O G tLong Term Resource Monitoring Program

Technical Report2005-T004

Spatial Structure and Temporal Variationof Fish Communities in the Upper Mississippi River System

DISTRIBUTION STATEMENTApproved for Public Release

Distribution Unlimited

May 2005

20050606 028

Long Term Resource Monitoring Program Technical Reportsprovide Long Term Resource Monitoring Program

partners with scientific and technical support.

All reports in this series receive anonymous.peer review.

Mention of trade names or commercial products does not constitute endorsement

or recommendation for use by the U.S. Department of the Interior, U.S. Geological Survey.

PrktW oa racYclod Wpt

Spatial Structure and Temporal Variation

of Fish Communities in the Upper Mississippi River System

BY

JOHN H. CHICK, BRAN S. ICKES, MAw A. PEGG, VAiE A. BARKo,

ROBERrA. HRABIK, AND DAVID P. HERZOG

1,i `UTiON STATEMENT AApproved for Public Release

Distribution Unlimited

U.S. Geological Survey

Upper Midwest Environmental Sciences Center

2630 Fanta Reed Road

La Crosse, Wisconsin 54603

May 2005

Suggested citation:

Chick, J. H., B. S. Ickes, M. A. Pegg, V. A. Barko, R. A. Hrabik, and D. P. Herzog. 2005. Spatial structure and temporal variationof fish communities in the Upper Mississippi River System. U.S. Geological Survey, Upper Midwest Environmental SciencesCenter, La Crosse, Wisconsin, May 2005. LTRMP Technical Report 2005-T004. 15 pp.

Additional copies of this report may be obtained from the National Technical Information Service, 5285 Port Royal Road,Springfield, VA 22161 (1-800-553-6847 or 703-487-4650). Also available to registered users from the Defense TechnicalInformation Center, Attn: Help Desk, 8725 Kingman Road, Suite 0944, Fort Belvoir, VA 22060-6218 (1-800-225-3842 or703-767-9050).

Contents

Page

P reface ............................................................................................................... v

A b stract ............................................................................................................ I

Introduction ......................................................................................................... I

M eth od s ............................................................................................................ 2

A n aly ses ...................................................................................................... 4

R esu lts ............................................................................................................... 5

C om m unity C om position ................................................................................. 5

C om m unity Structure ....................................................................................... 6

Community Structure-Environmental Relationships ................................................ 7

D iscussion ......................................................................................................... 8

Im plications for LTR M P ................................................................................. 14

A cknow ledgm ents ............................................................................................. 14

R eferences ...................................................................................................... 14

Tables

Number Page

1. Habitat variables routinely collected from each electrofishing site for the Long Term ResourceM onitoring Program ....................................................................................... 3

2. Common and scientific names for 19 fishes found only in the lower regional trend areas(Pool 26, La Grange Pool, and Open River Reach) .................................................... 6

3. Presence/absence data for 31 fishes contributing to compositional differences among the threelow er regional trend areas .................................................................................. 7

4. Mean catch-per-unit-effort (square root catch/15 min) of 14 species of fish accounting for morethan 90% of the dissimilarity among the groups A (Pool 8), B (Pools 4 and 13), C (Pool 26

and La Grange Pool), and D (Open River Reach) . ................................................... 9

Figures

Number Page

1. Map of the Upper Mississippi River System showing the six regional trend areas (Pools 4, 8,13, and 26, La Grange Pool, and Open River Reach) for the Long Term Resource MonitoringP rogram ................................................................................................ . ... 2

2. Nonmetric multidimensional scaling ordination of fish community composition data (presence/absence) for the Upper Mississippi River System collected by the Long Term ResourceM onitoring Program , 1994-2002 ........................................................................ 6

3. Nonmetric multidimensional scaling ordination of fish community structure data (square rootcatch/I5 min) for the Upper Mississippi River System collected by the Long Term ResourceM onitoring Program , 1994-2002 ...................................................................... 8

4. Nonmetric multidimensional scaling ordination of fish community structure data (square rootcatch/i15 min) for the Upper Mississippi River System collected by the Long Term ResourceM onitoring Program , 1994-2002 ...................................................................... 8

5. Nonmetric multidimensional scaling ordination of fish community structure data (square rootcatch/l 5 min) for the Upper Mississippi River System collected by the Long Term ResourceM onitoring Program , 1994-2002 ........................................................................ 9

6. Nonmetric multidimensional scaling ordination of fish community structure data and indexed bymultiple gears, for the Upper Mississippi River System collected by the Long Term ResourceM onitoring Program, 1994-2002 ..................................................................... 10

7. Nonmetric multidimensional scaling ordination of fish community structure data, indexed bymultiple gears, and averaged by year across all resource trend areas for the Upper MississippiRiver System collected by the Long Term Resource Monitoring Program, 1994-2002 ...... 10

8. Nonmetric multidimensional scaling ordination of fish community structure data, indexed bymultiple gears, and averaged by year across all resource trend areas for the Upper MississippiRiver System collected by the Long Term Resource Monitoring Program, 1994-2002 ...... 11

9. Nonmetric multidimensional scaling ordination of fish community structure data, indexed bymultiple gears, and averaged by year across all resource trend areas for the Upper MississippiRiver System collected by the Long Term Resource Monitoring Program, 1994-2002 ...... 12

10. Nonmetric multidimensional scaling ordination of fish community structure data (square rootcatch/I 5 min) for the Upper Mississippi River System collected in day electrofishing samplesby the Long Term Resource Monitoring Program, 1994-2002 .................................. 13

iv

Preface

The Long Term Resource Monitoring Program (LTRMP) was authorized under the WaterResources Development Act of 1986 (Public Law 99-662) as an element of the U.S. Army Corpsof Engineers' Environmental Management Program. The LTRMP is implemented by the UpperMidwest Environmental Sciences Center, a U.S. Geological Survey science center, in cooperationwith the five Upper Mississippi River System (UMRS) States of Illinois, Iowa, Minnesota, Missouri,and Wisconsin. The U.S. Army Corps of Engineers provides guidance and has overall Programresponsibility. The mode of operation and respective roles of the agencies are outlined in a 1988Memorandum of Agreement.

The UMRS encompasses the commercially navigable reaches of the Upper Mississippi River, aswell as the Illinois River and navigable portions of the Kaskaskia, Black, St. Croix, and MinnesotaRivers. Congress has declared the UMRS as both a nationally significant ecosystem and a nationallysignificant commercial navigation system. The mission of the LTRMP is to provide decision makerswith information for maintaining the UMRS as a sustainable large river ecosystem given its multusecharacter. The long-term goals of the Program are to understand the system, determine resource trendsand effects, develop management alternatives, manage information, and develop useful products.

This report supports Task 2.2.8 as specified in Goal 2, Monitor Resource Change, of the LTRMPOperating Plan (U.S. Fish and Wildlife Service 1993). This report was developed with fundingprovided by the LTRMP.

V

Spatial Structure and Temporal Variation

of Fish Communities in the Upper Mississippi River System

by

John H. Chick', Brian S. Ickes2, Mark A. Pegg3, Valerie A. Barko4,Robert A. Hrabik4, and David P. Herzog4

'Illinois Natural History Survey, Great Rivers Field Station8450 Montclair Avenue, Brighton, Illinois 62012

2U.S. Geological Survey, Upper Midwest Environmental Sciences Center2630 Fanta Reed Road, La Crosse, Wisconsin 54603

'Illinois Natural History Survey; Illinois River Biological Station704 North Schrader Havana, Illinois 62644

'Missouri Department of Conservation, Open River and Wetlands Field Station3815 E. Jackson Blvd., Jackson, Missouri 63755

Abstract: Variation in community composition (presence/absence data) and structure (relative abundance) ofUpper Mississippi River fishes was assessed using data from the Long Term Resource Monitoring Programcollected from 1994 to 2002. Community composition of fishes varied more in space than through time.We found substantial variation in community composition across two spatial scales: large-scale differencesbetween upper and lower river reaches and small-scale differences among individual regional trend areas(RTA). Community structure (relative abundance data) of fishes also varied more through space than throughtime. We found substantial variation in fish community structure at three spatial scales: (1) large-scaledifferences between upper and lower river reaches, (2) differences among individual RTA, and (3) differencesamong habitat strata, with backwaters having a distinct community structure relative to the main channel andside channels. When averaged across all RTA, fish community structure in 1994 and 1995 was distinct fromall other years, possibly as a result of the 1993 Flood. Fish community structure observations for each RTAand year correlated with the environmental variables measured at each sample site. A canonical approachrevealed that the combination of Secchi depth, water temperature, current velocity, and vegetation abundancehad the greatest correlation with community structure.

Key words: fish communities, LTRMP, ordination, spatial scale, Upper Mississippi River System

Introduction and trends of UMRS natural resources, (3) assistin the development and evaluation of management

The Long Term Resource Monitoring Program alternatives, and (4) manage and provide access

(LTRMP) was authorized by the Water Resources t ltinda, information and productssDeveopmnt At o 198 asan eemet ofthe to resulting data, information, and products (U.S.

Development Act of 1986 as an element of the

U.S. Army Corps of Engineers' Environmental Army Corps of Engineers 1997). A critical tool for

Management Program. The primary mission of achieving these goals is standardized monitoring

the LTRMP is to provide river managers and the of four key ecosystem components-water quality,aquatic vegetation, aquatic macroinvertebrates,

public with ecological information necessary to and fish-at ion atrn areRTA) on

maintain the Upper Mississippi River System the Mississippi River and one RTA on the Illinois

(UMRS) as a viable multiuse ecosystem. Four River C r o obeti of tRM are

long-term goals have been established for the

LTRMP: (1) increase understanding of how the the ability to detect long-term trends for these

river ecosystem operates, (2) monitor the status key components, and the ability to correlate these

trends with environmental variables to gain assess whether any patterns of the measuredinsight into possible cause-and-effect relations. environmental variables were correlated with

Fishes are one of the most important goods observed fish community patterns.and services that rivers provide to humans.Upper Mississippi River fishes are the subject Methodsof commercial and recreational fisheries, bothof which contribute substantially to local We analyzed fish data collected by theeconomies. For example, recreation on the Upper LTRMP from 1994 to 2002 (1993 was excludedMississippi River alone has been estimated to from these analyses because of incompleteprovide 18,000 jobs and generate $1.2 billion in data collection). This program monitors fishour economy per year and recreational fishing communities in six RTA in the Upper Mississippiis a key component of this economic activity River System: Pools 4, 8, 13, and 26, La Grange(Carlson et al. 1995; Sparks et al. 1998). Fish Pool of the Illinois River, and an open rivercommunities are frequently used as indicators of reach (Open River Reach; Figure 1). We reliedecological integrity for large-river ecosystems on two sets of data for our analyses: (1) daybecause of their diversity and their response electrofishing catch-per-unit-effort (CPUE;to environmental variation at multiple scales number collected per 15 min), which has been(Gammon and Simon 2000; Schiemer 2000; shown to have power to detect changes forSchmutz et al. 2000). Therefore, the ability the greatest number of species relative to allto detect variation in thecomposition and structure offish communities is a desirablefeature of long-term monitoringprograms in large-river w E

ecosystems. +We used LTRMP fish data Minnesota

from 1994 to 2002 to examineshifts in the composition(presence/absence of species)and structure (relative abundance Wisconsinof species) of fish communitiesin the UMRS. Our analyses of Poo

community composition reliedon presence/absence data froma combination of five gears: Poo°°13 Illinoisday electrofishing, large and Iowasmall hoop nets, fyke nets, andmini-fyke nets (Gutreuter etal. 1995). To assess patterns La G,in fish community structure, Missouri Pool

we relied on data from dayelectrofishing, and we also Pool 26

developed a multigear index 0 50 100 200 Miles

of community structure and I I I I I I I Open RivermrrrT-rmanalyzed this to complement 0 50 100 200 Kilometers

the information gained from themore conservative analysis ofday electrofishing data alone. Figure 1. Map of the Upper Mississippi River System showing the six regionalFinally, we used multivariate trend areas (Pools 4, 8, 13, and 26, La Grange Pool, and Open River Reach) in lightcorrelation techniques to grey for the Long Term Resource Monitoring Program.

2

the collection methods used in the LTRMP in diameter by 2.5 cm), and the mesh size was(Lubinski et al. 2001); and (2) a combination of 2.5 cm. Large and small hoop nets were baitedtotal catch data from day electrofishing, large with 3 kg of soybean cake and were deployed forand small hoop nets, fyke nets, and mini-fyke 48 hours. Hoop nets were set so that the open endnets to provide more complete data on species (first hoop) was facing downstream in water ofcomposition and community structure. Collection sufficient depth to submerge all of the throats.methodology are published elsewhere (Gutreuter Fyke and mini-fyke nets were Wisconsin-typeet al. 1995) and will be only briefly summarized trap nets comprised of three sections: (1) abelow. For all gears, data were collected using a rectangular frame, (2) a cab section within thestratified random design, where the main channel frame comprised of six hoops that led to the codborders, side channels, contiguous backwaters, end, and (3) a lead, which was a bar of meshand impounded areas constitute unique strata. that extended from the frame to the shoreline.Sites are selected at random from each stratum, For fyke nets, the lead was 15 m long and 1.3 mallowing for the computation of poolwide high, the frame was 1.8 x 6 m, the cab wasaverages that are weighted by the total area of formed from 0.9-m steel hoops, and the mesheach stratum within each RTA. size was 1.8 cm. For mini-fyke nets, the lead was

Day electrofishing was conducted using pulsed- 4.5 m long and 0.6 m high, the frame was 1.2 xDC output with two ring anodes, and the boat 3 m, the cab was formed by two 0.6-m diameterhull served as the cathode. Two dippers collected hoops, and the mesh size was 3 mm. Fyke andfishes, and voltage and amperage were adjusted mini-fyke nets were set with the lead extendedfor water temperature and conductivity to achieve perpendicular to the shoreline. Water depth at thea power output of 3,000 W. Day electrofishing frame had to be sufficient to submerge the throatswas conducted continuously along shorelines for and nets were fished for 24 hours.15 min at each sample site. For all collection methods, a series of standard

Large and small hoop nets were set in paired physical and chemical measurements weredeployments. Large hoop nets were 4.8 m long made at the initiation of sampling (Table 1).and included seven fiberglass hoops. The first Water depth was recorded from a depth-finder,hoop was 1.2 m in diameter and successive hoops and Secchi depth was recorded to the nearestdecreased in diameter incrementally by 2.5 cm. centimeter. Water temperature was measured toTwo throats were attached, one to the second the nearest 0.1 °C, conductivity was measuredhoop and one to the fourth hoop, and the mesh in pS/cm using a YSI Conductivity Metersize was 3.7 cm in diameter. Small hoop nets (YSI, Inc., Yellow Springs, OH), and currentwere 3 m long, had seven hoops (first hoop was velocity was measured to the nearest 0.01 m/s.0.6 in in diameter, successive hoops decreased Additionally, qualitative assessments of percent

Table 1. Habitat variables routinely collected from each electrofishing site for the Long Term Resource MonitoringProgram (Gutreuter et al. 1995).

Habitat factor Units ExplanationSecchi cmConductivity ItS/cmFlow (surface velocity) m/sWater temperature °CDepth m

0 = 0% coverage; 1 = 1-19% coverage 2 = 20-49% coverage;Vegetation coverage 0, 1, 2, 3 3=5%cvrg3 = 50% coverage

Vegetation density 0, 1, 2 0 = no vegetation; I = sparse; 2 = dense1 = silt; 2 = silt/clay/little sand 3 = sand/mostly sand;

4 = gravel/rock/hard clay

Woody structure presence/absence presence/absence of woody structure

Revetment presence/absence presence/absence of shoreline revetmentInlet/outlet presence/absence presence/absence on an inlet/outlet channel to a backwater lake

Flooded terrestrial vegetation presence/absence presence/absence of flooded terrestrial vegetation

3

aquatic vegetation coverage and density, substrate the species contributing to the dissimilaritytype, and other habitat factors were recorded among the groups identified with NMDS.(Table 1). Variation in community structure was

analyzed from day electrofishing CPUE dataAnalyses and from standardized total catch data from the

combination of day electrofishing, large andWe analyzed variation of community small hoop nets, fyke nets, and mini-fyke nets.

composition through space and time using For day electrofishing, we limited the speciespresence/absence data from a combination of day used to a group of 16 for which electrofishingelectrofishing, large and small hoop nets, fyke had power >0.80 to detect a 20% interannualnets, and mini-fyke nets. Because some species abundance change in at least one habitat stratumwere not well sampled by any of these gears, we of at least one RTA based on the Lubinski et al.eliminated any species where <20 individuals (2001) power analysis. This conservative criteriawere collected when summed across all RTA was adopted to help ensure that the patterns of

and years. This resulted in analyses being relative abundance used in these analyses reflectconducted on presence/absence data from a total true ecological patterns rather than samplingof 100 fishes. Hybrids and fishes not identified artifacts. Hybrids and fishes not identified to

to species were eliminated from these analyses. species were omitted from these analyses.Presence/absence data were summarized for Because of the large size of the UMRS and itseach RTA and year, and a similarity matrix physical complexity, no single gear effectivelywas constructed based on Euclidean distance. samples the entire UMRS fish community. Thus,All analyses were performed using the Primer we chose to include data from five gear typesversion 5 software package (Primer-E Ltd 2001). (day electrofishing, large and small hoop nets,

We used Analysis of Similarity (ANOSIM) fyke nets, and mini-fyke nets) simultaneouslyto test for significant variation among RTA and to permit the broadest definition of the UMRSyears. Analysis of similarity is analogous to fish community as possible. However, each gearunivariate Analysis of Variance (ANOVA) which differed notably in its selectivity characteristicstests for significant differences among groups. (Ickes and Burkhardt 2002), potentiallyUnlike ANOVA, however, ANOSIM is based complicating our approach. Our solutionon a similarity matrix rather than raw data, and capitalized on the highly standardized nature ofsignificance is based on comparisons of this the LTRMP sampling protocols. Within a RTA,matrix to random permutations of the matrix proportional gear allocations were constant(Clarke and Warwick 1994). Two test statistics over time (years). Although the individual gearsare provided by ANOSIM, an R statistic that used in our analyses differ in their selectivity,reflects the amount of dissimilarity associated the combined selectivity of the five gear typeswith each factor (analogous to the R2 statistic remains constant over time within a river reach.from ANOVA) and a P value that indicates By placing data from each gear on the samewhether R (range -1 to 1) is significantly scale (standardization) and calculating separatedifferent from zero. Both R and P are important multigear indexes for each study reach and year,to consider because it is possible for R to we ensure that no single gear overly influencedbe significantly different from zero but still our results while allowing the broadest definitioninconsequentially small (Clarke and Warwick of community as possible. Furthermore, the use1994). Our analyses tested for variation of the of poolwide estimates of mean CPUE weightedfish community among RTA when averaged by habitat strata should minimize differences inacross all years and variation among years abundance estimates arising from variation ofwhen averaged across all RTA. Nonmetric gear allocation among RTA.multidimensional scaling (NMDS) was used Annual mean CPUE estimates were compiledto identify groupings of observations, and a for each collected species from all RTA, years,similarity breakdown (SIMPER procedure in and gear types. For each gear, we limited thePrimer; Primer-E Ltd 2001) was used to identify species used in these analyses to those for

4

which power Ž0.80 to detect a 20% interannual Finally, we used the electrofishing data toabundance change in at least one habitat stratum examine whether spatial and temporal variationof at least one RTA (Lubinski et al. 2001). Under in fish community structure corresponded withthis criteria, a total of 37 fishes were included variation in the environmental factors collectedin these analyses. We arranged data from each from each sample site. The two categoricalgear in a matrix with species (n = 37) comprising vegetation measures-percent cover andthe rows and RTA (n = 6) and year (n = 10) density (Table 1)-were multiplied to formcombinations comprising the columns. We one variable representing overall abundance ofthen calculated a total catch for each row (i.e., aquatic vegetation. We calculated a normalizedsummed CPUE across species) and calculated (mean = 0, standard deviation = 1) Euclideanthe grand mean total catch (GMTC) for each distance matrix from the habitat variables, and agear type. To place CPUE data from each gear Mantel test was used to determine whether thison the same relative scale, we divided CPUE correlated with the Bray-Curtis similarity matrixfrom each species, RTA, and year combination from fish community structure data. A canonicalby the appropriate GMTC for that particular Mantel test (BioEnv procedure in Primer; Clarkegear. This standardization places all observations and Warwick 1994) was used to determine theon the same scale (proportion of GMTC) combination of habitat variables that provided thewhile maintaining all of the species abundance greatest correlation with community data.relations within a RTA-year combination andall differences among RTA-year combinations Resultswithin a species. Finally, we summed thestandardized mean CPUE estimates for allfive gear types together for each RTA and year Community Compositioncombination resulting in a 37 x 60 matrix,to arrive at a multigear index of community We found notable spatial variation ofstructure. community composition at two scales.

We used ANOSIM to test for variation Community composition varied significantlyin community structure (day electrofishing among river reaches (R = 0.92; P < 0.001) andCPUE and multigear index) among RTA when years (R = 0.13; P = 0.0 19), but the relativelyaveraged across years and for variation among small R value associated with among yearyears when averaged across RTA. We used differences suggests that most of the dissimilarityNMDS to identify groupings of observations, among observations was due to spatial variation.Analysis of similarity and NMDS were based on Nonmetric multidimensional scaling revealed

Bray-Curtis similarity matrices. Two analyses spatial groupings at two scales: upper and

were conducted examining different spatial lower river reaches and individual RTA. The

scales. In the first, observations consisted greatest variation was between upper and lower

of RTA-stratum-year combinations. In the river reaches (Figure 2). Also, community

second, observations consisted of RTA-year composition overlapped substantially among thecombinations. Furthermore, we examined three upper RTA whereas the three lower RTAcempombinpatirns. Fhrmor, wl b vexamined each formed separate and fairly distinct groupstemporal patterns across all RTA by averaging (Figure 2). Five fish species-burbot (Lota Iota),the standardized catch data from the multigear spotted sucker (Minytrema melanops), weedindex across RTA for each year. For the analysis shiner (Notropis texanus), western sand darterof day electrofishing data among RTA and years, (Ammocrypta clara), and central mudminnowa similarity breakdown was used to identify the (Umbra limi)-were collected only in upper riverspecies contributing most to the dissimilarity reaches. Nineteen species were collected only inamong groups. We did not conduct a similarity the lower river reaches (Table 2). The similaritybreakdown for the multigear index because the breakdown showed that 31 species contributedefficacy of this approach has not been examined more than 90% of the dissimilarity among theon a species-by-species basis. three lower RTA (Table 3).

5

Stress: 0.06 A 4 electrofishing varieds .0 significantly among RTA

[e e (R = 0.840; P < 0.001) andEJ 8 stratum (R = 0.532; P < 0.001).

Nonmetric multidimensional0 13 scaling revealed little overlap

A A A& between upper and lower riverAt A A A 26 reaches (Figure 3). Pool 8 was

"A 0o disassociated from the otherAA 8 LG upper RTA, and the Open

4 * River Reach was somewhat

1. * OR distinct from the other lower* reaches. Pools 4 and 13 showed

considerable overlap as didFigure 2. Nonmetric multidimensional scaling ordination of fish community Pool 26 and La Grange Poolcomposition data (presence/absence) for the Upper Mississippi River System (Figure 3). When the samplescollected by the Long Term Resource Monitoring Program, 1994-2002. Data were are coded according to stratum,from a combination of day electrofishing, large and small hoop nets, fyke nets, and gmini-fyke nets. Each point represents community composition for a single year backwaters were fairly distinctwithin the designated regional trend area. Ecological similarity was measured from the main channel bordersusing the Euclidean Distance metric. The upper resource trend areas (Pools 4, and side channels, which8, and 13) are represented by open symbols whereas the lower resource trend overlapped considerablyareas (Pool 26, La Grange Pool, and Open River Reach) are represented by shaded (Figure 4). A total of 12 speciessymbols. 4 = Pool 4, 8 = Pool 8, 13 = Pool 13, 26 = Pool 26, LG = La Grange Pool, and accounted for more than 90%OR = Open River Reach. of the dissimilarity between

Community Structure backwaters and the main channel

or side channels. Gizzard shad (Dorosoma

We found notable variation in community cepedianum), bluegill (Lepomis macrochirus),

structure at three spatial scales. Community largemouth bass (Micropterus salmoides),

structure of fishes collected using day common carp (Cyprinus carpio), smallmouthbuffalo (Ictiobus bubalus), black crappie

Table 2. Common and scientific names for 19 fishes found (Pomoxis nigromaculatus), bullhead minnow

only in the lower regional trend areas (Pool 26, La Grange (Pimephales vigilax), and freshwater drumPool, and Open River Reach). (Aplodinotus grunniens) were more abundant

Common name Scientific name in backwaters. Emerald shiner (Notropis

Bighead carp Hypopthalricht/ys nobilis atherinoides), spotfin shiner (Cyprinella

Blue catfish lctalurusfurcatus spiloptera), white bass (Morone chrysops), andBlacktail shiner Cyprinella venusta shorthead redhorse (Moxostoma macrolepidotum)Blackstripe topminnow Fundulus notatus were more abundant in the main channel bordersFreckled madtomn Noturus nocturnus and side channels.Goldfish Carassius auratusGrass carp Ctenophatyngodon idella We also found substantial variation inInland silverside Menidia beryllina community structure among river reachesLongear sunfish Lepornis niegalotis when annual poolwide averages of CPUEWestern mosquitofish Gambusia affinis were analyzed. Community structure variedRed shiner Cyprinella lutrensisRedear sunfish Leponis microlophus significantly among RTA (R = 0.83; P < 0.001)Silverband shiner Notropis shumardi and years (R = 0.226; P = 0.001), but theStriped bass Morone saxatilis relatively small R associated with years suggestsSkipjack herring Alosa chrysochloris that most of the dissimilarity among our dataSpotted bass Micropterus punctulat'usSilver carp Hycopothat-unichth s molirri was associated with differences among riverSilver carp H37othahnichthys molitrix

Threadfin shad Dorosoina petenense reaches. Nonmetric multidimensional scalingWhite perch Morone amnericana revealed little overlap between upper and lower

6

Table 3. Presence/absence data for 31 fishes contributing to compositional shorthead redhorse, silver redhorse,differences among the three lower regional trend areas (26 Pool 26, and black crappie. Group B hadLG = La Grange Pool, and OR = Open River Reach). the greatest abundance of emerald

Pools shiner, and group C had the greatestCommon name Scientific name collected in abundance of gizzard shad, common

Trout perch Percopsis omiscomaycus OR carp, freshwater drum, white bass,River redhorse Moxostoma carinatum OR smallmouth and bigmouth buffaloPugnose minnow Opsopoeodus emiliae ORBluntnose darter Etheostoma chlorosoma OR (Ictiobus cyprinellus).Mimic shiner Notropis volucellus OR Analysis of community structureInland silverside Menidia beryllina OR using the multigear index revealedBlacktail shiner Cyprinella venusta OR more defined differences amongSpotted bass Micropterus punctulatus ORSilver lamprey Ichthyomyzon unicuspis 26 the six RTA and greater temporalRock bass Ambloplites rupestris LG variation. Community compositionYellow perch Percaflavescens LG varied significantly among RTANorthern hog sucker Hypentelium nigricans LGPumpkinseed Lepomis gibbosus LG (R = 0.793; P < 0.001) and yearsHighfin carpsucker Carpiodes velifer LG (R = 0.324; P < 0.001). As with ourSilver redhorse Moxostoma anisuruin LG analyses of day electrofishing dataWhite perch Morone americana LG alone, NMDS revealed little overlapShovelnose sturgeon Scaphirhynchus platorynchus 26, OR of upper and lower river reachesSpeckled chub Macrhybopsis aestivalis 26, ORRiver darter Percina shumardi 26, OR and each of the six RTA were fairlyMississippi slivery minnow Hybognathus nuchalis 26, OR distinct (Figure 6). When data wereSpotfin shiner Cyprinella spiloptera 26, OR averaged by year across the sixChannel shiner Notropis wickliff! 26, ORGrass pickerel Esox americanus vermiculatus 26, LG RTA, 1994 was disassociated fromNorthern pike E. lucius 26, LG all other years (Figure 7), and aBrown bullhead Ameiurus nebulosus 26, LG group of seven species-commonStriped bass Morone saxatilis LG, OR carp, black crappie, channel catfish,White sucker Catostomus commersoni LG, OR carp, bakrapi chanel catsLongear sunfish Lepomis megalotis LG, OR bluegill, emerald shiner, gizzardJohnny darter Etheostoma nigrum LG, OR shad, and smallmouth buffalo-Fathead minnow Pinephales promelas LG, OR were associated strongly withPirate perch Aphredoderus sayanus LG, OR variation among years (Figure 8).

RTA (Figure 5). Furthermore, four groups were Some of these species decreasedapparent: A, Pool 8; B, Pools 4 and 13; C, after 1994, whereas others increasedPool 26 and La Grange Pool; and D, the Open after 1994 (Figure 9).River Reach. The similarity breakdown revealedthat 13 species accounted for more than 90% of Community Structure-Environmentaldissimilarity between upper and lower groups. RelationshipsEmerald shiner, bluegill, largemouth bass, spotfin

shiner, bullhead minnow, shorthead redhorse, Similarity among RTA and years in community

and silver redhorse (Moxostoma anisurum) structure was significantly correlated with

were more abundant in the upper river reaches, environmental variables (Mantel R = 0.60;

whereas gizzard shad, common carp, channel P < 0.001). Canonical Mantel correlations

catfish (Ictalurus punctatus), smallmouth buffalo, showed the greatest correlation (R = 0.76) with a

white bass, and freshwater drum were more combination of Secchi disk transparency, water

abundant in the lower river reaches. Fourteen temperature, current velocity, and vegetation

species accounted for more than 90% of abundance. Upper RTA had greater abundance

dissimilarity among groups (Table 4). Group A of aquatic vegetation and deeper Secchi depths.

had the greatest abundance of bluegill, spotfin Lower RTA had faster current velocity and higher

shiner, largemouth bass, bullhead minnow, temperature (Figure 10).

7

DiscussionStress:0.13 L 4

DO 0 Community structureJ f 0 O Mon [E 8 and composition of

tI 0 UMRS fishes varied^ 0 AO -. more in space than

ODU El 0 -. / 3 A A in time. Our analyses

E lO A t suggest a hierarchy[A • 26 of spatial variation.

l El E Observations firstEl r A O . LG grouped according

Lc to large-scale;Ifý. differences between

L A 0 OR upper and lower riverreaches, then grouped

at smaller scales

Figure 3. Nonmetric multidimensional scaling ordination of fish community structure data including individual

(square root catch/15 min of day electrofishing) for the Upper Mississippi River System RTA or groups ofcollected by the Long Term Resource Monitoring Program, 1994-2002. Each point represents RTA, and finallyfish community structure for a combination of year and habitat strata within the designated grouped according toregional trend area. Ecological similarity was measured using the Bray-Curtis Similarity habitat stratum. Formetric. The upper resource trend areas (Pools 4, 8, and 13) are represented by open thesymbols whereas the lower resource trend areas (Pool 26, La Grange Pool, and Open River se data, temporalReach) are represented by shaded symbols. 4 = Pool 4, 8 = Pool 8, 13 = Pool 13, 26 = Pool 26, patterns were largelyLG = La Grange Pool, and OR = Open River Reach. limited to variation

among observationswithin spatial

Stress: 0.13 A B groupings. If systemicA temporal trends of

a magnitude greaterA A A A A 0 * A than the observed

AP- A AA spatial variation had01 ~A AA A El M been prevalent, our

0 0 Ak A observations would

El El OA AAA 6(50have grouped first© Cf3 A E I .Eb by year and then by

Q] DE Oo (_ 0 g] 5 0 spatial groupings.00 cý 0•0,- 0ES Our observations

SM• Eothat spatial variation

tj of UMIRS fish0 El El communities were

predominate over

Figure 4. Nonmetric multidimensional scaling ordination of fish community structure temporal variation may

data (square root catch/1i5 min of day electrofishing) for the Upper Mississippi RiverSystem collected by the Long Term Resource Monitoring Program, 1994-2002. Each important implicationspoint represents fish community structure for a combination of year and habitat strata for understandingwithin the designated regional trend area. Ecological similarity was measured using the the ecology of thisBray-Curtis Similarity metric. The ordination is identical to Figure 3 (year x habitat strata system and for thex resource trend area) but with points coded by habitat strata (B = backwaters, M = main design of research andchannel, and S = side channel) rather than resource trend area. monitoring programs.

8

Stress: 0.1 A 4 found in a 1-year studythat extended LTRMPday electrofishing to

El 8 river reaches upstream

0 0 0 and downstream of three

0 13 RTA (Chick and Pegg

_1 0 0 A A 2004). Two previousE '0 0 0 A 4 A studies also found

[El El El A 26 distinct differences,l A , between upper and lower

A * * * LG UMRS reaches based on

0 * habitat variables (U.S.A 00 Geological Survey 1999;

* OR Koel 2001). Similar

spatial patterns of fish

Figure 5. Nonmetric multidimensional scaling ordination of fish community structure community structuredata (square root catch/15 min of day electrofishing) for the Upper Mississippi River have also been observedSystem collected bythe Long Term Resource Monitoring Program, 1994-2002. Each point in the Missouri andrepresents community structure (based on poolwide means) for a single year within the Illinois Rivers (Peggdesignated regional trend area. Ecological similarity was measured using the Bray-Curtis and Pierce 2002; PeggSimilarity metric. The upper resource trend areas (Pools 4, 8, and 13) are represented by and McClellend 2004).open symbols whereas the lower resource trend areas (Pool 26, La Grange Pool, and Open Geographic rangeRiver Reach) are represented by shaded symbols. 4 = Pool 4, 8 = Pool 8, 13 = Pool 13, 26 =Pool 26, LG = La Grange Pool and OR = Open River Reach. limitations of fishes

probably influencedFor example, the stratified random design used our community composition results, as severalin the LTRMP stratifies by habitat, whereas our species reach the northem or southern limits ofanalyses demonstrated that fish communities their range in the lower or upper portion of thevaried more across larger spatial scales (e.g., UMRS. Additionally, habitat factors and possiblyRTA and upper and lower reaches). contemporary and/or historical barriers to

The most consistent pattem observed in migration probably influenced differences in fishour analyses of community composition and composition and community structure betweenstructure was a clear separation of the upper three upper and lower reaches. Upper river reaches hadRTA from the lower RTA. Similar results were deeper Secchi depths and greater abundance of

Table 4. Mean catch-per-unit-effort (square root catch/15 min) of 14 species of fish accounting for more than90% of the dissimilarity among the groups (A = Pool 8; B = Pools 4 and 13; C = Pool 26 and La Grange Pool; andD = Open River Reach).

GroupSpecies Scientific name A B C D

Gizzard shad Dorosoma petenense 1.72 3.51 7.13 5.83Bluegill Lepomis macrochirus 5.21 3.04 1.43 0.22Spotfin shiner Cyprinella spiloptera 2.99 0.84 0.18 0.01Bullhead minnow Pirnephales vigilax 2.58 0.71 0.21 0.06Largemouth bass Micropterus salmoides 2.94 2.06 0.83 0.03Emerald shiner Notropis atherinoides 2.46 2.62 1.31 1.49Common carp Cyprinus carpio 1.43 2.22 3.11 1.55Shorthead redhorse Moxostoina macrolepidoturn 1.3 0.77 0.14 0.01Freshwater drum Aplodinotus grunniens 0.41 0.98 1.61 1.11Silver redhorse Moxostomna anisurum 0.99 0.48 0 0Smallmouth buffalo Ictiobus bubalus 0.09 0.38 1.4 0.45White bass Morone chrysops 0.49 0.84 1.64 0.95Bigmouth buffalo Ictiobus cyprinellus 0.05 0.15 0.68 0.15Black crappie Pomoxis nigromaculatus 0.8 0.76 0.52 0.04

9

stre5 0,12 A 4 In addition toS s 0differences between

0 8] upper and lower riverreaches, there were

[0 differences in fish0 13 communities among

A A RTA and stratum.0 000 A A A 26 Differences among RTA

0A0 0 *were significant in every0(A0 A kA A 0G analysis conducted.

A 0 0 Community composition

* was more variable among* 0 OR the three lower RTA than

among the three upperRTA. In our analyses

Figure 6. Nonmetric multidimensional scaling ordination of fish community structure data of day electrofishingand indexed by multiple gears for the Upper Mississippi River System collected by the data, some overlap inLong Term Resource Monitoring Program, 1994-2002. Each point represents community community structure wasstructure for a single year within the designated resource trend area. Ecological apparent for Pools 4 andsimilarity was measured using the Bray-Curtis Similarity metric. The upper resource 13, as well as for Pool 26trend areas (Pools 4, 8, and 13) are represented by open symbols whereas the lowerresource trend areas (Pool 26, La Grange Pool, and Open River Reach) are representedby shaded symbols. 4 = Pool 4, 8 Pool 8, 13 = Pool 13, 26 Pool 26, LG = La Grange Pool, used fairly conservativeand OR = Open River Reach. criteria for the inclusion

of species in this analysis.Our multigear analyses,

Strew 006 which included a total

2002 of 81 species, showedmore distinct groupingsof observations for each

200 of the RTA. As expected,

199 2000 fish community structure1998 differed among strata,

but this variation was less

1095 W6 important than variationbetween upper and

lower river reaches, andvariation among RTA.

1997 Classification systemsrecognizing major habitat

types, such as main

Figure 7. Nonmetric multidimensional scaling ordination of fish community channel borders, sidestructure data, indexed by multiple gears, and averaged by year across all channels, and backwaters,resource trend areas forthe Upper Mississippi River System collected by the are fundamental to theLong Term Resource Monitoring Program, 1994-2002. Ecological similarity was study of large riversmeasured using the Bray-Curtis Similarity metric. Labels reflect the averaged (Welcomme 1979, 1985).community structure for each year. As a result, monitoring

aquatic vegetation compared to lower river programs and ecological studies conducted on

reaches, whereas lower river reaches had faster large river systems (including the LTRMP) oftencurrent velocity and higher water temperature. explicitly incorporate these habitat types into

10

Stress: 0.02

CNSN

WLYE WTCP PCN

R DSN ~ d WTBR•bM q•RSL MjS HFHIOGR BKCP GZSDLGPSN~ Wres FWDM BC

GDEY

Figure 8. Nonmetric multidimensional scaling ordination of fish community structure data, indexed by multiplegears, and averaged by year across all resource trend areas for the Upper Mississippi River System collected bythe Long Term Resource Monitoring Program, 1994-2002. Ecological similarity was measured using the Bray-CurtisSimilarity metric. Each point represents one species, with information averaged across regional trend areas.Species grouping closely together varied similarly through time. Labels reflect the four-digit letter codes used toidentify species by the Long Term Resource Monitoring Program (Gutreuter et al. 1995). The circle encloses sevenfish species (BKCP = black crappie [Pomnoxis nigromaculatus], BLGL = bluegill [Lepomis macrochirus], CARP =common carp, lCyprinus carpio] CNCF = channel catfish [Ictalurus punctatus], ERSN = emerald shiner [Notropisatherinoides], GZSD = gizzard shad [Dorosomna cepedianum]) that are grouping disjunctively from all other species,suggesting stronger variation through time for these species.

their experimental design. Variation of fish incorporate this scale of variation into theircommunities at larger spatial scales is less experimental design?understood. Much of the variance in UMRS Temporal patterns of fish community variationfish communities was associated with large may reflect an effect of the 1993 Flood. The yearspatial scales, suggesting that our understanding 1994 was distinct from all other years withinof the dynamics of fish communities in the our time series, and the multigear index wasUMRS and other large river systems probably especially useful for identifying this pattern.would benefit from focused investigations Observations made during the 1993 Floodinto the factors contributing to these patterns. suggest that several fishes took advantage ofWhat environmental or biological variables are increased access to floodplain habitats for

influencing this variation? How can we separate feeding and reproduction, and many appearedthe influence of environmental variables on to produce exceptional year classes (Nationalthese spatial patterns from demographic factors Biological Service et al. 1994). In the timesuch as immigration/emigration, propagule/ series we examined (1994 to 2002), commonoffspring dispersal, zoogeography, etc.? If fish carp, freshwater drum, and black crappie hadcommunities vary at large spatial scales as much their peak abundance in 1994, possibly as aor more than they do among major habitat types, result of successful reproduction in 1993. Otherhow should monitoring and research studies species had their lowest abundance in 1994,

11

-0--

= co u-

a,

( .I_.) .,.- • = = .• " 0,

"-u .CO 0 &a, V (

,0 E . ,co

(1 CL) C" ) c

E "E

0)T ý

CV w C 0 1-- )

o (9 • - .

122

2) >" ii

ccc

.0.0

00ei

w, o

(~) _

C-~ Ei c

0 0.. o cEV E z e- t

E _r ai

m (j>W en i, co, _

L6 ca M u~ tZ3 u

____________________________________-1e

Co ifn 0 U

U1) Cl) 0 0rO E! i0CDZD -,

CIA~ 01 0 o

0 00'0 Wa

('m E 0 CL-

(C-4 " E0o

Li)

E , C' ciCO cn 0 a30'C)~' cc C).? 3)"V- CI00 0 CD

0)~ c0

-C lI

E co -6

'Io 0 ' oE2 a)

ca0 COcc

L~ 03 03 0I

0 ci,0

CD 0)0

E0)

c- 0a) E c

0 C1 C3cc >.03O C

oc

(0)a)

03~-C _ ) D03_ 0u

.00 m 0)u

C)0) .2 a--' c a

EL) - - t C) C.

un 0c co~3

13

possibly as a result of dramatic changes to habitat Environmental Sciences Center. Further supportassociated with the flood such as reductions was provided by the Minnesota Departmentin aquatic vegetation (Spink and Rogers 1996) of Natural Resources, Wisconsin Departmentand sedimentation effects (National Biological of Natural Resources, Iowa Department ofService et al. 1994). Given the magnitude of the Natural Resources, Missouri Department of1993 Flood, the relatively modest amount of Natural Resources, and Illinois Natural Historytemporal variation of UMRS fish communities Survey. We thank all of the field crews, teamfurther emphasizes the importance of the spatial leaders, and component specialists associatedvariation observed. with the LTRMP who were responsible for data

Our use of a multigear index of fish community collection and management. Barry Johnson andstructure was novel and should be further three anonymous reviewers provided valuableexamined as a potential useful method for comments on an earlier version of this report.examining community dynamics. Becauseseveral questions regarding the efficacy of this Referencesapproach need to be addressed, we limited ouruse of this index to a secondary and supportiverole to the more conservative analysis conducted Carlson, B. D., D.9B. Propst, D. J. Synes, andwith day electrofishing data alone. The results R.S. Jackson. 1995. Economic impact offrom the multigear analyses were, in general, re reao ne Upr MiSs rivercorroborative of those from day electrofishing. System. Prepared for the U.S. Army CorpsTherefore, we believe further investigation into of Engineers Waterways Experiment Station,this analysis technique is warranted. Vicksburg, Mississippi. Technical Report EL-95-16.

Implications for LTRMP Chick, J. H., and M. A. Pegg. 2004. LongTerm Resource Monitoring Program outpool

Our study demonstrated that LTRMP fish data fisheries analysis: Final report. U.S. Geologicalcan be used to detect variation in community Survey, Upper Midwest Environmentalcomposition and structure through space and Sciences Center, La Crosse, Wisconsin,time. We were able to detect temporal variation March 2004. LTRMP 2004-T001. 21 pp.that was consistent across all RTA, whilesimultaneously detecting large-scale spatial Clarke, K. R., and R. M. Warwick. 1994. Changevariation. Additionally, we were able to correlate in marine communities: an approach tolarge-scale spatial patterns with environmental statistical analysis and interpretation. Plymouthvariables measured locally within each river Marine Laboratories, Plymouth, Unitedreach. This last result points to the importance of Kingdom.the habitat data collected during fish monitoring, Gammon, J. R., and T. P. Simon. 2000. Variationincluding observational data (i.e., vegetation in a great river index of biotic integrity over acover and density). Finally, our analyses suggest 20-year period. Hydrobiologia 422/423:291-that further examination of spatial variability of 304.fish communities and habitat in the UMRS mayyield important insights into ecological dynamics Gutreuter, S., R. Burkhardt, and K. Lubinski.and function within this river-floodplain 1995. Long Term Resource Monitoringecosystem. Program procedures: Fish monitoring.

National Biological Service, Environmental

Acknowledgments Management Technical Center, Onalaska,Wisconsin. LTRMP 95-P002-1.

This study was funded by the U.S. Army 42 pp. + Appendixes A-J.

Corps of Engineers and administered by the Ickes, B. S., and R. W. Burkhardt. 2002.U.S. Geological Survey, Upper Midwest Evaluation and proposed refinement of the

14

sampling design for the Long Term Resource Schmutz, S., M. Kaufmann, B. Vogel, M.Monitoring Program's fish component. Jungwirth, and S. Muhar. 2000. A multi-U.S. Geological Survey, Upper Midwest level concept for fish-based, river-type-Environmental Sciences Center, La Crosse, specific assessment of ecological integrity.Wisconsin, October 2002. LTRMP 2002-TOOl. Hydrobiologia 422/423:279-289.17 pp. + Appendixes A-E. CD-ROM included.(NTIS # PB2003-500042) Sparks, R. E., J. C. Nelson, and Y. Yin. 1998.

Naturalization of the flood regime in regulatedKoel, T. M. 2001. Classification of Upper rivers: the case of the upper Mississippi River.

Mississippi River pools based on contiguous Bioscience 48:706-720.aquatic/geomorphic habitats. Journal ofFreshwater Ecology 16:159-170. Spink, A., and S. Rogers. 1996. The effects of

a record flood on the aquatic vegetation ofLubinski, K, R. Burkhardt, J. Sauer, D. Soballe, the Upper Mississippi River System: some

and Y. Yin. 2001. Initial analyses of change preliminary findings. Hydrobiologia 340:51-detection capabilities and data redundancies in 57.the Long Term Resource Monitoring Program. U.S. Army Corps of Engineers. 1997. ReportU.S. Geological Survey, Upper MidwestEnvironmental Sciences Center, La Crosse, to Congress, An evaluation of the UpperWisconsin, September 2001. LTRMP 2001- Mississippi River System EnvironmentalTOOL. 23 pp. + Appendixes A-E (NTIS # Management Program, December 1997.

PB2002-100123). Chapters 1-8 + Appendixes A-B.

National Biological Service, Illinois Natural U.S. Fish and Wildlife Service. 1993. Operating

History Survey, Iowa Department of Natural Plan for the Upper Mississippi River System

Resources, and Wisconsin Department of Long Term Resource Monitoring Program.

Natural Resources. 1994. Long Term Resource Environmental Management Technical Center,Monitoring Program 1993 Flood observations. Onalaska, Wisconsin, Revised September

National Biological Service, Environmental 1993. EMTC 91-P002R. 179 pp. (NTIS

Management Technical Center, Onalaska, #PB94-160199)

Wisconsin, December 1994. LTRMP 94-S011. U.S. Geological Survey. 1999. Ecological190 pp. (NTIS #PB95-181582). status and trends of the Upper Mississippi

Primer-E Ltd. 2001. Primer for Windows River System 1998: a report of the Long

Version 5.2.4. Plymouth, United Kingdom. Term Resource Monitoring Program.U.S. Geological Survey, Upper Midwest

Pegg, M. A., and C. L. Pierce. 2002. Fish Environmental Sciences Center, La Crosse,community structure in the Missouri and Wisconsin, April 1999. LTRMP 99-TOOl.Lower Yellowstone rivers in relation to flow 236 pp.characteristics. Hydrobiologia 479:155-167. Welcomme, R. L. 1979. Fisheries ecology of

Pegg, M. A., and M. A. McClelland. 2004. floodplain rivers. Longman Group Limited.Assessment of spatial and temporal fish London. 317 pp.community patterns in the Illinois River.Ecology of Freshwater Fish 13:125-135. Welcomme, R. L. 1985. River fisheries. Food

and Agriculture Organization, United Nations,Schiemer, F. 2000. Fish as indicators for the Rome. FAO Fisheries Technical Paper 262.

assessment of ecological integrity of largerivers. Hydrobiologia 422/423:271-278.

15

REPORT DOCUMENTATION PAGE Form approvedOIMB No. 0704-0188

Public reporting burden for this collection is estimated to average I hour per response, including time for reviewing instructions, searching existing data sources,gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any otheraspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations andReports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188)Washington, DC 20503.

1. AGENCY USE ONLY (Leave 2. REPORT DATE 3. REPORT TYPE AND DATES COVEREDBlank) May 2005

4. TITLE AND SUBTITLE 5. FUNDING NUMBERS

Spatial Structure and Temporal Variation of Fish Communities in the Upper Mississippi River System

6. AUTHOR(S) John H. Chick', Brian S. Ickes2, Mark A. Pegg3, Valerie A. Barko4, Robert A. Hrabik',

and David P. Herzog4

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESSES 8. PERFORMING ORGANIZATION'Illinois Natural History Survey, Great Rivers Field Station, 8450 Montclair Avenue, Brighton, Illinois 62012; REPORT NUMBER2U.S. Geological Survey, Upper Midwest Environmental Sciences Center, 2630 Fanta Reed Road, La Crosse,Wisconsin 54603; 3111inois Natural History Survey, Illinois River Biological Station, 704 North Schrader,Havana, Illinois 62644; 4Missouri Department of Conservation, Open River and Wetlands Field Station, 3815 E.JacksonBlvd.. Jackson. Missouri 637559. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESSES 10. SPONSORING, MONITORING

U.S. Geological Survey AGENCY REPORT NUMBER

Upper Midwest Environmental Sciences Center 2005-T0042630 Fanta Reed RoadLa Crosse, Wisconsin 54603

11. SUPPLEMENTARY NOTES

12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE

Release unlimited. Available from National Technical Information Service, 5285 Port Royal Road,Springfield, VA 22161 (1-800-553-6847 or 703-487-4650. Available to registered users from theDefense Technical Information Center, Attn: Help Desk, 8725 Kingman Road, Suite 0944, FortBelvoir, VA 22060-6218 (1-800-225-3842 or 703-767-9050).

13. ABSTRACT (Maximum 200 words)

Variation in community composition (presence/absence data) and structure (relative abundance) of Upper Mississippi River fishes wasassessed using data from the Long Term Resource Monitoring Program collected from 1994 to 2002. Community composition of fishesvaried more in space than through time. We found substantial variation in community composition across two spatial scales: large-scaledifferences between upper and lower river reaches and small-scale differences among individual regional trend areas (RTA). Communitystructure (relative abundance data) of fishes also varied more through space than through time. We found substantial variation in fishcommunity structure at three spatial scales: (1) large-scale differences between upper and lower river reaches, (2) differences amongindividual RTA, and (3) differences among habitat strata, with backwaters having a distinct community structure relative to the mainchannel and side channels. When averaged across all RTA, fish community structure in 1994 and 1995 was distinct from all other years,possibly as a result of the 1993 Flood. Fish community structure observations for each RTA and year correlated with the environmentalvariables measured at each sample site. A canonical approach revealed that the combination of Secchi depth, water temperature, currentvelocity, and vegetation abundance had the greatest correlation with community structure.

14. SUBJECT TERMS (Keywords) 15. NUMBER OF PAGES

fish communities, LTRMP, ordination, spatial scale, Upper Mississippi River System 15 pp.

16. PRICE CODE

17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION OF 19. SECURITY CLASSIFICATION OF 20. LIMITATION OF ABSTRACTOF REPORT THIS PAGE ABSTRACT

Unclassified Unclassified Unclassified

NSN 7540-01-280-5500 Standard Form 298 (rev. 2-89)Prescribed by ANSI Std. Z39-18298-102

SUlSUOOS! pue 'unossII 'elosauuIA 'eMAOI 'SiOU!ifi JO somS oqi put 'siao utu JosdioD SUUV "S'fl Oq) 'XOAjnS luo.oOjOD "S'fl oql Iq vojjo DAilladooo u St dIAIll-I O41aolo~o osnoldIlntw sj! UgAlii tU1SASOaa IOA.U -aXwj;)jq~uu!sns u SU tUJOs,ýS J A.•

!ddississiW ioddfl oqI 2UUUIEiW OJ UOTWUWIOJUI ql!M SJi•i•euU JOAUJ Opho.id ol si

dIAWIJLI aql jo uoissiuw atLL 'wROJd lutuo•WuIAI IlUtUwUOJAUt oq JO juatualUa u'S)SNz su 9861 JO IOV )uOwdoIoAo(] sJofnoso-I JoluA oqj .iapun pazTJoqml n SIM WtOSSS .IaA!-IS!~~~~~dditssIssliW saddfl oqj jOl (dlHI,'-I) twEJ2O~d Ou~uol!uoIA aainosoNI tuoL uo'-I -jtLL