Vol. 4 No 2, 103-117 2004 Assessment of anthropogenic influences on littoral-zone aquatic communities of Lake Texoma, Oklahoma-Texas, USA Michael A. Eggleton1, Keith B. Gido2, William J. Matthews3, Gary D. Schnell4 1Sam Noble Oklahoma Museum of Natural History, University of Oklahoma, 2401 Chautauqua, Norman 73072, Oklahoma USA. Current address: Aquaculture/Fisheries Center, University of Arkansas-Pine Bluff, Box 4912, Pine Bluff 71601, Arkansas USA. e-mail: meggleton@ uaex.edu. 2Division of Biology, Kansas State University, Ackert Hall, Manhattan, Kansas 66506, USA. e-mail: kgido@ksu.edu. 3Sam Noble Oklahoma Museum of Natural History, University of Oklahoma, 2401 Chautauqua, Norman 73072, Oklahoma USA/University of Oklahoma Biological Station, HC 71, Box 205, Kingston 73439, Oklahoma USA/Department of Zoology, University of Oklahoma, Norman 73019, Oklahoma USA. e-mail: wmatthews@ou.edu. 4Sam Noble Oklahoma Museum of Natural History, University of Oklahoma, 2401 Chautauqua, Norman 73072, Oklahoma USA/Department of Zoology, University of Oklahoma, Norman 73019, Oklahoma USA. e-mail: gschnell@ou.edu. Abstract From 1999-2001, we evaluated the effects of anthropogenic activities in and around Lake Texoma, Oklahoma-Texas, USA, on the structure of littoral-zone fish and benthic invertebrate assemblages at 20 potentially impacted sites relative to paired reference sites. Spatial structuring of both assemblages was strongly related to variables associated with water clarity, water-column chlorophyll-a levels, and degree of site exposure to wind and waves. Fish assemblages at reference sites and impact sites exhibited minor differences, but none that were considered indicative of severe anthropogenic stress. Conversely, benthic invertebrates exhibited greater differences between reference and impact sites. Procrustean analyses and Mantel tests indicated little concordance between reference site and impact site benthic invertebrate assemblages. Greater abundances of oligochaetes at impact sites and greater abundances of chironomids at reference sites contributed most to these differences, with the largest assemblage differences found at sites influenced by agriculture and sanitary dumps. Despite the fact that fishes and benthic invertebrates were structured along similar environmental gradients, little concordance was observed between assemblages. Wide annual fluctuations in the dominant taxa of each assemblage contributed most to the general discordance. Furthermore, discordant fish and invertebrate assemblages likely resulted because responses of each assemblage to anthropogenic impacts occurred at different scales of space and time. Key words: reservoirs, fishes, benthic macroinvertebrates, multitaxon assessment, assemblage structure, environmental impact, Procrustean analysis, Mantel test 104 M. A. Eggleton et al. 1. Introduction Concern over the condition or "ecological health" of reservoirs has increased over the past quarter century with a recognition that chemical standards have not protected aquatic resources (Adler et al. 1993). However, when studying environmental impact or biotic integrity (sensu Angermeier, Karr 1984), reservoirs provide a more difficult challenge than streams, rivers, or lakes because they are artificial systems that lack natural reference sites (Hickman, McDonough 1996). Furthermore, the temporary nature of reservoirs on an evolutionary time scale precludes expectations of natural communities (Jennings et al. 1995). These attributes have led some researchers to suggest the term "biotic integrity" may be inappropriate for reservoirs, recommending instead that effort be devoted to developing multimetric indices of reservoir health (e.g., "reservoir fish assemblage index"; Hickman, McDonough 1996). Despite the challenges presented in assessing the ecological health of reservoirs, they comprise a large proportion of freshwater resources in the United States (Miranda 1996). Although most contemporary methods focus on community-level assessments, several recent studies (e.g., Jackson, Harvey 1993; Allen et al. 1999a,b; O'Connor et al. 2000) have shown advantages in conducting community-level assessments on multiple assemblage types within the same environment. Measures of concordance between assemblage types implicit with this approach provide added insight into community-level responses to anthropogenic activities in aquatic systems (Jackson, Harvey 1993). Few studies have examined environmental impact in this manner, and none have been conducted in reservoirs. Our objective was to assess effects of anthropogenic activities on aquatic assemblages in the littoral-zone of a large reservoir. We assessed effects at the community level, considering only shifts in conventional assemblage measures (species richness, organism abundance, and assemblage structure) in response to anthropogenic influences. Most studies in reservoirs have focused on off-shore fish assemblages consisting of large sport fish that can move readily, whereas we focused on littoral-zone fish assemblages because they are an important functional component of aquatic systems (Northcote 1988). Littoral fishes tend to be small-bodied and have little capacity to move great distances and so were regarded as more likely to reflect local environmental stresses. Assemblages examined were those most likely to show demonstrable effects resulting from anthropogenic stressors that impinge on reservoir ecosystems: (1) juvenile/adult fishes (hereafter referred to as fishes), which include the littoral-zone "watercolumn" community (e.g., juvenile temperate basses, atherinids, shads, and minnows) and the "benthic nesting" community (mainly centrarchids); and (2) benthic invertebrates, which include substratedwelling macroinvertebrates such as chironomids, oligochaetes, burrowing mayflies, and molluscs. If anthropogenic stresses were prevalent in Lake Texoma, our general expectation was that strong differences would exist with fish and benthic invertebrate assemblages between reference and impact sites. Further, we expected that assemblages might respond to anthropogenic influences differently, with invertebrates responding to more local impacts (e.g., substrate differences associated with anthropogenic activities) and fishes responding to broader-scale influences (e.g., turbidity differences related to land use practices). In addition, because human activities alter multiple facets of the environment (Karr 1991), these assemblage types represent very different taxonomic and functional groups and were expected to provide a more complete view of how anthropogenic activities impinge on the Lake Texoma ecosystem. 2. Materials and methods Study area Lake Texoma, Oklahoma-Texas, USA is a 35 200 ha impoundment in the southern Great Plains region of the United States, located at the confluence of the Red and Washita rivers on the Oklahoma-Texas boarder. The reservoir was constructed in 1944 by the U.S. Army Corps of Engineers (USACE) for flood control and hydroelectric power production. The watershed of the reservoir encompasses about 103 000 km2; land use is predominantly agriculture, ranching, and forest with relatively low human population densities. Highly saline inflows from the Red River occur due to natural salt sources in the headwaters and tributaries. As a result, conductance values in Lake Texoma are high for freshwater (700 - 1200 µS cm-1; Gelwick, Matthews 1990) and distinctly different between river arms (Red River arm, usually >1000 µS cm-1; Washita River arm, usually <1000 µS cm-1). Secchi depths in the reservoir typically range from 50 to 125 cm (Matthews 1984) and are usually greater downlake near the dam. The reservoir thermally stratifies during the summer months and maximum depth is around 25 m. Reservoir stages vary about 2 m per year on average, but fluctuations may exceed 3 m in any given year (U.S. Army Corps of Engineers, Tulsa District website). The ecological health of Lake Texoma has significant repercussions on local economies because of its recreational fisheries, which are valued at over 25 million dollars (U.S.) annually (Schorr et al. 1991). Anthropogenic influences on aquatic communities Experimental design Statistically, the study was designed such that potentially "impacted" sites in Lake Texoma were matched or paired with physically similar "reference" sites for statistical validity. Twenty such impacted sites were identified in Lake Texoma by the U.S. Environmental Protection Agency (USEPA) and used in this evaluation. Anthropogenic stressors of interest to USEPA included: point-source chemical leachates (e.g., methyl tertbutyl ether [MTBE] and benzene-toluene-ethylbenzene-xylene [BTEX] compounds; An et al. 2002); nonpoint-source chemical leachates (e.g., herbicides and pesticides); nutrient enrichment resulting from elevated runoff of inorganic agricultural fertilisers; sedimentation; and activities associated with the alteration, destruction, or removal of aquatic habitats. In Lake Texoma, these major stressors were associated with one or more of the following anthropogenic activities: row-crop agriculture (mainly soy beans, corn, and cotton); rural septic runoff; marinas and related operations (e.g., boat repair/servicing and marine gas facilities); sanitary dumping grounds; developed public beaches; and oil production facilities (Schnell et al. 2002). In classifying impacted sites to a particular impact type, several factors were considered. For marinas, sanitary dumps, public beaches, and oil production facilities, the close proximity of the activity to the site in question was the main characteristic considered. In most cases, the source influence was located less than 200 m from the site or was contained throughout the site. For rowcrop agriculture and rural septic runoff, the prevalence of the land use within reservoir sub-basins and associated in situ characteristics were used to classify sites. For instance, agricultural sites typically contained greater than 40% of their drainage area in row crops and/or pasture lands, contained elevated water-column chlorophyll-a levels (usually greater than 15 µg dm-1, presumably related to nutrient enrichment), and had lower water transparency (presumably due to increased sedimentation). Sites exposed to rural septic runoff were those with greater than 3% of their drainage area constituting municipalities and/or moderate- to high-density residential or commercial development. Septic-influenced sites frequently also had elevated water-column and benthic chlorophyll-a levels. Principal components analysis (PCA) incorporating mainly site-specific physical variables (e.g., slope, aspect, site exposure to wind and waves) was used to pair each impact site with one of 60 potential reference sites. Each impact site was then paired with the reference site that was closest in multivariate space. This approach kept obvious physical differences within site pairs to a 105 minimum and yielded 20 site pairs that were used for community-level assessments. Site pairs were stratified within impact types as follows: agriculture (5 site pairs for evaluation), marinas (3 site pairs), developed public beaches (2 site pairs), sanitary dumping grounds (2 site pairs), rural septic runoff (4 site pairs), and oil production facilities (4 site pairs). The number of site pairs evaluated for each impact type was proportional to the prevalence of that influence in the reservoir. In 17 of 20 site pairings, impact sites were paired with reference sites located less than 3 km away (linear distance); between-site distances in the remaining three pairs ranged 6.4-11.2 km (Schnell et al. 2002). Sample collections Fish assemblages were assessed annually at all 40 sites from 1999 through 2001. Assessments were conducted during July of each year so that community compositions would not be skewed substantially by high numbers of young-of-theyear fishes (mostly fry). At each site, four adjacent 25 m reaches of shoreline were used for fish sampling. For each replicate reach, separate samples were taken using a 7.62 m by 1.8 m bag seine (4.8 mm mesh, one offshore haul) and a 4.6 m by 1.2 m (3.2 mm mesh, several inshore hauls) regular seine. Fishes from all inshore hauls and the offshore haul were then pooled to depict the fish assemblage for that 25 m reach. Gido et al. (2002) demonstrated that this procedure was adequate to depict fish assemblage structure in littoral-zone areas of Lake Texoma, and provide additional details on sampling and laboratory procedures. Following processing, fishes were preserved in 50% isopropyl alcohol and archived in the Sam Noble Oklahoma Museum of Natural History, University of Oklahoma. Benthic invertebrate assessments were made annually at the same 40 sites during June and/or July from 1999 through 2001. During each sampling period, three replicate samples were taken at each site using a 15 cm by 15 cm Ponar dredge in water 1.0-1.5 m deep. Following collection, each dredge sample was rinsed through a 500 µm sieve in the field to remove excess sediment and organic debris. In 1999 and 2000, samples were preserved in the field with 10% formalin containing rose bengal dye; in 2001, samples were preserved with 70% ethyl alcohol. In the laboratory, invertebrates were separated from the benthic detritus and other extraneous material by trained technicians using a dissecting microscope. Organisms were identified to the lowest practical taxonomic level (usually genus) using a compound microscope and standard taxonomic keys. Because most taxa collected (primarily chironomid "midges" [Insecta] and oligochaete "worms" [Annelida]) required that 106 M. A. Eggleton et al. slide mounts be made for identification, a project reference collection also was developed to aid in identifications. Data analysis As a first-order assessment of reference and impact sites, we used canonical correspondence analysis (CCA; ter Braak 1986) to evaluate relationships between fish and invertebrate assemblages and environmental variables. CCA is a direct gradient analysis that ordinates a species/taxa by sample data matrix within the constraint that scores be linear combinations of environmental variables (Palmer 1993). Twenty-four environmental variables, that included physical, chemical, biotic, land use, and hydrological factors, were assessed for each of the 40 sites (Eggleton et al. in press). Significance of the species/taxa -environment association was evaluated by comparing observed eigenvalues from the first three ordination axes to those generated from randomisation of the data (10 000 iterations). Rejection of this test meant that observed eigenvalues were greater than expected by chance (i.e. eigenvalues generated from randomised data), which indicated a significant association existed between assemblages and environmental variables. All CCA calculations were done on species/taxa by sample and environmental variable by sample data matrices using PC-ORD (McCune, Mefford 1999); all species data and environmental data measured on a continuous scale were square-root transformed prior to analysis (ter Braak 1986). Three main methods were used for assessment of assemblage differences between reference and impact sites. First, simple paired analyses of mean differences between reference and impact site assemblages were done using Wilcoxon signed-rank tests (Conover 1980). Assemblage measures that were assessed included species or taxa richness, organism abundance (as number of fish seine haul-1 or number of invertebrates dredge-1), and ordination site scores (CCA axis 1 and 2 only, as measures of assemblage structure). Site-pair differences were derived by subtracting values at reference sites from those of impact sites; thus, negative differences imply that reference site values were greater than impact site values. All paired analyses were done using Statistical Analysis Software (SAS Institute 2000). Second, Procrustean analysis (Gower 1971) was conducted using ordinations from correspondence analysis (CA; Legendre, Legendre 1998) done separately for reference and impact sites. Procrustean analysis superimposes ordinations from reference and impact sites using a rotationalfit algorithm that minimises the sum-of-squared residuals between the two matrices (Jackson 1995; Peres-Neto, Jackson 2001). Residuals between the original values and the best-fit solution are calculated for each site pair to identify outlying or deviant site pairs. The m2 value is the test statistic and also serves as a goodness-of-fit measure that describes the degree of concordance between reference site and impact site matrices. Low m2 values and low residual sum-of-squares represent high concordance between assemblage structures of the different matrices; high m2 values and high residual sum-of-squares represent low concordance between matrices. The statistical significance of the m2 statistic was evaluated using a simple reshuffling algorithm (20 000 randomisations) as proposed by Jackson (1995). All correspondence analysis ordinations were done on species/taxa by sample data matrices using PCORD (McCune, Mefford 1999); all species data were square-root transformed prior to analysis (ter Braak 1986). Procrustean analyses were performed using the first three CA axes with PROTEST (Jackson 1995). Third, we performed Mantel tests (Legendre, Legendre 1998) to assess the spatial concordance of reference site and impact site assemblages. Mantel tests were performed between species/taxa by sample matrices (square-root transformed raw counts converted to relative abundances) using the percent similarity distance measure. Mantel tests represent a more holistic assessment method because similarities among all reference sites and all impact sites (not just similarities within preselected site pairs) are considered. The standardised Mantel statistic (Z) is the test statistic, with higher values meaning greater concordance between reference and impact sites. The statistical significance of the Z-statistic was evaluated using a simple reshuffling algorithm as with Procrustean analysis (20 000 randomisations). All Mantel tests were conducted using NTSYSpc (Rohlf 2002). Significance for all analyses was declared at an alpha level of 0.05. However, to account for multiple comparisons in paired analyses separated by impact type, we employed a sequential Bonferroni adjustment of alpha when appropriate following Rice (1989) and Gelwick, Matthews (1992). 3. Results General patterns in assemblage structure Canonical correspondence analysis results indicated a significant association between fish assemblages and environmental variables (P=0.001). Fish assemblages were structured primarily along gradients of water transparency, water-column chlorophyll-a, and annual discharge (Fig. 1). These variables represented strong longitudinal gradients in Lake Texoma (i.e. varying greatly from uplake to downlake), and were simi- Anthropogenic influences on aquatic communities 107 2 Impact site Reference site Silt content 1 CCA axis 2 WC chlorophyll- a Annual discharge Residential area 0 Secchi reading -1 Site exposure Sand content -2 -2 -1 0 1 2 3 CCA axis 1 2 LABSIC FUNNOT LEPCYA POMANU LEPHUM CCA axis 2 1 POMNIG LEPMEG NOTCHR LEPMAC ICTBUB CARCAR LEPOSS APLGRU GAMAFF MACSTO ICTPUN PIMVIG CYPLUT HYBPLA DORPET DORCEP CTEIDE CYPCAR MENBER 0 NOTATH HIOALO MICSAL MICPUN PERMAC CYPVEN MICDOL MORCHR MACEST -1 MORSAX Fig. 1. Scatterplot of site (top) and species (bottom) scores from canonical correspondence analysis (CCA) of littoralzone fish assemblages at Lake Texoma, 1999-2001. Arrow lengths and directions represent relative strengths of gradients. First and second axes had eigenvalues of 0.306 and 0.233, respectively, and together explained 28.9% of variation in assemblage structure. Species codes: APLGRU = Aplodinotus grunniens, CARCAR = Carpiodes carpio, CTEIDE = Ctenopharyngodon idella, CYPCAR = Cyprinus carpio, CYPLUT = Cyprinella lutrensis, CYPVEN = C. venusta, DORCEP = Dorosoma cepedianum, DORPET = D. petenense, FUNNOT = Fundulus notatus, GAMAFF = Gambusia affinis, HIOALO = Hiodon alosoides, HYBPLA = Hybognathus placitus, ICTBUB = Ictiobus bubalus, ICTPUN = Ictalurus punctatus, LABSIC = Labidesthes sicculus, LEPCYA= Lepomis cyanellus, LEPHUM = L. humilis, LEPMAC = L. macrochirus, LEPMEG = L. megalotis, LEPOSS = Lepisosteus osseus, MACEST = Macrhybopsis aestivalis, MACSTO = M. storeriana, MENBER = Menidia beryllina, MICDOL = Micropterus dolomieu, MICPUN = M. punctatus, MICSAL = M. salmoides, MORCHR = Morone chrysops, MORSAX = M. saxatilis, NOTATH = Notropis atherinoides, NOTCHR = Notemigonus chrysoleucas, PERMAC = Percina macrolepida, PIMVIG = Pimephales vigilax, POMANU = Pomoxis annularis, POMNIG = P. nigromaculatus. -2 -2 -1 0 1 2 3 CCA axis 1 lar to those reported by Gido et al. (2002). Sites with high water transparency (i.e. Secchi readings, greater axis-1 scores) were located almost exclusively in the main reservoir near the dam and associated with species such as striped bass Morone saxatilis (MORSAX) and smallmouth bass Micropterus dolomieu (MICDOL; Fig. 1). Higher levels of water-column chlorophyll-a and annual discharge (lower axis-1 scores) were characteristic of the river arms and transitional areas and associated with species such as freshwater drum Aplodinotus grunniens (APLGRU), white crappie Pomoxis annularis (POMANU), bullhead minnow Pimephales vigilax (PIMVIG), emerald shiner Notropis athernoides (NOTATH), common carp Cyprinus carpio (CYPCAR), and river carpsucker Carpiodes carpio (CARCAR; Fig. 1). Fish assemblages also were structured along gradients related to the level of site exposure to wind and waves and the sand content of the substrate (lower axis-2 scores), and levels of residential development and substrate silt content (higher axis-2 scores). Species associated with high-exposure sites and increased sand content included striped bass, inland silverside Menidia beryllina (MENBER), and several cyprinid species. Species typically found at low-exposure sites with increased silt content included brook silverside Labidesthes sicculus (LABSIC), blackstripe topminnow Fundulus notatus (FUNNOT), bluegill Lepomis macrochirus (LEPMAC), longear sunfish L. megalotis (LEPMEG), and largemouth bass M. salmoides (MICSAL; Fig. 1). For benthic invertebrates, CCA also indicated a significant association between assemblage structure and environmental variables (P=0.001). However, unlike fishes, invertebrate assemblages exhibited a stronger spatial pattern within the reservoir. Invertebrate assemblages were mainly 108 M. A. Eggleton et al. 2 Impact site Reference site Site exposure Secchi reading CCA axis 2 1 Conductivity TKN WC Chlorophyll- a 0 -1 Collection year Wetland -2 -2 -1 0 1 2 3 CCA axis 1 2 1 LIMNO CRYP PSEUPOLY 0 CCA axis 2 CRYPT CLAD COEL PARAC DICR ENDO NEMA PUPA CRIC GASTR EPHEM TRICH GLYP ABLA PROB TANYT -1 AULO CHAO CHIR TANYP BRAN MICRO HIRUN Fig. 2. Scatterplot of site (top) and taxa (bottom) scores from canonical correspondence analysis (CCA) of littoral-zone benthic invertebrate assemblages at Lake Texoma, 1999-2001. Arrow lengths and directions represent relative strengths of gradients. First and second axes had eigenvalues of 0.304 and 0.212, respectively, and together explained 18.9% of variation in assemblage structure. Taxa codes: ABLA = Ablabesmyia, AULO = Aulodrilus, BRAN = Branchiura sowerbyi, CHAO = Chaoborus, CHIR = Chironomus, CLAD = Cladotanytarsus, COEL = Coelotanypus, COLEO = Coleoptera, CRIC = Crictopus, CRYP = Cryptochironomus, CRYPT = Cryptotendipes, DERO = Dero, DICR = Dicrotendipes, ENDO = Endochironomus, EPHEM = Ephemeroptera, GASTR = Gastropoda, GLYP = Glyptotendipes, HIRUN = Hirudinea, IMOLIG = Immature oligochaete, LARS = Larsia, LIMNO = Limnodrilus, MICRO = Microchironomus, NAIS = Nais, NEMA = Nematoda, ODON = Odonata, PARA = Parachironomus, PARAC = Paracladopelma, POLY = Polypedilum, PRIS = Pristina, PROB = Probezzia, PROC = Procladius, PSEU = Pseudochironomus, PUPA = Chironomidae pupae, TANYP = Tanypus, TANYT = Tanytarsus, TRICH = Trichoptera, TUBIF = Tubifex. PROC PARA TUBIF IMOLIGNAIS ODON DERO COLEO -2 PRIS LARS -3 -2 -1 0 1 2 3 CCA axis 1 structured along gradients of water transparency, water-column chlorophyll-a, and total Kjeldahl nitrogen (TKN; Fig. 2). Sites with higher water transparency (low axis-1 and high axis-2 scores) were located mostly in the main reservoir near the dam and associated with the chironomids Polypedilum (POLY), Pseudochironomus (PSEU), and Cryptochironomus (CRYP). Sites with low transparency and high levels of TKN (greater axis1 and axis-2 scores near zero) were characteristic of the turbid, more saline waters of the Red River arm. Typical taxa from these assemblages were the tubificid oligochaetes Limnodrilus (LIMNO) and Branchiura sowerbyi (BRAN), and the chirono- mids Tanypus (TANYP), Chironomus (CHIR), Cladotanytarsus (CLAD), Coelotanypus (COEL), Microchironomus (MICRO), and Paracladopelma (PARAC, Fig. 2). Secondarily, invertebrate assemblages were structured by collection year (low axis-2 scores), specific conductance (high axis-2 scores), and proportion of wetlands in the site drainage subbasin (low axis-2 scores; Fig. 2). Sites with high proportion of wetlands were located exclusively in the Washita River arm, and typically contained the oligochaetes Nais (NAIS) and Pristina (PRIS), and the chironomids Larsia (LARS), Glyptotendipes (GLYP), and Tanytarsus (TANYT; Fig. Anthropogenic influences on aquatic communities 109 measures were inadequate for assessing anthropogenic stress 0.6 with fishes and benthic inverte* brates in Lake Texoma. * In contrast, paired CCA * * scores indicated differences in assemblage structure between 0.4 reference and impact sites for both fishes and benthic invertebrates. With fishes, axis-1 differences were near zero and not 0.2 significant whether data were pooled across years (P=0.191) or not (P=0.418-0.528; Fig. 3). Conversely, CCA axis-2 scores were significantly different 0.0 during all 3 years (1999, Overall 1999 2000 2001 P=0.029; 2000, P=0.049; 2001, P=0.012) and when data were Fig. 3. Mean difference (difference = impact site - reference site) in CCA axis1 and axis-2 site scores of fish assemblages between paired reference and pooled across years (P=0.001; impact sites at Lake Texoma, 1999-2001. Asterisk denotes significant differFig. 3). This suggested that ence (P < 0.05). Whisker represents standard error. reference-impact structural differences with fishes were 2). However, the wetlands gradient was dominated more related to secondary environmental gradiby one site with a small drainage area, and was the ents in Lake Texoma, such as level of residential only site with a significant proportion of wetland development and substrate content, which may habitat in its sub-basin (35%, no other site with have varied significantly within site pairs. Surprismore than 5%). Unlike fishes, invertebrate assemingly, site exposure also was important in discrimblages differed not only between the main reserinating reference and impact sites along CCA axis voir and river arms on CCA axis 1, but also 2. Although small differences in site exposure did between the two different river arms on CCA axis exist within some site pairs, this observation was 2. not expected because the site-pairing procedure should have controlled for this effect. This observation was driven mostly by the increased abunReference-impact site comparisons dances of sunfishes at impact sites in marinas compared to their reference sites. Paired analyses. With benthic invertebrates, distribution of Overall, neither fish species richness (P=0.3070.584) nor fish abundance 1.0 (P=0.452-0.985) were significantly different between reference * 0.8 and impact sites. Similarly, no dif* ferences in fish richness or abun* 0.6 * dance were detected with any particular impact types (P=0.0980.4 0.854, Bonferroni -adjusted). Similar results were obtained for 0.2 benthic invertebrate assemblages (richness, P=0.132-0.277; abundance, P=0.087-0.992); no differ0.0 ences in invertebrate richness or abundance were detected with -0.2 any particular impact types CCA axis 1 CCA axis 2 (P=0.103-0.875, Bonferroni-0.4 adjusted). Results with both assemblages were identical Overall 1999 2000 2001 regardless of whether data were pooled across years or analysed Fig. 4. Mean difference (difference = impact site - reference site) in CCA by individual year. Furthermore, axis-1 and axis-2 site scores of benthic invertebrate assemblages between results indicated that species richpaired reference and impact sites at Lake Texoma, 1999-2001. Asterisk denotes significant difference (P < 0.05). Whisker represents standard error. ness and organism abundance Mean difference Mean difference CCA axis 1 CCA axis 2 110 M. A. Eggleton et al. 0.4 m 2 = 0.92 P = 0.008 Reference/impact Procrustean residual 1999 0.3 0.2 0.1 0.0 AGRIC BEACH DUMP MARINA OIL SEPTIC 0.4 m 2 = 0.47 P = 0.001 Reference/impact Procrustean residual 2000 0.3 0.2 0.1 0.0 AGRIC BEACH DUMP MARINA OIL SEPTIC 0.4 m 2 = 0.64 P = 0.001 Reference/impact Procrustean residual 2001 0.3 0.2 0.1 0.0 AGRIC BEACH DUMP MARINA OIL SEPTIC Fig. 5. Procrustean analysis residuals from comparison of fish assemblages at reference and impact sites in Lake Texoma, 19992001, averaged by impact type. Whisker represents standard error. CCA axis scores also indicated structural differences between reference and impact sites (Fig. 4). However, unlike fishes, differences were restricted to axis-1 scores. Differences in axis-1 scores between reference and impact sites were significant during all 3 years (1999, P=0.006; 2000, P=0.001; 2001, P=0.009) and when data were pooled across years (P=0.001; Fig. 4). Conversely, differences in axis-2 scores were not significant (individual years, P=0.176-0.474; years pooled, P=0.227). Thus, primary gradients such as watercolumn chlorophyll-a and Secchi reading (also identified from the fish CCA) were important in discriminating benthic invertebrate assemblages at reference and impact sites. Procrustean analyses. Procrustean analyses gave different results for fishes and invertebrates. With fishes, reference and impact site assemblages were highly concordant each year (P=0.001-0.008), meaning that the general structure of fish assemblages was consistently similar between reference and impact sites through time. Mean Procrustes residuals indicated that marinas usually contained the most different fish assemblages between reference and impact sites (Fig. 5), although the difference was not large enough to yield a nonsignificant test. Conversely, benthic invertebrate assemblages were highly discordant all 3 years (P=0.14-0.88), which indicated that reference-impact differences in assemblage structure approximated differences found between random communities (Jackson 1995). Mean Procrustes residual values indicated that sanitary dump (0.29±SE of 0.09) and agricultural (0.26±0.05) site pairs contained the most different assemblages (Fig. 6). However, mean residuals exceeded 0.20 for all impact types, which suggested consistent structural differences between reference and impact sites. Regardless of impact type, impact sites consistently contained greater abundances of oligochaetes (mean oligochaete-chironomid ratio 3.54±1.45), whereas greater abundances of chironomids were found at reference sites (0.42±0.10). Mantel tests. Mantel test results generally supported Procrustean analysis results for both fishes and benthic invertebrates. With fishes, significant positive concordance (P=0.001-0.027) was detected between the assemblage structures of reference and impact sites during 2 of 3 years (Fig. 7). During the year for which a nonsignificant result was obtained (2000), the association between reference and impact sites was still strongly concordant (P=0.079). With 111 Anthropogenic influences on aquatic communities Procrustean analysis and Mantel tests also were used to assess the concordance of fish and benthic invertebrate assemblages. Both analyses were done separately for references site and impact site assemblages. Mantel tests indicated no concordance between fish and invertebrate assemblages at reference (P=0.667-0.984) or impact (P=0.3020.736) sites. Procrustean analyses also indicated no concordance between assemblages during 1999 and 2000 (reference sites, P=0.130-0.229; impact sites, P=0.207-0.409). However, significant concordance between fish and invertebrates was detected during 2001 at both reference (P=0.001) and impact (P=0.023) sites, which indicated that the arrangement of site pairs in ordination space was closer than expected by chance (Jackson 1995). This concordance also corresponded with the one of three years that reference-impact invertebrate assemblages were judged concordant from Mantel tests. Although it is possible that inclusion of additional axes may have produced more significant tests, three axes usually contain most of the relevant biological information in ordination (Legendre, Legendre 1998). 4. Discussion Community responses to anthropogenic activities m 2 = 1.18 P = 0.137 Reference/impact Procrustean residual 1999 0.5 0.4 0.3 0.2 0.1 0.0 AGRIC BEACH DUMP MARINA OIL SEPTIC 0.6 2000 Reference/impact Procrustean residual Concordance of fish and invertebrate assemblages 0.6 m 2 = 1.53 P = 0.882 0.5 0.4 0.3 0.2 0.1 0.0 AGRIC BEACH DUMP MARINA OIL SEPTIC 0.5 2001 Reference/impact Procrustean residual benthic invertebrates, no concordance was detected between assemblage structures at reference and impact sites during 2 of 3 years (1999, P=0.401; 2000, P=0.248; 2001, P=0.018; Fig. 8). Nonsignificant results in 1999-2000 were largely due to high numbers of 0% similarities between reference and impact sites, which indicated no taxa in common within site pairs. m 2 = 1.28 P = 0.281 0.4 0.3 0.2 0.1 0.0 AGRIC BEACH DUMP MARINA OIL SEPTIC Fig. 6. Procrustean analysis residuals from comparison of benthic invertebrate assemblages at reference and impact sites in Lake Texoma, 1999-2001, averaged by impact type. Whisker represents standard error. Fishes. CCA axis scores identified some species/taxa from each assemblage type that were mostly associated with either reference or impact sites. With fishes, we did not observe a predominance of pollution-tolerant species (e.g., green sunfish L. cyanellus, golden shiner Notemigonus chrysoleucas, or bullheads Ameiurus spp.; Ohio EPA 1987) at impact sites in Lake Texoma. Overall, largemouth bass and threadfin shads Dorosoma petenense on average tended to be more abundant at reference sites, whereas white crappies, bluegills, brook silversides, and blackstripe topminnows were consistently more abundant at impact sites. The largest assemblage differences were observed at marinas and, to a lesser extent, agricultural sites. In marinas, reference-impact differences were due 112 M. A. Eggleton et al. D. cepedianum, bluegills, white crappies, and mosquitofishes Gambusia affinis were found at impact sites; greater abunZ = 0.427 0.8 dances of threadfin shads and bullhead P = 0.027 minnows were characteristic of reference sites. 0.6 Fish species associated with impact sites in Lake Texoma were not 0.4 characteristic of environmentally stressed conditions, and are not classified as tolerant species indicative of point-source 0.2 organic or heavy-metal pollution (Ohio EPA 1987; USEPA 1998). Rather, these particular species are mostly trophic spe0.0 0.0 0.2 0.4 0.6 0.8 1.0 cialists and their abundances reflected a wide array of human-induced influences 1.0 in Lake Texoma. Marina impact sites con2000 tain high levels of artificial structure (e.g., boat houses, docks, moorings) and shoreZ = 0.265 0.8 P = 0.079 line modifications, which created a gradient of habitat heterogeneity between impact sites and reference sites (e.g., Trial 0.6 et al. 2001). Observed assemblage shifts at marina impact sites towards a topmin0.4 now-brook silverside-sunfish assemblage, which are mostly bottom- and surface-feeding insectivores (Robison, 0.2 Buchanan 1988), were consistent with an increase in habitat complexity associated with proximity to marinas (Hosn, Down0.0 ing 1994). 0.0 0.2 0.4 0.6 0.8 1.0 Fish assemblage shifts at agricultural sites were less obvious. Differ1.0 ences in local habitat structure unrelated 2001 to agricultural practices may partially 0.8 explain assemblage differences. For Z = 0.554 P = 0.001 instance, one impact site contained high abundances of emergent vegetation not 0.6 found at other sites, which would correspond with high abundances of juvenile bluegills and white crappies (Robison, 0.4 Buchanan 1988), whereas another impact site was shallow (0.5 m) with low gradi0.2 ent (1%) and contained large areas of flooded grass, consistent with the high abundances of mosquitofishes (Robison, 0.0 Buchanan 1988). Given that fish assem0.0 0.2 0.4 0.6 0.8 1.0 blages respond rapidly and predictably to Reference site PSI (%) changes in habitat complexity (Gorman, Karr 1978; Angermeier, Karr 1984; JenFig. 7. Scatterplots of percent similarity values (PSI) from Mantel tests on fish assemblages at reference and impact sites in Lake nings et al. 1999), observed shifts at these impact sites were, as in marinas, largely a Texoma, 1999-2001. function of predictable habitat changes to increased abundances of blackstripe topminand not severe point- or nonpoint-source pollution nows, brook silversides, and several Lepomis sunin Lake Texoma. fishes at impact sites, and greater abundances of largemouth bass, blacktail shiners Cyprinella Benthic invertebrates. venusta, inland silversides, and bigscale logperch Benthic invertebrate assemblages at impact Percina macrolepida at reference sites. At agriculsites in Lake Texoma exhibited three main charactural sites, greater abundances of gizzard shads teristics. At impact sites, invertebrate richness was 1.0 Impact site PSI (%) Impact site PSI (%) Impact site PSI (%) 1999 Anthropogenic influences on aquatic communities 113 Impact site PSI (%) Impact site PSI (%) Impact site PSI (%) 1.0 on average about one taxon greater and abundances 40% greater than reference 1999 sites. Abundances of invertebrates have Z = 0.017 0.8 been shown to increase in environmenP = 0.401 tally degraded conditions, especially eutrophic conditions associated with 0.6 organic enrichment (Wiederholm 1984; Cyr, Downing 1988). Impact sites also showed a tendency towards an 0.4 oligochaete-dominated fauna, which is consistent with previous pollution-based 0.2 studies with invertebrates (Howmiller, Scott 1977; Wentsel et al. 1978; Winner et al. 1980). Specifically, we observed 0.0 greater abundances of tubificid 0.0 0.2 0.4 0.6 0.8 1.0 oligochaetes (Limnodrilus, Aulodrilus, and Branchiura sowerbyi), which are 1.0 characteristic of polluted environments 2000 (Milbrink 1978; Lauritsen et al. 1985), Z = 0.066 and may have been indicators of anthro0.8 P = 0.248 pogenic impact at some sites. Limnodrilus was consistently more abundant at all impact sites regardless of 0.6 impact type, whereas abundances of Aulodrilus and Branchiura sowerbyi 0.4 were greater only at agricultural and oil impact sites. All of these taxa are commonly associated with eutrophic condi0.2 tions or some form of nutrient enrichment (Wiederholm 1984), a scenario likely at many of the impact sites in 0.0 Lake Texoma due to the extensive row0.0 0.2 0.4 0.6 0.8 1.0 crop agriculture and pasture lands in the watershed. 1.0 Although the chironomid taxa col2001 lected were similar to those previously Z = 0.264 0.8 found in Lake Texoma (Vaughn 1982), P = 0.018 certain chironomids also may have indicated degraded environmental conditions 0.6 at some sites. The genus Cricotopus is commonly associated with heavy-metal pollution (Winner et al. 1980), but they 0.4 contributed only 0.4% of the total abundance of invertebrates, and heavy-metal concentrations were always below detec0.2 tion limits at sites where the taxon was collected. Abundances of Chironomus were consistently greater at agricultural 0.0 0.0 0.2 0.4 0.6 0.8 1.0 impact sites, and Tanypus abundances were consistently greater at agricultural Reference site PSI (%) and oil impact sites. Chironomus and Fig. 8. Scatterplots of percent similarity values (PSI) from Mantel Tanypus are common genera of which tests on benthic invertebrate assemblages at reference and impact some species have been associated with sites in Lake Texoma, 1999-2001. heavy sedimentation and nutrient enrichment (Wiederholm, Erickson 1979; Winnell, cations sufficient for assessing many environmenWhite 1985; Dawson, Hellenthal 1986). The lack tal impacts (Ferraro, Cole 1992). of species-level identifications precludes a more Overall, results suggest that observed shifts in thorough assessment. However, pollution tolerinvertebrate assemblages between reference and ances for many benthic invertebrates are fairly impact sites were likely related to anthropogenic similar across genera making genus-level identifiinfluences in Lake Texoma. However, the influ- 114 M. A. Eggleton et al. ences were those (e.g., row-crop agriculture) with the potential of producing widespread nonpointsource effects such as sedimentation and nutrient enrichment. No evidence was found that suggested assemblage shifts were in response to point-source pollutants such as industrial effluents or runoff, which is a priority concern for the USEPA. Concordance of fish and invertebrate assemblages We observed that fish and invertebrate assemblages in Lake Texoma were structured around similar variables, yet two conventional methods showed little concordance between fish and invertebrate assemblages at reference or impact sites. These results were opposite those obtained by Jackson, Harvey (1993), who found significant concordance between fish and invertebrate assemblages across 40 Canadian lakes, despite assemblages being structured along different environmental gradients. Fish assemblages in their study were structured around lake morphological measures (e.g., depth, area, volume), whereas invertebrate assemblages were related more to water chemistry measures (mainly pH, but also dissolved organic carbon, nitrate, sulphate). They speculated that observed patterns were the result of biotic processes within and between assemblages, most likely interactions between fish predators and invertebrate prey. So what factors might affect the concordance or discordance between assemblages in a given aquatic system? Multi-assemblage studies such as Jackson, Harvey (1993) are rare; thus, it is not widely established whether assemblages are likely to be discordant or concordant in aquatic systems. Furthermore, our study and theirs were conducted at considerably different scales (i.e. 40 lakes vs. 40 sites in one reservoir), so potential causative processes could well vary more among systems than within systems. In Lake Texoma, observed discordance between fish and invertebrates was partly attributable to the extreme temporal variability exhibited by the dominant taxa in each assemblage. With fish assemblages, the three most abundant species (88% of the total numbers) exhibited large fluctuations in abundance each year. Mean abundances of striped bass increased approximately 90-fold from 0.8±0.3 fish seine haul-1 in 1999 to 90±31 fish seine haul-1 in 2001. Threadfin shad abundance increased greater than 300% during 2000, whereas inland silversides exhibited a 90% increase between 1999 and 2000 followed by a 70% decrease in 2001 (Schnell et al. 2002). Similarly, annual fluctuations in dominant benthic invertebrate taxa also were observed. Abundances of two of the three most abundant oligochaetes (22% of the total numbers) declined substantially during the study. Abundances of Limnodrilus declined from 7.1±1.4 dredge-1 in 1999 to 2.7±1.3 dredge-1 in 2001, whereas Branchiura sowerbyi declined from 1.5±0.5 dredge-1 in 1999 to 0.5±0.2 dredge-1 in 2001 (Schnell et al. 2002). Additionally, three of the five most abundant chironomids (30% of the total numbers) also exhibited substantial increases in abundance during the same period: Polypedilum increased from 1.7±0.6 dredge-1 in 1999 to 6.9±1.4 dredge-1 in 2001; Dicrotendipes increased from 0.4±0.2 dredge-1 in 1999 to 5.2±1.8 dredge-1 in 2001; and Glyptotendipes increased from 0.05±0.02 dredge-1 in 1999 to 1.3±0.7 dredge-1 in 2001 (Schnell et al. 2002). Declining abundances of dominant taxa associated with impact sites (e.g., Limnodrilus and Branchiura sowerbyi) were likely related to the significant concordance of reference and impact invertebrate assemblages suggested by Mantel tests in 2001 (assemblages had been discordant in 1999 and 2000). In other words, declines in the abundances of taxa that made reference and impact assemblages different in 1999-2000 led to greater concordance in assemblages in 2001. Why are littoral-zone fish and macroinvertebrate assemblages in Lake Texoma so variable from year to year in terms of species or taxa abundance? The cause of this phenomenon is not entirely clear. Previous studies (e.g., Lienesch, Matthews 2000; Gido et al. 2002; Eggleton et al. in press) have suggested that littoral-zone areas in Lake Texoma are highly unstable environments for aquatic biota, especially for fishes and presumably (from this study) benthic invertebrates. Lienesch, Matthews (2000) indicated that wind velocity and wave height were strongly related to the structure of littoral-zone fish assemblages. Gido et al. (2002) showed significant relationships between fish assemblages and degree of northern exposure and silt/sand content, measures that are related to site exposure. In addition, Eggleton et al. (in press) reported that site exposure to wind and waves was significantly related to both larval and juvenile/adult fish assemblages in Lake Texoma. Substrates along highly-exposed shorelines in Lake Texoma are mostly medium and coarse sands, which are relatively unstable and tend to support lower densities of invertebrates (Minshall 1984). Although we can only speculate, exposure to wind and waves and associated physical changes (e.g., sand-dominated substrates) at many sites in Lake Texoma may be of such magnitude that common interactions purported to be involved in the structuring of assemblages in other types of systems (e.g., Lyons, Magnuson 1987; Hinch et al. 1991; Benson, Magnuson 1994; those suggested by Jackson, Harvey 1993) may be absent or intermittent and difficult to detect. Thus, the general absence of associations between fish and macroinvertebrate assemblages may not be Anthropogenic influences on aquatic communities surprising in Lake Texoma given the environmental variability in shoreline areas and associated variability in assemblages. We did expect that responses to anthropogenic influences, if detected, might be different between assemblage types, likely because fishes and benthic invertebrates respond to impacts at different scales of space and time (e.g., Allen et al. 1999b). For instance, fish and invertebrates were structured along similar primary gradients, specifically water clarity and water-column chlorophylla. However, paired reference-impact differences on CCA axis 1 were not significant for fishes but significant for invertebrates. This suggested that observed differences in these variables within reference-impact site pairs were not sufficient to affect fish assemblages. In other words, although fish assemblages may have may differed substantially between clear and turbid sites, water clarity itself varied little between reference and impact sites that usually were less than 3 km apart. However, because reference-impact invertebrate assemblages differed significantly along this same gradient (i.e., composed of mostly the same variables in ordination), differences in water clarity and water-column chlorophyll-a may have been substantial enough to influence invertebrate assemblages. Small within-site pair differences in water clarity, water-column chlorophyll-a, and/or TKN (an inorganic nutrient measure) that were negligible to fishes may have been significant to benthic invertebrates. An example would be the production of benthic algae, which provides food and habitat for many invertebrates (Merritt, Cummins 1996), and might be affected by subtle differences in water clarity or nutrients. Benthic invertebrates also have much less mobility than fishes, and would likely reflect environmental stress at the community level more readily. Thus, benthic invertebrates may have reflected anthropogenic impacts at finer scales than fishes, which might be indicators of larger-scale impacts in Lake Texoma. This suggestion also is consistent with rapid bioassessment protocols developed by the USEPA (Barbour et al. 1999), which presuppose such scale-related differences in assemblages in their monitoring programmes. 5. Conclusions Fish and benthic invertebrate assemblages in Lake Texoma did not exhibit high levels of degradation attributable to anthropogenic activities. Minor shifts in fish assemblage structure at some impact sites may have been related to anthropogenic influences, but shifts were not indicative of severe point-source organic or heavy-metal pollution. Rather, fish assemblage shifts appeared related to human modification of shoreline habi- 115 tats, especially in marinas. Benthic invertebrates exhibited greater differences in assemblage structure between reference and impact sites, presumably in response to nonpoint-source impacts associated with row-crop agriculture. Furthermore, our analyses suggested that species or taxa richness alone was inadequate for assessing anthropogenic stress on fish and invertebrate assemblages in Lake Texoma. This finding concurs with Allen et al. (1999a), who regarded richness as a somewhat ambiguous measure of biotic integrity and called for more intensive analyses that focused on other community structural attributes. We found little concordance between fishes and benthic invertebrates despite assemblages being structured along similar environmental gradients. The high annual variation in the abundance of several community-dominant fishes and invertebrates contributed most to this observation. Overall results suggested that fishes and invertebrates likely responded to anthropogenic influences at different scales of space and time in Lake Texoma. We do acknowledge that our study may have been conducted at such a broad scale that assemblage differences may have been difficult to detect with conventional approaches unless environmental impacts were of large magnitude. Nevertheless, our community-based results provide a thorough baseline assessment of present-day ecological conditions in Lake Texoma, and highlight areas in need of future monitoring and research. Acknowledgements Financial support for this study was provided by the U.S. Army Corps of Engineers and the U.S. Environmental Protection Agency. We are grateful to D. Cobb, R. Page, and L. Weider for the use and maintenance of equipment and facilities provided at the University of Oklahoma Biological Station, and M. Cook of the U.S. Environmental Protection Agency. Field and laboratory assistance was provided by R. Ramirez, C. Hargrave, F. March, M. Furuyama, and E. Johnson. 6. References Adler, R.W., Landman, J.C., Cameron, D.M. 1993. The Clean Water Act 20 years later. Island Press, Washington, DC, USA. Allen, A.P., Whittier, T.R., Kaufmann, P.R., Larsen, D.P., O'Connor, R.J., Hughes, R.M., Stemberger, R.S., Dixit, S.S., Brinkhurst, R.O., Herlihy, A.T., Paulsen, S.G. 1999a. Concordance of taxonomic richness patterns across multiple assemblages in lakes of the northeastern United States. Can. J. Fish. Aquat. Sci. 56, 739- 116 M. A. Eggleton et al. 747. Allen, A.P., Whittier, T.R., Larsen, D.P., Kaufmann, P.R., O'Connor, R.J., Hughes, R.M., Stemberger, R.S., Dixit, S.S., Brinkhurst, R.O., Herlihy, A.T., Paulsen, S.G. 1999b. Concordance of taxonomic composition patterns across multiple lake assemblages: effects of scale, body size, and land use. Can. J. Fish. Aquat. Sci. 56, 2029-2040. An, Y.J., Kampbell, D.H., Sewell, G.W. 2002. Water quality at five marinas in Lake Texoma as related to methyl tert-butyl ether (MTBE). Environ. Poll. 118, 331-336. Angermeier, P.L., Karr, J.R. 1984. Biological integrity versus biological diversity as policy directives. BioScience 44, 690-697. Barbour, M.T., Gerritsen, J., Snyder, B.D., Stribling, J.B. 1999. Rapid bioassessment protocols for use in streams and wadeable rivers: periphyton, benthic macroinvertebrates and fish, second edition. EPA 841-B-99-002. U.S. Environmental Protection Agency, Office of Water, Washington, DC, USA. Benson, B.J., Magnuson, J.J. 1994. Spatial heterogeneity of littoral fish assemblages in lakes: relation to species diversity and habitat structure. Can. J. Fish. Aquat. Sci. 49, 1493-1500. Conover, W.J. 1980. Practical nonparametric statistics, second edition. John Wiley & Sons, New York, New York, USA. Cyr, H., Downing, J.A. 1988. Empirical relationships of phytomacrofaunal abundance to plant biomass and macrophyte bed characteristics. Can. J. Fish. Aquat. Sci. 45, 976-984. Dawson, C.L., Hellenthal, R.A. 1986. A computerized system for the evaluation of aquatic habitats based on environmental requirements and pollution tolerance associations of resident organisms. EPA /600/S3-86. U.S. Environmental Protection Agency, Environmental Research Laboratory, Corvallis, Oregon, USA. Eggleton, M.A., Ramirez, R., Hargrave, C.W., Gido, K.B., Masoner, J.R., Schnell, G.D., Matthews, W.J. 2004. in press Predictability of littoral-zone fish assemblages through ontogeny in Lake Texoma, Oklahoma-Texas, USA. Env. Biol. Fish. 70, 000000. Ferraro, S.P., Cole, F.A. 1992. Taxonomic level sufficient for assessing a moderate impact on macrobenthic communities in Puget Sound, Washington, USA. Can. J. Fish. Aquat. Sci. 49, 1184-1188. Gelwick, F.P., Matthews, W.J. 1990. Temporal and spatial patterns in littoral-zone fish assemblages of a reservoir (Lake Texoma, Oklahoma-Texas, U.S.A.). Env. Biol. Fish. 27, 107-120. Gelwick, F.P., Matthews, W.J. 1992. Effects of an algivorous minnow on temperate stream ecosystem properties. Ecology 73, 1630-1645. Gido, K.B., Hargrave, C.W., Matthews, W.J., Schnell, G.D., Pogue, D.W., Sewell, G.W. 2002. Structure of littoral-zone fish communities in relation to habitat, physical, and chemical gradients in a southern reser- voir. Env. Biol. Fish. 63, 253-263. Gorman, O.T., Karr, J.R. 1978. Habitat structure and stream fish communities. Ecology 59, 507-515. Gower, J.C. 1971. Statistical methods for comparing different multivariate analyses of the same data. In: Hodson, F.R., Kendall, D.G., Tantu, P. [Eds]. Mathematics in the archaeological and historical sciences. Edinburgh University Press, Edinburgh, England, pp. 138-149. Hickman, G.D., McDonough, T.A. 1996. Assessing the reservoir fish assemblage index: a potential measure of reservoir quality. In: Miranda, L.E., DeVries, D.R. [Eds] Multidimensional approaches to reservoir management. American Fisheries Society, Bethesda, Maryland, USA, pp. 85-97. Hinch, S.G., Collins, N.C., Harvey, H.H. 1991. Relative abundance of littoral zone fishes: biotic interactions, abiotic factors, and postglacial colonization. Ecology 72, 1314-1324. Hosn, W.A., Downing, J.A. 1994. Influence of cover on the spatial distribution of littoral-zone fishes. Can. J. Fish. Aquat. Sci. 51, 1832-1838. Howmiller, R.P., Scott, M.A. 1977. An environmental index based on the relative abundance of oligochaete species. J. Water Poll. Contr. Fed. 49, 809-815. Jackson, D.A. 1995. PROTEST: a Procrustean randomization test of community environmentconcordance. Ecoscience 2, 297-303. Jackson, D.A., Harvey, H.H. 1993. Fish and benthic invertebrates: community concordance and communityenvironment relationships. Can. J. Fish. Aquat. Sci. 50, 2641-2651. Jennings, M.J., Fore, L.S., Karr, J.R. 1995. Biological monitoring of fish assemblages in Tennessee Valley reservoirs. Reg. Riv. Res. Manage. 11, 263-274. Jennings, M.J., Bozek, M.A., Hatzenbeler, G.R., Emmons, E.E., Staggs, M.D. 1999. Cumulative effects of incremental shoreline habitat modification on fish assemblages in North Temperate lakes. N. Am. J. Fish. Manage. 19, 18-27. Karr, J.R. 1991. Biological integrity: a long-neglected aspect of water resource management. Ecol. Appl. 1, 66-84. Lauritsen, D.D., Mazley, S.C., White, D.S. 1985. Distribution of oligochaetes in Lake Michigan and comments on their use as indices of pollution. J. Great Lakes Res. 11, 67-76. Legendre, P., Legendre, P. 1998. Numerical ecology, second edition. Elsevier Science, Amsterdam, Netherlands. Lienesch, P.W., Matthews, W.J. 2000. Daily fish and zooplankton abundances in the littoral zone of Lake Texoma, Oklahoma-Texas, in relation to abiotic factors. Env. Biol. Fish. 59, 271-283. Lyons, J., Magnuson, J.J. 1987. Effects of walleye predation on the population dynamics of small littoral-zone fishes in a northern Wisconsin lake. Trans. Am. Fish. Soc. 116, 29-39. Matthews, W.J. 1984. Influence of turbid inflows on ver- Anthropogenic influences on aquatic communities tical distribution of larval shad and freshwater drum. Trans. Am. Fish. Soc. 113, 192-198. McCune, B., Mefford, M.J. 1999. PC-ORD for Windows - Multivariate analysis of ecological data, version 4.01. MJM Software, Inc., Gleneden Beach, Oregon, USA. Merritt, R.W., Cummins, K.W. [Eds] 1996. An introduction to the aquatic insects of North America, third edition. Kendall/Hunt Publishing Company, Dubuque, Iowa, USA. Milbrink, G. 1978. Indicator communities of oligochaetes in Scandinavian lakes. Verh. Inter. Verein. Limnol. 20, 2406-2411. Minshall, G.W. 1984. Aquatic insect-substratum relationships. In: Resh, V.H., Rosenberg, D.M. [Eds] The ecology of aquatic insects. Praeger Publishers, New York, New York, USA, pp. 358-400. Miranda, L.E. 1996. Development of reservoir fisheries management paradigms in the twentieth century. In: Miranda, L.E., DeVries, D.R. [Eds] Multidimensional approaches to reservoir management. American Fisheries Society, Bethesda, Maryland, USA, pp. 3-11. Northcote, T.J. 1988. Fish in the structure of and function of freshwater ecosystems: a "top-down" view. Can. J. Fish. Aquat. Sci. 45, 361-379. Ohio Environmental Protection Agency (Ohio EPA). 1987. Biological criteria for the protection of aquatic life. Ohio Environmental Protection Agency, Columbus, Ohio, USA. O'Connor, R.J., Walls, T.E., Hughes, R.M. 2000. Using multiple taxonomic groups to index the ecological condition of lakes. Env. Mon. Assess. 61, 207-228. Palmer, M.W. 1993. Putting things in even better order: the advantages of canonical correspondence analysis. Ecology 74, 2215-2230. Peres-Neto, P.R., Jackson, D.A. 2001. How well do multivariate data sets match? The advantages of a Procrustean superimposition approach over Mantel test. Oecologia 129, 169-178. Rice, W.R. 1989. Analyzing tables of statistical tests. Evolution 43, 223-225. Robison, H.W., Buchanan, T.M. 1988. Fishes of Arkansas. University of Arkansas Press, Fayetteville, Arkansas, USA. Rohlf, F.J. 2002. NTSYSpc: Numerical taxonomy and multivariate analysis system, version 2.11a. Exeter Software, Setauket, New York, USA. SAS Institute, Inc. 2000. Statistical analysis system, version 8.0. SAS Institute, Inc., Cary, North Carolina, USA. 117 Schnell, G.D., Matthews, W.J., Eggleton, M.A., Gido, K.B., Pogue, D.W. 2002. System assimilative capacity (SAC) study, Lake Texoma, Oklahoma and Texas: determination of anthropogenic effects on communities. Interim project report. U.S. Army Corps of Engineers, Tulsa District, Tulsa, Oklahoma, USA/U.S. Environmental Protection Agency, National Risk Management Research Laboratory, Ada, Oklahoma, USA. Schorr, M.S., Sah, J.S., Schreiner, D.F., Meador, M.R., Hill, L.G. 1991. Lake Texoma striped bass fishery: economic impact and cast net evaluation - economic impact of Lake Texoma fishery. Federal Aid in Fish Restoration Project F-49-R, Job No. 2. Final project report. Oklahoma Department of Wildlife Conservation, Oklahoma City, Oklahoma, USA. ter Braak, C.J.F. 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67, 11671179. Trial, P.F., Gelwick, F.P., Webb, M.A. 2001. Effects of shoreline urbanization on littoral fish assemblages. J. Lake Res. Manage. 17, 127-138. U.S. Environmental Protection Agency (USEPA). 1998. Lake and reservoir bioassessment and biocriteria - technical guidance document. U.S. Environmental Protection Agency, Office of Water, Washington, DC, USA. Vaughn, C.C. 1982. Distribution of chironomids in the littoral zone of Lake Texoma, Oklahoma and Texas. Hydrobiologia 89, 177-188. Wentsel, R., McIntosh, A., McCafferty, W.P. 1978. Emergence of the midge Chironomus tentans when exposed to heavy metal contaminated sediment. Hydrobiologia 57, 195-196. Wiederholm, T. 1984. Responses of aquatic insects to environmental pollution. In: Resh, V.H., Rosenberg, D.M. [Eds] The ecology of aquatic insects. Praeger Publishers, New York, New York, USA, pp. 508-557. Wiederholm, T., Eriksson, L. 1979. Subfossil chironomids as evidence of eutrophication in Ekoln Bay, central Sweden. Hydrobiologia 62, 195-208. Winnell, M.H., White, D.S. 1985. Trophic status of southeastern Lake Michigan based on the Chironomidae (Diptera). J. Great Lakes Res. 11, 540548. Winner, R.W., Boesel, M.W., Farrell, M.P. 1980. Insect community structure as an index of heavy-metal pollution in lotic ecosystem. Can. J. Fish. Aquat. Sci. 37, 647-655.