Transactions of the American Fisheries Society 122:63-73, 1993

A Quantitative Health Assessment Index forRapid Evaluation of Fish Condition in the Field

S. MARSHALL ADAMSEnvironmental Sciences Division. Oak Ridge National Laboratory

Oak Ridge, Tennessee 37831, USA

ALLEN M. BROWNAquatic Biology Laboratory, Tennessee Valley Authority

Norris, Tennessee 37828, USA

RONALD W. GOEDEUtah Division of Wildlife Resources, Fisheries Experiment Station

Logan, Utah 84321, USA

Abstract.—The health assessment index (HAI) is an extension and refinement of a previouslypublished field necropsy system. The HAI is a quantitative index that allows statistical comparisonsof fish health among data sets. Index variables are assigned numerical values based on the degreeof severity or damage incurred by an organ or tissue from environmental stressors. This approachhas been used to evaluate the general health status offish populations in a wide range of reservoirtypes in the Tennessee River basin (North Carolina, Tennessee, Alabama, Kentucky), in HartwellReservoir (Georgia, South Carolina) that is contaminated by polychlorinated biphenyls, and inthe Pigeon River (Tennessee, North Carolina) that receives effluents from a bleached kraft mill.The ability of the HAI to accurately characterize the health of fish in these systems was evaluatedby comparing this index to other types of fish health measures (contaminant, bioindicator, andreproductive analysis) made at the same time as the HAI. In all cases, the HAI demonstrated thesame pattern offish health status between sites as did each of the other more sophisticated healthassessment methods. The HAI has proven to be a simple and inexpensive means of rapidlyassessing general fish health in field situations.

The biotic integrity of an ecological system isoften reflected by the health of organisms that re-side in that system. In aquatic ecosystems, fish,and particularly those species near the top of thefood chain, are generally regarded as representa-tive indicators of overall system health. Becauseof their position in the food chain, fish integratethe effects of many biotic and abiotic variablesacting in the system and reflect secondary impactsof chronic stress mediated through the food chain(Larkin 1978; Adams and McLean 1985).

Fish in their natural environments are typicallysubjected to numerous stressors including unfa-vorable or fluctuating temperatures, high watervelocities and sediment loads, low dissolved ox-ygen concentrations, limited food availability, andother types of episodic variables. In addition, an-thropogenic stressors such as contaminant loadingcan add to the insults that fish may already ex-perience in many systems. All these factors, in-dividually or together, can impose considerablestress on physiological systems offish and impairtheir health (Wedemeyer et al. 1984). When anorganism is challenged by environmental stress-

ors, energy is required to deal with that stress,diverting physiologically useful energy away fromthe critical functions of growth and reproduction(Barton and Schreck 1987; Wedemeyer et al. 1990).Depending on its severity, stress can load or limitphysiological systems, reduce growth, impair re-production, predispose fish to disease, and reducethe capacity offish to tolerate additional stressors(Adams 1990b; Barton and Iwama 1991).

A variety of approaches has been used to eval-uate the effects of stress on the health of fish pop-ulations. These approaches have been applied tomany types of aquatic ecosystems that experiencea variety of environmental stressors. Some of themore commonly used approaches for assessing fishhealth are age and growth analysis and the con-dition factor (Le Cren 1951; Bagenal and Tesch1978; Busacker et al. 1990), various condition ororganosomatic indices (Goede and Barton 1990),and numerous measures of biochemical, physio-logical, and pathological condition (Neff 1985;Adams 1990a; Niimi 1990). Each of these typesof health measure has its own set of advantagesand limitations, depending on the objectives of

63

64 ADAMS ET AL.

the particular study, but most of them cannot berapidly and inexpensively applied to field studies.In many cases, samples have to be processed andanalyzed in a laboratory, which requires varyingdegrees of time and expense. For example, deter-mination of condition indices may involve onlysimple length and weight measures, whereas bio-chemical, physiological, and pathological analysesoften require specialized training, instrumenta-tion, and possibly large time commitments.

As a rapid and inexpensive alternative to thesemore sophisticated approaches for evaluating fishhealth and condition, Goede and Barton (1990)developed a field necropsy method that providesa health profile offish based on the percentages ofanomalies observed in the tissues and organs ofindividuals sampled from a population. The ma-jor purpose of the necropsy method is to detectgross changes in the health offish poulations earlyenough for corrective or remedial actions to betaken. The primary value of this approach is thatit provides a means of establishing a data base fordetecting trends in the health of a fish populationover time. If, by applying this method, a changeor trend is indicated in the health of members ofa population, then more detailed and specializedprocedures can be applied to the problem. Theconcept of this strategy is analogous to that usedby the medical profession. A physician may use abattery of generalized tests on a patient to helpdiagnose an illness. If the results of a general healthscreening help to focus on the nature of the prob-lem, then more specific measures can be used forfurther diagnosis.

The major limitation of the current necropsy orhealth profile method is that it does not providequantitative results that are amenable to statisti-cal comparison of data among sites, species, oryears. This method also does not currently ac-count for severity or degree of damage in some ofthe variables within the necropsy system. Our ob-jectives in this study, therefore, were to modifyand refine the necropsy-based approach to (1) pro-vide a quantitative index so that statistical com-parisons can be made between data sets, (2) in-clude variables in the health index that reflect thedegree of damage incurred as a result of environ-mental stressors, and (3) provide examples of theuse of this index in different types of aquatic sys-tems to demonstrate and validate its applicability.

ApproachThe necropsy method of Goede and Barton

(1990) currently consists of 16 variables that can

be grouped into the following categories: (1) threeblood parameters (hematocrit, leukocrit, and plas-ma protein); (2) length, weight, and condition fac-tor; (3) percentage offish with normal and abnor-mal eyes, gills, pseudobranchs, spleens, kidneys,and livers; and (4) index values of damage to skin,fins, thymus, hindgut inflammation, fat deposits,and bile color. Analysis and evaluation of the fishnecropsy data involve a summary of the meansfor variables in group 4, a summary of the per-centage of normals in group 3, means and vari-ances of blood constituents in group 1, and lengthand weight data in group 2. Even though the nec-ropsy method provides a health status profile ofa fish population, there is no quantitative basis forcomparing statistically the entire index with all itsvariables to another population sample either intime or space. In addition, past experience withthe original necropsy method has shown that somevariables within the index, such as conditions ofthe skin, gills, and fins, demonstrate varying levelsof damage that should be taken into account whenthe health of a fish population is evaluated. Thehealth assessment index (HAI), which we presenthere, is intended to minimize these limitations ofthe necropsy method by rendering it quantitativefor statistical analysis and comparisons among datasets.

Variable QuantificationFor the HAI to have a statistical basis, all vari-

ables within the index must be assigned a numer-ical value. To assign a numerical value of condi-tion to each variable within the HAI, all variablesare first given a field code designation accordingto the original necropsy classification criteria ofGoede and Barton (1990). A numerical substitu-tion is then made for each assigned code (Table1). All codes that represent a normal condition arereplaced by a zero, and all codes that represent anabnormal condition or anomaly assume a valueof 30. For example, the normal liver color forcentrarchids such as black basses Micropterus spp.and sun fish Lepomis spp. is dark or light red, whichis therefore assigned a value of zero (Table 1).Other liver conditions that are considered to beabnormal are assigned a value of 30 (Table 1). Amaximum value of 30 was chosen to provide asuitable range for value ranking. The criteria bywhich variables in the HAI are assigned variousconditions or rankings are admittedly somewhatsubjective. The nature of this variable ranking sys-tem is consistent, however, with the principal pur-pose and objective of the HAI approach, which is

QUANTITATIVE FIELD HEALTH ASSESSMENT INDEX 65

to provide a rapid and simple field methodologyfor characterizing the general health of membersof a fish population.

All variables except bile color and the mesen-teric fat level were used in the calculation of theHAL Bile color could not be assigned a numericalvalue based on normality because bile can take ondiffering colors depending on feeding regimes ofthe fish (time since last meal, meal quantity andquality, etc.). Mesenteric fat deposits in the fish,or the lipid index, can vary widely depending notonly on food availability and feeding regimes, buton other interacting factors such as fish size, sex,time of year, and stress level. Because of theseinteracting variables, the lipid index could not beassigned normal or abnormal condition values.For hematocrit and plasma protein, the normalranges should be established for each species orgroups of similar species for major geographicalareas of the country (e.g., southeast, northwest).

Variable RankingTo account for differences in severity of damage

or level of effect, some variables of the HAI areassigned values of 10, 20, or 30, depending on theextent of the abnormality or observed damage(Table 1). For example, if fins with light activeerosion are noted, a value of 10 is assigned. A fincondition characterized by moderate active ero-sion with some hemorrhaging rates a value of 20,and severe erosion with hemorrhaging receives aranking of 30. Other variables of the HAI thatreceive assigned values based on the severity ofdamage or condition are parasite loads, thymuscondition, hindgut inflammation, skin condition,and hematocrit and plasma protein levels (Table1).

Calculation of the HAITo calculate an HAI for each fish within a sam-

ple, numerical values for all variables are summed.The HAI for a sample population is then calcu-lated by summing all individual fish HAI valuesand dividing by the total number offish examinedfor that sample. A standard deviation for eachsample is calculated as

SDN- 1

N = number of fish per site;X = average index for each site;K/ = index value for fish /.

The coefficient of variation (CV) is calculated asCV = lOO-SD/A'.

The HAI for the sample population is com-posed of multiple HAI observations calculated foreach fish within that sample. Therefore, statisticalcomparisons can be made among sample sites,sample times for the same site, and even species.Since the HAI value for a given site or system isa mean based on the number of sampled fish inthe population, the central limit theorem is a jus-tification for using parametric procedures for sta-tistical comparisons among data sets. The HAI foreach sample site is a mean based on a relativelylarge sample size (e.g., N = 30); therefore, the dis-tribution of individual HAI values for a site tendsto be normally distributed according to the centrallimit theorem (Snedecor and Cochran 1980). Inaddition to these parametric procedures, non-parametric rank statistics could also be used fortesting differences among data sets.

In addition, the CV of each sample can be cal-culated and used to indicate the level or degree ofstress experienced by a fish population. For fishexperiencing stress resulting from poor waterquality, for example, we might except the CV tobe lower for that sample than for a reference pop-ulation because all individuals within the stressedpopulation should be equally exposed to waterquality stressors. In contrast, when fish are ex-posed to bacterial infection, the vulnerability ofindividuals may differ according to the compe-tence of their immune systems and other inherentbiochemical and physiological factors. In this sit-uation, variability in a stress response should behigher among members of the stressed than of thereference population. Caution must be exercisedin the interpretation of the CV because other fac-tors such as fish size distribution, sex, and speciesmay influence the inherent variability of stress re-sponses in fish.

Table 2 provides an example of how an HAI iscalculated for a sample fish population. An indexis first determined for each fish based on the sum-mation of the anomaly values for that fish. Theindividual index values are then used to calculatea population sample mean for the HAI. Letter andnumber codes for each variable in Table 2 followthose in Table 1. The parenthetical numbers inTable 2 are the values (10, 20, or 30) assigned tothe observed anomaly that are summed to cal-culate an individual fish index value.

In the example given in Table 2, pathologieswere most often observed in the liver and spleen;

66 ADAMS ET AL.

TABLE 1.—Description of variables used in the health assessment index (HAI). Values are assigned to each ofthese variables according to the type and severity of the observed anomoly (modified from Goede and Barton 1990).

Variable

Thymus

Fins

Spleen

Hindgut

Kidney

Skin

Liver

Eyes

Gills

Pseudobranchs

Parasites

Variable conditionNo hemorrhageMild hemorrhageModerate hemorrhageSevere hemorrhageNo active erosionLight active erosionModerate active erosion with some hcmorrhagingSevere active erosion with hemorrhagingNormal; black, very dark red, or redNormal; granular, rough appearance of spleenNodular; containing fistulas or nodules of varying sizesEnlarged; noticeably enlargedOther; gross aberrations not Biting above categoriesNormal; no inflammation or reddeningSlight inflammation or reddeningModerate inflammation or reddeningSevere inflammation or reddeningNormal; firm dark red color, lying relatively flat along the length

of the vertebral columnSwollen; enlarged or swollen wholly or in partMottled; gray discolorationGranular; granular appearance and textureUrolithiasis or ncphrocalcinosis: white or cream-

colored mineral material in kidney tubulesOther; any aberrations not fitting previous categoriesNormal; no aberrationsMild skin aberrationsModerate skin aberrationsSevere skin aberrationsNormal; solid red or light red color"Fatty" liver; "coffee with cream" colorNodules in the liver; cysts or nodulesFocal discoloration; distinct localized color changesGeneral discoloration; color change in whole liverOther; deviation in liver not fitting other categoriesNo aberrations; good "clear" eyeGenerally, an opaque eye (one or both)Swollen, protruding eye (one or both)Hemorrhaging or bleeding in the eye (one or both)Missing one or both eyesOther, any manifestation not fitting the aboveNormal; no apparent aberrationsFrayed; erosion of tips of gill lamellae resulting in "ragged" gillsClubbed; swelling of the tips of the gill lamellaeMarginate; gills with light, discolored margin along tips of the la-

mellaePale; very light in colorOther; any observation not fitting aboveNormal; flat, containing no aberrationsSwollen; convex in aspectLithic; mineral deposits, white, somewhat amorphous spotsSwollen and lithicInflamed; redness, hemorrhage, or otherOther; any condition not covered aboveNo observed parasitesFew observed parasites

Originalfield

designation

01230123BGDEOT0123

NSMG

UOT0123ACDEFOTNBEHMOTNFC

MPOTNSLS&LIOT01

Substi-tuted

value forthe HAI

01020300

10203000

3030300

102030

0303030

30300

1020300

30303030300

30303030300

3030

3030300

30303030300

10

QUANTITATIVE FIELD HEALTH ASSESSMENT INDEX 67

TABLE 1.—Continued.

Variable

Hemaiocriia

Leukocrit

Plasma proteinb

Variable condition

Moderate parasite infestationNumerous parasitesNormal rangeAbove normal rangeBelow normal rangeBelow normal rangeRange defined as normalOutside the normal rangeNormal rangeAbove normal rangeBelow normal range

Originalfield

designation

23

30-45%>45%

19-29%<18%<4%>4%

30-69 mg/dL>70mg/dLb

<30 mg/dL

Substi-tuted

value forthe HAI

20300

1020300

300

1030

a Normal ranges for ccntrarchid species such as largemouth bass and redbreast sunfish.b Values greater than 70 mg/dL are generally inaccurate because of factors that interfere with the protein analysis such as elevated

serum lipids.

lesser frequencies of anomalies were recorded forkidney, parasite loads, plasma protein concentra-tions, and leukocrit. All index variables wereanomalous for at least one fish in the sample. Fish3 and 11 were in very poor health, as indicatedby their high index values. For example, fish 3with an index value of 190 had a fatty liver (30),moderate skin aberrations (20), light active finerosion (10), a swollen pseudobranch (30), mildhemorrhaging of the thymus (10), moderate in-flammation of the hindgut (20), a granular kidney(30), a few internal parasites particularly in theheart and liver (10), and an abnormally high leu-kocrit (30). The condition of most of the fish inthis sample, however, was not this poor, as indi-cated by the population mean HAI of 97.3.

An HAI for the sample population can be cal-culated in the field within 5 min after the healthassessment procedure is completed. Data for eachfish are entered into a portable computer, and aspreadsheet program with a series of macros as-signs each observed anomaly a numerical valueas in Table 1, tabulates the HAI for each fish, andcalculates a mean, SD, and CV for the samplepopulation. Comparisons of the HAIs can then bemade among sample sites, between differentaquatic systems, or at the same site over time toestablish temporal patterns in fish populationhealth. Statistical comparisons can also be madeamong indices and environmental conditions todetermine correlations or relationships betweenfish health and environmental variables such astemperature, dissolved oxygen, or turbidity. Nor-mal or expected HAI values for a population are

established by sampling unaffected areas over timeand comparing results to affected sites.

ApplicationThe use of the HAI to assess the health of fish

populations in the field is demonstrated in thissection by three examples of the index's successfulapplication. This approach is currently being usedby the Tennessee Valley Authority (TVA) to eval-uate the general health status of fish in a widerange of reservoir types over the entire TennesseeValley (North Carolina, Tennessee, Alabama.Kentucky). The HAI has also been applied in res-ervoirs and rivers to assess the effects on fish healthof polychlorinated biphenyl (PCB) contaminationin Hartwell Reservoir (South Carolina, Georgia)and of pulp and paper effluents in the Pigeon River(North Carolina, Tennessee).

TV A Reservoir SurveyThe TVA conducts an annual monitoring sur-

vey on 28 reservoirs, including 11 mainstreamsystems and 17 tributary storage impoundments.The objectives of this "vital signs" survey are toprovide basic information on reservoir health orintegrity and to obtain screening-level informa-tion that is used to evaluate each reservoir withrespect to the mandates of the U.S. Clean WaterAct. The HAI is used along with other measuresof reservoir condition such as physical and chem-ical characterization of the water and sediment,surveys of benthic macroinvertebrates, and acutetoxicity screening to assess the overall health ofthe reservoir ecosystem. To obtain information

68 ADAMS ET AL.

TABLE 2.—An example of the calculation of the health assessment index based on a sample of 15 fish. Letter andnumber designations for each variable of the HAI follow those given in Table 1; the first entry is the field designationand the number in parentheses is the value assigned to an observed anomaly. Parenthetic numbers are summed tocalculate an index value for each fish. Fat and bile designations do not enter the index.

Fishnumber

or .sta-tistic

123456789

101112131415

MeanSD

Health assessment index variable

Liver

A(0)F(30)C(30)E(30)A(0)A(0)C(30)C(30)A(0)F(30)C<30)A(0)A(0)A(0)F(30)

Eye

N(0)N(0)N(0)N(0)B(30)N(0)N(0)N(0)N(0)N(0)B(30)N(0)H(30)N(0)N(0)

Skin

0(0)0(0)2(20)0(0)0(0)0(0)0(0)0(0)2(20)0(0)2(20)0(0)0(0)0(0)0(0)

Fin

0(0)0(0)1(10)0(0)0(0)0(0)0(0)2(20)0(0)0(0)3(30)0(0)0(0)0(0)0(0)

Pseudo-branchS(30)N(0)S(30)N(0)N(0)N(0)N(0)N(0)N(0)N(0)N(0)S(30)N(0)N(0)N(0)

Thymus

0(0)0(0)1(10)0(0)0(0)0(0)0(0)2(20)0(0)0(0)0(0)0(0)0(0)0(0)0(0)

Fat

321340214230132

SpleenD(30)G(0)G(0)

OT(30)D(30)E(30)D(30)E(30)0(0)B(0)E(30)D(30)B(0)E(30)G(0)

Hindgut

0(0)0(0)2(20)0(0)0(0)0(0)1(10)0(0)0(0)0(0)0(0)0(0)0(0)1(10)0(0)

Kidney

N(0)N(0)G(30)S(30)S(30)S(30)N(0)N(0)N(0)N(0)N(0)N(0)N(0)N(0)S(30)

Bile

220313003330210

for the HAI, 30 largemouth bass Micropterus sal-moides, ranging in size from 0.5 to 2.5 kg, werecollected by boat electrofishing at three areas ineach reservoir, including the upper section, tran-sition area (midlake), and the lower section nearthe dam. An HAI was then calculated for eachfish, for each area of the reservoir, and for theentire reservoir.

The average HAI for all reservoirs in the TVAvalleywide survey was 62; Watts Bar (63) repre-sented the average impoundment in the system(Table 3). Watts Bar is a large mainstream systemcharacterized by moderate levels of primary andsecondary productivity and high standing crops offorage fish. The range of reservoir HAI values forthe 22 impoundments in 1991 was 17 (best) forthe relatively pristine Watauga system to 79 (worst)for Chickamauga Reservoir. Watauga Reservoir,the system with the healthiest largemouth basspopulation, is a mountain headwater impound-ment that supports a well-balanced warm water andcoolwater fishery. As reflected by the low HAIscore, water quality in this system is very goodand there are no major nutrient or contaminantinputs into the system. Conversely, ChickamaugaReservoir, a large mainstream impoundment be-low Watts Bar on the Tennessee River, had thehighest HAI score, reflecting a largemouth bass

population that is in relatively poor conditioncompared to the other TVA impoundments. Thisreservoir receives contaminants from numeroussources, the most notable being the Hiwassee Riv-er into which effluents from a pulp and paper plantare discharged.

Hart well ReservoirHartwell Reservoir is a 24,400-hectare U.S.

Army Corps of Engineers impoundment of theSavannah River along the South Carolina-Geor-gia border. The fish in the upper northeast sectorof the reservoir in South Carolina contain highlevels of PCBs (Gaymon 1988, 1990); body bur-dens generally decrease downstream toward thedam. The Tugaloo River arm, primarily in Geor-gia, is relatively free of contaminants and servedas a reference site for this study. At each of threesites in this reservoir, 30 largemouth bass, rangingin size from 0.5 to 2.0 kg, were collected by boatelectrofishing and the HAI was determined. Sitelocations and mean PCB concentrations (wet-weight basis) in the fillets of largemouth bass were(1) 12 Mile Creek (21 /*g/g), (2) 18 Mile Creek (2.0Mg/g), and (3) Tugaloo River arm (0.3 Mg/g).

The mean value of the HAI for the largemouthbass sample population was highest at 12 MileCreek, lowest at the reference site, and interme-

QUANTITATIVE FIELD HEALTH ASSESSMENT INDEX 69

TABLE 2.—Extended.

Fishnum-beror .

sta-tistic

123456789

10I I12131415

MeanSD

Health assessment index variablePara-sites0(0)0(0)1(10)3(30)0(0)2(20)2(20)0(0)1(10)0(0)0(0)0(0)3(30)1(10)0(0)

Plasmaprotein

34(0)41(0)45(0)51(0)37(0)28 (30)35(0)52(0)26 (30)45(0)32(0)33(0)36(0)40(0)29 (30)

Leuko-crit

MO)3(0)4(30)3(0)3(0)2(0)1(0)1(0)4(30)3(0)4(30)3(0)MO)1(0)1(0)

Hemat-ocrit

33(0)36(0)40(0)37(0)37(0)42(0)29 (30)40(0)37(0)27 (30)28 (30)41(0)43(0)29 (30)44(0)

Indexvalue

6030

19012090

110no1009060

20060609090

97.346.5

diate at 18 Mile Creek (Table 3), reflecting thegradient of PCB contamination at these sites.Anomalies in three of the index variables (liver,gill, and kidney) were primarily responsible forinfluencing the HAI of largemouth bass in Har-twell Reservoir (Table 4). Liver condition, whichwas based primarily on color, was responsible for50-57% of all abnormal observations at the 12Mile and 18 Mile creek sites and 17% of the anom-alies at the reference site. Many of the abnormallivers for fish at 12 Mile and 18 Mile creeks hada light or coffee-cream color, indicative of high fatdeposition. A fatty liver usually is a pathologicalstate attributable to excessive accumulation of lip-ids in cellular cytoplasm (Roberts 1978). Fat ac-cumulation can result from an inability to convertlipids in hepatocytes to a phospholipid form suit-able for use (Runnells et al. 1965). This pathologyhas been observed in fish exposed to PCBs andother organic compounds (Lipsky et al. 1978;Klaunig et al. 1979). Spleens and kidneys also pre-sented important anomalies, primarily in large-mouth bass from the 12 Mile Creek site. At 12Mile Creek, kidneys were swollen, which is a gen-eral indicator of pathology (Goede and Barton1990). Spleen abnormalities in fish from both 12Mile and 18 Mile creeks were primarily the resultof enlargement, whereas abnormal spleens from

TABLE 3.—Health assessment index (HAI) values forlargemouth bass from the Tennessee Valley Authority(TVA) survey and sites in Hartwell Reservoir (Georgia-South Carolina), and for redbreast sunfish from the Pi-geon River (North Carolina-Tennessee) study.

Aquatic system

Coeffi-cient

of vari-HAI SD ation

TVA valleywidcMean of all reservoirsAverage reservoir (Watts Bar)Healthiest reservoir (Watauga)Worst reservoir (Chickamauga)

Hartwell ReservoirReference site (Tugaloo River)Intermediate site (18 Mile Creek)Contaminated site (12 Mile Creek)

Pigeon RiverPigeon River, km 95Pigeon River, km 35Little River (reference)Little Pigeon River (reference)

62 15.6 25.263 8.6 13.717 3.5 20.179 3.2 4.1

42 36.0 85.764 36.9 57.774 30.8 41.8

60 35.6 59.751 35.7 69.521 23.2 108.835 25.4 72.6

the reference fish were caused by nodules. Becausethe spleen serves a hemopoietic function, enlarge-ment could indicate bacterial infections or dis-ease. Abnormal levels of plasma protein and he-matocrit also contributed to the higher overall HAIat the 12 Mile and 18 Mile creek sites. The lowhematocrit values at these two sites may indicatea disease or pathogenic infection (Cardwell andSmith 1971).

Pigeon RiverThe Pigeon River is a high-gradient stream that

originates in the mountains of western North Car-olina, receives effluents from a bleached kraft millat river kilometer 105, and then flows northwestbefore joining the French Broad River near Knox-ville, Tennessee. Along the entire length of thePigeon River below the mill, the water has a darktea color, due to staining by dissolved organics,and contains chlorinated phenols and resin acidscharacteristic of bleached kraft mill effluents (Leachand Thakore 1973). At sites 10 and 70 km down-stream from the mill outfall, 35 adult redbreastsunfish Lepomis auritus were sampled by boatelectrofishing. From two reference streams (LittleRiver and Little Pigeon River, which are similarin size, flow, bottom substrate, and gradient to thePigeon River), equal numbers of redbreast sunfishwere also collected. Fish collected from all threestreams were examined, and an HAI was calcu-lated for the sample population at each site.

70 ADAMS ET AL.

TABLE 4.—Percentage offish with tissue and organ anomalies in a sample collected from sites in Hartwell Reservoir(Georgia-South Carolina) and the Pigeon River (North Carolina-Tennessee).

SitePercentage of fish with anomalies in

Liver Gill Kidney Spleen Eyes Fins

Hartwell ReservoirReference site18 Mile Creek12 Mile Creek

Pigeon RiverPigeon River, km 95Pigeon River, km 35Little River (reference)Little Pigeon River (reference)

175057

47501727

73317

53500

20

202743

20302317

151330

20637

000

13300

007

0600

Fish collected 10 km downstream from the pa-per mill (river km 95) had the highest HAI (poor-est health), and the mean index decreased slightlyfor redbreast sunfish sampled 70 km downstreamof the mill (river km 35) (Table 3). The mean HAIvalues for both reference sites, however, were con-siderably lower than those from the two PigeonRiver sites. The mean HAI for redbreast sunfishfrom the Little River was less than 50% of thevalue of the two Pigeon River sites, and the indexfrom the Little Pigeon River was bout 35% lowerthan at the contaminated sites (Table 3).

Of the 14 variables included in the calculationof the HAI, 6 were responsible for most of theabnormalities observed in fish from the PigeonRiver and reference sites (Table 4). Changes in theliver, gill, and kidney were the main anomaliesobserved; lesser effects were recorded for thespleen, eyes, and fins. As with the anomalies ob-served in the fish exposed to PCBs in HartwellReservoir, many of the livers in the fish collectedfrom the Pigeon River were characterized by highfat levels and focal discoloration. Focal discolor-ation can be due to various factors including focalnecrosis caused by bacterial infections. Gill dam-age is highly characteristic of fish exposed to pulpmill effluents (Lehtinen et al. 1984; Couillard etal. 1988). Effects on gills include hyperplasia, club-bing of lamellae, and lamellar fusion that not onlyimpairs gas exchange but also affects osmoregu-latory and excretory functions in fish. Anomaliesin the kidney were observed in 20-30% of the fishexamined from the Pigeon River (Table 4). Rel-atively high proportions (17-23%) offish from thereference sites also had abnormal kidneys, but theanomalies there were attributed primarily toswelling, probably caused by parasitic infestation.In addition to swelling and parasitic infestation,abnormal kidneys in fish from the Pigeon River

could have resulted from tubule damage and ne-crosis caused by the pulp mill effluents (Santos etal. 1990).

In both the Pigeon River and Hartwell Reser-voir studies, the CVs were higher for fish from thereference sites than for those from the contami-nated sites (Table 3). One possible interpretationof this finding is that fish living in degraded en-vironments are all exposed equally to constantstressors such as poor water quality and, therefore,the variability in physiological condition in fishfrom a contaminant-exposed population may tendto be less than for fish in unstressed environments.

Evaluation of the HAIThe ability of the HAI to characterize the health

profile of a fish population can be evaluated bycomparing this index to other measures of fishhealth taken at the same time as the HAI. In Hart-well Reservoir, Adams and Greeley (1991) usedthree other biomonitoring approaches, in additionto the HAI, for assessing the condition of thelargemouth bass population: (1) levels of PCBs inthe flesh and gonads; (2) biological indicators in-cluding detoxification enzymes, histopathology,blood chemistries, and bioenergetic indices; and(3) reproductive competence. All four methodsprovided similar conclusions relative to the healthstatus of largemouth bass at the three samplingsites in Hartwell Reservoir (Table 5). The healthprofile of the largemouth bass population in Hart-well Reservoir, regardless of the method of anal-ysis, follows an inverse gradient with PCB levelsin fish, being best at the reference site and worstin the 12 Mile Creek section of the reservoir. Onearea in which the HAI differs from the three othermethods, however, is in the ability of the indexto distinguish the condition offish at 12 Mile Creekfrom the health offish at 18 Mile Creek. The HAI

QUANTITATIVE FIELD HEALTH ASSESSMENT INDEX 71

TABLE 5.—Comparison of four biomonitoring techniques used in the Hartwell Reservoir study and their relativeability to distinguish the health status of largemouth bass at each sampling site.

Biomonitoringtechnique

Relative health status of largcmouth bass

Best Intermediate Worst Comments

Health assessmentindex

Reference 18 Mile Creek 12 Mile Creek PCBs highest at 12 Mile Creek in filletsand ovaries

Reference 18 Mile Creek 12 Mile Creek Based primarily on integrated responseanalysis with biochemical, physiolog-ical, histopathological, and bioener-getic measures

Based primarily on abundance of vitel-logenic oocytes. ovary size, and estro-gen receptor data

12 Mile Creek 12 Mile and 18 Mile creek HAI valuessimilar; HAI only indicates depar-tures from normal: not diagnostic

Contaminant analysis

Bioindicator analysis

Reproductive analysis Reference 18 Mile Creek 12 Mile Creek

Reference 18 Mile Creek

indicates that the health status of largemouth bassfrom these two sites was similar (e.g., HAI valuesof 74 and 64 for 12 Mile and 18 Mile creeks,respectively), whereas the other methods suggestlower levels of effects at the 18 Mile Creek site.This apparent discrepancy implies that the HAIis either a much more sensitive early warning in-dex of health effects than the other techniques orthat it could be overestimating the condition sta-tus offish, particularly at 18 Mile Creek.

In the Pigeon River, several health assessmentmethods were also compared with the HAI toevaluate the ability of this index to characterizepopulation health. This study included indicatorsof (1) biochemical and physiological changes, (2)reproductive abnormalities, (3) population effects,and (4) community-level disturbances. The HAIdemonstrated the same patterns of fish health sta-tus among sites as did each of the other assessmentmethods (Adams et al. 1992). For example, thesite with the highest HAI value (poorest health)nearest the pulp mill outfall was correlated withthe highest levels of detoxification enzyme induc-tion (an indicator of contaminant exposure), thelargest number of reproduction-related anomalies,an abnormal size and age structure distribution ofthe population, and the lowest index of biotic in-tegrity (IBI; a measure of fish community health)(Adams et al. 1992).

Results from these two case histories demon-strate that the HAI can be a reliable method forassessing the general health status of a fish pop-ulation. Furthermore, the HAI has the added ad-vantage over many other approaches of being rel-atively simple, rapid, and cost-effective. Thesimplicity of this approach, however, may also beone of its primary limitations because of its weak-

ness as a diagnostic tool. The HAI is not designedto be diagnostic in character, but to provide a first-level assessment of the health profile of a fish pop-ulation. If the HAI identifies a general healthproblem in a population, then more specific as-sessment approaches could follow, such as the ap-plication of biological indicators (Adams et al.1989; Adams 1990a). Another potential applica-tion of the HAI is its inclusion as an additionalmetric in the IBI. Currently, the IBI consists ofabout 12 metrics that are collectively used to ob-tain a single index score on the ecological statusof a fish community. The HAI could be includedas an additional metric in the IBI that would serveas an indicator of the health of individual fish inthat population or community.

The particular approach taken to determine thehealth status of a fish population depends on theobjectives and needs of a particular user group(e.g., fish culture, environmental quality, or fish-ery management) and the resources available(technical expertise, funding, etc.). For issues deal-ing with fish culture and fishery management, thenecropsy method of Goede and Barton (1990) isuseful in helping to diagnose the nature of a spe-cific problem. For example, a fish culturist maywant to determine the success of various types offeeding programs or evaluate the effects of diseaseon condition. The objectives of the fishery man-ager may be to evaluate the results of a stockingprogram or the bioenergetic status offish. In issuesdealing with water quality, the HAI could be ap-plied to assessing the general health of fish pop-ulations. It should be emphasized that the HAIand the necropsy method are designed to comple-ment each other in the evaluation of fish health.The degree to which they should be used sepa-

72 ADAMS ET AL.

rately or together depends primarily on the needsof a particular user group. In this regard, futurerefinements in the HAI and necropsy method willinvolve developing additional components of thesetwo approaches that will address more specificconcerns related to environmental quality, fishculture, or fishery management.

AcknowledgmentsWe are grateful to B. A. Barton and J. B. Hunn

for their many helpful suggestions and comments.We also thank C. C. Coutant and M. G. Ryon forreviewing the original manuscript and W. D.Crumby for his help in field collections and fishhealth assessment. Oak Ridge National Labora-tory is managed by Martin Marietta Energy Sys-tems, Inc. under contract DE-AC05-84OR21400with the U.S. Department of Energy. This is pub-lication 3942, Environmental Sciences Division,Oak Ridge National Laboratory.

ReferencesAdams, S. M., editor. 1990a. Biological indicators of

stress in fish. American Fisheries Society Sympo-sium 8.

Adams, S. M. 1990b. Status and use of biological in-dicators for evaluating the effects of stress on fish.American Fisheries Society Symposium 8:1-8.

Adams, S. M., W. D. Crumby, M. S. Greeley Jr., andL. R. Shugart. 1992. Responses of fish populationsand communities to pulp mill effluents: a holisticassessment. Ecotoxicology and Environmental Safety24:347-360.

Adams, S. M., and M. S. Greeley, Jr. 1991. Assessmentand evaluation of the ecological health of fish pop-ulations exposed to PCBs in Hartwell Reservoir.Tennessee Valley Authority, Report TVA/WR/AB—91/14, Chattanooga, Tennessee.

Adams, S. M., and R. B. McLean. 1985. Estimationof largemouth bass, Micropterus salmoides La-cepede, growth using the liver somatic index andphysiological variables. Journal of Fish Biology 26:111-126.

Adams, S. M., K. L. Shepard, M. S. Greeley, B. D. Ji-menez, M. G. Ryon, and L. R. Shugart. 1989. Theuse of bioindicators for assessing the effects of pol-lutant stress on fish. Marine Environmental Re-search 28:459-464.

Bagenal, T. B., and F. W. Tesch. 1978. Age and growth.Pages 101-136 in T. B. Bagenal, editor. Methodsfor assessment offish production in fresh water, 3rdedition. Black well Scientific Publications, Oxford,UK.

Barton, B. A., and G. K. Iwama. 1991. Physiologicalchanges in fish from stress in aquaculturc with em-phasis on the response and effects of corticosteroids.Annual Review of Fish Diseases 1:3-26.

Barton, B. A., and C. B. Schreck. 1987. Metabolic cost

of acute physical stress in juvenile steelhead. Trans-actions of the American Fisheries Society 116:257-263.

Busacker, G. P., I. R. Adelman, and E. M. Goolish.1990. Growth. Pages 363-387 in C. B. Schreck andP. B. Moyle, editors. Methods for fish biology.American Fisheries Society, Bethesda, Maryland.

Cardwell, R. D., and L. S. Smith. 1971. Hematologicalmanifestations of vibriosis upon juvenile chinooksalmon. Progressive Fish-Culturist 33:232-235.

Couillard, C. M., R. A. Berman, and J. C. Panisset.1988. Histopathology of rainbow trout exposed toa bleached kraft pulp mill effluent. Archives of En-vironmental Contamination and Toxicology 17:319-323.

Gaymon, H. 1988. Polychlorinated biphenyls (PCBs)in fish and sediment of Lake Hartwell—1986, An-derson, Oconee and Pickens Counties, South Car-olina. South Carolina Department of Health andEnvironmental Control, Division of Water QualityAssessment and Enforcement, Technical Report001-88, Columbia.

Gaymon, H. 1990. Polychlorinated biphenyls (PCBs)in fish and sediment of Lake Hartwell—1989, An-derson, Oconee and Pickens Counties, South Car-olina. South Carolina Department of Health andEnvironmental Control, Division of Water QualityAssessment and Enforcement, Technical Report004-90, Columbia.

Goede, R. W., and B. A. Barton. 1990. Organismicindices and an autopsy-based assessment as indi-cators of health and condition of fish. AmericanFisheries Society Symposium 8:93-108.

Klaunig, J. E., M. M. Lipsky. and B. F. Trump. 1979.Biochemical and ultrastructural changes in teleostliver following subacute exposure to PCB. Journalof Environmental Pathology and Toxicology 2:953-963.

Larkin, P. A. 1978. Fisheries management: an essayfor ecologists. Annual Review of Ecology and Sys-tematics 9:57-73.

Leach, J. M., and A. N. Thakore. 1973. Identificationof the constituents of kraft pulping effluent that aretoxic to juvenile coho salmon (Oncorhynchus ki-sutch). Journal of the Fisheries Research Board ofCanada 30:479^484.

Le Cren. E. D. 1951. The length-weight relationshipand seasonal cycle in gonad weight and conditionin the perch (Perca fluviatilis). Journal of AnimalEcology 20:201-219.

Lehtinen, K. J., M. Notini, and L. Landner. 1984. Tis-sue damage and parasite frequency in flounders,Platichthys flesus (L.), chronically exposed tobleached kraft pulp mill effluents. Annales ZoologiciFennici 21:23-28.

Lipsky, M. M., J. E. Klaunig, and D. E. Hinton. 1978.Comparison of acute response to PCB in liver of ratand channel catfish: a biochemical and morpholog-ical study. Journal of Toxicology and Environmen-tal Health 4:107-121.

Neff, J. M. 1985. Use of biochemical measurementsto detect pollutant-mediated damage to fish. Amer-

QUANTITATIVE FIELD HEALTH ASSESSMENT INDEX 73

ican Society for Testing and Materials Special Tech-nical Publication 854:155-183.

Niimi, A. J. 1990. Review of biochemical methodsand other indicators to assess fish health in aquaticecosystems containing toxic substances. Journal ofGreat Lakes Research 16:529-541.

Roberts, R. J. 1978. The pathophysiology and system-atic pathology of teleosts. Pages 55-91 in R. J. Rob-erts, editor. Fish pathology. Bailliere Tindall, Lon-don.

Runnells, R. A., W. S. Monlux, and A. W. Monlux.1965. Principles of veterinary pathology, 7th edi-tion. Iowa State University Press, Ames.

Santos, M. A., F. Pires, and A. Hall. 1990. Metaboliceffects of kraft mill effluents on the eel Anguilla an-guilla L. Ecotoxicology and Environmental Safety20:10-19.

Snedecor, G. W., and W. G. Cochran. 1980. Statisticalmethods, 7th edition. Iowa State University Press,Ames.

Wedemeyer, G. A., G. A. Barton, and D. J. McLeay.1990. Stress and acclimation. Pages 451-477 in C.B. Schreck and P. B. Moyle, editors. Methods forfish biology. American Fisheries Society, Bethesda,Maryland.

Wedemeyer, G. A., D. J. McLeay, and C. P. Goodyear.1984. Assessing the tolerance of fish and fish pop-ulations to environmental stress: the problems andmethods of monitoring. Pages 163-195 in V. W.Cairns, P. V. Hodson, and J. O. Nriagu, editors.Contaminated effects on fisheries. Wiley, New York.

Received May 18, 1992Accepted August 17, 1992