Ten Towns Committee
Macroinvertebrate Sampling - 2000 - (Original)
This page contains two documents:
Protocol for Macroinvertebrate Sampling

Leland W. Pollock, Ph.D.
Department of Biology
Drew University


During the 1990's, an annual survey of macroinvertebrate communities inhabiting the 5 streams flowing into the Great Swamp National Wildlife Refuge was conducted by the U.S. Department of Agriculture in cooperation with the Morris County Soil Conservation District and Somerset-Union Soil Conservation District (SCS). The attached Figure 1, from a recent SCS report, shows their 17 sampling locations. Survey results form an annual monitor of water quality in these streams through the calculation of a pollution tolerance metric, Beck’s Biotic Index, for each site. Although this tool was devised for use in Florida, its comparative simplicity, classifying organisms in rough groupings according to previously determined group-sensitivity to pollutants, has made Beck’s Biotic Index useful in other localities. Unfortunately, the agencies initially involved in this work are unable to continue the survey beyond the 1999 season.

As an instructor in aquatic ecology at Drew University, I accompanied the Soil Conservation Service survey team to some of their collection sites on 2 occasions in the past to observe their techniques in order to produce a comparable laboratory exercise for my students in Biol 186/Freshwater Ecology at Drew. As a result, in 1994, 1997, and 1998, my classes have conducted parallel surveys at 13 of the SCS’s 17 sites during the fall season when the class meets (as opposed to the early summer SCS sampling period). We have used a somewhat more quantitative approach (following the EPA’s Rapid Bioassessment Protocols for Use in Streams and Rivers) by collecting composite samples at each site comprised of >100 individuals, isolating animals in the field but returning them to the laboratory for identification at the dissecting microscope level. I have the students sort their preserved samples as best they can, but I check and/or make all final identifications to insure consistency in interpretations. The advantage of this technique is the generation of data which can be used to calculate additional water quality metrics. Our studies have generated a composite reference collection of 118 species from the 5 streams. Results are beginning to provide us with enough “history” at these sites to begin to distinguish important changes in community from more routine “noise”.

We have analyzed the data through the construction of a linked spreadsheet, which, as data are entered, calculates the Beck’s Biotic Index, the Index of Biological Integrity (based on 13 biological/ecological characteristics of each community), and the New Jersey Index of Tolerance (another multi-component metric). An independent study student at Drew this year is using a suite of environmental data gathered at each site to attempt to identify principal parameters which correlate with differences observed in community structure. This spring, a Drew student intern with the Great Swamp Watershed Association will attempt to find similar correlations between data from the SCS stream studies and associated NJDOT water analyses.

While we plan to continue our student studies during the fall, it is important to maintain the continuity of the base which the SCS team has created in early summer. Seasonal differences in community structure and environmental conditions require that samples be done under similar conditions to remain comparable. I would like to provide that continuity through the study proposed here.


In early summer, 2000 and 2001, I propose to study the same 17 sampling locations used by SCS (Figure 1) and to follow protocols we use in our fall field studies (which I feel match or exceed those currently employed by SCS–not intended as a criticism of their procedures). At each collection site, we will make the following environmental observations: stream width & mean depth, mean current flow - discharge volume, dissolved oxygen, pH, temperature, turbidity, and fill out an EPA Habitat Assessment Field Data Sheet.

The longer the run of stream monitoring data the greater its value in distinguishing between routine variability and more significant problem-driven change. Continuity between the past SCS studies and the proposed series can be maintained by integrating their data into my data storage and analytic protocols, and by determining the comparative quantitative relationship between their observations and mine.

An annual survey of results will be delivered to the Ten Towns Committee on a schedule and in a format to be agreed upon. Although I hope that this program would be an on-going source of data through extension beyond the two years proposed here, I would like to use experience during this initial 2-year period to shape any necessary adjustments in procedures and allocations for subsequent years.

Project Director’s Qualifications

I am in my 26th year teaching aquatic ecology at Drew. I have an extensive background in the identification of invertebrates in general, and marine ones in particular, having recently published, A Practical Guide to the Marine Animals of Northeastern North America (Rutgers University Press, 1998). My published primary research deals with the systematics and ecology of the microscopic members of a phylum known as Tardigrada or “water bears” which live in marine beaches. These interests equip me well to work with the identification of both macroscopic and microscope animals.

Within the past 1˝ years, my interest in stream benthic communities has blossomed through my involvement with your Volunteer Stream Monitoring Teams, especially helping to instruct these interested citizens in the sampling and identification of stream invertebrates. I have recently produced an internet-based Pictorial Key to the Families of Mayflies of Northern New Jersey, which I intend to be the first of a series of such identification aids for students and volunteer citizen monitors.


A Report to the Ten Towns Great Swamp Management Committee

The Condition of the Great Swamp Streams
Summer, 2000

Leland W. Pollock, Ph.D.
Department of Biology
Drew University
October 25, 2000

1. Introduction

The types and numbers of macroinvertebrates (mostly insect larvae/nymphs) that form the biological community at a particular stream location are shaped by the composite environmental conditions flowing by the site during the recent past. As the drainage focus for the upstream landscape, in-stream conditions are a direct reflection of the degree of environmental stress in the surrounding area. Therefore, a strategically placed collection of macroinvertebrate samples can provide a relatively easy method for evaluating water quality of the entire watershed and for pin-pointing specific problem areas within it.

From 1992 through 1999, the National Resources and Conservation Service (NRCS) of the Morris County Soil Conservation Service conducted an annual, early-summer survey of macroinvertebrate communities at 17 sites scattered throughout the five major streams traversing the Great Swamp Watershed. Following the expiration of NJ DOT funding support for this project, the NRCS decided to terminate their studies with their 1999 survey. I have been engaged by the Ten Towns Great Swamp Management Committee to continue this valuable data series. This report summarizes results from the June, 2000 sampling period.

2. Sampling Methods:

Standard field and laboratory methods, adapted from the Environmental Protection Agency’s Rapid Bioassessment Protocols (EPA, 1999), were applied to the 17 sample locations shown in Figure 1. These sites match those used in previous NRCS surveys with a few adjustments in location designed to improve sample comparability. I chose riffle habitats (i.e., turbulent water flow over stretches of shallow, cobble substrate) for this survey because such habitats are generally acknowledged to support the greatest community richness. Following field exploration and discussion with the original NRCS sampling team, we agreed that one previously used sample site (GB1) should be abandoned. This location, south of the Pleasant Plains bridge within the Refuge itself, consists of deeper water flowing over fine particulate substrate. With no riffle habitat available, the biotic community here can not include the riffle-dependent organisms upon which all other comparisons are made and thus, this site is not directly comparable to the other sites. A new site was added on Indian Grave Brook (IG1) just above the confluence of this tributary with the Passaic River. Including this site extends the survey to a new, minimally disturbed portion of the watershed. Five other sites were shifted (by no more than 100 yards) from their previous locations, either to utilize substrates more comparable to the remaining sites (at BB1, LB4, and GB4) or to provide a reasonable distance to permit community recovery below the significant sources of disturbance created by dams/impoundments (at LB2 and PR1). At each field site, using a Surber sampler (1 foot square sampling frame, 500 µm mesh), my assistant and I collected and preserved a composite of three independent samples representative of the range of water flow conditions present within the riffle habitat.

In the laboratory, we used a random-numbers technique to isolate from the original composite a subsample including at least 200 individuals, which then were identified and enumerated. Karr & Chu (1999) strongly criticize the accuracy of community assessments based on the 100 individual standard suggested by the EPA protocol (EPA, 1999). While they encourage a >400 individual sample, I chose to use a >200 individual sample size as a logistical compromise. This provides more comprehensive assessments of communities than do the (often considerably) less than100 individual samples used previously. The remaining composite then was surveyed, and single individuals representing species not included in the 200+ individual-subsample were added to it. This step never resulted in the addition of more than 6 individuals and permitted us to note the presence of potentially important indicator species in the sample. Many stream macroinvertebrates are small in size. Often closely related genera are difficult to distinguish in the field. Consequently, in this study, all specimens were preserved in the field and later identified in the laboratory to the lowest taxonomic level possible, usually to the genus or species level using dissecting and compound microscopy.

Differences in field and laboratory techniques used here are likely to influence the direct comparability of data between this survey and the NRCS surveys that preceded it. Accordingly, during the same period in June, 1999 that the NRCS was conducting their annual survey, I collected a set of samples at each of the 5 streams at the sites closest to the border of the Refuge (i.e., BB1, LB1, GB2, PB1, PR1).

To add a physical-chemical-habitat context to this survey, on June 20, 2000, we monitored environmental conditions at all 17 sites during a 7 hour period. Conducting our survey of environmental factors in this fashion provides a better comprehensive comparison of site conditions “at one moment in time” than would conducting individual site measurements at the time of specimen collection, spread over three weeks of widely variable conditions. At each site, we recorded metered-readings of dissolved oxygen (Hanna Instruments Model HI91410), temperature and pH (Hanna Instruments Model H9024C), total dissolved substances (Oakton Model WD-35661-57), turbidity (Hanna Instruments Model HI93703), current speed (Swoffer Model 2100) and water depth at 3 evenly spaced locations forming a transect across the stream. The latter two factors were combined to calculate the stream discharge rate at that location. We also filled out a standard Habitat Assessment Field Data Sheet for High Gradient Streams (EPA, 1999), which included the coding of 10 habitat parameters, each on a scale of 0 (low) to 20 (high), guided by photographs and descriptions (EPA, 1999). These parameters include diversity of epifaunal substrate, available cover, degree of embeddedness of stones, velocity/depth regimes, extent of sediment deposition, status of channel flow, degree of channel alteration, frequency of riffles, stability of banks, extent of vegetation cover, and width of riparian zones.

3. Metrics and Statistical Treatments.

Metrics are quantitative representations of single (simple metric) or combined (multi-metric) characteristics of a sampled biological community. To match this survey’s results with the metric used in past studies, the Beck Biotic Index, was calculated to characterize the biotic community found at each site. Based on family-level identifications, three categories of macroinvertebrate families were tallied using criteria described in the SCS Water Quality Indicators Guide: Surface Waters. Class 1 organisms are “sensitive or intolerant of pollution”; Class 2 organisms are “facultative” and thus can tolerate varied conditions, perhaps including moderate levels of pollution; Class 3 organisms are “pollution tolerant”. An undisturbed community should be comprised mostly of Class 1 and perhaps a few Class 2 or 3 species. Settings heavily impacted by organic pollution will be dominated by Class 3 animals. Locales in which Class 2 organisms are the most abundant reflect at least some significant source of stress, either natural or human caused. Beck Index for a particular site was calculated by doubling the number of Class 1 families found there and adding to that the number of Class 2 families. Class 3 families are not included in the calculation.

The formula for Beck Index is:
       BI = 2n1 + n2
       BI = Beck Index
       n1 = number of Class 1 families identified
       n2 = number of Class 2 families identified

Beck Index values calculated at the family level correspond to the following water quality assessments.

Beck Index Water Quality Assessment
0 Grossly polluted
1-5 Moderately polluted
6-9 Clean but limited habitat quality
10 or higher Clean

Although it has the advantage of being easy to calculate, Beck Index also has potential drawbacks. It was developed to help categorize macroinvertebrates in Florida streams and to especially to determine the impact of organic pollution on their community composition. There are other, more widely employed metrics that may be even more useful for our local circumstances.

The Benthic Index of Biological Integrity (B-IBI) is one such multi-metric. It combines several distinctive, stress-influenced community characteristics into a single aggregate value that can be used to compare the level of stress evidenced by communities from different stream localities. This Index is based on comparison with community characteristics found at regional “reference” sites that show minimal levels of environmental stress. These characteristics might include presence of specific indicator species or pollution intolerant groups, richness of genera, dominance by the most common species, etc. A B-IBI metric can be tailored to a particular setting by selecting for inclusion in the measure those community characteristics which correlate most closely with a sequence of sampling sites arrayed by personal observation along a known gradient from least to most disturbed (Karr & Chung, 1999). For this study, data gathered by Drew students and faculty during fall, 1999 were used to place 13 of the 17 sampling sites treated in the current study into a sequence reflecting increasing stress: (least stressed) PR3, PB2, PB3, PB1, GB3, PR1, PR2, GB4, GB2, LB2, LB1, LB3, BB1 (most stressed). Thirty-seven associated community characteristics were plotted following this array, and those showing strongest correlation with the stress gradient were selected for inclusion in a refined, Great Swamp version of the B-IBI. The resulting nine community characteristics that best reflect a gradient of low-to-high stressful conditions at Great Swamp Watershed sites are shown in Table 1.

The range of potential values for each of these characteristics is divided into 3 sub-ranges representing values expected from least stressed (“reference” sites), intermediate, and most stressed communities respectively. Then, depending on the range into which a specific characteristic at a particular site falls, it is assigned a score of 5, 3, or 1 respectively. The B-IBI value is the sum of these scores with a maximal (least stressed) potential of 9 x 5 = 45 and a minimal value (most stressed) or 9 x 1 = 9. B-IBI values were calculated for each site in this study.

Viewing metrics from one collection period in the context of other comparable observations can help to distinguish “normal” variations and overall trends of change. Below in this study, results from recent NRCS surveys (1998 and 1999) are placed in relation to their total data set (early summers, 1992-1999). Also, results obtained in the present survey are compared to comparable surveys conducted by Drew faculty and students at 13 of these 17 sites (fall, 1994, 1997-1999). This technique can also provide warning when the conditions of a particular site approach or fall below previous minimal values.

The environmental correlates of these groups of sites were explored using the multi-variate statistical technique, canonical correspondence analysis (CCA). In this procedure, site data for species (i.e., Table 2) is entered as the main file and site environmental data (i.e., Table 3) is entered as the secondary file. A full explanation of this procedure can be found in the works of Ter Braak (1986, 1987). A detailed accounting here surpasses our need. The resulting 2-dimensional plot arrays all species in best-fit positions relative to the environmental variables present. Then positions of all species included in a particular sampling site form a site-specific “constellation” of points within the graph space, and a symbol for each site is located at the centroid of its particular species cluster. This places all of the sample sites in relative position to the major environmental gradients as well.

4. Results and Discussion:

A total of 3512 individuals including 122 distinct entities (either species or genera) were identified among the 17 sites in this survey. Table 2 lists the count data.

Figure 2 shows family-level Beck Index values derived from parallel surveys conducted at 5 stream sites during the June, 1999 by the NRCS team using their standard techniques and by me using techniques described above. While my techniques made little difference in the most stressed locality where comparatively few macroinvertebrate types occur (BB1), the differences were more significant in the other settings, especially at LB1, GB2, and PB1.


Similar patterns can be seen in Figure 3 in which June, 1999 Beck Index data (from NRCS) are compared to June, 2000 Beck Index data from this study. Higher Beck Index values in 2000 at LB2 and GB4 probably in part reflect adjustments in those sample site locations made to find more comparable substrates or to avoid proximity to an obvious disturbance. But the substantial differences in Beck values found in Passaic River and Primrose Brook sites probably reflect a more thorough characterization of the communities present at those species-rich locales rather than any substantive improvement or change in actual field conditions between 1999 and 2000. The Indian Grave Brook (IG1) site was newly added during this study.


A comparison between traditional Beck Index and the new B-IBI values for June 2000 data is shown in Figure 4. The fact that the patterns of the two metrics are quite similar suggests that the Beck Index used in the past works quite well to characterize the Great Swamp streams, even if absolute values for less stressed communities may have been underestimated in the past. Because the B-IBI is based on more ecologically sound and locally tailored characteristics, it represents a more convincing metric for use in subsequent comparisons.


The last two sets of NRCS observations, for 1998 and 1999, are shown in Figure 5 in relation to the range of conditions recorded at each site during the past eight years. As their 1999 report to the Ten Towns Committee notes (NRCS, 1999), Beck Index values at many sites were lower in 1999 than in 1998, and are, in fact, near the bottom of all previous measures. They attributed this to unusual drought conditions preceding and during their observation period. The dramatic 1999 rise in Beck Index at the GB3 site is difficult to understand since that site has been heavily impacted by the recent dredging/dam repair project at Silver Lake immediately upstream.


Comparisons between this summer’s observations (2000) and those made by Drew students last fall (1999) appear as Figure 6, for Beck Index,


and Figure 7, for B-IBI. In both cases, values are seen within the context of maximal/minimal ranges observed at each site during falls, 1994, 1997-1999 and can be seen to show very similar trends. Matching the observation made by the NRCS team earlier in 1999, in many cases, fall 1999 values were at or near minimal values recorded so far. Drought conditions during late 1998 and 1999 are the likely cause. Encouragingly, in all but 2 cases, summer 2000 data showed modest to considerable improvement. Using B-IBI data in Figure 7, only 2 locations showed no change in the 1999-2000 period, while two others declined in quality. The Primrose Brook location (PB1) immediately downstream from the recent Blue Mill Road bridge replacement project was understandably impacted by this activity. Recovery of this normally rich community should be rapid. For reasons that are not clear, the site closest to the Refuge boundary on Loantaka Brook, LB1, showed slight decline in the 1999-2000 period. In Great Brook, the site immediately below the recent Silver Lake dredging and dam restoration project remains at the bottom end of its normal range. Recovery from heavy silting from the same project at the next downstream site (GB2) is encouraging. At least some of the considerable mussel population at the GB2 site appear to have survived this disturbance. The Passaic River location (PR2) just downstream from I-287 also remained at the low end of its range despite improvement at the sites above and below it.

Environmental variables recorded during our survey of June 20, 2000 are displayed in Table 3. Combinations of physico-chemical conditions distinguish the array of 17 sampling sites studied here into logical groupings. Stressed sites are generally matched by more extreme environmental parameters. However, in many cases, the range in conditions observed is surprisingly narrow. Temperatures showed a mean of 20.9 C and a range of only 18.8 C (at LB1 and GB2, surprising since both are somewhat downstream from exposed, slow-water impoundments) to 23.8 C (at PR1, just below Osbourne Pond). Dissolved oxygen is often a key variable for macroinvertebrates. But in our streams, a range of 7.28 mg/L (at relatively sluggish, warm LB4) to 9.26 mg/L (at GB5, just below a shallow, eutrophic impoundment) surrounded a mean of 8.31 mg/L – all levels considered to be well above stressful for most organisms. The variation in pH was also modest, 7.34 units (at BB1, downstream from a golf course) to 8.68 units (again at GB5, probably reflecting the same eutrophic productivity in the impoundment immediately upstream). Larger variations were observed in Total Dissolved Substances (TDS) and turbidity. TDS ranged from 93.2 units (at relatively pristine PB3) to 790 units (at LB4 in a channelized reach next to the Morris Township swimming pool). Mean TDS of 341.8 units separated our most stressed streams (Black Brook, Loantaka Brook, and upper Great Brook, all above the mean) from the remainder which are less stressed. Turbidity was lowest, 0.62 FTU, at the PR2 site, perhaps surprisingly from its location just downstream to I-287. On the other hand, current speeds at PR2 tend to favor sandier deposits rather than fine, suspendable silt. Turbidity was highest, 10.6 FTU, at stressed LB1 (only slightly above 10.43 FTU at BB2, just below the Chatham Township sewage treatment plant). Again, the mean of 4.86 FTU separates the more stressed sites (higher) from those less stressed, with the interesting exceptions of LB3, just below the Morris Township sewage treatment plant, with 2.98 FTU much lower than LB4 or LB2; and PR1, at oddly high 6.47 FTU, probably related to its position just below Osbourne Pond.

The results of canonical correspondence analysis are shown in Figures 8 and 9. Figure 8 shows the relationship of species observed to important environmental gradients, including dissolved oxygen, total dissolved substances (TDS), temperature, turbidity, and pH. Note that in several instances groups of species occurred at virtually the same spot. To make the graph more readable, in those cases, the term “group” followed by a group number is shown on the graph, and its constituent members are listed to the lower left. This figure can help to determine which species make useful indicators of particular environmental conditions. Look in the direction of arrow heads representing increasing values or arrow tails showing decreasing values for particular variables. Species found toward those extremes are best choices as indicators. Note that frequently species beginning with “Dip” (standing for Order Diptera, Family Chironomidae) occur there, with different “Dip” species near each arrow head and tail. This demonstrates the potential value of including species-level detail for this difficult group.

Figure 9 indicates the placement stream sites within the gradients of environmental factors. Despite modest absolute ranges in dissolved oxygen and temperature, both species and site distributions are scattered most strongly along these gradients (top right to lower left). This includes the more pristine sites (IG1, PR3, PB3, PB2) which are found toward the upper right. Higher total dissolved solids (TDS) levels (toward the upper left) characterize upper Loantaka Brook sites, while the position of LB1, GB5, and GB3 reflects higher turbidity at those sites. , while more stressed sites appear toward the periphery of several environmental gradient arrows suggesting more stressed conditions - Higher pH, Higher Turbidity (GB5, LB1); Lower Dissolved Oxygen, Higher Temperatures (BG5, BB2, LB4); Higher TDS (LB3). LB3 (below the Morris Township sewage treatment plant), LB4 (next to the Township’s swimming pool), and BB2 (next to Chatham Township’s sewage treatment plant) appear to be the sites located farthest along the lines of increasing stress. The position of BB1 in this figure is something of a mystery in that it appears in the higher quality quadrant of the diagram and yet is obvious as a site under considerable stress.

Finally, on the basis of my several years observing the five streams within the Watershed, I prepared first rough draft of what appears to me to be the most likely sources of environmental stress at each stream site. In Table 4, I listed 6 most likely sources of stress for our region and indicated for each site the relative likelihood of significant impact from that source (++ high likelihood, + likelihood, 0 average or mild impact, - unlikely to be a significant source of stress). These are subjective speculations, based on observations but not on direct, quantitative assessments. In several cases, the most likely, but again speculative, source of the problem is assigned. For each site, the sum of all “+”s was calculated. This analysis reinforces the idea that Black Brook and the upper portion of Loantaka Brook face the strongest and most varied sources of stress, and Indian Grave, Primrose, and upper parts of the Passaic Rivers suffer least. Clearly however I must e missing some important stress(es) impacting the LB1 location.

5. Stream Summaries:

All in all our stream communities reflect considerable improvement over their condition during the stressful 1999 drought period. Our more pristine areas, including Indian Grave Brook, Primrose Brook, and the upper Passaic River, are at all-time highs, for the most part. Compared to more stressed streams, such as Black Brook, Loantaka Brook, and much of Great Brook, these stream sites tend be wider, with greater riffle zone frequency, higher in dissolved oxygen, the variety of habitat options, and the extent and condition of the riparian zone, but lower in total dissolved substances and turbidity.

Indian Grave Brook: As a new site in this survey, Indian Grave Brook serves as a “reference” stream, hosting the highest quality biological community in the area.

Primrose Brook: The upper Primrose Brook sites that appear to have recovered from the severe silting incident they suffered in 1997, although PB1 has dropped some in quality coincident with the summer, 2000 Blue Mill Road bridge replacement project immediately above this site. While this comparatively pristine stream appears to be quite resilient to insult, it is worth noting that we have challenged all three of its sampling sites strongly with siltation stress during the past 5 years.

Passaic River: Our PR3 site in the upper Passaic River provides a lovely example of clean-water macroinvertebrate community structure. However, the lower Passaic River sites appear to struggle from obvious sources of temperature, channelization, and sedimentation stress, probably in association with I-287 at PR2 and with Osbourne Pond just above PR1.

More disturbing is the fact that some of our already embattled localities appear to be losing ground, with BB1, LB1, LB4, the lower (swampward) half of Great Brook, and I287-impacted PR2 at or near the lower end of their 6 year range.

Great Brook: Lower Great Brook either is (GB2) or hopefully is about to begin (GB3) recovering from sedimentation resulting from the recent Silver Lake dredging project. But all of Great Brook in general appears to suffer from excessive sedimentation and poor riparian protection, and high total dissolved substances mark upper Great Brook sites as well.

Loantaka Brook: Loantaka Brook 4 and 2 may be edging upward, although high sediments, turbidity, and total dissolved substances, and a lack of riffle habitat appear to limit recovery prospects there. The biotic community in the lower Loantaka Brook area (LB1) suggests that conditions are not good there, although it is not clear which environmental variables are primarily responsible.

Black Brook: Characterization of this stream has been limited to small, slow moving, channelized tributaries in this and previous surveys. Variable water levels, high total dissolved substances and turbidity, and poor macroinvertebrate habitat availability stress these communities. Most of the more substantial flow of Black Brook lies within the Refuge itself and thus is beyond the boundaries for this survey. While its several tributaries eventually coalesce into a substantial stream, its generally low gradient results in slow velocity and a silty bottom in which riffle habitat is not found.

6. The Future: Group Membership & Linear Discriminant Analysis

Too late for thorough incorporation into this report, I have become aware of a potentially useful approach to address the important question of when is a status change at a particular site significant. The Department of Environmental Protection for the state of Maine (Davies & Tsomides, 1997, Davies et al., 1995, 1999) assigns each of their sampling sites to one of four groups which have water quality characteristics defined legislatively. For each group, this produces a collection of similar macroinvertebrate communities. Using the statistical procedure of linear discriminant analysis, they have been able to quantify group-specific ranges of conditions pertaining to the same types of community structure characteristics as we used above to construct a local B-IBI. These group limits are built into a model against which subsequently sampled communities (or sample sites) can be tested for fit. Similarly, it is possible to detect when repeat sampling of a particular community no longer qualifies under the limits for the group it previously belonged to. This becomes a powerful, objective way to identify sites slipping in quality.


To explore the applicability of this technique for the Great Swamp sites, I have used the multi-variate statistical procedure, detrended correspondence analysis (DECORANA), to identify natural groupings of sites that display similar biological community structure. Data documenting the species composition of every site was entered and this analysis compared similarities and variations among all of their compositions. The sites are then arrayed along an axis that reflects the relative position of each site with regard to the compositional feature identified as most responsible for the variance observed in species composition among the sites. In DECORANA, the cause of this source of variance is not determined, only its effect on resultant community structure. Then, with this primary source of variance subtracted from the array of sites, the second most significant source of variance is found, with sites arrayed appropriately along a second axis. The product of DECORANA is Figure 10 that places each site (= each community) within the two-dimensional space represented by these axes, relative in position to one another. This technique was used to identify three groups of sites with similar composition, presumably experiencing similar environmental conditions and stresses.

Group 1 sites (IG1, PR3, PB1, PB2, PB3) share the characteristic of being most pristine, least disturbed or stressed. Group 3 sites (BB1, BB2, LB3, LB4, GB5) lie at the opposite end of the spectrum, displaying significantly stressed community structure. Group 2 sites (PR1, PR2, GB2, GB3, GB4, LB1, LB2) show intermediate features. The DECORANA groupings matched my subjectively derived stress-gradient sequence of sites used above to construct the B-IBI metric.


Figure 11 shows mean B-IBI values for each of the three Groups for each year in which there is Drew-derived data present. Once again the patterns seem clear and supportive. Using stress-source data in Table 4, the mean score for the sites included in Group 1 above was 6.4; for Group 2 sites was 4.4; and for Group 3 sites was 1.6. While perhaps less than fully convincing as evidence, this matching distinction on the basis of likely degree of stress is at least reassuring. And finally, mean values for each of the 9 community characters used to create the B-IBI measure for the Great Swamp follow clear group-related patterns, again suggesting that these three Groups appear to represent entities distinguishable by community composition as well as by environmental factors.

It remains to use linear discriminant analysis to construct a working model for our watershed sites. But if this is possible to accomplish, I believe that it may provide a useful tool for objectively evaluating subsequently collected samples in the future within the watershed.

8. Literature Cited

  • Barbour, M.T., J.Gerritsen, B.D.Snyder, and J.B.Stribling. 1999. Rapid Bioassessment Protocols for Use in Streams and Wadeable Rivers: Periphyton, Benthic Macroinvertebrates and Fish, Second Edition. (EPA 841-B-99-002). U.S. Environmental Protection Agency; Office of Water; Washington, D.C.
  • Davies, S.P. and L.Tsomides. 1997. Methods for Biological Sampling and Analysis of Maine’s Inland Waters. Maine Department of Environmental Protection, Bureau of Land and Water Quality, Division of Environmental Assessment, Augusta, Maine.(DEP-LW107-A97). 29 pp.
  • Davies, S.P., L.Tsomides, D.L.Courtemanch, and F.Drummond. 1995. Maine Biological Monitoring and Biocriteria Development Program. Maine Department of Environmental Protection, Bureau of Land and Water Quality, Division of Environmental Assessment, Augusta, Maine. 61 pp.
  • Davies, S.P., L.Tsomides, J.L.DiFranco and D.L.Courtemanch. 1999. Biomonitoring Retrospective: Fifteen Year Summary for Maine Rivers and Streams. Division of Environmental Assessment, Bureau of Land and Water Quality, Augusta, Me. (Maine DEPLW 1999-26). 190 pp.
  • Karr, J.R. and E.W.Chu. 1999. Restoring Life in Running Waters: Better Biological Monitoring. Island Press, Washington. 206 pp.
  • NRCS. 1999. 1999 Water Quality Inventory: Macroinvertebrates of the Great Swamp Tributaries. USDA Natural Resources Conservation Service, Somerset, NJ. 13 pp.
  • Ter Braak, C.J.F. 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology, 67:1167-1179.
  • Ter Braak, C.J.F. 1987. The analysis of vegetation-environment relationships by canonical correspondence analysis. Vegetatio 69:69-77.

9. Acknowledgements

I wish to express my appreciation of the Ten Towns Great Swamp Watershed Management Committee’s understanding of the value of maintaining this on-going data base that uses macroinvertebrate communities to document water quality conditions throughout the watershed. In particularm, I am grateful for their funding of this particular study. In addition I acknowledge the generous, in-kind support of my home institution, Drew University, and faculty colleagues who have provided encouragement as well as the equipment and facilities necessary to the project’s completion. And finally, I am especially indebted to Kristine Joppe-Mercure, my indefatigible Drew student assistant during this project, for being such a capable “quick study” in adopting the techniques used here.

Table 4. Potential Sources of Stress in Great Swamp Streams.

Monitoring Station Organic/ Nutrient Sediments Point- Source Toxins Non-point Source Toxins S/low Water Flow High Temps Score
BB1 ++ glf - + glf? ++ pav, pest? ++ + imp 8
BB2 ++ stp + + stp? + pav 0 + stp 6
LB1 + imp, liv + - + pav 0 - 3
LB2 ++ stp,imp + - + pav, pest? 0 + imp 5
LB3 ++ stp ++hab + stp? 0 + + stp 7
LB4 0 + + pool? ++ pav, hou ++ + opn 7
GB2 ++ liv ++ dred - - - - 4
GB3 + ++dred - + pav 0 + imp 4
GB4 - ++hab - ++ pav ++ 0 6
GB5 ++ imp - - + pav 0 + imp 4
PB1 + + - + - - 3
PB2 - + - + - - 2
PB3 + 0 - + - + imp 3
PR1 + + - + - + imp 4
PR2 - ++ - ++ pav - + imp 5
PR3 - - - - - - 0
IG1 - - - - - - 0

dred = pond dredging hou = high density housing pav = pavement
glf = golf course imp = impoundment pest = pesticides
hab = poor habitat liv = livestock stp = sewage treatment plant

Return to text

Table 1.  Components of the Great Swamp B-IBI (Benthic Index of Biological Integrity)

B-IBI Component Explanation Response to Stress
EPTspecies Number of species of intolerant Ephemeroptera (mayflies) + Plecoptera (stoneflies) + Trichoptera (caddisflies) declines
#Eph Total individuals of Ephemeroptera declines
#Plec Total individuals of Plecoptera declines
#Trich Total individuals of Trichoptera declines
IndSpp Number of indicator species present from: Caenidae, Ephemerellidae, Heptageniidae, Leuctridae, and the genus, Dolophilodes declines
PerlAb Total individuals of the plecopteran family Perlidae declines
DOM Proportion of total individuals included in the two most abundant species increases
TAXA Number of individual taxa (genera or species) declines
Pamph Proportion of total individual included in the crustacean order Amphipoda (“scud”) increases

Return to text

Printable Version
Please send comments to the webmaster from this or any other page.
Macroinvertebrate Sampling Study - 2000 - Original Format