1231Q 2002 Estuarine Research Federation

Estuaries Vol. 25, No. 6A, p. 1231–1242 December 2002

An Estuarine Benthic Index of Biotic Integrity for the Mid-Atlantic

Region of the United States. II. Index Development

ROBERTO J. LLANSO1,*, LISA C. SCOTT1, JEFFREY L. HYLAND2, DANIEL M. DAUER3, DAVID E.RUSSELL4, and FREDERICK W. KUTZ4

1 Versar, Inc., 9200 Rumsey Road, Columbia, Maryland 21045-19342 National Oceanic and Atmospheric Administration, National Ocean Service, 219 Fort Johnson

Road, Charleston, South Carolina 29412-91103 Department of Biological Sciences, Old Dominion University, Norfolk, Virginia 235294 U.S. Environmental Protection Agency, Region III, Environmental Science Center, 701 Mapes

Road, Ft. Meade, Maryland 20755-5350

ABSTRACT: A benthic index of biotic integrity was developed for use in estuaries of the mid-Atlantic region of theUnited States (Delaware Bay estuary through Albemarle-Pamlico Sound). The index was developed for the Mid-AtlanticIntegrated Assessment Program (MAIA) of the U.S. Environmental Protection Agency using procedures similar to thoseapplied previously in Chesapeake Bay and southeastern estuaries, and was based on sampling in July through earlyOctober. Data from seven federal and state sampling programs were used to categorize sites as degraded or non-degradedbased on dissolved oxygen, sediment contaminant, and sediment toxicity criteria. Various metrics of benthic communitystructure and function that distinguished between degraded and reference (non-degraded) sites were selected for eachof five major habitat types defined by classification analysis of assemblages. Each metric was scored according to thresh-olds established on the distribution of values at reference sites, so that sites with low scoring metrics would be expectedto show signs of degradation. For each habitat, metrics that correctly classified at least 50% of the degraded sites in thecalibration data set were selected whenever possible to derive the index. The final index integrated the average scoreof the combination of metrics that performed best according to several criteria. Selected metrics included measures ofproductivity (abundance), diversity (number of taxa, Shannon-Wiener diversity, percent dominance), species compositionand life history (percent abundance of pollution-indicative taxa, percent abundance of pollution-sensitive taxa, percentabundance of Bivalvia, Tanypodinae-Chironomidae abundance ratio), and trophic composition (percent abundance ofdeep-deposit feeders). The index correctly classified 82% of all sites in an independent data set. Classification efficienciesof sites were higher in the mesohaline and polyhaline habitats (81–92%) than in the oligohaline (71%) and the tidalfreshwater (61%). Although application of the index to low salinity habitats should be done with caution, the MAIA indexappeared to be quite reliable with a high likelihood of correctly identifying both degraded and non-degraded conditions.The index is expected to be of great utility in regional assessments as a tool for evaluating the integrity of benthicassemblages and tracking their condition over time.

IntroductionIn 1995, the U.S. Environmental Protection

Agency implemented a research, monitoring, andassessment program to gather information on theextent and condition of natural resources in themid-Atlantic region of the United States. This pro-gram, the Mid-Atlantic Integrated Assessment(MAIA), covers the watersheds of the DelawareBay, Chesapeake Bay, the coastal bays of the Del-marva peninsula, and the Albemarle-PamlicoSound. Since its establishment, MAIA has focusedon a variety of issues that affect estuaries, streams,groundwater, and landscapes in the mid-Atlanticregion (Bradley and Landy 2000). With regard toestuaries, MAIA has focused on various impacts re-

* Corresponding author; tele: 410/964-9200; e-mail:llansorob@versar.com.

sulting from population growth. Between 1950 and1990, population in the MAIA region grew from13 to 21 million (Culliton et al. 1990). The pro-gressive urbanization and industrialization of wa-tersheds in the MAIA region have caused a varietyof impacts that have changed the quality of theseestuaries over the years. The Delaware estuary, forexample, is affected by the lack of water clarity andby toxic contaminants. It is one of the most nutri-ent-enriched estuaries in the world and has one ofthe highest levels of chemical contaminants in fishand shellfish in the nation (National Oceanic andAtmospheric Administration 1994; Pennock et al.1994). The Chesapeake Bay suffers from nutrientover-enrichment, low dissolved oxygen, and re-duced water clarity (Malone 1987; Tuttle et al.1987; Harding 1994; Malone et al. 1996; Bricker etal. 1999). Albemarle-Pamlico Sound and neighbor-

1232 R. J. Llanso et al.

ing tributaries have been affected by multiplestressors, including low-oxygen conditions due toorganic over-enrichment (Paerl et al. 1998) andsediment contamination (Hyland et al. 2000).

Regional assessments of environmental condi-tions are necessary for effective restoration of nat-ural resources and maintenance of estuarine en-vironmental quality. The objective is to use a com-mon set of measurements that can be applied uni-formly over the entire region. A recent approachto environmental characterization in estuaries hasbeen the development and application of indicesof biotic integrity that combine several measuresof benthic community response (Weisberg et al.1997; Adams et al. 1998; Engle and Summers 1999;Van Dolah et al. 1999; Paul et al. 2001). These in-dices focus on benthic communities because thebenthos respond predictably to natural and an-thropogenic stress (Pearson and Rosenberg 1978;Dauer 1993; Wilson and Jeffrey 1994). Two meth-ods have been used. One method combines step-wise and linear discriminant analyses to produce amulti-variate index (Weisberg et al. 1993; Engle etal. 1994; Engle and Summers 1999; Paul et al.2001). The discriminant approach produces acombination of measures that discriminate opti-mally between reference and degraded conditions.Components are normalized to account for the ef-fects of natural variability, but the normalizationprocess can be complex.

The second method defines community charac-teristics expected at reference sites with no evi-dence of anthropogenic stress. A scoring systembased on the distribution of values of key benthicattributes at these sites is then used to evaluate thecondition of a site. Variability due to natural factorssuch as salinity or sediment type is accounted forby partitioning the data into habitat classes. Thismethod was first applied to fish assemblages infreshwater environments by Karr (1981) and isknown as the index of biotic integrity (IBI). Weis-berg et al. (1997) modified the method to developa benthic index of biotic integrity (B-IBI) for usein the Chesapeake Bay. The method has also beenapplied in the development of benthic indices forNew York-New Jersey Harbor (Adams et al. 1998)and for estuaries of the southeastern U.S. (Van Do-lah et al. 1999). The B-IBI approach is easy to un-derstand and interpret, and an advantage over oth-er methods resides in its ability to evaluate ecolog-ical condition of a sample by comparing values ofkey benthic attributes to reference values expectedunder non-degraded conditions in similar habitattypes.

In this paper, we develop an index of benthiccommunity condition for use in the MAIA regionusing the B-IBI approach. Results of initial steps to

identify habitat types have been reported by Llansoet al. (2002). Here we describe the index devel-opment process and test the efficiency of the indexat classifying samples as degraded vs. non-degrad-ed.

MethodsThe index development process consisted of sev-

en steps: data compilation and standardization,classification of sites into reference (non-degrad-ed) and degraded categories, partitioning of datainto calibration and validation data sets segregatedaccording to habitat types, testing of metrics fordifferences between reference and degraded sitesfor each of the major habitat types, calculation ofreference-range thresholds and scoring of metrics,selection of metrics according to site classificationcriteria, and combination of metrics to representthe MAIA index. Each step is described in detailbelow.

DATA COMPILATION AND STANDARDIZATION

Data from seven sampling programs were assem-bled for this study. Information on the samplingprograms were presented in Llanso et al. (2002).Data from federal (MAIA, Environmental Monitor-ing and Assessment Program, National Status andTrends) and state (Maryland and Virginia) moni-toring programs that have sampled estuaries in theregion from as early as 1984 were assembled in adatabase. Only sites that included benthic inverte-brate data and were sampled during the summer,defined as July 1 through October 7, were used inthis study. Data for sites sampled more than onceduring the summer were averaged. All samplingprograms used a 440-cm2 Young grab (one to threegrabs per site), sieved samples through a 0.5-mmmesh screen, and identified organisms to the low-est possible taxonomic level.

Information collected in conjunction with thebenthic samples included water column measures(salinity, water depth, and bottom dissolved oxy-gen) and sediment quality measures (silt-clay con-tent, percent organic carbon, contaminant concen-trations, and sediment toxicity). Sediment toxicitywas based on two laboratory bioassays, a 10-d acuteamphipod bioassay using Ampelisca abdita (Ameri-can Society for Testing Materials 1993) and the Mi-crotox bioassay, which measures changes in bacte-rial luminescence as an indicator of acute sublethaleffects in sediment elutriates (Bulich et al. 1981).Details on specific sampling protocols and mea-surements taken can be found in Chaillou et al.(1996), Dauer et al. (1998), Hyland et al. (1998,2000), Strobel (1998), Paul et al. (1999), and Ran-asinghe et al. (1999). Prior to analysis, data werestandardized to ensure common species nomencla-

Benthic Index Development 1233

ture and uniformity in station designation, variabledesignation, and units of measure. See Llanso etal. (2002) for the species standardization protocol.The database consisted of 1,999 sites, 19 of whichwere sampled in multiple years yielding a total of2,083 sampling events. Because the majority of thesites were sampled only once, we refer to all sam-pling events as sites.

CLASSIFICATION OF SITES

Reference (non-degraded) and degraded siteswere selected according to dissolved oxygen, sedi-ment contaminant, and sediment toxicity criteria.Sites were defined as reference if for all samplingevents dissolved oxygen concentrations were great-er than 3.0 ppm, no chemical contaminant con-centration exceeded Long et al.’s (1995) effectsrange-median concentrations, no more than twochemical contaminants exceeded Long et al.’s(1995) effects range-low concentrations, and sedi-ments were not toxic in Ampelisca or Microtox bio-assays. Amphipod bioassays followed guidelinesprovided in American Society for Testing Materials(1993) protocols, and were considered to indicatetoxicity when mean test survival was significantlydifferent from (a 5 0.05) and less than 80% ofcontrol survival. Microtox bioassays were consid-ered to indicate toxicity when the EC50 of test sed-iments (sediment concentration that reduces bac-terial light production by 50% relative to watercontrols) was # 0.2% for sediments with silt-claycontent $ 20%, the EC50 was # 0.5% for sedi-ments with silt-clay content , 20% (Ringwood etal. 1997), or the EC50 of test sediments was signif-icantly different from controls. Sites were definedas degraded if any of the following criteria weremet: dissolved oxygen concentrations were less orequal to 2.0 ppm, any chemical contaminant con-centration exceeded Long et al.’s (1995) effectsrange-median concentrations, or sediments weretoxic in the Ampelisca or Microtox bioassays. Sitesthat did not meet the classification criteria for ei-ther the reference or the degraded condition, ordid not have associated dissolved oxygen or chem-istry data, were classified as intermediate.

PARTITIONING OF DATA

The classification analysis of benthic assemblagesdistinguished nine habitat classes as a combinationof six salinity classes, two sediment types, and theseparation of North Carolina and Delaware-Ches-apeake Bay polyhaline sites (Llanso et al. 2002).Based on these results, sites were partitioned intohabitat types according to salinity, silt-clay content,and geographical location. Two thirds of the siteswere selected randomly to represent the calibra-tion data set. The remaining one third represented

the validation data set. The calibration data set wasused to develop the index. The validation data setwas intentionally withheld and used to evaluate in-dependently the performance of various possibleindices resulting from different metric combina-tions (see below). Sites were classified into two datasets (calibration and validation) for each of threeconditions (reference, degraded, and intermedi-ate) and nine habitat groups (Table 1). The num-ber of degraded sites available for both index cal-ibration and validation were very few, particularlyfor polyhaline and euhaline habitats. For theNorth Carolina polyhaline sand habitat, no sitescould be identified as degraded (Table 1). We con-fronted this problem by merging all high-salinitysites into one polyhaline-euhaline group. Althoughthis solution was less than optimal, we felt it wasreasonable given the high degree of overlap indi-cated between sediment types, and the lack ofunique species representative of North Carolina es-tuaries (Llanso et al. 2002). The validation rate wasnonetheless high for the combined euhaline andpolyhaline sites (see Results). The final site classi-fication consisted of two data sets, three condi-tions, and five habitat classes defined according tosalinity: tidal freshwater (0–0.5‰), oligohaline(0.5–5‰), low mesohaline (5–12‰), high meso-haline (12–18‰), and polyhaline ($ 18‰; Table1). Intermediate sites were excluded from furtherconsideration.

TESTING OF METRICS

Twenty-three attributes of benthic communitystructure and function (referred to as metrics)were tested with the calibration data set to deter-mine those that differed significantly between ref-erence and degraded sites within each of the hab-itat classes (Table 2). The candidate metric list in-cluded a variety of measures of productivity, diver-sity, species composition, and trophic composition.The list included metrics that were found to beuseful indicators of benthic condition in other in-dex development efforts (Weisberg et al. 1997; Ad-ams et al. 1998; Engle and Summers 1999; Van Do-lah et al. 1999; Paul et al. 2001), as well as somenovel metrics. The productivity and diversity met-rics are widely used benthic community metrics.Percent dominance, defined as 100 minus the per-cent abundance contribution of the top two nu-merically-dominant taxa in the community, was se-lected in the development of the B-IBI for south-eastern U.S. estuaries (Van Dolah et al. 1999). Sen-sitive taxa that showed strong relationships tosediment contaminant levels in that study were alsoused in the present analysis. Pollution-sensitiveNorth Carolina taxa (Group C, Van Dolah et al.1999) were Ampeliscidae, Haustoriidae, Tellinidae,

1234 R. J. Llanso et al.

TA

BL

E1.

Nu

mbe

rof

site

sav

aila

ble

for

ind

exd

evel

opm

ent

(cal

ibra

tion

dat

ase

t)an

dva

lidat

ion

for

each

ofth

ree

con

dit

ion

s(r

efer

ence

,d

egra

ded

,an

din

term

edia

te)

and

nin

eh

abit

atgr

oup

s.In

the

fin

alsi

tecl

assi

fica

tion

,si

tes

ingr

oup

s(5

)th

rou

gh(9

)w

ere

com

bin

edin

toon

ep

olyh

alin

e($

18‰

)gr

oup

(see

text

).In

term

edia

tesi

tes

did

not

mee

tth

ecl

assi

fica

tion

crit

eria

for

eith

erth

ere

fere

nce

(non

-deg

rad

ed)

orth

ed

egra

ded

con

dit

ion

,an

dw

ere

excl

ud

edfr

oman

alys

is.

Inte

rmed

iate

site

sal

soin

clu

de

site

sw

ith

out

eith

erd

isso

lved

oxyg

enor

chem

istr

yin

form

atio

n.

NC

5N

orth

Car

olin

a;D

E-C

BA

Y5

Del

awar

e-C

hes

apea

keB

ay;

San

d5

0–10

%si

lt-c

lay

con

ten

t(p

arti

cles

,62

m)

byw

eigh

t;Si

lt-c

lay

510

–100

%si

lt-c

lay

con

ten

tby

wei

ght.

Dat

aSe

tC

ond

itio

n

Hab

itat

(1)

Tid

alFr

esh

wat

er(2

)O

ligoh

alin

e

(3)

Low

Mes

ohal

ine

(4)

Hig

hM

esoh

alin

e

(5)

NC

Pol

yhal

ine

San

d

(6)

NC

Pol

yhal

ine

Silt

-cla

y

(7)

DE

-CB

AY

Pol

yhal

ine

San

d

(8)

DE

-CB

AY

Pol

yhal

ine

Silt

-cla

y

(9)

DE

-CB

AY

Eu

hal

ine

(5)–

(9)

Com

bin

edP

olyh

alin

e

Cal

ibra

tion

Valid

atio

n

Ref

eren

ceD

egra

ded

Inte

rmed

iate

Ref

eren

ceD

egra

ded

24 21 87 12 11

25 15 97 13 8

37 37 230 19 19

87 65 241 44 33

21 0 1 10 0

13 3 7 7 1

18 2 18 9 1

30 18 72 15 9

45 413

8 22 2

127 27 236 63 13

Inte

rmed

iate

4552

100

116

13

1030

6911

3

Lucinidae, Hesionidae, Cirratulidae, Cyathura poli-ta, and Cyathura burbancki. Two other pollution-in-dicative and sensitive species lists were modifiedfrom those used in the Chesapeake Bay B-IBI(Weisberg et al. 1997) and were based either onevidence in the literature or on statistical testingconducted during the Chesapeake Bay B-IBI de-velopment. The trophic categories were those usedby Weisberg et al. (1997).

One recently published metric (North CarolinaSensitivity Index) and two previously unpublishedmetrics (Tolerance Score and Tanypodinae-Chi-ronomidae percent abundance ratio) were testedas well. The North Carolina Sensitivity Index (Ea-ton 2001) is a weighted abundance average fortaxa classified according to sensitiveness to pollu-tion. Tolerance values for . 200 estuarine taxawere used as weights in the index (range 5 1–5).The Tolerance Score is very similar in constructionto the North Carolina Sensitivity Index, and wasexpanded from Lenat (1993) to include freshwatertaxa from piedmont and coastal streams in theMAIA region. The Tanypodinae-Chironomidaepercent abundance ratio is a measure of the rela-tive contribution of the Tanypodinae to all the oth-er midges found in a sample. The Tanypodinae areconsidered tolerant to pollution (Lenat 1993), andthe ratio is expected to increase in perturbed ar-eas. Similar ratios have been used in other studies(Barbour et al. 1996).

A metric is defined as an attribute of the benthicassemblage that changes in some predictable waywith increased human influence (Karr et al. 1986).For a metric to be useful, differences between ref-erence and degraded sites should be statisticallysignificant and the differences should be in a di-rection that is consistent with established ecologi-cal principles. These principles are based on par-adigms which predict changes in stressed ecosys-tems, such as reductions in the abundance and di-versity of organisms and shifts in dominance frompollution sensitive to pollution tolerant species(Pearson and Rosenberg 1978; Odum 1985; Dauer1993). Based on consistency with these ecologicalprinciples and Mann-Whitney U tests for differenc-es in means between reference and degraded sites,21 metrics were selected for further testing (Table2). A probability level of 0.1 was used to reducethe risk of Type II errors of declaring degradedsites as non-degraded. To avoid additional statisti-cal variance that may reduce the power of the test,sites with no fauna (azoic) were excluded fromtesting.

CALCULATION OF THRESHOLDS AND SCORINGOF METRICS

The next step in the index development processwas the calculation of thresholds for defining ref-

Benthic Index Development 1235

TABLE 2. Candidate metrics and mean values at reference (top number) and degraded (bottom number) sites in the Mid-AtlanticIntegrated Assessment (MAIA) region by habitat class. Bold numbers indicate pair is different (p , 0.1) by Mann-Whitney U test.Asterisks indicate metrics selected for further testing. See text for metric definition. na 5 not applicable.

MetricTital

Freshwater OligohalineLow

MesohalineHigh

Mesohaline Polyhaline

ProductivityAbundance* (no. m22)

Biomass* (g AFDW m22)

3,937.513,365.8

7.22.2

2,758.22,355.6

6.08.4

2,890.3908.3

2.03.5

3,609.11,108.2

3.21.1

4,343.61,430.9

4.75.5

DiversityShannon-Wiener diversity* (log2)

Number of taxa*

1.91.9

10.08.9

2.01.48.05.8

2.30.7

10.02.6

2.61.2

12.24.5

3.01.6

19.28.8

Percent dominance*

Species CompositionPercent abundance of pollution-indicative taxa*

Percent abundance of pollution-sensitive taxa*

22.925.3

58.880.32.7a

3.4

28.116.8

35.671.327.712.6

35.215.2

16.143.422.823.0

40.919.4

21.050.825.312.3

47.030.0

9.922.242.623.8

Percent abundance of pollution-sensitive NC taxa*(Group C, Van Dolah et al. 1999)

Percent abundance of pollution-indicative oligohaline taxa*

Percent abundance of pollution-sensitive oligohaline taxa*

1.4a

2.7na

na

7.81.8

49.883.321.06.3

26.411.0na

na

17.09.8na

na

16.712.7na

na

North Carolina Sensitivity Index*

Tolerance Score*

Tanypodinae-Chironomidae percent abundance ratio*

1.81.38.39.2

27.250.8

1.81.76.28.5

18.552.1

1.71.2na

na

1.61.3na

na

1.71.3na

na

Percent abundance of Limnodrilus

Percent abundance of Tubificidae*

Percent abundance of Spionidae*

13.614.852.572.21.20.9

5.613.433.161.821.87.8

na

7.611.816.933.1

na

3.414.515.131.3

na

0.40.6

11.324.1

Percent abundance of Capitellidae

Percent abundance of Amphipoda*

0.00.10.50.1

4.0b

0.05.7

11.4

12.8b

9.511.57.0

16.5b

9.35.54.6

17.9b

12.17.72.6

Percent abundance of Bivalvia*

Trophic CompositionPercent abundance of carnivore-omnivores*

15.59.3

29.416.6

8.15.0

26.013.4

20.013.6

21.88.6

21.310.7

22.212.7

14.33.5

28.437.2

Percent abundance of interface feeders*

Percent abundance of deep-deposit feeders*

Percent abundance of suspension feeders*

1.91.1

53.272.914.59.3

29.719.737.362.07.04.9

48.349.125.734.23.14.8

33.4c

43.429.735.614.27.1

28.630.225.919.216.96.2

a Result contrary to ecological principle.b Cannot interpret metric direction; the species Mediomastus ambiseta and the genus Capitella respond differently to pollution.c Cannot interpret metric direction; interface feeders include spionids and some bivalves (e.g., Macoma) that respond differently to

pollution.

erence-value ranges for each metric-habitat com-bination and the application of a scoring system tothese ranges. Procedures were based on modifica-tions of those described by Weisberg et al. (1997).Threshold values were established as the 10th and50th (median) percentile values of reference sites

for each metric-habitat combination. This proce-dure differed from Weisberg et al. (1997), whoused 5th and 50th percentiles. For each metric, avalue falling below the 10th percentile thresholdwas considered to deviate strongly from values atthe best reference sites in the same habitat. An up-

1236 R. J. Llanso et al.

per threshold corresponding to the 90th percen-tile was used for some metrics (e.g., percent abun-dance of pollution-indicative taxa) because the di-rection of the response is such that higher per-centages are expected in degraded sites than inreference sites. Fifth and 95th percentile thresh-olds as used by Weisberg et al. (1997) were alsocalculated, but a high percentage of misclassifica-tions for various metrics were found when thesethresholds were applied.

Abundance and biomass respond bimodally topollution (Pearson and Rosenberg 1978; Weisberget al. 1997). An increase in the abundance and/orbiomass of organisms is expected at polluted siteswhen the stress from pollution is moderate, suchat sites where there is organic enrichment of thesediment. A decrease in the abundance and bio-mass of organisms is expected at sites with highdegrees of stress from pollution. For these two met-rics, an upper threshold corresponding to the 90thpercentile value of reference sites was establishedin addition to the lower threshold correspondingto the 10th percentile.

A score of 1 was used if the value of the metricfor the site being evaluated was below the 10th per-centile of corresponding reference values, a scoreof 3 was used for values between the 10th percen-tile and the median, and a score of 5 was used forvalues above the median. For metrics with upperthresholds (see above), the scoring was reversed sothat a score of 1 was used for values above the 90thpercentile, a score of 3 for values between the 90thpercentile and the median, and a score of 5 forvalues below the median of the corresponding ref-erence values. For abundance and biomass, a scoreof 1 was used if the value of the metric for the sitebeing evaluated was below the 10th percentile orabove the 90th percentile, a score of 3 for valuesbetween the 10th and 25th or between the 75thand 90th percentiles, and a score of 5 for valuesbetween the 25th and 75th percentiles of corre-sponding reference values. A score of 1 was con-sidered to indicate impaired condition.

In some cases, two modifications to the abovescoring system were introduced. These proceduresalso differed from procedures described by Weis-berg et al. (1997). First, metrics based on percent-ages or tolerance values (all of the metrics listedin Table 2 except abundance, biomass, Shannondiversity, and number of taxa) were not scoredwhen there was no fauna. We did this because theinclusion of azoic sites would have exaggerated theresponse of metrics that are based on percentages.Second, pollution-sensitive taxa metrics (i.e., per-cent abundance of pollution-sensitive taxa, percentabundance of pollution-sensitive NC taxa, and per-cent abundance of pollution-sensitive oligohaline

taxa) that would score 3 or 5 when the overallabundance at a site is low (i.e., # 20% of the lowerabundance threshold) were not scored. This wasdone to avoid high scores due to the presence ofa few individuals of pollution-sensitive speciesfound among a small number of total individualswithin a sample (Weisberg et al. 1997).

SELECTION OF METRICS

Using the scoring system described above, clas-sification efficiencies (percent sites correctly clas-sified) for each of the metrics selected during sta-tistical testing were examined to determine whichmetrics performed best (Table 3). Most metricsclassified non-degraded sites very well, typicallybetter than 90% correct classifications, but degrad-ed sites were generally classified at much lower lev-els. Those metrics with at least 50% correct classi-fications for the degraded sites were selected forfinal evaluation. This approach reduced the risk ofType II errors (i.e., degraded stations classified asnon-degraded). The exception to this rule weremetrics in the tidal freshwater habitat, for whichthis criterion was relaxed due to the generally low-er classification efficiencies observed for this hab-itat. Percent abundance of pollution-indicative oli-gohaline taxa and percent abundance of Tubifici-dae which met this rule were considered redun-dant of other metrics in the oligohaline habitatand were excluded from further consideration.Twelve metrics, some performing well in morethan one habitat (Table 3), were selected to deter-mine which combination of metrics representedthe best index.

COMBINATION OF METRICS INTO AN INDEX

In the final phase of the index development pro-cess, the calibration and validation data sets wereused to evaluate which combination of metrics bestdistinguished between degraded and non-degrad-ed conditions for each of the five habitats. For eachhabitat, metrics were combined in all possible waysand the resulting combinations of metrics were ar-ranged in order according to how well they cor-rectly classified sites as degraded and non-degrad-ed in the calibration and validation data sets. Fourcriteria of acceptance were then established. Themetric combination that correctly classified sites inthe calibration data set within 5% of the most ef-ficient combination, correctly classified sites in thevalidation data set within 10% of the most efficientcombination, had the most number of metrics, andhad a variety of functional categories, was selectedto represent the final index. Only combinationsthat did not include biomass were considered. Bio-mass was excluded because it was not determinedby all the sampling programs in the MAIA region,

Benthic Index Development 1237

TABLE 3. Classification efficiencies (percent of sites correctly classified) on calibration data for each of the metrics that distinguishedbetween reference and degraded sites in a direction consistent with established ecological principles. Asterisks indicate metrics selectedfor final testing in various combinations (see text).

Habitat/Metric

Percent Sites Correctly Classified

Non-degraded Degraded Total

Tidal FreshwaterAbundance*Percent abundance of pollution-indicative taxaNorth Carolina Sensitivity Index*Tolerance Score

79.291.791.791.7

76.214.338.123.8

77.855.666.760.0

Tanypodinae-Chironomidae percent abundance ratio*Percent abundance of TubificidaePercent abundance of AmphipodaPercent abundance of carnivore-omnivoresPercent abundance of interface feeders

91.791.7

100.091.7

100.0

38.128.60.0

28.60.0

66.762.253.362.253.3

Percent abundance of deep-deposit feeders*Oligohaline

Shannon-Wiener diversityNumber of taxaPercent dominance*Percent abundance of pollution-indicative taxa*

91.7

92.092.092.092.0

38.1

46.726.753.353.3

66.7

75.067.577.577.5

Percent abundance of pollution-sensitive taxaPercent abundance of pollution-sensitive NC taxaPercent abundance of pollution-indicative oligohaline taxaPercent abundance of pollution-sensitive oligohaline taxaTolerance Score

91.791.792.0

100.092.0

46.746.753.30.0

46.7

74.474.477.561.575.0

Tanypodinae-Chironomidae percent abundance ratio*Percent abundance of TubificidaePercent abundance of SpionidaePercent abundance of carnivore-omnivoresPercent abundance of deep-deposit feeders*

92.092.0

100.092.092.0

53.360.00.0

46.760.0

77.580.062.575.080.0

Low MesohalineAbundance*Biomass*Shannon-Wiener diversity*Number of taxa*

81.182.191.991.9

73.094.381.186.5

77.088.986.589.2

Percent dominance*Percent abundance of pollution-indicative taxa*Percent abundance of pollution-sensitive NC taxaNorth Carolina Sensitivity Index*Percent abundance of AmphipodaPercent abundance of Bivalvia

91.991.991.991.9

100.091.9

70.050.047.470.00.0

40.0

84.277.276.884.264.973.7

Percent abundance of carnivore-omnivoresHigh Mesohaline

Abundance*Biomass *Shannon-Wiener diversity*

91.9

80.580.090.8

40.0

55.469.072.3

73.7

69.775.082.9

Number of taxa*Percent dominance*Percent abundance of pollution-indicative taxaPercent abundance of pollution-sensitive taxa*Percent abundance of pollution-sensitive NC taxa

90.890.890.890.790.7

67.762.843.162.549.0

80.980.473.280.675.6

North Carolina Sensitivity Index*Percent abundance of TubificidaePercent abundance of AmphipodaPercent abundance of BivalviaPercent abundance of carnivore-omnivores

90.890.8

100.090.890.8

51.025.50.0

39.241.2

76.166.763.071.772.5

Percent abundance of suspension feedersPolyhaline

Abundance*Biomass*

100.0

80.379.4

0.0

66.770.8

63.0

77.977.6

Shannon-Wiener diversity*Number of taxa*Percent dominancePercent abundance of pollution-sensitive taxa

90.692.990.690.4

59.366.747.835.3

85.188.384.083.8

Percent abundance of pollution-sensitive NC taxaNorth Carolina Sensitivity IndexPercent abundance of AmphipodaPercent abundance of Bivalvia*Percent abundance of deep-deposit feedersPercent abundance of suspension feeders*

90.390.6

100.090.690.690.6

38.943.50.0

69.626.152.2

83.883.384.787.380.784.7

1238 R. J. Llanso et al.

and therefore some data sets did not include bio-mass measurements.

The index value for a site was computed by av-eraging the scores of the individual metrics (range:1–5). Index values , 3.0 were considered to indi-cate stressed benthic assemblages indicative of de-graded conditions.

Results and DiscussionThe index selected for use in the MAIA region

consisted of three metrics in each of the tidal fresh-water and oligohaline habitats, five metrics in eachof the low mesohaline and high mesohaline habi-tats, and four metrics in the polyhaline habitat. Ta-ble 4 lists the selected metrics and thresholds. Twometrics were percent abundance contributions ofspecies indicators: percent abundance of pollution-sensitive taxa (Table 5) and percent abundance ofpollution-indicative taxa (Table 6). One metric wasthe percent abundance contribution of deep-de-posit feeders: all the Oligochaeta and the poly-chaetes Heteromastus filiformis and Mediomastus spp.This last metric provided information about thetrophic composition of the community.

The index selected for the MAIA region cor-rectly classified 83% of the sites in the calibrationdata set and 82% of the sites in the validation dataset. The lowest overall classification efficiency byhabitat was for the oligohaline using the calibra-tion data set (77% of the sites correctly classified)and for the tidal freshwater using the validationdata set (61% of the sites correctly classified; Table7). Although we used both calibration and valida-tion data in the metric combination selection pro-cess, validation data were independent of thoseused to derive the metric thresholds. The useful-ness of the index should then be evaluated accord-ing to how well it classifies sites of known sedimentquality in the independent data set. The index cor-rectly classified a high percentage of sites in theindependent data set, but classification efficienciesvaried substantially across habitats. Classificationefficiencies of sites during the validation phasewere higher in the high-salinity habitats (low me-sohaline, high mesohaline, and polyhaline; range5 81–92%) than in the low-salinity habitats (tidalfreshwater and oligohaline; range 5 61–71%).Chesapeake Bay B-IBI estimates were also found tobe low for sites in tidal freshwater and oligohalinehabitats (Weisberg et al. 1997; Alden et al. 2002).The development and application of biocriteria ismost difficult in the tidal freshwater and oligoha-line regions because of the high variability in ben-thic community composition in those habitats(Dauer 1993). This variability may be partially dueto the various natural stresses in low-salinity habi-tats, which make the application of biological in-

dices less reliable than those established for moresaline habitats. The classification performance forsingle metrics in the tidal freshwater and oligoha-line habitats was relatively poor, but it improvedwhen the metrics were combined. Classification ef-ficiencies on validation data in these habitats werelow for the degraded sites (Table 7), about what itwould be expected from chance alone. One pos-sible factor contributing to the relatively low clas-sification efficiency in the low-salinity habitats issample size. The number of samples available forselection of thresholds in the tidal freshwater andoligohaline habitats was particularly low. Better re-sults may be achieved if a larger array of sampleswere used for the reference distributions, whichmight improve threshold definition and reducethe number of misclassifications.

In the tidal freshwater and oligohaline habitats,the fauna is often dominated by tubificid oligo-chaetes, and a variety of midges are present andcan be abundant. In the MAIA index and otherindex development efforts (Weisberg et al. 1993,1997; Engle et al. 1994; Eaton 2001; Paul et al.2001) tubificids and many chironomids have beencategorized as indicative of organic and toxic pol-lution. This does not mean that low salinity habi-tats are of lesser quality than high salinity habitatswhere tubificids are less abundant and chirono-mids usually absent. What matters is the relativepercentage of these taxa in a sample. Indicator spe-cies usually dominate heavily the community in sit-uations of organic enrichment or where toxic pol-lution eliminates other species. In the MAIA index,the relative abundance of pollution-indicative taxain the oligohaline habitat must be . 75% for thismetric to indicate impaired conditions, and in thetidal freshwater the abundance of deep-depositfeeders (mostly tubificids) must be . 90%. Theclassification of most oligochaetes and many chi-ronomids as indicative of pollution may add to thelow efficiency problem of biological indices in lowsalinity environments. Although a wide range ofsensitivities to pollution has been recognized forinsect larvae (Hilsenhoff 1987; Lenat 1993; Eaton2001), a better understanding of the biology of ol-igochaetes and chironomids in relation to pollu-tion might help improve assessments in estuaries.

The combinations of metrics that were used torepresent the index were among those with thehighest percentages of correct classifications. Theywere not the best performing in terms of highestnumber of sites correctly classified as degraded ornon-degraded. Top classification efficiencies onvalidation data were 70% in the tidal freshwater,76% in the oligohaline, 92% in the low mesoha-line, 84% in the high mesohaline, and 92% in thepolyhaline habitat. Combinations with slightly low-

Benthic Index Development 1239

TA

BL

E4.

Th

resh

old

su

sed

tosc

ore

each

met

ric

ofth

eM

id-A

tlan

tic

Inte

grat

edA

sses

smen

t(M

AIA

)in

dex

.

Hab

itat

/M

etri

c

Scor

ing

Cri

teri

a

13

5

Tid

alFr

esh

wat

erA

bun

dan

ce(n

o.m

22 )

Tan

ypod

inae

-Ch

iron

omid

aep

erce

nt

abu

nd

ance

rati

oP

erce

nt

abu

nd

ance

ofd

eep

-dep

osit

feed

ers

Olig

ohal

ine

Per

cen

td

omin

ance

,1,

401.

5or

$7,

068.

2.

69.8

.90

.6

,10

.4

1,40

1.5–

1,77

2.7

or$

5,25

3.8–

7,06

8.2

20.9

–69.

861

.7–9

0.6

10.4

–30.

0

$1,

772.

7–5,

253.

8#

20.9

#61

.7

$30

.0P

erce

nt

abu

nd

ance

ofp

ollu

tion

-ind

icat

ive

taxa

Per

cen

tab

un

dan

ceof

dee

p-d

epos

itfe

eder

sL

owM

esoh

alin

eA

bun

dan

ce(n

o.m

22 )

Shan

non

-Wie

ner

div

ersi

ty(l

og2)

.75

.8.

64.3

,81

8.2

or$

5,15

1.5

,1.

5

35.5

–75.

838

.9–6

4.3

818.

2–1,

590.

9or

$3,

659.

1–5,

151.

51.

5–2.

4

#35

.5#

38.9

$1,

590.

9–3,

659.

1$

2.4

Nu

mbe

rof

taxa

Per

cen

td

omin

ance

Per

cen

tab

un

dan

ceof

pol

luti

on-in

dic

ativ

eta

xaH

igh

Mes

ohal

ine

Abu

nd

ance

(no.

m2

2 )

,6.

3,

18.7

.39

.5

,63

6.4

or$

6,90

9.1

6.3–

10.3

18.7

–35.

87.

5–39

.5

636.

4–1,

174.

2or

$4,

159.

1–6,

909.

1

$10

.3$

35.8

#7.

5

$1,

174.

2–4,

159.

1Sh

ann

on-W

ien

erd

iver

sity

(log

2)N

um

ber

ofta

xaP

erce

nt

dom

inan

ceP

erce

nt

abu

nd

ance

ofp

ollu

tion

-sen

siti

veta

xaP

olyh

alin

e

,1.

7,

6.0

,22

.4,

2.9

1.7–

2.6

6.0–

11.7

22.4

–41.

92.

9–21

.3

$2.

6$

11.7

$41

.9$

21.3

Abu

nd

ance

(no.

m2

2 )Sh

ann

on-W

ien

erd

iver

sity

(log

2)N

um

ber

ofta

xaP

erce

nt

abu

nd

ance

ofB

ival

via

,64

7.7

or$

8,71

9.7

,2.

0,

7.0

,0.

44

647.

7–1,

454.

5or

$5,

704.

5–8,

719.

72.

0–3.

17.

0–19

.30.

44–7

.5

$1,

454.

5–5,

704.

5$

3.1

$19

.3$

7.5

1240 R. J. Llanso et al.

TABLE 5. Taxa defined as pollution-sensitive in the Mid-Atlan-tic Integrated Assessment (MAIA) index. Taxa listed at the ge-nus level include all species.

Annelida: PolychaetaBhawania heterosetaChaetopterus variopedatusClymenella torquataDiopatra cuprea

Listriella clymenellaeMollusca

Acteocina canaliculataAnadara ovalisAnadara transversa

Glycera americanaGlycinde solitariaLoimia medusaMacroclymene zonalisMarenzelleria viridis

Ensis directusMacoma balthicaMercenaria mercenariaRangia cuneataSpisula solidissima

Mediomastus ambisetaMesochaetopterus tayloriNephtys pictaSabaco elongatusSpiochaetopterus costarum

Tagelus divisusTagelus plebeiusTellina agilisUnionidae (all species)

PhoronidaSpiophanes bombyx

ArthropodaCyathura politaAmpelisca verrilli

Phoronis spp.Echinodermata

Microphiopholis atra

TABLE 6. Taxa defined as pollution-indicative in the Mid-At-lantic Integrated Assessment (MAIA) index. Taxa listed at thegenus level include all species.

Annelida: PolychaetaEteone heteropodaLeitoscoloplos fragilisParaprionospio pinnataStreblospio benedicti

MolluscaMulinia lateralis

Arthropoda: InsectaAblabesmyia parajantaChaoborus spp.

Annelida: OligochaetaBranchiura sowerbyiDero spp.Enchytraeidae (all species)Ilyodrilus templetoni

Chironomus spp.Clinotanypus spp.Coelotanypus spp.Cryptochironomus spp.Dicrotendipes spp.

Isochaetides spp.Limnodrilus spp.Quistradrilus multisetosusTelmatodrilus vejdovskyiTubifex spp.

Endochironomus spp.Glyptotendipes spp.Harnischia spp.Kiefferulus spp.Microchironomus spp.

Tubificidae immature withcapilliform chaetae

Tubificidae immature withoutcapilliform chaetae

Tubificoides spp.

Parachironomus spp.Procladius spp.Tanypus spp.

TABLE 7. Classification efficiencies (percent of sites correctlyclassified) of the Mid-Atlantic Integrated Assessment (MAIA)index for sites from the calibration and validation data sets with-in each habitat class.

Data Set Habitat/Metric

Percent Sites Correctly Classified

Non-degraded Degraded Total

Calibration Tidal FreshwaterOligohalineLow MesohalineHigh MesohalinePolyhaline

91.792.083.886.285.0

71.453.389.278.570.4

82.277.586.582.982.5

Validation Tidal FreshwaterOligohalineLow MesohalineHigh MesohalinePolyhaline

66.784.689.584.188.9

54.650.094.775.876.9

60.971.492.180.586.8

er classification efficiencies but with larger numberof components and categories of biotic responsewere favored over combinations with fewer metrics.We reasoned that the use of more metrics was bet-ter than fewer metrics because the benthos oftenshow a staged response to stress in which the var-ious components of the assemblage respond dif-ferently to different levels of disturbance (Pearsonand Rosenberg 1978). Categories that indicate dif-ferent aspects of benthic community structure andfunction provide more robust information aboutthe developmental stage of the community.

While applying the final selection criteria, we se-lected against metric combinations that includedbiomass. Biomass measurements are very useful forvarious reasons. For example, biomass data provideinformation about the presence of large-sized spe-cies in the community, which is often indicative ofa past history of good sediment quality conditions(Warwick 1986; Dauer 1993). Because biomassdata were unavailable for North Carolina estuariesor Delaware Bay, the reference range distributionof biomass values excluded a large portion of theMAIA region. This would represent a problem if aMAIA index that included biomass as a metricwere used in a region for which the referencerange distribution had been inappropriately esti-mated. If the index included biomass and the bio-mass component were dropped in a subsequentfield application, the resulting partial index mightend up having an unacceptable high rate of mis-classifications. Good alternative metric combina-tions with high classification efficiencies were avail-able and were used in the selection process. Bio-mass was one of the least effective metrics at dis-tinguishing between degraded and non-degraded

conditions. Inclusion of biomass in the final indexwould have resulted only in small improvements.Most effective metrics on validation data werenumber of taxa and Shannon diversity in the highsalinity habitats, deep-deposit feeding taxa in thetidal freshwater, and pollution-indicative taxa inthe oligohaline.

One aspect of the index development processthat has been under emphasized in other studiesis the classification performance for the degradedsites. In the MAIA index development, special at-tention was given to attaining percent correct clas-sifications for the degraded sites of at least 50%.Classification efficiencies for the reference sites inthe calibration phase of the index were expectedto be high around 90%, as the thresholds used toscore the metrics were based on the distribution ofvalues at reference sites. Classification efficienciesfor the degraded sites are usually much lower be-

Benthic Index Development 1241

cause not all values of a metric at a degraded sitewould be expected to fall below the 10th percentileof corresponding reference values, especially incommunities that are stressed by natural factorsand show responses to natural disturbance thatcannot be distinguished from those due to pollu-tion. If only overall classifications were examined,it would be quite possible to select a metric thatcorrectly classified most reference sites while fail-ing to identify every single degraded site. Thismethod would be neither sensitive nor useful toenvironmental managers since they would be mostinterested in knowing where the problems oc-curred. Although classification efficiencies for theoligohaline and tidal freshwater habitats were notas high as we had hoped, the MAIA index appearsto be quite reliable in the more saline habitats,with a high likelihood of correctly identifying bothdegraded and non-degraded conditions. Cautionshould be used when applying the index to sam-ples collected from oligohaline and tidal freshwa-ter environments.

The MAIA index provides an uniform measurewith which to make comparisons of benthic con-dition in different estuaries within the mid-Atlanticregion of the U.S. Development of this new indexis based on benthic data from several estuaries inthe region, thereby widening its potential applica-tion beyond that of any index previously developedin the mid-Atlantic. The application of indices ofbiotic integrity beyond the region for which theywere developed may be compromised by the largenatural variability inherent in estuarine systems. In-dex development efforts should incorporate datathat are representative of the full range of biolog-ical, physical, chemical, and hydrodynamic char-acteristics of the region of application. Existing in-dices (e.g., Weisberg et al. 1993, 1997; Van Dolahet al. 1999; Paul et al. 2001) do not cover all partsof the MAIA region. The MAIA index has incor-porated geographic variation by using data collect-ed by several state and federal monitoring pro-grams that have sampled benthic assemblagesthroughout the region and in a wide variety of es-tuarine environments. The usefulness of the MAIAindex is that it combines wide geographical appli-cation with the simplicity of approach, easiness ofcalculation and interpretation, and effectiveness ofthe B-IBI method. The index is expected to be ofgreat utility in regional assessments as a tool forevaluating the integrity of benthic assemblages andtracking their condition over time.

ACKNOWLEDGMENTS

The U.S. Environmental Protection Agency (EPA), throughits Office of Research and Development, funded and managedthe research described here under contract number 68-D-98-

002 to Technology Planning and Management Corporation viaa subcontract with Versar, Inc. Participants of a benthic work-shop helped clarify issues and provided valuable input in theindex development process. We especially thank Richard Batiuk(Chesapeake Bay Program), Ananda Ranasinghe (Southern Cal-ifornia Coastal Research Program), and Elizabeth Ebersole(Maryland Department of Natural Resources) for critically re-viewing the manuscript. Thuzar Myint managed the databaseand conducted the analysis of data. Virginia Engle (EPA GulfEcology Division) and an anonymous reviewer provided valu-able peer-review comments. This manuscript has been subjectedto the EPA’s review and approved for publication. Mention oftrade names or commercial products does not constitute en-dorsement or recommendation by EPA for use.

LITERATURE CITED

ADAMS, D. A., J. S. O’CONNOR, AND S. B. WEISBERG. 1998. Sedi-ment quality of the NY/NJ Harbor system. Final report. Aninvestigation under the Regional Environmental Monitoringand Assessment Program (REMAP). EPA/902/R-98/001. U.S.Environmental Protection Agency—Region 2, Division of En-vironmental Science and Assessment, Edison, New Jersey.

ALDEN, III, R. W., D. M. DAUER, J. A. RANASINGHE, L. C. SCOTT,AND R. J. LLANSO. 2002. Statistical verification of the Chesa-peake Bay Benthic Index of Biotic Integrity. Environmetrics 13:473–498.

AMERICAN SOCIETY FOR TESTING MATERIALS. 1993. Guide for con-ducting 10-day static sediment toxicity tests with marine andestuarine infaunal amphipods. ASTM E-1367-92. AmericanSociety for Testing Materials, Philadelphia, Pennsylvania.

BARBOUR, M. T., J. GERRITSEN, G. E. GRIFFITH, R. FRYDENBORG,E. MCCARRON, J. S. WHITE, AND M. L. BASTIAN. 1996. A frame-work for biological criteria for Florida streams using benthicmacroinvertebrates. Journal of the North American BenthologicalSociety 15:185–211.

BRADLEY, M. P. AND R. B. LANDY. 2000. The mid-Atlantic inte-grated assessment (MAIA). Environmental Monitoring and As-sessment 63:1–13.

BRICKER, S. B., C. G. CLEMENT, D. E. PIRHALLA, S. P. ORLANDO,AND D. R. G. FARROW. 1999. National Estuarine Eutrophica-tion Assessment: Effects of Nutrient Enrichment in the Na-tion’s Estuaries. National Oceanic and Atmospheric Admin-istration, National Ocean Service, Silver Spring, Maryland.

BULICH, A. A., M. W. GREENE, AND D. L. ISENBERG. 1981. Reli-ability of the bacterial luminescence assay for determiningtoxicity of pure compounds and complex effluent, p. 338–347. In D. R. Branson and K. L. Dickson (eds.). Aquatic Tox-icology and Hazard Assessment, Proceedings of the FourthAnnual Symposium, ASTM STP 737. American Society forTesting and Materials, Philadelphia, Pennsylvania.

CHAILLOU, J. C., S. B. WEISBERG, F. W. KUTZ, T. E. DEMOSS, L.MANGIARACINA, R. MAGNIEN, R. ESKIN, J. MAXTED, K. PRICE,AND J. K. SUMMERS. 1996. Assessment of the ecological con-dition of the Delaware and Maryland coastal bays. EPA/620/R-96/004. U.S. Environmental Protection Agency, Office ofResearch and Development, Washington, D.C.

CULLITON, T. J., M. A. WARREN, T. R. GOODSPEED, D. G. REMER,C. M. BLACKWELL, AND J. J. MCDONOUGH, III. 1990. 50 yearsof population change along the nation’s coast. Second reportof a Coastal Trend Series. National Oceanic and AtmosphericAdministration, National Ocean Service, Rockville, Maryland.

DAUER, D. M. 1993. Biological criteria, environmental healthand estuarine macrobenthic community structure. Marine Pol-lution Bulletin 26:249–257.

DAUER, D. M., M. F. LANE, H. G. MARSHALL, AND K. E. CARPEN-TER. 1998. Status and trends in water quality and living re-sources in the Virginia Chesapeake Bay: 1985–1997. Preparedfor the Virginia Department of Environmental Quality, Rich-mond, Virginia. Old Dominion University, Norfolk, Virginia.

1242 R. J. Llanso et al.

EATON, L. 2001. Development and validation of biocriteria usingbenthic macroinvertebrates for North Carolina estuarine wa-ters. Marine Pollution Bulletin 42:23–30.

ENGLE, V. D. AND J. K. SUMMERS. 1999. Refinement, validation,and application of a benthic condition index for NorthernGulf of Mexico estuaries. Estuaries 22:624–635.

ENGLE, V. D., J. K. SUMMERS, AND G. G. GASTON. 1994. A benthicindex of environmental condition of Gulf of Mexico estuaries.Estuaries 17:372–384.

HARDING, L. W. 1994. Long-term trends in the distribution ofphytoplankton in Chesapeake Bay: Roles of light, nutrientsand stream flow. Marine Ecology Progress Series 104:267–291.

HILSENHOFF, W. L. 1987. An improved biotic index of organicstream pollution. Great Lakes Entomologist 20:31–39.

HYLAND, J. L., L. BALTHIS, C. T. HACKNEY, G. MCRAE, A. H. RING-WOOD, T. R. SNOOTS, R. F. VAN DOLAH, AND T. L. WADE. 1998.Environmental quality of estuaries of the Carolinian Province:1995. Annual statistical summary for the 1995 EMAP-EstuariesDemonstration Project in the Carolinian Province. NOAATechnical Memorandum NOS ORCA 123. National Oceanicand Atmospheric Administration, National Ocean Service,Office of Ocean Resources Conservation and Assessment, Sil-ver Spring, Maryland.

HYLAND, J. L., W. L. BALTHIS, C. T. HACKNEY, AND M. POSEY. 2000.Sediment quality of North Carolina estuaries: An integrativeassessment of sediment contamination, toxicity, and conditionof benthic fauna. Journal of Aquatic Ecosystem Stress and Recovery8:107–124.

KARR, J. R. 1981. Assessment of biotic integrity using fish com-munities. Fisheries 6:21–27.

KARR, J. R., K. D. FAUSCH, P. L. ANGERMEIER, P. R. YANT, AND I.J. SCHLOSSER. 1986. Assessing biological integrity in runningwaters: A method and its rationale. Special Publication 5. Il-linois Natural History Survey, Urbana, Illinois.

LENAT, D. R. 1993. A biotic index for the southeastern UnitedStates: Derivation and list of tolerance values, with criteria forassigning water-quality ratings. Journal of the North AmericanBenthological Society 12:279–290.

LLANSO, R. J., L. C. SCOTT, D. M. DAUER, J. L. HYLAND, AND D.E. RUSSELL. 2002. An estuarine benthic index of biotic integ-rity for the mid-Atlantic region of the United States. I. Clas-sification of assemblages and habitat definition. Estuaries 25:1219–1230.

LONG, E. R., D. D. MACDONALD, S. L. SMITH, AND F. D. CALDER.1995. Incidence of adverse environmental effects within rang-es of chemical concentrations in marine and estuarine sedi-ments. Environmental Management 19:81–97.

MALONE, T. C. 1987. Seasonal oxygen depletion and phytoplank-ton production in Chesapeake Bay: Preliminary results of1985–86 field studies, p. 54–60. In G. B. Mackiernan (ed.).Dissolved Oxygen in the Chesapeake Bay: Processes and Ef-fects. Maryland Sea Grant, College Park, Maryland.

MALONE, T. C., D. J. CONLEY, T. R. FISHER, P. M. GILBERT, L. W.HARDING, AND K. G. SELLNER. 1996. Scales of nutrient-limitedphytoplankton productivity in Chesapeake Bay. Estuaries 19:371–385.

NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION (NOAA).1994. Assessment of Chemical Contaminants in the Chesa-peake and Delaware Bays. NOAA, Coastal Monitoring andBioeffects Assessment Division, National Status and TrendsProgram, Silver Spring, Maryland.

ODUM, E. P. 1985. Trends expected in stressed ecosystems. Bio-science 35:419–422.

PAERL, H. W., J. L. PINCKNEY, J. M. FEAR, AND B. L. PEIERLS. 1998.Ecosystem responses to internal and watershed organic mat-ter loading: Consequences for hypoxia in the eutrophying

Neuse River estuary, North Carolina, USA. Marine Ecology Pro-gress Series 166:17–25.

PAUL, J. F., J. H. GENTILE, K. J. SCOTT, S. C. SCHIMMEL, D. E.CAMPBELL, AND R. W. LATIMER. 1999. EMAP-Virginian Prov-ince four-year assessment report (1990–93). EPA 600/R-99/004. U.S. Environmental Protection Agency, Atlantic EcologyDivision, Narragansett, Rhode Island.

PAUL, J. F., K. J. SCOTT, D. E. CAMPBELL, J. H. GENTILE, C. S.STROBEL, R. M. VALENTE, S. B. WEISBERG, A. F. HOLLAND, AND

J. A. RANASINGHE. 2001. Developing and applying a benthicindex of estuarine condition for the Virginian BiogeographicProvince. Ecological Indicators 1:83–99.

PEARSON, T. H. AND R. ROSENBERG. 1978. Macrobenthic succes-sion in relation to organic enrichment and pollution of themarine environment. Oceanography and Marine Biology AnnualReview 16:229–311.

PENNOCK, J. R., J. H. SHARP, AND W. W. SCHROEDER. 1994. Whatcontrols the expression of estuarine eutrophication? Casestudies of nutrient enrichment in the Delaware Bay and Mo-bile Bay estuaries, USA, p. 139–146. In K. R. Dyer and R. J.Orth (eds.). Changes in Fluxes in Estuaries: Implicationsfrom Science to Management. Olsen and Olsen, Fredensborg,Denmark.

RANASINGHE, J. A., L. C. SCOTT, AND F. S. KELLEY. 1999. Chesa-peake Bay water quality monitoring program, long-term ben-thic monitoring and assessment component. Level 1 Compre-hensive report, July 1984–December 1998. Prepared for Mary-land Department of Natural Resources, Tidewater EcosystemAssessments. Versar, Inc., Columbia, Maryland.

RINGWOOD, A. H., M. E. DELORENZO, P. E. ROSS, AND A. F. HOL-LAND. 1997. Interpretation of Microtox solid-phase toxicitytests: The effects of sediment composition. Environmental Tox-icology and Chemistry 16:1135–1140.

STROBEL, C. J. 1998. Environmental Monitoring and AssessmentProgram, Mid-Atlantic Integrated Assessment, Estuaries Com-ponent. Field Operations and Safety Manual. U.S. Environ-mental Protection Agency, Atlantic Ecology Division, Narra-gansett, Rhode Island.

TUTTLE, J. H., R. B. JONAS, AND T. C. MALONE. 1987. Origin,development and significance of Chesapeake Bay anoxia, p.442–472. In S. K. Majumdar, L. W. Hall, Jr., and H. M. Austin(eds.). Contaminant Problems and Management of LivingChesapeake Bay Resources. Pennsylvania Academy of Science,Philadelphia, Pennsylvania.

VAN DOLAH, R. F., J. L. HYLAND, A. F. HOLLAND, J. S. ROSEN, AND

T. R. SNOOTS. 1999. A benthic index of biological integrity forassessing habitat quality in estuaries of the southeastern USA.Marine Environmental Research 48:269–283.

WARWICK, R. M. 1986. A new method for detecting pollutioneffects on marine macrobenthic communities. Marine Biology92:557–562.

WEISBERG, S. B., J. B. FRITHSEN, A. F. HOLLAND, J. F. PAUL, K. J.SCOTT, K. J. SUMMERS, H. T. WILSON, R. M. VALENTE, D. G.HEIMBUCH, J. GERRITSEN, S. C. SCHIMMEL, AND R. W. LATIMER.1993. EMAP-Estuaries Virginian Province 1990 Demonstra-tion Project Report. EPA/620/R-93/006. U.S. EnvironmentalProtection Agency, Narragansett, Rhode Island.

WEISBERG, S. B., J. A. RANASINGHE, D. M. DAUER, L. C. SCHAFF-NER, R. J. DIAZ, AND J. B. FRITHSEN. 1997. An estuarine benthicindex of biotic integrity (B-IBI) for the Chesapeake Bay. Es-tuaries 20:149–158.

WILSON, J. G. AND D. W. JEFFREY. 1994. Benthic biological pol-lution indices in estuaries, p. 311–327. In J. M. Krammer(ed.). Biomonitoring of Coastal Waters and Estuaries. CRCPress, Boca Raton, Florida.

Received for consideration, September 26, 2001Accepted for publication, May 17, 2002