Professional Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Society for Freshwater Science is collaborating with JSTOR to digitize, preserve and extend access to Journal
of the North American Benthological Society.
http://www.jstor.org
Benthic macroinvertebrates are often collect- stream size, and season. The EPT taxa richness
ed to help evaluate water quality and/or habitat criteria have been in use for many years, but
quality. This task involves collection of repre- we have only recently derived biotic index cri-
sentative samples, accurate taxonomy, and some teria. This paper describes the method used to
system to convert invertebrate data into water- produce the North Carolina Biotic Index (NCBI)
quality ratings. North American stream ecolo- and gives tolerance values for taxa in south-
gists have struggled to establish such ratings eastern USA. The emphasis was on the objective
for over 50 years, and a bewildering array of derivation of tolerance values, because other
methods have been proposed (Cairns and Pratt investigators have already demonstrated the
1993). Currently, more than one type of data utility and accuracy of a biotic index (Chutter
summary (metric) is often used to summarize 1972, Hilsenhoff 1982, 1987, 1988, Narf et al.
invertebrate data, with a final evaluation of wa- 1984, Jones et al. 1981).
ter quality based on several independent meth- The first North Carolina biotic index was es-
ods. Many of the proposed metrics, however, tablished in 1980, using "expert opinions" to
are applicable only to a particular type of col- assign tolerance values for each taxon. Trial val-
lection, a single habitat, or a single geographic ues were assigned by NC DEM biologists, and
area. adjusted following discussion at a meeting of
The most widely used metrics are taxa rich- benthic ecologists from North Carolina, South
ness and "biotic indices". Taxa richness is as- Carolina and Georgia. This index (with poten-
sumed to be inversely related to the degree of tial values ranging from 0-5) was modeled on
stress, whereas biotic indices attempt to sum- that of Hilsenhoff (1977), which was, in turn,
marize information on the tolerance of the mac- derived from Chutter's (1972) index:
roinvertebrate community. The North Carolina Sum TV,N,
(NC) Division of Environmental Management Total N
(DEM) uses taxa richness of the most intolerant
invertebrate groups (Ephemeroptera + Plecop- where TV, is the tolerance value of the ith taxa,
tera + Trichoptera, or "EPT")and a biotic index N, is the abundance of the ith taxa, and Total
similar to that of Hilsenhoff (1987). Both metrics N is the number of individuals in the sample.
are weighted equally in assigning water-quality Total N may be based on either actual densities
ratings (NC DEM 1992). We have invested sub- (No./m2, etc.) or abundance categories (see
stantial efforts in testing and evaluating both Methods).
methods, both in deriving criteria and in mak- Ideally, such an index would be used to target
ing adjustments for the effects of ecoregions, specific kinds of stress. For example, a species
279
might be sensitive to metal toxicity, but tolerant the computer system, respectively, as 1, 3 or 10.
of low dissolved oxygen. Rather than a single Calculations of mean abundance for this study
index, an array of indices would be used to also included zero values, since I used all sam-
assess water quality and/or habitat quality. Un- ples in a particular water-quality category, not
fortunately, this level of information for species just samples where this taxon occurred. Similar
tolerance is not available. Dr. Hilsenhoff tar- abundance categories are used with the Chan-
geted his index at organic pollution, although dler Score, a European biotic index (Chandler
it is used by other investigators to look at many 1970).
other kinds of stress. The NCBI is intended for The primary measurement used to compute
examination of the general level of pollution, the tolerance value for any taxon was the av-
regardless of source. erage abundance (range = 0-10) in each water-
The utility and accuracy of such an index is quality category. I computed how far each taxon
only as good as the system used to derive the (starting from Excellent) extended into the areas
tolerance values. The use of expert opinion val- of poorer water quality. The abundance values
ues proved to be unreliable in North Carolina, (by water-quality category) were converted into
as biotic index ratings frequently conflicted with cumulative percentiles and graphed vs. the wa-
ratings from EPT taxa richness values or ratings ter-quality score (1 = Excellent, 2 = Good, 3 =
based on professional judgment. Unbiased as- Good-Fair, 4 = Fair, 5 = Poor). I tried a 50th
signment of tolerance values requires a large percentile, a 75th percentile, and a 90th per-
number of invertebrate collections that have centile value; the 75th percentile produced the
already received some kind of water-quality rat- greatest separation of intolerant and tolerant
ing. To my knowledge, only Hilsenhoff (1987) species (Table 1), while still leaving the facul-
has used such a method to derive tolerance val- tative species with values near the midrange.
ues. Other lists of invertebrate tolerance in use Calculations of such percentiles usually re-
within the United States are modifications of quired the use of simple linear interpolation
Hilsenhoff's list with "expert opinion" modi- between categories. For example, the mean
fications for taxa not found in Wisconsin. North abundance data (number per sample, by water-
Carolina biologists saw a need for indepen- quality class) for Cricotopusbicinctus in Table 1
dently deriving tolerance values for southeast- is: Excellent (1): 1.0, Good (2): 2.4, Good-Fair
ern species, as well as deriving criteria for biotic (3): 3.1, Fair (4): 4.0, and Poor (5): 3.7. Convert-
index values that could be adjusted for differ- ing these abundance values to a cumulative per-
ences in ecoregion, season, and stream size. centage produces: 7.0, 23.9, 45.8, 73.9, 100.0. This
indicates that the 75th percentile for the water-
Methods quality score lies between Fair (4) and Poor (5)
and yields by interpolation a value of 4.04.
The NC DEM has a data set of >2000 mac- This procedure produced a range of prelim-
roinvertebrate stream samples collected during inary tolerance values generally between 1.0
water-quality surveys between 1983 and 1992. and 4.5, but the desired range (to be comparable
These samples have been assigned to five water- with Hilsenhoff 1987) was 0-10. If the 1-4.5
quality categories (bioclassifications): Poor (n = range is graphed vs. the 0-10 range (with 1 and
299), Fair (n = 392), Good-Fair (n = 481), Good 4.5 being the x and y intercepts), a conversion
(n = 492), and Excellent (n = 424); they include formula for tolerance value (TV) is derived:
both standardized qualitative samples (Lenat
Final TV = 2(1.43 x Preliminary TV - 1.43).
1988) and collections limited to Ephemeroptera,
Plecoptera, and Trichoptera (Eaton and Lenat The 75th percentile technique was best suited
1991). Classification is based on EPT taxa rich- for taxa with substantial amounts of data. If only
ness and an earlier version of the NCBI. a few data points were available, then the 75th
A variety of information is associated with percentile value often was too high. Review of
each sample, including ecoregion, date, and our data base suggested that the 75th percentile
stream size. The abundance information for each was suitable for taxa with at least 25 observa-
taxon is semiquantitative, being tabulated as ei- tions (out of the 2000 samples). For taxa with
ther Rare (1-2/sample), Common (3-9/sample) only 10-24 collections, a 50th percentile was
or Abundant (>10/sample), and it is coded in used, and the old "expert opinion" values were
TABLE1. Average abundance values (range = 0-10) by water-quality rating (five categories) of represen-
tative taxa, and North Carolina tolerance values (TV) calculated from this information. "+" = <0.1, N = No.
of records for taxon.
Group 1: Intolerant
Drunella wayah 0.4 0.1 + + - 51 0.0
Rhithrogenaspp. 0.7 0.2 0.1 - 155 0.4
Chimarraspp. 2.7 2.9 1.4 0.4 + 650 2.8
Micrasema wataga 0.8 0.8 0.5 0.1 + 280 3.2
Goera spp. 0.6 0.2 + + - 172 0.4
Brachycentrusspinae 0.8 0.1 + + + 73 0.0
Pteronarcys dorsata 0.8 0.7 0.1 + 186 1.8
Acroneuriaabnormis 6.2 4.6 1.7 0.4 + 1031 2.2
Mean TV (?1 SD) 1.4 (+1.3)
Group 2: Facultative
Stenonema modestum 6.6 8.1 8.2 6.9 1.1 1660 5.8
Ephemerellacatawba gr. 3.8 1.9 1.7 0.7 0.4 550 4.0
Eurylophellatemporalis 1.6 1.5 1.2 0.6 0.2 518 4.6
Cheumatopsychespp. 5.5 6.9 7.1 6.6 2.5 1680 6.6
Hydropsyche venularis 1.8 2.7 2.4 1.5 0.3 656 5.4
Perlesta placida 1.5 1.9 1.8 1.1 + 524 5.0
Ancyronyx variegata 1.2 2.3 2.8 2.5 1.0 726 7.0
Polypedilum convictum 2.3 3.1 3.0 1.8 0.4 622 5.2
Mean TV (?1 SD) 5.4 (?1.0)
Group 3: Tolerant
Cricotopusbicinctus 1.0 2.4 3.1 4.0 3.7 744 8.8
C. tremulus gr. 0.3 0.6 1.1 1.3 1.4 279 9.0
Chironomusspp. 0.6 1.0 1.5 3.1 5.0 711 9.8
Polypedilum illinoense 1.4 2.8 2.9 3.9 5.2 932 9.2
Physella spp. 0.8 2.0 2.3 3.5 3.8 825 9.0
Argia spp. 1.7 2.7 3.3 4.0 4.1 914 8.6
Limnodrilushoffmeisteri 0.6 0.9 1.3 2.3 4.0 561 9.8
Asellus spp. 0.5 0.6 1.2 1.4 2.0 390 9.4
Mean TV (?1 SD) 9.2 (?0.4)
used for taxa with <10 records. This system is ber-November; Winter: December-February;
designed to be compatible with existing biotic and Spring: March-May.
indices currently in use for the southeastern For the ANOVA calculations, the water-qual-
United States. ity rating again used a score from 1-5. There
Average biotic index values were calculated were few differences between biotic index val-
for all standardized qualitative samples, and ues for flowing streams in the Piedmont and
analysis of variance methods (two-way and Coastal Plain ecoregions, so these two areas were
three-way ANOVA) were used to examine the combined in subsequent calculations. After
effects of three ecoregions (Mountain, Pied- making seasonal corrections (see Results), the
mont, Coastal Plain), four seasons, four stream prior water-quality ratings were then used to
width categories (1-2 m, 3-7 m, 8-20 m, >20 derive classification criteria for the NCBI. For
m) and five water-quality ratings (Excellent, initial tests of tolerance values, 24 taxa were
Good, Good-Fair, Fair, Poor). Seasons were de- chosen that could be roughly divided into In-
fined as Summer: June-August; Fall: Septem- tolerant, Facultative, or Tolerant categories.
TABLE2. Seasonally correcteda average NCBI by for each ecoregion in North Carolina. I made
ecoregion and water-quality rating, with ANOVA re- tables for biotic index values vs. season for each
sults. of the ecoregions, and attempted simple addi-
tive corrections that would equalize the average
Water-quality rating NCBI values for all seasons. Fall, winter, and
Excel- Good- spring values were corrected to the average
Ecoregion lent Good Fair Fair Poor summer value. The following system proved
Mountain 3.78 4.58 5.52 6.32 7.80 satisfactory. For the Mountain ecoregion, 0.4
Piedmont/Coastal 5.02 5.65 6.26 7.09 8.32 was added to the fall (September-November)
biotic index, and 0.5 was added to the spring/
Two-way ANOVA winter (December-May) biotic index. For the
Mean
Source df square F-value Piedmont/Coastal Plain ecoregions, 0.2 was
added to the spring/winter biotic index. After
Ecoregion 2 88.5 307.8
4 making these corrections to the NCBI, season
Water-quality score 265.8 924.0
could be eliminated as a significant variable
Ecoregion x score 8 2.4 8.5
Residual 1467 0.2 - when the effects of season and water-quality
class were examined within specific ecoregions
a For mountain streams: 0.5 added to
winter/spring (p = 0.29 for Mountain ecoregion, p = 0.36 for
collections, 0.4 added to fall collections. For Pied- Piedmont/Coastal Plain).
mont/Coastal streams: 0.2 added to winter/spring
collections. Three-way ANOVA examining the effects of
stream width, ecoregion, and water-quality class
on NCBI values (seasonally corrected) indicated
These a priori choices were based on profes- that all three factors significantly affected biotic
sional judgment and were intended to be non- index values (p = 0.001), but that width effects
controversial, but I included taxa known to be were very small (F = 6.8), relative to Ecoregion
tolerant of both organic and toxic materials. (F = 228.4) and Water-quality Score (F = 831.4).
Criteria for each water-quality class were then
derived using mean values (Table 2) as the mid-
Results
point for each class. This procedure showed that
For the 24 taxa chosen for initial testing, the NCBI values for Piedmont/Coastal streams were
distribution of average abundance values across almost one unit higher than for Mountain
five water-quality categories always produced streams of comparable water quality (Table 3).
unimodal patterns, and confirmed the initial
assignment of these taxa as intolerant, faculta- Discussion
tive, and tolerant (Table 1). The calculation of
tolerance values (using the 75th percentile) pro- North Carolina had a data set of 2000+ stream
duced low values for the intolerant group (mean macroinvertebrate samples which were divided
= 1.4), high values for tolerant taxa (mean = into five water-quality ratings. This data set was
9.2), and intermediate values for the facultative used to derive preliminary tolerance values for
group (mean = 5.4). over 500 taxa. A regression equation then was
Appendix 1 is a complete list of tolerance
values for North Carolina stream invertebrates.
TABLE3. Criteriafor North CarolinaBiotic Index
Nomenclature generally follows that of Brig- after seasonal corrections.
ham et al. (1982). Comparable values from a
Wisconsin data base (Hilsenhoff 1987) also have
been listed. Some taxa have not been included Water Ecoregion
in this list, either because we have insufficient quality Piedmont/
class Mountain Coastal
data to calculate a tolerance value (<10 collec-
Excellent <4.18 <5.24
tions), or because a taxon can only be identified
with unpublished keys for particular North Good 4.17-5.09 5.25-5.95
Good-Fair 5.10-5.91 5.96-6.67
Carolina taxa. Fair 5.92-7.05 6.68-7.70
Two-way ANOVA showed that differences in Poor >7.05 >7.71
season significantly affected biotic index values
used to stretch the initial range of water-quality mayfly flies Stenonema modestumand species in
scores (1-4.5) to a 0-10 range (NCBI). I then the Ephemerellacatawbagroup are fairly tolerant,
tested for the effects of season, stream size, and abundant, and widespread in North Carolina.
ecoregion on NCBI. Simple additive seasonal Wisconsin data, however, had suggested that
corrections were derived, correcting spring, fall these are very intolerant species. It is possible
and winter values to the mean summer NCBI that these "populations" actually are different
values. Stream size did not have a large effect species that are not separable with current tax-
on NCBI values (relative to ecoregion and water onomic keys. These problems illustrate the per-
quality), but Piedmont/Coastal streams had ils of using a biotic index outside its intended
NCBI values about one unit higher than Moun- geographic range.
tain streams (Table 2). Mean tolerance values for the major taxonom-
Hilsenhoff (1987) used similar procedures to ic groups were similar for the North Carolina
derive tolerance values for Wisconsin streams and Wisconsin data sets (Table 4). In particular,
from a data set of 2000+ invertebrate collec- both the tolerance values and the rankings for
tions. His initial tolerance values were based on Ephemeroptera, Plecoptera and Trichoptera
expert opinion, but were later modified based were very close. Chironomidae (as a group) were
on the mean biotic index value for sites where ranked somewhat differently for the two data
any given taxon was collected. This procedure sets, but comparable taxa were often similar in
produced a range of values from 1.4-3.9 on a their tolerance values. The greatest between-
possible scale of 0-5 (W. L. Hilsenhoff, Univer- region differences occurred for Megaloptera and
sity of Wisconsin, personal communication). Odonata.
These numbers were then "stretched" to cover Some of the differences between North Car-
the entire range of a 0-10 scale. Dr. Hilsenhoff's olina and Wisconsin tolerance values may re-
values, however, were based on means, while flect differences in collection methods. The Wis-
the NC values used a 75th percentile figure. consin samples were limited to riffle samples,
Furthermore, the Hilsenhoff Biotic Index (HBI) whereas North Carolina uses multiple-habitat
was only intended to monitor organic pollu- collection methods. More slow-water species
tion, while the NCBI measures any stress that (especially Odonata) would occur in the mul-
either lowers EPT taxa richness or promotes the tiple-habitat collections, and possibly affect the
development of tolerant species. assignment of tolerance values.
The North Carolina tolerance values func- Both our data and Hilsenhoff's data showed
tioned well to separate examples of intolerant, that biotic index values vary with season. Hil-
facultative, and tolerant taxa (24 taxa: Table 1). senhoff's values are standardized for spring and
Comparisons of these tolerance values with Hil- fall collections, with a subtraction of 0.5 during
senhoff's (1987) data were complicated by a summer months (Hilsenhoff 1988). North Car-
number of problems. Sometimes no comparable olina data, however, are standardized for sum-
values could be found (3 taxa), sometimes com- mer collections, with some corrections for other
parisons could only be made by using different periods of the year. For mountain streams, I
species in the same genus (2 taxa), and some- have suggested adding 0.4-0.5 during non-
times tolerance values were listed at a species summer months, similar to the corrections pro-
level in one data set, but at the generic level in posed for Wisconsin streams. During periods of
the other data set (9 taxa). Using the most com- expected change (late spring, fall), stream rat-
parable information from the Wisconsin data ings should be made with caution.
set, mean tolerance values for the Intolerant, Biotic index values also appear to vary with
Facultative and Tolerant groups in Table 1 were ecoregion. This may be related to differences in
0.6, 4.1, and 6.1. Although the trend for the stream temperature, although other variables
mean tolerance values is similar to the North also may affect between-ecoregion differences.
Carolina data (Table 1), the values for the Tol- North Carolina mountain streams and Hilsen-
erant examples showed considerable difference: hoff's Wisconsin streams had similar criteria
6.1 vs. 9.2. ranges; both areas support cold-water fish spe-
Some species had North Carolina populations cies (especially trout). If the Wisconsin data are
that were rated as more tolerant than compa- seasonally corrected for summer collections
rable Wisconsin populations. For example, the (subtract 0.5, Hilsenhoff 1988), then the NC
TABLE4. Comparisons of tolerance values (TV), by taxonomic group (ranked by Mean TV), for North
Carolina and Wisconsin (Hilsenhoff 1987). N = No. of records/group.
for the biotic index. Great Lakes Entomologist 21: NARF, R. P., E. L. LANGE, AND R. C. WILDMAN. 1984.
9-13. Statistical procedures for applying Hilsenhoff's
JONES, J. R., B. H. TRACY,J. L. SEBAUGH, D. H. Biotic Index. Journal of Freshwater Ecology 2:441-
HAZELWOOD, AND M. M. SMART. 1981. Biotic in- 448.
dex tested for ability to assess water quality of NC DIVISION OFENVIRONMENTAL MANAGEMENT. 1992.
Missouri Ozark Streams. Transactions of the Standard operating procedures, biological mon-
American Fisheries Society 110:627-637. itoring. Raleigh, North Carolina.
LENAT,D. R. 1988. Water quality assessment of
streams using a qualitative collection method for Received: 4 November 1992
benthic macroinvertebrates. Journal of the North Accepted: 4 May 1993
American Benthological Society 7:222-233.
APPENDIX 1. Tolerance values for North Carolina stream macroinvertebrates, with comparable values for
Wisconsin taxa (Hilsenhoff 1987). * = based on less than 25 collections for North Carolina data or less than
5 collections for Wisconsin data. ** = comparisons based on differing levels of taxonomy. NC taxonomy
according to Brigham et al. (1982).
APPENDIX 1. Continued.
APPENDIX 1. Continued.
APPENDIX 1. Continued.
APPENDIX 1. Continued.
APPENDIX 1. Continued.