases invertebrate densities in wetlands, but perhaps only for about a year after initiation of flooding (Figure 8). For example, flooding of Manitoba marshes containing cat-tail, hardstem bulrush, and common reed to a level 1 m above normal caused a major year-long increase in numbers of nektonic invertebrates in both the vegetation and in open water areas. Biomass of nektonic invertebrates increased only in the vegetated areas. Densities of benthic invertebrates increased in flooded vegetation but not in open areas. On a year-round basis, invertebrate biomass and production is probably greatest in semipermanent wetlands (Duffy and Birkelo 1993; Nelson 1989, 1993; Bataille and Baldassarre 1993), but sometimes can reach greater seasonal peaks in temporary and permanent wetlands.

It has been suggested that the density and viability of dormant stages of some invertebrates could be used to determine in advance whether (and how rapidly) the restoration of a drained wetland will restore its functional characteristics, e.g., part of its invertebrate community (N. Euliss, personal communication, NPSC, Jamestown, ND). The eggs or other dormant stages of several invertebrate taxa -- notably the cladocerans, anostracans, and conchostracans -- may hatch in response to certain conditions during an 8-week incubation in laboratory aquaria. If incubated sediment samples from a farmed wetland fail to produce such hatchings, it could be assumed that degradation has been so severe as to make full functional restoration impractical, just as lack of seed viability of seed banks is sometimes interpreted.

4.4.2 Invertebrates as Indicators of Changes in Changes in vegetative cover

Species Composition

An increase in the ratio of algae-consuming species (e.g., certain mayflies) to detrivorous species (e.g., certain worms, isopods, amphipods) and in the ratio of open-water forms (e.g., zooplankton, water striders, midges) to vegetation-dwelling forms (e.g., amphipods, snails, some mayflies) is expected if a wetland has been exposed to herbicides, grazing, fire, flooding, or other vegetation removal processes. This is because such disturbances, by

Figure 8. Response of density and biomass of nektonic and benthic invertebrates in a prairie wetland to water level manipulations.

  Invertebrate Component and Its Response
Season Condition Nekton: Density and Biomass Benthos: Density and Biomass
Spring Unflooded Greater in emergent than open water habitats About equal in emergent and open water habitats
Spring Flooded Greater in emergent than open water habitats. Density in both habitats is greater than it is in these habitats in unflooded wetland, but only density (not biomass) is greater in flooded open water than in unflooded open water. Greater in emergent than in open water habitats, and in emergents, is also greater than in unflooded condition. Density and biomass in flooded open water are no greater than in unflooded open water.
Summer Unflooded Greater in open water than emergent habitat. About equal to levels in open water and emergent habitat.
Summer Flooded
(year 1)
Mostly greater than in unflooded condition in both habitats, but biomass in flooded open water differs little from biomass in unflooded open water. Greater in emergent than in open water habitats.
Summer Flooded
(year 2)
In emergent habitat, a decline as compared to levels in this habitat the first year after flooding. Decline results in levels similar to those in unflooded emergent habitat (except in stands of Sclochloa). Increased densities continue from first post-flooding year in some emergent habitat (common reed and hardstem bulrush), but density changes only slightly in open water habitat.

The unflooded condition describes the normal water level of the experimental wetland, and the flooded condition describes the water level in an otherwise similar wetland that had been raised about 1 m above normal at the beginning of the growing season and maintained at that level throughout the season. These conclusions were drawn from studies of the Delta Marsh (Murkin and Kadlec 1986b, Neckles et al. 1990, Murkin et al. 1991, Murkin et al. 1992). thinning the canopy of emergent plants, create open water areas where algae and submersed macrophytes proliferate (Overland et al. 1993).

Data from the Delta Marsh (Murkin et al. 1991) suggest that the ratio of predatory to herbivorous-detrivorous invertebrates might be used to indicate changes in cover conditions. Predatory s tend to dwell in emergent vegetation, but invade open water areas to some degree in midsummer as submersed and floating-leaved plants develop. At that time, their densities are similar to their densities in emergent stands in the springtime. Field studies that have examined invertebrate responses to changes in vegetation cover are indexed in Appendix G.

Species Richness

High invertebrate richness in prairie wetlands is associated with presence of vegetation. Beyond some point, however, vegetation stands can become so dense that invertebrate richness declines (e.g., Broschart and Linder 1986; Kaminski and Prince 1981a,b), perhaps due to development of anoxic conditions.

The variety of invertebrate families, especially of larger invertebrates, can be greater in wetlands that have been mowed than in otherwise similar wetlands that have not, especially if the hay is not removed (Kaminski and Prince 1981a,b; Beck et al. 1987). Surprisingly, a wetland whose emergent cover had been rototilled had a greater variety of invertebrates than an otherwise similar undisturbed wetland (Kaminski and Prince 1981b). This may have been a short-term response attributable to more rapid warming of the disturbed soils in the spring (H. Murkin, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba).

Data from the Delta Marsh also suggest that during normal springtime conditions, open water and emergent vegetation differ little with regard to their variety of nektonic and benthic invertebrates. By midsummer, open water sites contain submersed plants, and consequently can support a greater variety of nektonic invertebrates. If nearby emergent vegetation has been flooded within the last 1-2 years, it can support as many or more nektonic and benthic species (Murkin et al. 1991). An exception might be stands of cat-tail. In these, flooding seems to have little effect on the variety of benthic invertebrates (Murkin and Kadlec 1986b). Other data from the Delta Marsh (Kaminski and Prince 1981a,b) suggest that invertebrate richness is not always greater in wetland units that have relatively equal proportions of open water and emergent vegetation, compared with those that do not.

Density and Biomass

Invertebrate biomass in prairie wetlands is strongly linked to plant biomass (McCrady et al. 1986) and thus to aerial cover of vegetation. This is because vegetation provides submersed habitat space that macroinvertebrates colonize at a far greater density than open water areas (Engel 1990). Throughout the early growing season, an undisturbed seasonal wetland in North Dakota had much greater densities of invertebrates than a flooded summer-fallow wetland (Swanson et al. 1974). Invertebrates also were more abundant in the undisturbed wetland than in a flooded grain-stubble wetland, except in late June.

Beyond some point, however, vegetation stands in prairie wetlands can become so dense that invertebrate density and biomass decline (e.g., Broschart and Linder 1986; Kaminski and Prince 1981a,b; Murkin et al. 1982; Murkin and Kadlec 1986b). At least in some prairie wetlands, the paucity of some invertebrates in dense stands is due more to the lack of oxygen in sediments of these habitats, than to cooler temperatures or lack of algal foods in the stands, both of which are due to shading (Wrubleski 1987, Murkin et al. 1992). A common amphipod in prairie wetlands, Hyalella azteca, requires an oxygen concentration of at least 1.2 mg/L over the span of a month to achieve full reproductive and growth potential (Nebeker et al. 1992).

Wetlands containing open-water areas interspersed with relatively equal areas of dense vegetation often have the greatest invertebrate biomass (Kaminski and Prince 1981a,b). However, a study in one prairie wetland (Murkin et al. 1992) indicated that peaks in the horizontal distribution of invertebrates do not generally occur at the ecotone between the open water patches and stands of macrophytes, as was commonly assumed (e.g., Voigts 1976). Rather, peaks in invertebrate abundance seem to occur wherever the greatest variety of substrates occurs, and this variety is not always greatest along the ecotone between open water and vegetation.

Data from the Delta Marsh (Kaminski and Prince 1981a,b) indicate that cover ratio has less influence on invertebrate abundance and biomass than does the type of cover manipulation that has occurred (mowing, rototilling, etc.). In particular, practices that allow large amounts of plant litter to decay over the winter seem to support exceptional abundance and biomass of invertebrates the following spring (Kaminski and Prince 1981a,b; Ball and Nudds 1989). Thus, the degree to which vegetation removal has a neutral or beneficial effect on invertebrates seems to depend partly on the type of removal process (e.g., mechanical thinning, ditching, burning, herbicides, herbivore introduction), the type of vegetation, and the particular spatial patterns that are created (Nelson and Kadlec 1984).

Invertebrate association with open water or emergent vegetation varies by season. Data from the Delta Marsh suggest that early in the growing season, numbers of nektonic and benthic invertebrates are greater within emergent vegetation stands than in open water areas (Murkin et al. 1992). By midsummer, open water sites containing submersed plants have more nektonic invertebrates, and about equal numbers of benthic invertebrates, as compared to sites with emergent vegetation. However, if the emergent vegetation has been recently flooded, densities of benthic invertebrates (but not necessarily nektonic invertebrates) will be greater in emergent vegetation than in open water areas (Murkin and Kadlec 1986b, Murkin et al. 1991). Also, data from four North Dakota semipermanent wetlands show that stands of cat-tails supported greater densities of invertebrates than open water patches, whether natural or created by herbicide application (Solberg and Higgins 1993a).

Macroinvertebrate densities in a South Dakota prairie wetland (McCrady et al. 1986) were greatest in Ceratophyllum demersum, and zooplankton densities were greatest in Lemna minor, compared with other species of non-emergent plants. Higher summertime densities of invertebrates in open water areas (containing submersed plants) than in stands of emergent vegetation have also been documented in an Iowa marsh (Voigts 1975).

The occurrence of consistently high densities of invertebrates throughout a wetland is likely a sign of hydrologic and vegetational conditions that are spatially diverse. This is because different invertebrates (and different life stages of the same invertebrate species) require different cover densities, hydroperiods, and types of vegetation at different seasons (Murkin et al. 1992).

Interannual Variability of Density and Biomass

In the Delta Marsh, interannual changes in abundance and biomass of invertebrates were least in areas that had been disturbed by mowing (Kaminski and Prince 1981b).

4.4.3 Invertebrates as Indicators of Wetland Salinity

Species Composition

Certain invertebrates are characteristic of hypersaline prairie wetlands. These include the Anostracan brine shrimp (Artemia), brine flies (Ephydra), ostracods, and a few species of midges and aquatic worms. Other taxa known to be relatively tolerant (up to < 30,000 mg/L salt) include certain species of midges, mosquitoes, aquatic worms, dragonflies, water beetles (especially the Dytiscidae and Hydrophilidae), and water bugs (Kreis and Johnson 1968, Swanson et al. 1974). Although some of the salt-tolerant species in these s also occur in less saline wetlands, their abundance is typically greater in saline wetlands. Thresholds of 80 and 5000 µS/cm specific conductance might be of ecophysiological significance for some wetland invertebrates, because these levels seem to represent disjunct points in the spatial distribution of water beetle (Coleoptera) species distribution in Canada (Lancaster and Scudder 1987). Above a salinity of 50 g/L, the usual numerical dominance of chironomid midges in prairie lakes shifts to dominance by dolichopodids and ephydrids ("brine flies") (Timms et al. 1986). Midge species composition seems rather unaffected within the lower range of salinities present in prairie wetlands (Driver 1977). Among semipermanent wetlands, most gastropods occur only where specific conductance is less than about 5000 µS/cm (Swanson et al. 1988), org/L (Timms et al. 1986). Salinity ranges of dozens of prairie benthic invertebrates, as determined from their distribution among many lakes, are reported by Larson (1975), Timms et al. (1986), and Timms and Hammer (1988); these data have largely been incorporated in Appendix B.

Species Richness

The variety of invertebrate species within major taxonomic assemblages generally declines in prairie wetlands with increasing salinity and/or with increasing specific conductance, in part because the biomass of most submersed plants decreases (Lancaster and Scudder 1987, Timms et al. 1986, Hartland-Rowe 1966), and in part because fewer taxa are physiologically adapted to higher salinity levels. However, the range of tolerances is likely to be wide, as evidenced by studies of aquatic beetles, for which the correlation between salinity and species richness is not strong (Timms and Hammer 1988).

Density and Biomass

The few species that tolerate highly saline conditions often occur at very great densities in prairie wetlands. This is partly attributable to reduced pressures from competition and predation. Water beetle populations in highly saline lakes also appear to have smaller body sizes and fewer size classes (Lancaster and Scudder 1987).

4.4.4 Invertebrates as Indicators of Sedimentation and Turbidity

Species Composition

A shift from herbivorous and filter-feeding species (many midges, zooplankters, and mayflies) to sediment-burrowing species (many aquatic worms) can indicate that major turbidity and sedimentation incidents have occurred or are continuing. This is because a reduction in light penetration kills submersed plants and attached algae, and these plants contain a characteristic assemblage of herbivorous species. Burrowing species meanwhile can continue to exploit soft sediments. Excessive sedimentation might be expected to have different effects on species that overwinter in wetland sediments as eggs, as opposed to overwintering as diapausing adults, but this apparently has not been investigated. Long-term changes in wetland turbidity and sedimentation might be inferred by collecting decay-resistant remains of invertebrates from sediment cores or settling traps, and determining if the historical species are ones that characteristically prefer turbid, silty, anoxic environments (see Section 4.6.3).

Species Richness

A diminished variety of invertebrates is another sign that turbidity and sedimentation conditions have been severe, for the reasons just given. Species richness is particularly likely to decline in semipermanent and permanent wetlands, where sediments are most likely to become anoxic.

Density and Biomass

Total density or biomass of invertebrates is a poor indicator of sedimentation, because either increases or decreases can occur in response to increased sedimentation. Increases often occur when some species of burrowing aquatic worms that tolerate low oxygen conditions are able to proliferate and, in the absence of intense predation, come to dominate the aquatic community.

4.4.5 Invertebrates as Indicators of Excessive Nutrient Loads and Anoxia

Species Composition

Particular assemblages of invertebrate species have commonly been reported to be useful indicators of lake trophic state (as categorized in the book by Rosenberg and Resh 1993) and might be similarly useful for signaling wetlands that have received excessive nutrients. In general, the proportion of "scrapers" (species that characteristically graze on algae) increases with eutrophication, at least during the early stages of enrichment. Specifically, increases in the ratios of (a) tubificid worms to sedentary aquatic insects, (b) the midge subfamilies Tanypodinae and/or Chironomini to the subfamily Orthocladiinae, (c) non-burrowing mayflies to burrowing invertebrates, and/or (d) cladocerans to rotifers, have been reported to indicate excessive nutrient loading of wetlands or other water bodies (Wiederholm 1980, Ferrington and Crisp 1989, Kansanen et al. 1984, Radwan and Popiolek 1989, Rosenberg et al. 1984, Jones and Clark 1987). Over 600 invertebrate species are categorized according to their association with a particular water body's nutrient status in a literature-based table in Rosenberg and Resh (1993). Long-term changes in wetland nutrient status might be inferred by collecting decay-resistant remains of invertebrates from sediment cores or settling traps, and using the Rosenberg and Resh (1993) table to determine what proportion of the found species are ones that characteristically prefer enriched environments (see Section 4.6.3).

Species Richness

 

Species richness of invertebrates can decrease (Wiederholm and Eriksson 1979, Sedana 1987) or increase (Tucker 1958) in response to enrichment. In lakes, zooplankton richness initially increases with increasing phytoplankton production, then it decreases as production continues to rise (Dodson 1992).

Density and Biomass

Density and/or biomass of invertebrates, especially midges, increases with larger increases in wetland fertility (Ferrington and Crisp 1989, Johnson and McNeil 1988, Murkin et al. 1991). Indeed, the density of midges (as measured using emergence traps) has been recommended as an efficient indicator in some situations of secondary production in lakes (Welch et al. 1988). Although data from other regions show invertebrate density increasing in response to increased nutrients (e.g., Cyr and Downing 1988, Tucker 1958, Belanger and Couture 1988, Sedana 1987), substantial and chronic nutrient additions are needed to cause this response (Gabor et al. 1994, Murkin et al. 1994a), and at some point the response of the whole invertebrate community changes from an increase to a decrease in density. This occurs as plant litter accumulates faster than it can be processed effectively and oxygen is depleted from the sediments and water column (Almazan and Boyd 1978), causing a reduction in densities of many invertebrates (Hartland-Rowe and Wright 1975, Pezeshik 1987, Schwartz and Gruendling 1985). Based on results of two experimental nutrient loading studies in prairie wetlands, Murkin et al. (1994a) suggest that "ideal nutrient addition levels" which balance positive fertilization effects against adverse oxygen depletion are between 60 and 200 mg/L for phosphorus and between 1600 and 2400 mg/L for nitrogen, added biweekly during summer.

4.4.6 Invertebrates as Indicators of Pesticide and Heavy Metal Contamination

Species Composition

Among insecticides, the synthetic pyrethroids (especially deltamethrin) are generally more toxic to invertebrates than the organochlorine, organophosphorus, and carbamate pesticides (Sheehan et al. 1987). Mollusks are a possible exception to this ranking. Mayflies and amphipods tend to be more sensitive to most insecticides than are midges and adult water beetles. In one of the few biassays conducted in a prairie wetland, Johnson (1986) found the insecticides carbofuran, fonofos, and phorate to be highly toxic to two invertebrates (Daphnia and a midge species). Carbofuran's toxicity to aquatic invertebrates was corroborated in Wayland and Boag's (1990, 1995) prairie wetland bioassays. When applied to a Minnesota wetland at typical field concentrations, the pesticides temephos, chlorpyrifos, and dursban have been demonstrated to kill copepods, cladocerans, and phantom midges (Helgen et al. 1988).

In general, herbicides are not as acutely lethal to invertebrates as are insecticides (e.g., Buhl and Faerber 1990). Perhaps the most toxic herbicides are the triazines, including the commonly used herbicide atrazine. Atrazine has been shown to cause shifts in community composition and emergence times of aquatic insects at a concentration of 2 mg/L (Dewey 1986), and as little as 0.230 mg/L reduced the development of midges (Macek et al. 1976). The herbicide triallate is also quite toxic to prairie invertebrates (Johnson 1986, Arts et al. 1995). Other herbicides used in wetlands have been shown to increase the dominance of invertebrates (e.g., many aquatic worms) that are tolerant of low dissolved oxygen, a result related to the large oxygen deficit caused by decay of massive amounts of plants (Scorgie 1980). Herbicides can also increase the dominance of open-water forms (e.g., cladocerans) as their algal food base blooms after the reduction of shading aquatic vegetation.

A host of factors influence toxicity and mortality, and are sufficient to change the generic rankings of insecticide toxicity as well as lethal thresholds. These can include:

Environmental factors: water temperature, organic content, pH, alkalinity, suspended solids.

Dose factors: concentration, the specific formulation (inert ingredients), frequency of application, duration of exposure.

Biotic factors: the invertebrate species, its life stage, proximity of unexposed microhabitat patches, degree of simultaneous stress from other factors that may be related (e.g., oxygen stress and enrichment from plant decomposition and photosynthetic inhibition for 1-2 weeks after herbicide application) or unrelated (e.g., drought).

The availability of vegetation can be particularly important to invertebrate survival in wetlands having sediments that are contaminated chemically or that are persistently anoxic or saline. In such situations vegetation provides a colonization surface isolated from the sediments, where contaminants often are concentrated (Nebeker et al. 1988). Richness and abundance of epiphytic and nektonic (swimming) invertebrate s can thus remain high in some contaminated but well-vegetated wetlands (McLachlan 1975).

In other regions and in sediments exposed to some herbicides or severely contaminated by heavy metals, investigators have noted a shift from a community of aquatic insects and toward a community dominated by certain oligochaetes (aquatic worms) (e.g., Wentsel et al. 1978, Howmiller and Scott 1977, Winner et al. 1980). In non-wetland water bodies, areas that are at least moderately contaminated often are dominated by chironomid midges (Winner et al. 1980, Cushman and Goyert 1984, Rosas et al. 1985, Waterhouse and Farrell 1985, McCarthy and Henry 1993) and other aquatic invertebrate species whose adults have wings and short life cycles, e.g., water bugs (Hemiptera) and water beetles (Coleoptera) (Borthwick 1988, Courtemanch and Gibbs 1979, Gibbs et al. 1981). Wetland amphipods (Gammarus, Hyallela), clam shrimp (Lynceus brachyurus), and many zooplankton species, appear to be very sensitive to certain pesticides, whereas most aquatic snails and worms are less sensitive (Sheehan et al. 1987, McCarthy and Henry 1993). Amphipods are especially useful as indicators of contamination because they are relatively stationary (i.e., because they do not emerge and fly away like aquatic insects, their presence can be more indicative of the longer-term conditions of a wetland). Dosed populations may require at least a year to recover (Gibbs et al. 1981, McCarthy and Henry 1993). Amphipods occur in most wetlands with relatively persistent standing water, and their response to pesticides has been documented in prairie pothole wetlands specifically (Borthwick 1988). Pesticide bioassays in prairie wetlands by Ruelle and Henry (1993) indicated greater sensitivity of Daphnia magna than Hyalella azteca, and greater sensitivity among younger than older individuals of both.

In wetlands that lack surface water, nematodes can be particularly sensitive indicators of contaminant toxicity; those of the subclass Adenophorea tend to be more sensitive than those of the subclass Secernentea (Bongers 1990, Platt et al. 1984, Zullini and Peretti 1986). The nematode suborder Dorylaimina, oribatid mites, and many ground beetles (Carabiidae) are highly sensitive. Apparently the least sensitive organisms in such habitats are the soft-bodied invertebrates such as earthworms, terrestrial herbivores such as ants and weevils, and invertebrates that inhabit the upper soil layers such as springtails (Collembola) (Bengtsson and Tranvik 1989).

For protecting soil invertebrates, Bengtsson and Tranvik (1989) suggest maximum allowable concentrations for lead of less than 100-200 mg/kg; less than 100 mg/kg for copper; less than 500 mg/kg for zinc, and less than 10-50 mg/kg for cadmium. Concentrations of metals, pesticides, and other substances toxic to invertebrates are tabulated rather comprehensively in USEPA's AQUIRE data base.

Species Richness, Density, and Biomass

Depressed richness and density of aquatic invertebrates is sometimes suggestive of past or ongoing exposure to pesticides, heavy metals, or other contaminants in both permanently flooded (Ferrington et al. 1988, Krueger et al. 1988, Winner et al. 1975, Marshall and Rutschsky 1974) and drier (Bengtsson and Tranvik 1989) habitats. Richness of a pond invertebrate community was reduced following application of the herbicide linuron (Stephenson and Kane 1984). Richness and density of invertebrates can decline even at levels of phenols and oil-water ratios not known to be toxic in laboratory studies (Cushman and Goyert 1984).

Bioaccumulation

Bioaccumulation of some substances appears to be greater among sand-dwelling invertebrates than mud-dwelling invertebrates (Muir et al. 1983). Several aquatic invertebrate taxa effectively accumulate certain heavy metals in lakes (e.g., Hare et al. 1991) and probably wetlands, but few data are available specifically from prairie wetlands.

Physical and Genetic Deformities

Physical deformities of individuals often accompany severe pollution. For example, midges with deformed mouth parts were noted in areas of synthetic-coal-derived oil pollution (Cushman and Goyert 1984). This indicator is difficult to recognize objectively, and has not been examined in prairie wetlands. Perhaps more objective would be the application of electrophoresis techniques in genetic analysis. Such an approach might be able to rapidly detect past exposure of an invertebrate population to a pesticide, assuming that surviving organisms have a different gene frequency than the parent population (M. Brinkman and W. Duffy, personal communication, South Dakota State Univ., Brookings).

4.5 Monitoring Techniques

4.5.1 General Considerations

Methods and equipment for field-sampling of invertebrates, including wetland taxa, are reviewed comprehensively by Murkin et al. (1994b). Other useful summaries are provided by Edmondson and Winberg (1971), Downing and Rigler (1984), Isom (1986), Fredrickson and Reid (1988b), Ross and Murkin (1989), Staley and Rope (1993), and Rosenberg and Resh (1993).

Larvae of most aquatic invertebrates can be found in wetlands throughout the year. However, particular s (e.g., midges, mayflies) are more evident from about May until September, whereas others (e.g., Anostraca, some Trichoptera) are more abundant earlier in the spring and can be found outside the usual growing season (Swanson et al. 1974). If wetlands can be sampled only once, then the late wet season or beginning of the dry season are usually the recommended time, as density and richness in many wetlands tend to be greatest then (Marchant 1982). Depending on the study objective, the sampling schedule may need to be adjusted to coincide with phenologies of particular taxa (Resh 1979, Sklar 1985). For example, one might want to avoid sampling immediately after a synchronous emergence of the usually dominant species (i.e., a day or week when nearly all individuals of a species emerge at once). Maximum information is often obtained when most invertebrates are within a size range (later, larger instars) retained by nets used to sample them, and can be identified with greatest confidence. Estimates of macroinvertebrate production or seasonal change in standing crop generally require that samples be collected at least monthly and preferably biweekly.

4.5.2 Sampling Equipment

The choice of equipment depends largely on the wetland microhabitat to be sampled. Different assemblages of wetland invertebrates inhabit sediments (benthos), rooted plants or algae (phytomacrofauna), open water (nekton), and the surface film (neuston).

A significant problem in analyzing wetland invertebrate data arises from difficulties in determining the spatial dimensions of the area from which a sample was drawn. Accurate estimates of density (individuals per unit area) are difficult to achieve due to difficulties in accurately measuring the complex wetland substrate (submersed plants, emergent plant stems, etc.). To address this, some investigators have removed the substrate along with the collected sample, measured both, and reported density as weight (or number of organisms) per unit of weight or area of substrate (e.g., Mrachek 1966). In some cases regression coefficients have been calculated to convert plant weights to plant area, which can be further converted to estimates of invertebrate density using previously determined empirical relationships (Downing and Cyr 1985, Downing 1986).

The most common method of sampling invertebrates in prairie wetlands has involved use of sweep nets (or modified dip nets). These are the familiar long-handled, inexpensive insect nets. They can be used in water or air, except in mostly robust or dense stands of vegetation. They are either swept through a standard length of vegetation, or placed on the bottom and hauled vertically through the water column in a rapid stroke. User variability can be a concern, but sweep nets are convenient to use and are particularly suited for capturing large (e.g., crayfish) or quick-moving species such as adult dragonflies and water striders that are not collected by other methods, Samples are not strictly quantitative because the unit of area swept is difficult to determine accurately (Adamus 1984, Plafkin et al. 1989). Also, measured species composition is strongly influenced by mesh size. In trial comparisons against a modified Gerking sampler (see below), Kaminski and Murkin (1981) found sweep nets to be just as effective in sampling water-column taxa. Other researchers who have described results from use of sweep nets in prairie wetlands include Hanson (1952), Voigts (1975), Lancaster and Scudder (1987), McCrady et al. (1986), Kreil and Crawford (1986), LaGrange and Dinsmore (1989b), and Swanson (1984). Sweep nets are one device being used to sample prairie wetlands in the EMAP effort as well as in the effort conducted by the State of Montana.

A method suitable to use anywhere the water is at least a few inches deep involves use of activity (funnel) traps. Most trap designs follow descriptions of Murkin et al. (1983) and Swanson (1987b). Traps are positioned below the water level or on the bottom for hours or days, and nektonic invertebrates that enter the funnel-shaped trap cannot escape. Traps can be positioned vertically (more likely to capture emerging insects) or horizontally. Collections contain only nektonic invertebrates or, if placed on the bottom, only non-sedentary invertebrates (e.g., adult clams are usually missed). Non-insect invertebrates (e.g., Hyalella) as well as aquatic insects are passively collected. Traps can be fitted with lights to increase their attraction to some insects (Lancaster and Scudder 1987). Use is limited to wetlands with standing water, and traps are probably more effective when placed in open water areas or at the edge of vegetation patches, than if placed within dense stands. If traps are set for more than a few days, it sometimes is necessary to move them as water levels recede. Samples are not strictly quantitative because it is impossible to tell what size area the organisms came from, and because some invertebrates and fish caught in the trap prey on other captives during the holding time (Murkin et al. 1983). However, the samplers are lightweight and inexpensive, and sample processing time is less than for some other methods because samples are mostly free of plant material and sediment. Because activity traps accumulate organisms over time, fewer species are missed as a result of ephemeral factors that cause avoidance.

From two years of biweekly samples from 10 wetland sites, Murkin et al. (1983) found significant correlation between the total numbers of invertebrates collected in activity traps and the number collected using sweep nets, although species composition differed somewhat (e.g., traps attract predatory invertebrates disproportionately). In a study of midges, Welch et al. (1988) found no difference in total catch between 0.142-m2 and 0.283-m2 trap sizes. Traps with inverted funnels inserted in the jar necks caught more pupae than traps without funnels, and total catch in the traps without funnels was 58% of the catch in traps with funnels. Other researchers who describe results of using activity or funnel traps in prairie wetlands include Armstrong and Nudds (1985), Kreil and Crawford (1986), Helgen et al. (1988), Hanson and Swanson (1989), Neckles et al. (1990), Murkin et al. (1991), and Bataille and Baldassarre (1993). Activity traps are also being used in the EMAP effort.

If the objective is to sample invertebrates that inhabit mainly the water column, tube samplers (e.g., "water column samplers," "Swanson samplers," "Gerking samplers", "stovepipes", "box samplers") can be used. These are plexiglass cylinders about 6 cm wide that enclose a standard area of bottom substrate. Like corers (see below) they may sample some benthic organisms, but they are not designed to effectively penetrate the sediment (the device described by Euliss et al. 1992 is possibly an exception). In some (e.g., the Gerking sampler), the bottom can be sealed off with a sliding door, plug, or similar feature once the sampler is in place. Some have been fitted with a reinforced cutting edge on the bottom. Tube samplers are not effective in dense vegetation or for catching quick-moving organisms, burrowing species, very large organisms, or many epiphytic species. A major advantage is that, by usually sampling the entire vertical extent of the water column, they capture diurnally-migrating species that are missed by samplers that can sample only at a particular depth.

The most popular type of tube sampler for use in prairie wetlands seems to be the design of Swanson (1978a, 1983), which has been used by Neckles et al. (1990), Broschart and Linder (1986), and LaBaugh and Swanson (1988). A refined and expanded version of this sampler is described by Euliss et al. (1992). Also, Gates et al. (1987) described a type of tube sampler that simultaneously collects invertebrates on plants and in the sediment. They found this to give results for plant invertebrates at least as precise and sometimes more accurate than obtained by clipping macrophytes (see below). Other designs are described by Freeman et al. (1984), Gerking (1957), Korinkova (1971), Hiley et al. (1981), Legner et al. (1975), Mackay and Qadri (1971), Martin and Shireman (1976), and Minto (1977).

Emergence and aerial light traps are another option. They consist of floating nets or funnels, covering an area of 0.1 or 0.5 m2, that are anchored at and just above the water surface or (rarely) are submerged. They are left in place for a specified period of time, during which they are checked daily. They passively collect aquatic insects that are developing from larval to winged adult stage (i.e., emerging from the water column), and thus do not collect non-emerging invertebrates which sometimes are a dominant component of the invertebrate community. Use of emergence traps is limited to wetlands containing open patches of surface water during the growing season when most insects emerge. They can be placed over either open water or short vegetation. If traps are set for more than a few days, they should be moved periodically to avoid altering the sampled habitat by shading it (H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba). Traps also must sometimes be moved as water levels recede. A popular type of emergence trap appears to be the design of LeSage and Harrison (1979), as modified by Wrubleski (1984). A design by M. Butler, described by Nelson and Butler (1987) is also used. Results of using emergence traps in prairie wetlands are described by Nelson and Butler (1987), Driver (1977), Ross and Murkin (1993), Neckles et al. (1990), Wrubleski (1989), and Bataille and Baldassarre (1993).

Because emergence traps are left in place for up to several weeks, they reduce the problem encountered by other samplers of missing key species due to inappropriate time of visit. Sample processing times are favorable because organisms do not have to be separated from sediments and plant material. Because emerging insects come from a variety of microenvironments, emergence and aerial light traps can integrate well the extreme spatial heterogeneity within many wetlands. On the other hand, this variety makes it impossible to standardize or determine the unit of area from which the organisms originated. Moreover, some trapped organisms may prey upon each other, confounding any quantitative estimates, if traps are not emptied often. Initial purchase and maintenance of traps can be costly, and vandalism can be a problem. If emergence traps are unavailable, many species of emerging aquatic insects can be identified and densities grossly estimated from exuviae (shed remains) that are easily sieved from the water surface (McGill et al. 1979).

Another choice for sampling invertebrates that normally attach to wetland plants involves use of artificial substrates. Plants are not sampled directly, but rather, plastic plants or other sterile surfaces (e.g., Hester-Dendy plate samplers) are totally submersed in the wetland water column and allowed to be colonized over a period of at least a month (Macan and Kitching 1972, Cairns 1982). Because such substrates standardize the surface area and texture, collections from substrate samplers are highly comparable to each other, making them attractive for use in monitoring of water column water quality. They also are lightweight and sample processing is relatively easy. However, disadvantages include the fact that a return trip to the wetland is required, vandalism can be a problem, their use is limited to wetlands with surface water, they sample only epiphytic species, and representativeness is sometimes uncertain (Adamus 1984). In stands of submersed vegetation, Gerrish and Bristow (1979) used plastic mimics of the pondweed, Potamogeton richardsonii, interspersed among live experimental plants. Although this yielded no significantly different numbers of invertebrates or species per unit of surface area than were found on real plants, aquatic worms were significantly more common on the artificial substrates. Natural substrates initially devoid of organisms can also be used as colonization substrates. For example, plant litter of measured area or volume can be placed in wetlands to allow colonization by detrivorous species over a specified period of time.

If the objective is to sample invertebrate communities inhabiting relatively unvegetated wetland sediments, then dredges -- also called grab samplers (Ekman, Ponar, etc.) -- are often used. They consist of a box with jaws that is lowered onto the sediment. The jaws enclose a specified area of bottom, and retrieve sediments and associated organisms to a sediment depth of about 5 cm. Dredges are used only where surface waters of at least 0.5 m in depth are present, and they are not effective where there are aquatic plants that jam the jaws and prevent full closure. Because an unknown number of organisms subsequently escape and the exact area and sediment depth of the spot being sampled is never certain, estimates of density are only crudely quantitative. Large organisms (e.g., crayfish), water column organisms, and fast-moving species in particular are sampled poorly. Dredges are cumbersome and relatively expensive, and their samples are time-consuming to sort, but they have been used in prairie studies by Timms et al. (1986) and Driver (1977).

Another option for sampling sediments is to use core samplers. Unlike grabs, corers do not have jaws, and instead rely on compactive force or suction to retrieve sediments, sometimes to a depth of about 15 cm. They suffer the same disadvantages as dredges. Samples usually are more precisely quantitative, but the mean size of organisms effectively captured is often smaller, due to the narrow dimensions of corers. Core samplers are sometimes the only option for quantitatively sampling sediment organisms in wetlands that lack surface water. Where aquatic plants interfere with core sampler operation, some investigators have suggested welding a saw blade to the leading edge of the corer, for clipping heavy roots and stems (Murkin and Kadlec 1986b). The corer design that seems to have been used the most in prairie wetlands is that of Swanson (1978c); a similar design is proposed by Bay and Caton (1969). Results of using corers in prairie wetlands are described by Neckles et al. (1990), Talent et al. (1982), Murkin et al. (1982), Murkin and Kadlec (1985a), Broschart and Linder (1986), Kreil and Crawford (1986), Nelson and Butler (1987), and Murkin and Kadlec (1986b).

During periods when sediments or soils are not covered by water, pitfall traps and soil extraction techniques can be used, and sometimes yield the highest densities and species richness (Coulson and Butterfield 1985). If only plant-dwelling invertebrates need to be sampled, another approach involves directly clipping the vegetation while confining it in an enclosed box. Clipped vegetation is then carefully examined for invertebrates in the laboratory. This might provide more precise quantification than does use of sweep nets, although nektonic invertebrates are seldom captured. Downing and Cyr (1985) found the most cost-effective quadrat size for clipping to be 500 cm2. Plants were enclosed in a 6-liter plastic box. Clipping aquatic macrophtyes in quadrats of varying sizes yielded five times higher populations of invertebrates than did sampling with some tube samplers (Gerking, Macan, or Minto samplers). Vacuum suction also can be used to help remove small invertebrates from foliage in the field (Southwood 1981).

4.5.3 Time-Integrating Methods

The above methods are used primarily to sample living organisms. Sometimes, the sclerotized, decay-resistant remains of particular invertebrate s (chironomid head capsules, and exoskeletons and eggs of snails, ostracods, daphnids, and conchostracans) persist in an undecomposed condition in wetland sediments for months, years, or even decades and centuries. Settled remains can be collected by a variety of devices (e.g., "sediment traps") and sieved to separate the remains. This, along with identification and enumeration of body parts, can be a difficult, laborious, and somewhat subjective process, but the resulting species composition data provide clues to the wetland's previous long-term environmental conditions. Examples are demonstrated generally by Walker et al. (1991), Walker (1993), and Streever and Crisman (1993), and in prairie lakes specifically by Synerholm (1979) and Euliss et al. (1993). In wetlands exposed to mixing winds, the resulting resuspension of decay-resistant remains from many time periods can complicate data interpretation, unless samples are collected for comparison simultaneously using other methods (N. Euliss, personal communication, NPSC, Jamestown, ND). As part of 1992-1994 pilot studies in the prairie region, EMAP sponsored the development and testing of methods for accurately sampling decay-resistant remains of invertebrates.

4.5.4 Bioassay Methods

A review of laboratory, outdoor mesocosm, or in situ bioassay methods involving invertebrates is beyond the scope of this report. Use of bioassays to explore toxicity in prairie wetlands has been relatively limited. Examples include studies by Johnson (1986, Borthwick (1988), Helgen et al. (1988), Wayland and Boag (1990), and Ruelle and Henry (1993).

4.5.5 Bioaccumulation

Methods for collecting wetland invertebrates and assessing bioaccumulation of contaminants in their tissues are described in Staley and Rope (1993).

4.6 Variability and Reference Points

Numerical estimates cited in the following Sections are difficult or impossible to compare with one another, because they are based on samples collected with a variety of equipment and methods, in different wetlands, and for different time periods. They are cited only to provide order-of-magnitude illustration of levels of various parameters that have been encountered in prairie wetlands. For a specific listing of the methods behind each cited value and study, see Appendix J.

4.6.1 Spatial Variability

Species Richness

The true species richness of invertebrates in prairie wetlands is generally unknown because taxonomic knowledge and resources have nearly always been insufficient to make species-level determinations of specimens. This is suggested by the fact that a collection from four prairie wetlands of 2594 individuals representing just a single invertebrate -- the aquatic Coleoptera -- yielded 57 species. Remarkably, these four wetlands contained almost half the aquatic Coleopteran species ever found in North Dakota (Hanson and Swanson 1989). Similarly, weekly collections of midges emerging from a single pond within the Delta Marsh yielded a total of 84 species (Wrubleski and Rosenberg 1990).

Although comparisons among studies are hindered by the fact that levels of resolution in taxonomic determinations have varied greatly, some results are as follows:

Location, citation Basin type Sampling approach Number of taxa
Nebraska (Rainwater Basin);
Gordon et al. 1990
Seasonal
(n = 8),
1 field season
One field season 39 (cumulative)
South Dakota;
Duffy & Birkelo 1993
Semipermanent Cores, biweekly 74
(mean = 44/basin, range = 31-58)
South Dakota;
Broschart & Linder 1986
Lakeside marsh Tube samples and cores, one field season Benthos: 5.7 families/basin
Nekton: 7.0
families/basin
Iowa;
Voigts 1975
Lakeside marsh   20-32 families
Iowa;
LaGrange & Dinsmore 1989
Restored semipermanent Sweep net samples, single visit 4-16 families/basin
Iowa;
Hemesath 1991
Restored
semipermanent
(n = 17)
Sweep net samples,
3/basin in June
7-17 families/basin, cumulative richness of 32 families
North Dakota;
Euliss, pers. comm., NPSC, Jamestown, ND
Semipermanent
(n = 18)
Sweep net samples,
2 field seasons
29 families cumulatively;
Median = 18/basin (range, 9-25)
North Dakota (Cottonwood Lakes);
Semipermanent & Seasonal
(n = 5)
  9-19 taxa/basin
(rotifers, copepods, and cladocerans only)
Manitoba (Delta Marsh);
Murkin et al. 1991, Neckles et al. 1990
Lakeside, unmanipulated Funnel trap (24-hr sets, 5 years) 54 families;
Max./sample = 18;
Median = 8
Manitoba (Delta Marsh);
Kaminski & Prince 1981 a,b
Lakeside   1.2 families/m3 (one year);
1.9 families/m3
(another year)
Manitoba;
Bataille & Baldassarre 1993
Semipermanent
(n = 3)
Activity traps, weekly 26 families/500 samples
  Emergence traps, weekly 50 families/500 samples

Richness varies significantly within wetlands as well. When Delta Marsh samples from various periods and years were pooled, the zone with cumulatively the most families was the hardstem bulrush (Scirpus acutus) zone (47 families), followed by the open water zone (44 families), and cat-tail zone (36 families). When manipulated parts of the Delta Marsh were included as well, the zone with cumulatively the most families was the whitetop (Scolochloa festucacea) zone (50 families), followed by cat-tail (48 families), hardstem bulrush and red goosefoot (47 families), softstem bulrush (Scirpus validus) (46 families), rayless aster (Aster brachyactis) (45 families), and the open water zone (44 families) (H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba). However, in both manipulated and unmanipulated wetlands, the relative rankings of zones based on their species richness varied by season and year.

Species Composition

Species that are most numerous or constitute the greatest biomass are often the most ecologically influential. The species of invertebrates that dominate prairie wetlands vary from wetland to wetland; those reported in the literature to dominate at least one prairie wetland are shown in Appendix B. Species that are most sensitive to environmental change are often those with the narrowest habitat requirements, and species with narrow habitat requirements can often be identified as those with locally restricted distributions. Of the cumulative total of 74 taxa found by biweekly core sampling of four South Dakota semipermanent wetlands, 33 taxa were found in only a single wetland (W. Duffy, personal communication, South Dakota St. Univ., Brookings, SD). Of 32 families found collectively in the 17 restored prairie wetlands in Iowa, five families were found only in one wetland (Hemesath 1991, Hemesath and Dinsmore 1993). Of 54 families found in funnel traps in four differently-manipulated units of the Delta Marsh over a 5-year period, 19 were present in only one unit. The hydrologically unmanipulated unit had the most taxa (8) that occurred nowhere else; these included several mayflies (H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba). In the survey of 18 semipermanent wetlands in North Dakota which found 29 invertebrate groups, the groups with the most restricted distribution (as sampled with a sweep net) were fairy shrimp, amphipods (2 wetlands each), and broad-shouldered water striders (1 wetland) (N. Euliss, personal communication, NPSC, Jamestown, ND).

Density and Biomass

Macroinvertebrate densities as high as 36,000/m2 are reported from the Delta Marsh by Neckles et al. (1990), although densities of just the benthic invertebrates reached a seasonal peak of about 1200/m2 (Murkin et al. 1982, Murkin and Kadlec 1986b). In a lakeside prairie marsh in South Dakota, Broschart and Linder (1986) reported summertime means of 3534 and 7898/m2 in ditched and unditched areas respectively. Another lakeside marsh in Iowa had a maximum of about 15,000/m2, mainly associated with submersed plants (Voigts 1975). Midge larvae alone were present at densities of up to 10,092/m2 in prairie wetlands used by mallards with broods, whereas samples from 16 randomly selected wetlands had a maximum of 5337/m2 (Talent et al. 1982). In contrast, when a corer was used, the mean densities from eight seasonal wetlands in the Rainwater Basin of Nebraska ranged from only 28/m2 in one wetland to 86/m2 in the most productive wetland (Gordon et al. 1990).

In sweep net samples fromundisturbed seasonal wetlands in North Dakota, Swanson et al. (1974) reported a mean of about 6500 individuals per m3, whereas densities in water column samples from eight seasonal wetlands in the Rainwater Basin of Nebraska ranged from 106,000/m3 in one wetland to 1,636,000/m3 in another (Gordon et al. 1990).

After core-sampling five natural wetlands of eastern North Dakota bimonthly in summer for two years, Kreil and Crawford (1986) reported a mean of 2686 individuals/m3 in the poorest wetland and a mean of 89,460 individuals/m3 per wetland. Using a core-type sampler in the Delta Marsh, Kaminski and Prince (1981a,b) reported mean densities of benthic invertebrates of 6993 and 18,906/m3 (depending on year), whereas Neckles et al. (1990), using vertical activity traps, reported a mean of 210,000/m3 from the Delta Marsh. The following year, the mean invertebrate density was only 6993/m3. Mean density of core-sampled invertebrates among four South Dakota semipermanent wetlands ranged from 4782/m2 to 20,063/m2 (Duffy and Birkelo 1993). Using a corer elsewhere in North Dakota, Nelson (1989) found densities of 78,000/m2 in semipermanent wetlands and 2400/m2 in seasonal wetlands.

In tube samples from the lakeside prairie marsh in South Dakota, Broschart and Linder (1986) reported means from tube samples of 9687 and 15,194 macroinvertebrates per m3 in unditched and ditched areas respectively. Zooplankton densities averaged about 700,000/m3. In a shallow prairie lake in Minnesota, zooplankton densities peaked at over 450,000/m3 in early autumn (Hanson and Butler 1990). In seasonal wetlands in North Dakota, they peaked at > 2,000,000/m3 but peaked at < 1,000,000/m3 in semipermanent wetlands in the same area (LaBaugh and Swanson 1993). Zooplankton densities also exceed 1,000,000/m3 in the Delta Marsh (Collias and Collias 1963) and in seasonal wetlands of the Rainwater Basin of Nebraska (Gordon et al. 1990).

In five natural wetlands in eastern North Dakota, Kreil and Crawford (1986) found an average of between 53 and > 3000 individuals per trap. Activity traps in three prairie wetlands in western Minnesota caught an average of between 156 and 1323 individuals, with seasonal peaks of between 738 and 4448 individuals (per sample per wetland per sampling period). Wetlands that had the largest numbers one year did not necessarily have similar rank among the three wetlands the following year (M. Hanson, personal communication, Minnesota Dept. Natural Resources, Bemidji, MN).

Numbers of nektonic invertebrates found in part of the Delta Marsh during 24-hr sets of each underwater funnel trap ranged up to about 180 (Murkin et al. 1991), whereas in nearby potholes, Bataille and Baldassarre (1993) found up to 2094 per trap. In unmanipulated parts of the Delta Marsh, when samples from various periods and years were pooled, the zone with the greatest numbers of individuals per activity trap was the cat-tail zone, followed distantly by the hardstem bulrush and open water zones. When manipulated parts of the Delta Marsh were considered instead, the zone with the greatest numbers was the red goosefoot (Chenopodium rubrum) zone, followed by the whitetop, cat-tail, rayless aster, softstem bulrush, and open water zones (H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba). However, in both manipulated and unmanipulated wetlands, the relative rankings of zones based on their invertebrate densities varied by season and year.

Published data from emergence traps are more limited. From 0.5-m2 emergence traps left in the Delta Marsh for 29 days in late summer, the mean density of the 10 commonest midges was 179 per trap, and the maximum was 640/m2 (Wrubleski 1989). Emergence sampling of a nearby permanent wetland found a maximum density of 977/m2 (Bataille and Baldassarre 1993). On a total annual basis, emergence of midges in the part of Delta Marsh studied by Wrubleski varied from 2322 to 15,400 individuals/m2, depending on vegetation type and cover ratio (Wrubleski 1989, Wrubleski and Rosenberg 1990).

Biomass estimates are influenced by mesh size, inclusion/exclusion of snail and clam shells, and sampling equipment. From part of the Delta Marsh, Kaminski and Prince (1981a,b) reported a mean biomass of invertebrates of 11,161 mg/m3 during one year and 2843 mg/m3 during another. Based on sweep net samples, the median invertebrate biomass of 18 semipermanent wetlands in North Dakota was about 1300 mg/m3 (N. Euliss, personal communication, NPSC, Jamestown, ND). In tube samples from the lakeside prairie marsh in South Dakota, Broschart and Linder (1986) reported means of 8524 and 6564 mg/m3 from ditched and unditched areas respectively.

On a per-area basis, mean biomass of core-sampled invertebrates in four South Dakota semipermanent wetlands ranged from 1543 to 5428 mg/m2, and production ranged from 4604 to 21,800 mg/m2 (Duffy and Birkelo 1993). In unmanipulated parts of the Delta Marsh, benthic biomass peaked at about 8000 mg/m2 (Murkin et al. 1982, Murkin and Kadlec 1986b). Activity trap samples from four Minnesota semipermanent wetlands yielded a biomass per sample of from 0.38 g in one wetland to 3.23 g in another; seasonal peak biomass ranged from 0.89 g in one wetland to 7.48 g in another (M. Hanson, personal communication, Minnesota Dept. Natural Resources, Bemidji, MN). In a lakeside prairie marsh in South Dakota, Broschart and Linder (1986) reported biomass means of 1746 and 1314 mg/m2 from ditched and unditched areas respectively. Data from submersed vegetation beds in 11 eastern Canadian lakes showed a range of invertebrate densities of 1000-2900 mg/m2 (Lalonde and Downing 1992).

In unmanipulated parts of the Delta Marsh, when samples from various periods and years were pooled, the zone with the greatest biomass per activity trap was the cat-tail zone, followed by the softstem bulrush and open water zones (i.e., the same as when based on number of individuals). Samples from the open water zone weighed only one-quarter the weight of those from the cat-tail zone. When manipulated parts of the Delta Marsh were considered instead, the zone with the greatest biomass was the red goosefoot (Chenopodium rubrum) zone, followed by the softstem bulrush, whitetop, rayless aster, cat-tail, and hardstem bulrush zones; mean invertebrate biomass in the last of these was about half that in the first (H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba). However, as noted earlier, in both manipulated and unmanipulated wetlands the relative rankings of zones based on their invertebrate densities varied by season and year. Considering just the midge component of the invertebrate community, numbers and biomass of emerging individuals were greater in beds of pondweed (Potamogeton pectinatus) than in cat-tail or bulrush stands during a 2-year study in an unmanipulated part of the Delta Marsh (Wrubleski and Rosenberg 1990).

4.6.2 Temporal Variability

Species Composition

Within a season, species composition of invertebrates changes markedly. Water bugs (Hemiptera), water beetles (Coleoptera), and snails (Gastropoda) seem to be more evident later in the growing season in some prairie wetlands (Bartonek and Hickey 1969, Swanson et al. 1974). In unmanipulated parts of the Delta Marsh, data from activity traps and artificial substrates showed that seasonal peaks were attained mostly in late spring or early summer by mosquitoes, ostracods, and water mites; in mid-summer by lymnaeid snails and non-predacious midges; and in late summer by planorbid and physid snails, cladocerans, copepods, and amphipods (H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba). In three western Minnesota semipermanent wetlands, seasonal peaks were attained mostly in late spring or early summer by clam shrimps (Conchostraca), water beetles, dragonflies, and cladocerans; in mid-summer by mayflies, snails, leeches, and water mites; and in late summer by water bugs (Hemiptera), amphipods, and copepods. However, the same taxa did not necessarily show the same seasonal patterns in all wetlands, or even in the same wetland between years (M. Hanson, personal communication, Minnesota Dept. Natural Resources, Bemidji, MN).

Species Richness

In the Delta Marsh, Murkin et al. (1991) documented variation in taxonomic richness within a year. Richness in activity traps ranged from an average of about 11 families in late summer to near 0 during October sampling (the latest sampling of the year). Similarly, when samples from all years and zones within the unmanipulated wetlands were pooled, the data show relatively constant richness until early to mid-September, at which time richness drops. Richness in the unmanipulated wetland ranged from a high of 33 families per seasonal period to a low of 11. None of the manipulated units had more than 31 families per seasonal period, and most had no more than 16. Some manipulated units had only two orfamilies per seasonal period (H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba).

Interannual variation also can be extreme. Of 54 families collected by activity traps in unmanipulated parts of Delta Marsh over a 5-year period, no more than 51 were collected in any single year, and in one year only 38 were found (H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba). In one of 18 semipermanent wetlands in North Dakota, the number of invertebrate families changed from 4 to 16 families between just two years, whereas in several other wetlands the number of families remained stable or decreased between years (N. Euliss, personal communication, NPSC, Jamestown, ND). In six prairie wetlands in southern Saskatchewan, the number of midge species over a 3-year period ranged from 9 in one wetland to 20-24 in another (Driver 1977). Annual species extinction rates varied from 0 to 44% (of all midge species present during the period); immigration rates varied from 0 to 58%; and turnover rates (the difference between immigration and extinction rates) varied from 11 to 100%. As expected, immigration and species replacement rates were greater in temporary wetlands than in semipermanent or seasonal wetlands.

Density and Biomass

The seasonal variation in availability of invertebrates in prairie wetlands is a critical factor affecting waterbird use. Although temporary wetlands (as noted above) harbor generally fewer species of invertebrates than do semipermanent wetlands, the species that are present typically occur in enormous quantities at a season when invertebrate biomass in more permanently flooded wetlands is relatively small or unavailable due to persisting ice cover.

In sweep net samples fromundisturbed seasonal wetlands in North Dakota, invertebrates varied seasonally within the growing period from a maximum of about 18,964 individuals per m3 in one wetland in April to about 1000/m3 in another wetland in June (Swanson et al. 1974). In South Dakota, the density and biomass of benthic invertebrates in three seasonal wetlands was greatest in late June just before the wetlands dried up, whereas in a semipermanent wetland, density and biomass increased as the growing season progressed, reaching highest levels during the last sampling on September 21 (W. Duffy, personal communication, South Dakota St. Univ., Brookings, SD). In a group of four semipermanent wetlands in western Minnesota, peak biomass in activity trap samples occurred in late spring - early summer (M. Hanson, personal communication, Minnesota Dept. Natural Resources, Bemidji, MN).

In semipermanent wetlands of Delta Marsh, Murkin et al. (1991) and Bataille and Baldassarre (1993) documented considerable variation in density and biomass by season. Total numbers of individuals in activity traps ranged from about 2094 in early summer (Bataille and Baldassarre 1993) and 300 in late summer (Murkin et al. 1991) to near 0 during October sampling (the latest sampling of the year). Biomass ranged from about 150 mg in summer to near 0 mg in October. Peaks in invertebrate abundance and biomass generally occurred in spring, often coinciding with the period when waterfowl were laying eggs (Bataille and Baldassarre 1993). A second peak, especially of zooplankton, sometimes occurs in late summer or early fall when inputs of plant litter to the water column peaked (Murkin et al. 1991). There is some interannual variability in the timing of seasonal peaks; a late summer peak was noticeable in some years and habitats but not in others, and the spring peak occurred earlier in some years than in others (Wrubleski and Rosenberg 1990, and H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba). Cladocerans in particular experienced a midsummer depression in numbers, coincident with an increase in cover of submersed plants (Murkin et al. 1991).

Interannual variation can be tremendous. Between just two years, the number of individuals in each of several semipermanent wetlands in North Dakota varied byorders of magnitude, whereas in others it varied only slightly (N. Euliss, personal communication, NPSC, Jamestown, ND). The invertebrate density (mean number of individuals per sample per period) in two western Minnesota semipermanent wetlands varied three- and five-fold between years, but did not change significantly in a third (M. Hanson, personal communication, Minnesota Dept. Natural Resources, Bemidji, MN). In unmanipulated parts of the Delta Marsh, numbers of invertebrates caught in activity traps varied sevenfold over a five-year period, and biomass varied by a factor of 3. In emergence traps from the same general area, the mean abundance of midges varied from 21,499/m2 one year to 29,627/m2 another, while biomass changed little (Wrubleski and Rosenberg 1990). In manipulated parts of the Marsh, total invertebrate abundance varied tenfold over the five years, and biomass varied twofold (H. Murkin, personal communication, Institute for Wetland and Waterfowl Research, Oak Hammock Marsh, Manitoba). In one part of the Delta Marsh, interannual changes in abundance and biomass of invertebrates were least in areas that had been disturbed by mowing (Kaminski and Prince 1981b). Interannual fluctuations in the amphipod, Hyalella, appear to be particularly great (Voigts 1975), in one instance ranging between 0 and 71% of all invertebrates sampled in a wetland, depending on the year (Bartonek and Hickey 1969). Interannual changes are probably a reflection of changing vegetation and water conditions.

The part of the season during which the peak occurs in a particular wetland's invertebrate population is not necessarily consistent between years (Wrubleski and Rosenberg 1990). For example, in the same set of Minnesota wetlands, invertebrates in two of the wetlands appeared to peak in late summer one year but early summer the next, whereas in a third wetland, the peaks both years occurred in early summer.

Bioaccumulation

In a lake in eastern Canada, bioaccumulation of heavy metals in aquatic insects followed a seasonal periodicity, and in no case varied more than sixfold over the course of a year (Hare and Campbell 1992). Bioaccumulation depends on characteristics of the particular contaminant, the abiotic environment, and characteristics of the invertebrate species, e.g., its feeding habits, body size, and microhabitat preferences (Krantzberg 1989, van Hattum et al. 1991, Hare et al. 1991).

4.6.3 Spatial vs. Temporal Variability

In a four-year study of diving beetle communities in an Alberta lake, Aiken (1991) found that species composition varied less between years than between zones within the lake (zones were: sedge, cat-tail, willow, willow-cat-tail, mixed, and open water). In a two-year study of midges emerging fromhabitats (pondweed, cat-tail, bulrush) of the Delta Marsh, Wrubleski and Rosenberg (1990) found that numbers and biomass varied less between years than among these habitats.

4.7 Collection of Ancillary Data

It is easier to separate the anthropogenic from the natural causes of impairment of community structure if data are collected or inferred simultaneously on the following variables of particular importance to wetland invertebrates:

age of wetland and its successional status, water or saturation depth, conductivity and baseline chemistry of waters and sediments (especially pH, alkalinity or calcium, and organic carbon), sediment type, presence of fish and salamanders, density, type, and form of vegetation (particularly, total surface area), cover ratio, and the duration, frequency, and seasonal timing of regular inundation, as well as time elapsed since the last severe inundation or drought.

All of these features vary to a large degree naturally as well as in response to human activities such as soil tillage, compaction, and erosion; fertilizer and pesticide application; and water regime modification.

4.8 Sampling Design and Required Level of Sampling Effort

4.8.1 General Considerations

Locations within a wetland from which invertebrate samples are collected can be chosen according to many of the designs described for sampling wetland vegetation (Section 4.8, p. ?). EMAP effort's has collected invertebrate samples randomly from transects radiating in four compass directions from the center of each wetland.

The time required to collect an invertebrate sample varies somewhat among the sampling methods (sweep net, corers, etc.), and is about 1-3 minutes per sample, not including travel time to the sample site. The largest collection-time differences among methods relate to the number of samples each method requires to achieve a prespecified level of precision, and the sorting times required to separate invertebrates from debris in the collected samples. If sorting is done at the wetland site, a screen such as designed by Swanson (1977b) and modified by Euliss and Swanson (1989) can expedite the process. More often, samples are sorted in a laboratory under high-intensity lights or in direct sunlight. Sorting time is about 15-30 minutes for samples from emergence and activity traps and artificial substrates, about 15-45 minutes for sweep net and tube samples, and at leasthours for some core and dredge samples (Murkin et al. 1983, Resh et al. 1985). These estimates depend, of course, on how numerous and cryptic the sampled organisms are, how large and "dirty" the sample is, and how completely one wishes to process the sample. Samples collected with a fine-mesh net (e.g., < 80 microns) will naturally contain many more individuals than samples collected with coarser nets.

Many investigators have used sugar flotation methods to separate invertebrates from debris; apparently only a few have used rose bengal stain to increase the detection of individuals in samples. Core samples are routinely sieved before sorting. Except for very small samples, live-sorting (sorting of samples in the field prior to preservation) is unlikely to succeed in removing more than a small proportion of individuals, which are usually the largest and most active forms and thus not necessarily representative of the invertebrate community. Because the time that would be required to find every individual in a dirty sample seems almost limitless, some investigators have used a single, standard sorting time for all samples, but most have simply exercised judgement as to when they believe they have found nearly all specimens. If species richness is to be determined, then sorting only a fixed number of individuals is inadvisable because of a bias toward selecting larger and less cryptic individuals. More preferable is the complete counting of all individuals within randomly-selected subsamples; subsamples may be delineated by drawing a grid on the bottom of the sorting pan. Once all individuals have been sorted, tallies of species richness are justifiably based on a fixed number of individuals (generally 100-1000) that are chosen randomly from within a sample.

If biomass determinations will be made, samples are typically dried in a drying oven to a constant weight. In some cases, snail and clam shells are first dissolved with acid and caddisflies are removed from their cases, which are not weighed.

An important consideration that affects sample costs is the desired level of taxonomic identification. Identifying aquatic invertebrates to the species level usually allows the investigator to make more refined statements about the condition of a wetland, but can increase sample processing time at least fourfold and requires advanced training, experience, and the availability of region-specific taxonomic keys. There are no data to indicate whether, and under what conditions, identification of organisms only to the family level would be sufficient to define the ecological integrity of a prairie wetland. Determination of invertebrates to genus or species takes at least three times longer than when identification is made only to the family level (assuming the identifier is familiar with keys of all taxa). Processing of samples (sorting, identification to family, data entry) from sweep nets and activity traps can be accomplished in 2-3 hours, whereas about 4 hours are required for samples collected using corers and tube samplers (N. Euliss, Jamestown, NPSC, Jamestown, ND).

Sampling costs are determined not only by sample collection, sorting, and identification times, but also by the number of samples collected. This should depend on expected variability (coefficient of variation) and the desired precision. Although ecologists commonly consider acceptable a confidence level of 90% and standard errors less than 20% of the mean, Downing's (1979) review of the literature on benthic sampling in lakes revealed that fewer than 3% of published studies attained this goal. For their work in a prairie wetland, Murkin and Kadlec (1986a) stated beforehand that they would accept estimates within + or - 30% of a mean, and as a result they were able to specify the number of samples needed to achieve this precision. Among prairie wetland researchers, they are apparently alone in using and reporting such relevant information.

Interpreting the collected data poses many other methodological challenges that cannot be addressed in this report. In particular, separating the component of variability that is due to natural causes (weather, vegetation, etc.) from the component that is due to anthropogenic causes is always challenging. From an analysis of stream invertebrate data from one California stream, Resh and McElravy (1993) found that the chance of detecting a significant interannual difference due to natural variability alone was at least 22% when species richness is used, 23-35% when invertebrate density is used, at least 5% when a similarity index (Simpson's) is used, and 20-35% when a purported "indicator species" was used. They also found that, for that stream, a sampling regime in which five replicates were collected at a single site would result in an ability to detect a 56% interannual (2-yr) change in species richness and a 73% change in the similarity index, whereas the same regime would only be sensitive enough to detect a > 200% change in invertebrate density and a > 300% change in the indicator species (with a 95% chance of being correct at the 5% level of significance).

From a review of 46 studies that monitored benthic invertebrates in lakes, Voshell et al. (1989) found thatwas the usual number of replicates collected. Review of the prairie wetland literature indicates that previous studies (those intended to characterize the invertebrates or algae) usually collected, in each wetland and at each point in time, 2-4 samples per zone (and in a few cases, 2-3 replicates of each sample). For invertebrate biomass estimates in Minnesota lakes, Hanson and Butler (1990) reported that samples collected at 4- and 6- week intervals were very similar to those based on nine biweekly collections.

4.8.2 Asymptotic Richness: Results of Analysis

For this report, we analyzed invertebrate taxonomic richness from three data sets from prairie wetlands. One data set consisted of replicates collected from four semipermanent wetlands during a single growing season (Duffy, unpublished data). Data from four corer replicates collected within a wetland were first combined into a single list containing all taxa for each wetland-date combination. Two of the wetlands were sampled four times before they dried up, and analysis of data from each showed that half the total number of species (from all four dates) could have been detected in collections from any two dates, but that to detect 95% of the species, sampling on all four dates was required (Appendix O). For a wetland that was sampled on six dates, analysis suggested that half the species could have been found in samples from any two dates, but that to detect 95% of the species, sampling on all six dates was required. Finally, for a more persistently flooded wetland that was sampled on nine dates, the analysis indicated that half the species could have been found (as before) in samples from any two dates, but that to detect 95% of the species, sampling on all nine dates was required.

A second data set (Euliss, unpub. data) consisted of 381 invertebrate sweep-net samples collected from multiple transects in 19 prairie wetlands during a two-year period. Without pooling any of the samples, our analysis indicated that half the 29 taxa that were present collectively could have been detected with only five samples, but to detect 90% of the taxa, at least 178 samples would be required (Appendix O). We then examined data just from the wetland that had the greatest richness (wetland 28-III), and determined that any two samples would produce half the 25 taxa found in all 26 samples, whereas 21 samples would be required to find 95% of the species.

Finally, we examined a very large data set from invertebrate activity traps used in the MERP research effort on Delta Marsh, Manitoba (Murkin, unpub. data). Samples had been collected at various depths in various zones and during various weeks over a 5-year period(1)

. We conducted the following analyses:

Number of Seasonal Periods. Just the data for the two zone-year combinations that had the greatest richness in the unmanipulated "reference" unit of the Delta Marsh were compiled. These combinations were: zone 4 (Scirpus acutus zone) in 1985 (31 taxa), and zone 5 (cat-tail zone) in the same year (30 taxa). In the first instance, half the taxa that were present in the samples from 10 periods could have been detected if only two periods (sampling weeks) had been covered, and 95% of the taxa could have been detected from nine periods. In the second instance, half the taxa that were present in the samples from 20 periods could have been detected if onlyperiods had been sampled, and 95% taxa could have been detected from 15 periods.

Number of Years. Just the data for the two zone-year combinations that had the greatest richness in the unmanipulated "reference" unit of the Delta Marsh were compiled. These combinations were: zone 4 during period 4 (last week of May), and the same zone during period seven (late June). In both instances, half the 25 taxa that were present in the samples from all five years could have been detected if only two years had been sampled, but to detect 95% of the taxa, the full five years are required. Species accumulated at a slightly more rapid rate (i.e., interannual conditions were more similar) in late May than in late June.

Number of "Replicates". Data from all 246 sampled combinations of year, period, and zone were examined, from the unmanipulated "reference" unit. Half the 53 taxa that were present collectively in the 246 samples could have been detected if only 13 samples had been collected, and 225 samples would be needed to detect 95% of the taxa.

4.8.3 Power of Detection: Results of Analysis

The Components of Variance approach, as described in Section 1.5, was applied to invertebrate data from four data sets. These data sets (Duffy, Euliss, Hanson, and MERP) are described in Appendix L.

Taxonomic Richness

Core sampling, as implemented by the Duffy study, was better able to detect interwetland differences in the total sampled number of individuals than in the total sampled number of taxa. The Hanson, MERP, and Euliss studies showed the converse. The apparent results of these comparisons might be due less to the type of sampler used than to the relative intensities of sampling. Analysis of the Duffy core data suggests that a sample size of 10 wetlands would allow detection of interwetland differences of seven taxa, whereas if activity traps or sweep nets were used, sampling the same number of wetlands would allow detection of interwetland differences of two andtaxa, respectively. These taxa would likely be different because different sampler types capture different taxa. For wetland types, sampler types, and experimental designs similar to those of thesestudies, sampling additional wetlands beyond 6-13 wetlands has little effect on increasing the precision of the richness estimates.

Total Number of Sampled Individuals

Analysis of the data suggests that a sample size of 10 wetlands would allow detection of interwetland differences of 9200 individual organisms per core sample. Sampling the same number of wetlands with activity traps would allow detection of interwetland differences of 1600 individuals (Hanson data) or 6200 individuals (MERP data), whereas sampling with sweep nets or sediment traps would allow detection of differences of 1300 or 120 individuals, respectively. The sediment trap data also show that 10 transects would allow detection of intertransect differences of 67 total individuals. For wetland types and experimental designs similar to those of these four studies, sampling additional wetlands beyond 5-10 wetlands has little effect on increasing the precision of the density estimates. For the sediment trap approach, sampling more than 14 transects brings diminishing returns with regard to precision of estimates of the number of individuals.

Biomass

The data suggest a sample size of 10 wetlands allows detection of differences between total biomass means ofg (using corer samples or activity traps), 2.1 g (using a sediment sampler), 0.7 g (sweep net sampling), or 0.180 g (more intensive activity trap sampling). Data also indicate that beyond a sample size of about 6-12 wetlands, adding additional wetlands has little effect on increasing the precision of estimates (i.e., the ability to distinguish between means of the total sample biomass of any two of the region's wetlands). For the sediment trap approach, sampling more than 10 wetlands or 12 transects brings diminishing returns with regard to precision of biomass estimates.

Numbers of Individuals: Specific Taxa

Core sampling, as implemented by the Duffy study, was best able to detect interwetland differences in the total sampled number of individuals of (in decreasing order of power of detection): Chironomidae, Anostraca, Conchostraca, Amphipoda, Ostracoda, and total. The data suggested that a corer sample size of 10 wetlands would allow detection of interwetland differences of between 11 (Chironomidae) and 6000 (Ostracoda) individuals. For wetland types and experimental designs similar to those of the Duffy study, core-sampling additional wetlands beyond about nine wetlands has little effect on increasing the precision of estimates of most taxa.

Activity trap sampling, as implemented by the MERP project, was best able to detect interwetland differences in the total sampled number of individuals of (in decreasing order of power of detection): Tanytarsini (a Chironomid group), Amphipoda, Ostracoda, Physidae, Cladocera, and total. In comparison, Hanson's activity trap data were best able to detect interwetland differences in the sampled number of individuals of (in decreasing order of power of detection): Hirudinea, Amphipoda, Conchostraca, Ostracoda, Copepoda, and Cladocera. The MERP and Hanson data suggested that sampling 10 wetlands with activity traps would allow detection of interwetland differences of between two (Tanytarsini) and 2100 (Cladocera) individuals. For wetland types and experimental designs similar to those of the MERP and Hanson studies, placing activity traps in additional wetlands beyond 5-12 wetlands has little effect on increasing the precision of estimates of total sampled numbers of most taxa.

Sweep net sampling, as represented by the Euliss data, was best able to detect interwetland differences in the total sampled number of individuals of (in decreasing order of power of detection): Ephemeroptera, Physidae, Conchostraca, Lymnaeidae, and Chironomidae. The data suggest that sampling 10 wetlands with sweep nets in the manner of Euliss' study would allow detection of interwetland differences of between(Ephemeroptera) and 500 (Chironomidae) individuals. For wetland types and experimental designs similar to those of the Euliss study, conducting sweep net sampling of additional wetlands beyond about six wetlands has little effect on increasing the precision of estimates of total sampled numbers of individuals of a particular taxon.

The sediment trap approach, as used by Euliss, was best able to detect interwetland differences in the total sampled number of individuals of (in decreasing order of power of detection): Lymnaeidae, Cladocera, and Ostracoda. The data suggest that sampling 10 wetlands with sediment traps in the manner of Euliss' study would allow detection of interwetland differences of between 1.6 (Lymnaeidae) and 100 (Ostracoda) individuals.

Biomass: Specific Taxa

Activity trap sampling, as implemented by the MERP project, was best able to detect interwetland differences in the total sampled biomass of individuals of (in decreasing order of power of detection): Amphipoda, Ostracoda, and Cladocera. The data suggested that sampling 10 wetlands with activity traps would allow detection of interwetland biomass differences of, at best, between 0.011 g (Amphipoda) and 0.066 g (Cladocera). For wetland types and experimental designs similar to those of the MERP and Hanson studies, placing activity traps in more than 5-12 wetlands has little effect on increasing the precision of estimates of the sampled biomass of most taxa. Similarly, for designs similar to the Euliss study, placing sediment samplers in more than 8-13 wetlands or 11-14 transects per wetland does little to increase the precision of estimates of the sampled biomass of most taxa.

4.9 Summary

The species composition of invertebrate communities, and to a lesser degree their species richness, demonstrates diagnostic responses to changes in prairie wetland salinity, water regime, and sedimentation/turbidity (Table 3). Invertebrates also respond sensitively to changing vegetative cover, nutrient levels, and presence of some contaminants, but existing information is too limited and confounding effects are too prevalent to currently allow widespread use of invertebrates to diagnose impairment of prairie wetlands from these stressors. Even for the better-known responses, few thresholds have been documented consistently, and the ability to use invertebrates to distinguish natural from anthropogenic levels of stressors is currently limited.

Invertebrate communities are being monitored with increasing frequency in prairie wetlands partly because of their recognized importance as food for waterbirds. Invertebrates that appear to be sensitive to the widest variety of stressors include amphipods, mayflies, clam shrimp, and fairy shrimp. Because of their high dispersal abilities and reproductive capacity, prairie wetland invertebrate communities appear to recover quickly (within weeks or months) from the direct effects of acute nonpersistent stressors. Because of this, they are poor temporal integrators of prairie wetland condition, unless the expense of frequent sampling is acceptable, or a systematic analysis of decay-resistant remains found in sediments is implemented. Results of such an analysis of decay-resistant remains can help establish "reference conditions" for development of regional water quality standards, but further information is first required on the tolerance thresholds of the taxa most commonly found decay-resistantly.

Individual prairie wetlands that are semipermanently flooded generally contain about 20-40 invertebrate families, at densities of 1-20,000 organisms/m2. Estimates of species composition, richness, and density are strongly influenced by the type of sampling gear and by sampling design. Several studies have quantified the interwetland and interannual variability of invertebrate communities in prairie wetlands. Variability spanning several orders of magnitude is often strongly linked to long-term wet-dry cycles and associated vegetation changes in individual wetlands.

Additional research is needed to document invertebrate response thresholds to all stressors, but particularly to sedimentation and water level change. Before biocriteria can be fully developed, information is also needed on the potential loss or gain of information resulting from various levels of specimen identification and use of various sampling protocols.

Table 3. Summary Eval uations of Possible Invertebrate Indicators of Stressors in Prairie Wetlands.

Evaluations are based on technical considerations, not cost or practicality. A rating of FAIR or POOR is assigned when too few data (FD) suggest potential as an indicator, or when confounding effects (CE) of other variables often overshadow those of the listed stressor, with regard to effects on the indicator.

Stressors Possible Indicators Evaluation
Hydrologic stressors Species Composition
Richness
Density, Biomass
GOOD
GOOD
FAIR (CE)
Changes in vegetative cover conditions Species Composition
Richness
Density, Biomass
GOOD
FAIR (CE)
GOOD
Salinity Species Composition
Richness
Density, Biomass
GOOD
FAIR (CE)
POOR
Sedimentation & turbidity Species Composition
Richness
Density, Biomass
FAIR (FD)
FAIR (FD)
POOR
Excessive nutrients & anoxia Species Composition
Richness
Density, Biomass
GOOD
POOR (CE)
FAIR (CE)
Herbicides Species Composition
Richness
Density, Biomass
FAIR (FD)
FAIR (FD)
FAIR (FD)
Insecticides Species Composition
Richness
Density, Biomass
GOOD
FAIR (FD)
GOOD
Heavy Metals Species Composition
Richness
Density, Biomass
FAIR (CE)
POOR (FD)
POOR (FD)

1. 1     See Appendix 12 for full description of the monitoring design and data structure.


Jump to main content.