C Copyntni by ihe Amencjn Fisheries Socieiy 2(

•I the American Natural and Anthropogenic Influences on the

"sages for pink 1 DistributioU1SU ru n of the Threatened Neosho Madtom ^ in C. Busack i in a Midwestern Warmwater Stream diversity units a jonid fishes in • MARK L. WILDHABER,Wi LD * ANN L. ALLERT, AND CHRISTOPHER J. SCHMITT "Department of 1 • RAD 95-02. I u.U.Ss. GGeological Survey, Columbia Environmental Research Center. 4204 0 New Haven Road. Columbia. Missouri 65201. USA •' B. A. White. 1 VERNO\ N M. TABOR AND DANIEL MULHERN •v :lbaum). from (miitt U.S. Fish and Wildlife Sen-ice. md potential fl^ fcTO=M&t>\flfcQ8:\L-fYcmr&dlVir a•WK \Houston Street. Suiie E. Manhattan. Kansas 66502. USA urnal of Fish WD KENNETH L. POWELL larcini. J. M. ¥"'••flt; "~ i 0 •'97. Ethnic- W)ttien L^P O1^^ Wesrvi'i/od Professional Semices, Inc., 99 Anagram Drive. Eden Prairie. Minnesota 55344. USA ition-specific • M, D

I SCOTT P. SOWA £r. and R. J. M Missourii CooperativCoopt'ra e Fish and Wildlife Research Unit. The School of Natural Resources, salmon (On- 9 Universityy of Missouri. 302 AB\R Building. Columbia. Missouri 65211. USA 'i the Fraser M Jtellite DNA fl -s- a Abstract. — We aattemptet eh d to discern the contributions of physical habitat, water chemistry, nu- in recovery 9 tnentstrients., and contacontaminantm in s from historic lead-zinc mining activities on the riffle-dwelling benihic •n Biology g; 19 fish communityy off thmadtoms. seem to be limited by the presence of cadmium, lead, and zinc in water and in benihic invertebrate food sources and by Sinauer As- physical habitat. The population density and community structure of fish in the Spring River also seem to be related 10 water chemistry and nutrients. Concurrently, diminished food availability iimating F- may be limiting fish populations at some sites where Neosho madtoms are not found. Many of n structure. the natural factors that may be limiting Neosho madlpm.and other riffle-dwelling fish populations in the Spring River probably are characteristic of the physiographic regiort drained by the upper reach and many of the tributaries of the Spring River. Our results indicate that competition between the Neosho madtom and other species within the riffle-Swelling fish cnrnrnuniiy is jj cause of Neosho madtom^popuraTfon TrrnitatiorrTiT tlie Spiini! RTver7

Relationships between stream fish communities dance of stream fishes (Layher and Maughan 1985; and their habitats have been well documented (An- Layher et al. 1987; Maret et al. 1997). Habitat has germeier and Karr 1984; Matthews and Heins been the primary focus of studies that target factors I/ 1987: Kessler and Thorp 1993). Physical habitat limiting the distribution and density of stream fish- f complexity has been correlated with fish species es, especially threatened and endangered fishes diversity (Gorman and Karr 1978). Habitat factors (Kessler and Thorp 1993; Freeman and Freeman such as waier depth, velocity, and substrate com- 1994). position are important to stream fishes (Aadland The Neosho madtom Noturus p!acidns is a small 1993). Moreover, habitat utilization by stream fish- (<75 mm total length! ictalurid first described as es varies with community composition (Fausch a species in 1969 (Taylor 1969). Neosho madtoms and White 1981: Finger 1982), and water chem- have been found in the highest numbers in riffles istry and nutrients affect the distribution and abun- during daylight in late summer and early fall, after young of the year are estimated to have recruited to the population (Moss 1983; Luttrell el al. 1992; * Corresponding author: mark_wildhaher

243 Spring River Study Area For Enlargement (see Figure 2)

North Fork of the Spring River

FIGURE 1.—Sampling sites on the Neosho. Couonwood. and Spring rivers in 1994. Triangles represent siies where Neosho mudtoms were collected: squares represent sites where they were not collected. and gravel, moderate to slow flows, and depths as supporting 20 fishes not found anywhere else averaging 0.23 m (Moss 1983). Neosho madioms in Kansas. Except for one small population just feed on larval insects among stones at night (Cross upstream-"of Baxter Springs, Kansas (Pflieger and Collins 1995). The Neo'sho madlom was listed ' 1975; Barks 1977; Wilkinson et al. 1996), Neosho as threatened by the U.S. Fish and Wildlife Service madtoms have only been collected from the Spring (USFWS) in May 1990, and a recovery plan was River upstream of the primary sources of pollution approved in September 1991 (USFWS 1991). The from lead fPb) and zinc (Zn) mining (Figure 2). USFWS (1991) hypothesized that habitat and po- Studies of the effects of contaminants on fish tential fish competitors of the Neosho madlom, populations have generally focused on the contam- such as other ictalurids, darters (Percidae), and inants (McCormick et al. 1994) and given little other riffle-dwelling benthic fishes, may limit Ne- attention to other concurrent factors (Neves and osho madlom populations. Currently, Neosho Angermeier 1990, Hall et al. 1996, Scoti and Hall madioms are found in main stems of the Neosho, 1997). Hall et al. (1996) assessed habitat factors Cottonwood, and Spring rivers in Kansas, Mis- along with contaminants; however, they empha- souri, and Oklahoma (Luttrell et al. 1992: Cross sized overall ecological health and biological in- and Collins 1995; Wilkinson et al. 1996) (Figure tegrity of the fish community, not specific popu- 1). The density of Neosho madtoms is much great- lations of fish. Contaminants and physicochemicaJ er in the Neosho system (i.e., the Neosho and Cot- characteristics differ between the Neosho and Cot- lonwood rivers combined) than in the Spring River ton wood rivers (Neosho system) and the Spring (Moss 1983, Wilkinson et al. 1996). Cross and River (Moss 1983; Spruill 1987; Allen and Black- Collins (1995) described the Spring River drainage ford 1995). All are affected by similar anthropo- ver \rea --ment re 2)

3 KUwn«««n

Missouri FIGURE 2.—Enlargement of the Spring River study area with sampling sites in 1994. Triangles represent sites where Neosho madtoms were collected: squares represent sites where they were nol collected. Lead-zinc mining Arkansas and processing occurred within shaded areas.

:'epresem sites genie factors (agriculturaJ runoff and municipal considered in any comprehensive evaluation of waste inputs) (Allen and Blackford 1995; Kiner et these river systems. Previous studies (Moss 1983) nywhere else al. 1997). The Spring River is also impacted by indicate the Spring Rjver tends to be less turbid ipulation just runoff from historic Pb—Zn mining and related ac- and has lower un-ionizcd ammonia (NH3), chloride sas (Pflieger tivities that have resulted in elevated levels of Pb, (Cl), and sulfate (SO,,) concentrations than the Ne- 996), Neosho Zn, and cadmium (Cd) (Barks 1977; Czarneski osho system. Turbidity may provide protection to ;m the Spring 1985; Spruill 1987; Smith 1988; Schmitt et al. the Neosho madtom from predators; NH3, Cl, and s of pollution 1993) and by industrial inputs from chemical man- SO4 may have both natural and anthropogenic • (Figure 21. ufacturing and industrial facilities (Kiner et al. sources (Wetzel 1983). lants on fish 1997). Lead, Zn, Cd. arsenic (As), iron (Fe), mer- Detrimental effects of Pb, Zn, and Cd on fish • theconiam- cury (Hg), and manganese (Mn) are also a concern have been well documented", and all three can be i given little in the Neosho system. However, concentrations of "acutely toxic (USEPA 1986; Eisler 1988). Effects < (Neves and Pb and Zn in fish and sediments of the Neosho have been documented for waterborne (Eisler and L-ott and Hall system are much lower than those historically Hennekey 1977; Weber 1993; Bryan et al. 1995) Libitat factors found in Center and Turkey creeks, tributaries of and dietary exposures (Thomas and Juedes 1992; they empha- the Spring River. Further, As. Fe, Hg, and Mn have Woodward et al. 1994). Lead affects heme syn- Mo/ogicai in- relatively low concentrations in the Neosho system thesis (Johansson-Sjobeck and Larsson 1979), res- pecific popu- (Spruill 1987; Smith 1988: Schmitt et al. 1993; piration (Somero et al. 1977), and reproductive sicochemical Allen and Blackford 1995). Most of the metals of behavior (Weber 1993) of fishes. High concentra- )sho and Cot- concern in the Neosho and Spring River systems tions of Zn cause hyperglycemia (Wagner and d the Spring can be toxic to fish, and water quality standards McKeown 1982), behavioral avoidance (Wood- •n and Bla^ck- for protection of aquatic life have been established ward et al. 1995, 1997), increased heterozygosity lar anthropo- for them (USEPA 1986); therefore, they must be of specific allozymes (Roark and Brown 1996), 246

and reduced survival (Eisler and Hennekey 1977). associated with Neosho madtoms in the Neosho gions. B Cadmzum can affect the immune system (Lemaire- system to compare them with those in the mining- an ecot( Gony et al. 1995), the kjdney (Gill et al. 1989). affected Spring River. We collected data on the effectivt and behavior (Bryan et al. 1995). aquatic community (fish and invertebrate species We st The primary objective of this paper is to eval- richness and density of potential competitors), and Spr uate natural and anthropogenic factors that may be physical habitat (depth, velocity, and substrate of collet limiting the Neosho madtom and other riffle- size), water chemistry (temperature, turbidity, pH. tern. 12 dwelling benthic fishes in the Spring River. We dissolved oxygen, hardness, alkalinity, conductiv- general! wanted to determine if lower densities of Neosho ity, S04, and CD nutrients (un-ionized NH,, nitrite harbor madioms in the Spring River than in the Neosho plus nitrate [NO; +• NO,], and phosphate [PO4]), USFWS system were a result of metals contamination, low- and metals and metalloids (As, Cd. Fe. Hg, Mn, lions (I. er-quality physicochemical habitat, biotic inter- Pb. and Zn) in water, invertebrates, or both. This and Co actions, or some combination of these factors. lisi of measurements was compiled from what oth- Eleven : er researchers had previously identified as factors were aa. Study Area of concern in the Neosho and Spring River sys- 20 grav. The study area included the main stems of the tems, as discussed in the introduction (e.g., Barks and Gra Neosho XGrand) and Cottonwood rivers in Kansas 1977; Spruill 1987: Smith 1988; Allen and Black- in the ri and Oklahoma and the Spring River in Kansas, ford 1995; Kiner et al. 1997). We used an empirical pling at Missouri, and Oklahoma (Figures 1, 2). All are model based on physical habitat, water chemistry, August part of the Arkansas River system. Pan or all of and nutrients measured in the Neosho system dur- At e: the main stems of these rivers are in the Prairie ing 1991 to predict the Neosho madtom distribu- pendicu Parkland Province (Bailey 1995) and the Central tion for that system and Spring River in 1994 with- ly from Irregular Plains (Omernik 1987). The Neosho sys- out information on metals or metalloids. We then of the < tem and the lower Spring River drain mainly compared predicted and observed values from both were si- mixed-grass prairie wilh mature riparian vegeta- river systems and different years to assess the ex- each in tion along some sections, whereas the upper Spring tent to which basic environmental quality arid met- lished \ River and many of its tributaries primarily drain als contamination limited Neosho madiom distri- wide 01 deciduous forests of the Ozark Uplands Province bution in the Spring River. We also used the 1994 (>1.2? ecoregion (Moss 1983). The Spring River and its data to compare the Neosho system to the Spring pled in tributaries drain parts of the Tri-State Mining Dis- River and to compare sites on the Spring River each tr trict in Missouri, Kansas, and Oklahoma (Spruill with Neosho madtoms (madtom sites) to sites on their di 1987). which was mined for Pb and Zn from 1850 the Spring River without Neosho madtoms (no- impact; to the 1960s (Barks 1977). The Spring River drains madtom sites). We compared differences in habitat ceeded approximately half the land area, has 70% of the and benthic communities between the Neosho sys- es, sub mean annual discharge and 1.7 limes the gradient tem and the Spring Rjver relative to differences in face v. of the Neosho system; however, all three rivers in madtom versus no-rnadtom sites within the Spring area b^ this study possess similar riffle-pool habitat (Moss River in an attempt to separate system differences 3 m up 1983: Kiner et al. 1997). The Cottonwood and Ne- from within-Spring River differences. and pr- osho rivers join near Emporia, Kansas; the Neosho The methods we used, to model the Neosho sys- rids, ii and Spring rivers join near Miami, Oklahoma, in tem are supported by the work of others (Layher (Pflieg what is now Grand Lake of the Cherokees (Figure and Maughan 1985: Leftwich et al. 1997). Based Substr: 1). The Cottonwood River. Neosho River upstream on previous research (Moss 1983; USFVv'S 1991, adjacei of its confluence with the Cottonwood River, and Luttrel) et al. 1992; Fuselier and Edds 1994). we cm-de. Spring River are fifth-order streams. Downstream assumed that the abundance of Neosho madtoms pier. 'I of its confluence with the Cottonwood, the Neosho on gravel bars during daylight in late summer- gorize' River is a sixth-order stream. The Neosho and Cot- early autumn is an index of their overall abundance mm, 9 tonwood rivers are regulated by reservoirs. The at a site. The discrete nature of the summsr-fall which Spring River is essentially unregulated until its distribution of the Neosho madtom and its. com- velocii confluence with Shoal Creek in Cherokee County. paratively specialized habitat requirements facil- a Mar Kansas, in a power plant cooling reservoir. itated investigation and habitat modeling. Layher al) stai and Maughan (1985) stated that habitai models are surfaci Methods generally more successful for species with narrow alyzed We quantified'Neosho madiom distribution. Ne- niche requirements than for generalise, and that orimei osho madtom habitat, and the benthic communities they are better applied within than across ecore- turbidi INFLUENCES ON NEOSHO MADTOM DISTRIBUTION 247

gions. Because the lower Spring River represents In 1994, the 1991 sampling procedures were re- an ecotone, the model we developed should be peated except that pore waters and benthic inver- effective. tebrates also were collected, and these samples We selected sites on the Neosho, Cottonwood, along with surface water samples were analyzed and Spring rivers that maximized the probability for metals. Access to certain sites was limited and of collecting Neosho madtoms. In the Neosho sys- we obtained complete data for only 6 of the 11 tem, 12 shoreline gravel bars comprising stones Neosho system sites. At each station, sampling generally less than 38-mm diameter and known to proceeded in the following order to minimize im- harbor Neosho madtoms were selected by the pacts of samples on each other: fishes, benthic in- USFWS for monitoring Neosho madtom popula- vertebrates, substrate, pore water, water depth, and tions (USFWS 1991). All 12 sites on the Neosho water velocity. As in 1991. all ictalurids were iden- and Cotton wood rivers were sampled in 1991. tified in the field and released. Voucher specimens Eleven sites, many the same sites sampled in 1991, of other taxa and unidentifiable fishes were pre- were again sampled in 1994. In the Spring River, served in ethanol for later identification. Benthic 20 gravel bars between the North Fork confluence invertebrates were collected in undisturbed sub- and Grand Lake of the Cherokees (most of the bars strate adjacent to the fished area with a modified in the river) were selected. In 1991 and 1994, sam- Hess sampler (0.1- or 0.037-m: bottom area; the pling at all sites occurred during daylight between smaller one was used for water depths generally August and October. shallower than 0.19 m) with a 0.3-mm-mesh col- At each 1991 site, three to five transects per- lection bag. Substrate within the Hess sampler was ; pendicular to the river channel were spaced equal- disturbed for 2 min. Benthic invertebrates were ly from downstream to upstream along the length preserved in 80% ethanol for later identification of the gravel bar. In most instances, five stations to the lowest taxonomic level possible (Merrill and were spaced equally but at least-2 m apart along Cummins 1984) except chironomids and oligo- each transect. Fewer than five stations were estab- chaetes were not identified below the family level. lished when the river channel was less than 10 m Pore water was extracted directly from undisturbed wide or when a station would be too deep to seine substrate with a vacuum pump system upstream of (> 1.25 m). Transects on each gravel bar were sam- the Hess sample collection site and adjacent to the pled in order from downstream to upstream. On fish collection site. A Hydrolab Surveyer 11 was each transect, stations were sampled in order of used to measure temperature, dissolved oxygen. their distance from the gravel bar. To minimize pH. and conductivity during pore-water extraction. impacts of samples on each other, sampling pro- Pore-water samples were compoiited by transect ceeded in the following order at each station: fish- for subsequent analyses. Each composite sample es, substrate, water depth, water velocity, and sur- was distributed between two acid-cleaned, high- face water. Fishes were collected from a 4.5-m: density polyethylene bottles. One subsample was area by disturbing the gravel substrate. We started analyzed by inductively coupled argon plasma 3 m upstream of a stationary seine (3.0-mm2 mesh) transmission spectroscopy (ICAP) for As, Cd, Fe, and proceeded downstream to the seine. All ictalu- Hg, Mn, Pb, and Zn. The second subsample was rids, including Neosho madtoms, were identified analyzed for alkalinity, hardness, and Cl by titra- .(Pflieger '1~975) and released back into the river. tion;for turbidity with a Hach 2100A turbidimeter, Substrate was collected from an undisturbed area for NH3 with an Orion EA940 meter; and for NO; adjacent to the fish sampling location with a 13- + NO,, SO4, and PO4 with a Hach DR 2000 spec- cm-deep x 10-cm-diameter cylindrical grab sam- trophotometer (APHA et al. 1992). All pore-water pler. The substrate sample was sieved and cate- sampling equipment was acid-cleaned between gorized into five size-classes (<2 mm. 2 to <9 sites. Water velocity at 60% of water depth was mm, 9 to < 19 mm. 19 to < 38 mm, and £38 mm), measured with a Swoffer Instruments model 2100 which were then weighed. Water depth and water current meter. velocity at 60% of water depth were measured with In 1994, after all station samples were collected a Marsh-McBirney model 201 current meter. After at a site, we collected surface water and benthic all station samples were collected at a site, a single invertebrate samples for metals analyses and mea- surface water grab sample was collected and an- sured geospatial coordinates. A surface water grab alyzed with a Hach model DREL/1C portable col- sample was collected from the midpoint of the orimeter for pH. alkalinity, hardness, conductivity, center transect for analysis of rnetals and water turbidity, NHj, NO: + NO3, SO4, PO.,, and Cl. chemistry. Because pore water was extracted on 248 W1LUHABER ET AL.

TABLE 1.—Riffle-dwelling fish laxa collected in the Ne- "Megaloptera" (dobsonflies). and "others" (gen- We osho, Cotionwood. and Spring rivers thai were assumed to erally molluscs). Although Neosho madtoms of i be benthic competitors of the Neosho madtom based on would not eat adults of these large taxa, these taxa fish habitat use and feeding descriptions as given by Pflieger site (1975). were selected to represent concentrations of toxic metals in detritivorous and predatory invertebrates use '•:' Family and scientific name Common name upon which they do feed. Benthic invertebrate con ;.-•= Catostomidae samples of less than 5 g were analyzed for metals V nes Cycleptus elongalus Blue sucker without partitioning. Geospatial coordinates of the & the Hyptnlelium nigricans Nonhem hog sucker gravel bar were determined with a Trimble Path- :'• efai Moxosloma duquesnei Black redhorse ™.y Mtixosloma erylhrurum Golden redhorse finder Plus geographical positioning system. all i Moxosioma macrolepidotum Shonhead redhorse Statistical Analyses •>; of 1 Maxjtsloma spp. cul; We analyzed the data at the site level to assess '.V Sciaemdae by: Aplodinotus grunniens Freshuaifr drum differences between the Neosho system and the Cyprinidae Spring River and between madtom and no-madtom F Erimysltu A-punctaius Gravel chub sites in the Spring River. Arithmetic site means at e Nbtropis spp. or Pimephales spp. i'•*f• ego Phenacobius mirabilis Suckermoulh minnow were calculated for depth, velocity, and pore-water *?•VSfi•; alsc Pimephales nolaluj Bluntnose minnow chemistry and metals. For each metal, we included m Pimephales tenellus Slim minnow only samples with concentrations above the de- /¥*• free Pimephales vigilax Bullhead minnow •*£ buti Ictaluhdae tection limit in the mean because we considered & by Iclolurus punclants Channel caifish these samples a measure of the maximum possible AC Nolurus cxilis Slender madlom exposure at a site. For benthic invertebrates, we stra Noiurus flavus Stonecal calculated species richness and Epherneroptera. indi Noturus miurus Bnndled madlom Plecoptera, and Trichoptera richness (EPT) at each i•$» wat .A/orurur nflctumus Freckled madlom ; : Pviodictis olivaris Flathead catfish site. Previous studies have demonstrated the ef- ';V|' oxy Cottidae fectiveness of these metrics for documenting en- 3... con Coitus Carolina* Banded sculpin vironmental impacts (Kerans and Karr 1994). We ale\ Percidae I Etheosioma blennioides Greeaside darter calculated site densities of Neosho madtoms and, Cot Etheosioma flabellare Fanlail daner as a group, potential competitors (Table 1). We 'j!.1 ben VT Elheosioma nigrwn Johnny darter calculated fish densities by dividing the total num- ^ ' the Elheosioma sligmaeum Speckled daner ber of Neosho madtoms or potential competitors <; ma: Elheosioma speclabile Orangediroal darter '.'^ Elheosinma whipplei Redfin daner collected at a site by the total area sampled with non Elheosioma zonale Banded darter the kick seine. We determined the list of potential r bicii Percina caprodes Logperch competitors based on habitat preferences and food V Percina copeiandi Channel dartct habits of each species, as described by Pfljeger Percina phoxocephala _Slendcrt>ead darter thei Percina shumardi River daner (1975). For each site, we" calculated species rich- ten, ness as a general measure of the natural and an- me; thropogenic impacts on the fish community. Be- the gravel bars from the same interstitial spaces where cause species richness values depend highly on the Sp. Neosho madioms are found and because surface level of effort (sampling time, area, or both), we water and pore-water measurements were similar also calculated species rarefaction, which adjusts (see Wildhaber et al. 1996), only pore-water con- species richness estimates to a constant level of effort (Hurlbert 1971; James and Rathbun 1981), centrations are presented here. However, surface as suggested by Ludwig and Reynolds (1988): water measurements were incorporated into esti- mates of Neosho madlom densities because no pore-water measurements were collected in 1991. Benthic invertebrates for metals analyses were col- E(S) = lected from seines used for fish sampling, aug- mented with kick-net collections when necessary. Invertebrates were placed in acid-washed plastic bags with acid-cleaned, Teflon-coated forceps. £(5) = expected number of species; They were analyzed by 1CAP for the same metals n = total number of fish collected; as pore waters except As and Hg were not ana- n, = total number of fish collected in species i; lyzed. For metals, benthic invertebrate samples N = sample size; were partitioned into "Decapoda" ectec INFLUENCES ON NEOSHO MADTOM DISTRIBUTION 249

-s" (gen- We used rarefaction to calculate expected number k.al-Wallis tests were used to compare observed madtoms of species at a site [£(5)| when a given number of and predicted densities of Neosho madtoms from ;ese laxa fish l.V) are collected. The number of stations per the Spring River in 1994 based on observed den- . of toxic site sampled for fish ranged from 10 to 25, so we sities of Neosho madtoms from the Neosho system lebrates used species rarefaction to make species richness in 1991 (SAS Institute 1990). Stepwise multiple .rtebrate comparable among sites. Comparable species rich- linear regression with forward selection was used T metals ness values among sites were produced by using to develop a model based on physical habitat, wa- es of the the same sample size (N) for each site in all rar- . ter chemistry, and nutrient measures from 1991 ole Path- efaction calculations. The sample size (AO used in Neosho system data. The variable list used in- all rarefaction calculations was the lowest number cluded depth, water velocity, substrate size cate- of fish collected at any one site. We did not cal- gories, geometric mean of substrate size, fredle culate any similarity indices; these were reported index, and surface water chemistry. Inclusion of :o assess individual variables in the model was based on an and the by Schmitt et al. (1997). •madtom • For substrate, we calculated size category means a = 0.15 criterion and a final model in which ail e means al each site by dividing total weight of a size cat- variables were significant at a = 0.05. The model •re-water egory by total weight of all size categories. We based on the 1991 data were used to estimate Ne- included also calculated the substrate geometric mean and osho madtoms densities at sites sampled in 1994. .: the de- fredle index (geometric mean adjusted for distri- Kruskal-Wallis tests were used to validate the >nsidered bution of particle sizes) at each site, as suggested USFWS 1991 model from 1994 Neosho system i possible by McMahon et al. (1996), to characterize sub- data and to assess distributional differences be- rates, we strate suitability for Neosho madtoms. The fredle tween observed and predicted 1994 Neosho mad- leroptera. index relates potential permeability of sediment to tom densities in the Spring River upstream and 0 at each water and hence is an indirect index of dissolved downstream of Center Creek (i.e., most upstream :d the ef- oxygen transport within sediment, and it has been source of mining-derived contaminants). .nting en- correlated with the emergence success of salmonid The statistical methods used to make primary i994). We alevins (Platts et al. 1983, citing other sources). comparisons within 1994 data included analysis of loms and, Composite site means for metal concentrations in variance (ANOVA), correlation analysis, multi- ie 1). We benthic invertebrates were calculated by summing variate A.NOVA (MANOVA), principal compo- total num- the product of metal concentration (fig/g) and bio- nents analysis (PCA), and discriminant analysis •mpeiitors mass of a taxonomic category (g) over all taxo- (SAS Institute 1990). Separate one-way ANOVAs pled with nomjc categories and dividing the sum by the total were performed on site means tor each variable ! potential biomass of all taxonomic categories combined (g). between river systems and between madtom and s and food We first checked site means for normality and no-madtom sites. Along with testing the composite y Pflieger then tested homogeneity of variance for river sys- metal concentration for benthic invertebrates, we ecies rich- tem differences using Levene's test, as recom- tested if either of the major groups, Decapoda and al and an- mended by Milliken and Johnson (1984). In 1994, Megaloptera, biased our composite results. We unity. Be- the number of madtom and no-madtom sites in the tested for differences in metal concentration be- >hly on the Spring River were almost equal (9 and 11 sites, tween Decapoda and Megaloptera at sites where both), we respectively). Therefore, for tests between madtom both groups were represented. We also tested for •ch adjusts and no-madtom sites, we assumed that F-statistics significant differences between river systems arid u level of and i tests for comparisons of normally distributed between, madtom and no-madtom sites for Decap- •un 1981), variables would be effective whether or not vari- oda and Megaloptera concentrations separately. (1988): ances were equal, as suggested by Milliken and Because log)0-transformation of NO2 + NO3 pore- Johnson (1984). Any variable with nonnormal site water concentrations did not produce equal vari-

means was logi(,-transformed. The absence of Ne- ances between nver systems, NO: + NO3 concen- osbo madtoms (density = 0) at 11 of 20 sites in trations were analyzed with a Welch (1951) vari- the Spring River in 1994 made it impossible to ance-weighted ANOVA. Correlation analyses normalize densities through transformation even were used to assess relationships between nonzero with the addition of a constant before transfor- Neosho madtom densities and other variables. We mation. Thus, for 1994 data, we restricted corre- used the multivanate tests to verify the results of lation and regression analyses to madtom sites, the individual ANOVA tests and to determine if n species r. which precluded development of multiple-regres- the significant differences identified by ANOVA sion models. effectively characterized river system and mad- •d. Siepwise multiple linear regression and Krus- tom-no-madtom differences. In our discriminant 10 100 Predicted Neosho madtom density (fish/100 m2)

* Naosho River system G Spring River below Center Creek ° Spring River above Center Creek — Predicted equate observed Neosho FIGURE 3.—Predicted versus observed Neosho madtoms densities in Neosho and Spring river systems. Predicted Fishe values are based on a regression model developed from 1991 USFWS data Observed values were obtained during munitie 1994 collections. Observed densities of the Neosho and Cottonwood rivers combined (Neosho system) wt:re sia- differed nificanllv higher than densities found the Spring River below Center Creek (P = 0.0007; N = 21: all comparison* and inv here are from KruskaJ-Wallis tests), whereas predicted values did not differ between river systems (P = 0.28. A 2). Non = 16). Observed and predicted Neosho madlom densities in the Neosho system were not significantly different for observed and predicted densities, respectively, at Spring River sites above Center Creek (P = 0.057. A' = 21: in the 1 P = 0.19. N = 16). Observed Neosho madlom densities at Spring River sites above Center Creek were significant!) (Table higher than densities at sites below Center Creek (P - 0.008: A' = 20). whereas predicted values were not (P = collecte 0.94; A' = 20). as oppo Density in the I\ analyses, we used stepwise discriminant analyses D = density of Neosho madtoms (number/ contrasi with forward selection followed by removal to pro- 100 rrr); Spring duce a discriminant function. We then tested how C38 = weight proportion of substraten > 38 species well the resulting discriminant function described mm; nor EP'l the observed data. Cl = chloride ion concentration (mg/L). Physi Because we were required.to have more obser- : ems.—'I vations than variables before we did any multi- For the equation, r = 0.72; N = 11: P < 0.017 in their variate analyses, we shortened the list of variables for G3; ajid P < 0.0026 for Cl. Based on a Bon- trient c used for MANOVA, PCA, and discriminant anal- ferroni-adjusted a = 0.0025 (0.05/20 compari- sureme: ysis in three ways. First, we excluded from mul- sons), Cl was highly correlated with SO4 (r = 0.89: substrat tivariate analyses any metal that was detected at P = 0.0003: N = 11), conductivity (r = 0.83: P Neosho fewer than 75% of our sites. Second, we used the =• 0.0015; N = ] 1), and hardness {r = 0.82: P = osho sy. fredle index to represent all substrate categories. 0.0022: N - 11). As a result of these strong cor- and SO Third, we used only variables with P-values less relations, only Cl significantly added to the vari- live, all than 0.05 in one-way ANOVAs. ance in the 199) data that was accounted for b\ the overall regression model. River i" Results At the six 1994 sites where water quality and those oi kaline ' Predicted Neosho Madlom .-Densities subsirale composition were measured in the Ne- osbo system, predicted Neosho madtom densities hard (!.' Stepwise regression with forward selection of : panicu! 1991 USFWS data from the Neosho system pro- ranged from 12.1/100 m- less lo 42.6/100 m more than observed densities (Figure 3). Despite the and vei duced the following equation for predicting Ne- No dou wide range, observed and predicied 1994 densities osho madlom densities from physical habitat, wa- occurrir for the Neosho system were not significantly dif- ter chemistry, and nutrient measurements: 1987). | ferent i Kruskal-Wallis lesl for disiribulional dif- more th. ferences: P = 0.92; df = )). Likewise, predicied INFLUENCES ON NEOSHO MADTOM DISTRIBUTION 251 and observed densities of Neosho madtoms at (132-145 mg/L) than in the Neosho River (49-58 Spring River sites above Center Creek did not dif- mg/L) and more than threefold greater than in fer significantly. Below Center Creek., however, reaches of the Spring River and its tributaries not observed densities were markedly lower than pre- affected by mining (Schmitt et al. 1997). In con- dicted. Of the combined 26 sites on the Neosho. trast, NO: + NOj concentrations were greater in Cottonwood, and Spring rivers, the Spring River the Spring River than the Neosho system (Table sites at the mouth of Center Creek and downstream 2). Depth, velocity, substrate 9 to <19 mm, and ai Willow Creek were predicted to have the two pore-water pH. dissolved oxygen. PO^, and Cl did highest densities of Neosho madtoms. The ob- not differ between river systems (Table 2). served average density of Neosho madtoms was Metals.—Concentrations of various metals in 100% of rhe predicted average density above Cen- pore waters and benthic invertebrates differed sig- ter Creek but only 1% of prediction below Center nificantly between the Neosho system and Spring Creek. Above Center Creek, the predicted average River. Concentrations of Fe and Mn \n pore water density of Neosho madtoms was 13% of that pre- were higher in the Neosho system, Cd was only dicted for the Neosho system, whereas below Cen- detected in Spring River pore waters at the mouth ter Creek the average predicted density was 364% of Turkey Creek, and Pb was not detected in any of the density predicted for the Neosho system. pore-water sample (Table 2). Concentrations of Cd and Pb were higher in composite samples of ben- Neosho System versus Spring River thic invertebrates from the Spring River than in Fishes and invertebrates.— The aquatic com- those from the Neosho system (Table 2). Delect- munities of the Neosho system and Spring River able concentrations of As, Hg, and Zn in pore wa- differed, as illustrated by fish densities and by fish ters and of Fe and Zn in composite Invertebrate and invertebrate community composition (Table samples did not differ significantly between river 2). Nonzero Neosho madtom densities were higher systems (Table 2). in the Neosho system than in the Spring River Except for a few inconsistencies among com- (Table 2). Furthermore, Neosho madtoms were posite. Decapoda. and Megaloptera metal concen- collected at only 9 of 20 sites in the Spring River trations, taxonomic group analyses generally sup- as opposed to 10 of 1 ] sites in the Neosho system. ported the results of river system comparisons Density of potential competitors was also greater based on composite samples (Table 21. Concen- ill the Neosho system than in the Spring River. In trations of Fe and Mn were significantly higher in contrast, fish species rarefaction was greater in the MegaJoptera than in Decapoda (respectively: F - Spring River than in the Neosho system. Neither 62.57 and 9.86; P = 0.0001 and 0.0032; ;V = 42). species richness of fish and benthic invertebrates Concentrations of Cd in Decapoda were not quite nor EPT differed between river systems (Table 2). significantly different between river systems. Lead Physical habitat, water chemistry, and nutri- was detected in the Neosho system at only one site ents.—The Neosho system and Spring River differ for Decapoda and at no sites for Megaloptera. Con- in their physical habitat, water chemistry, and nu- centrations of Zn in Megaloptera were greater in trient concentrations. Most of the substrate mea- the Spring River than in the Neosho system surements and indices indicate that Spring River Physical habitat, water chemistry, nutrients, and substrate consists of coarser gravel than thai of the metals combined.—The shortened list of variables Neosho system (Table 2). Pore waters of the Ne- used in multivariate analyses included pore-water osho system were warmer, harder, had higher NH, temperature, conductivity, turbidity, alkalinity, and SQ4 concentrations, and were more conduc- and hardness; NH,, NO. + NO,, SO.,, and Mn tive, alkaline, and turbid than those of the Spring pore-water concentrations; the fredle index; and River (Table 2). Pore waters from all sites except Cd. Mn. and "Zn concentrations in composite in- those on the Cottonwood River were typicaJly al- vertebrate samples. Results of MANOVA dem- kaline (pH 7.5-8.5. alkalinity 100-160 mg/L) and onstrated a significant difference between river hard (150-220 mg/L); the Cottonwood River was systems (Wilks' lambda: P < 0.002; /V = 22). Prin- particularly high in alkalinity (about 200 mg/L) cipal components analysis of the same variables and very hard (>330 mg/L) (Schmitt et al. 1997). accounted for more than 63% of the variability in No doubt reflecting the dissolution of naturally the data with only the first two principal compo- occurring gypsum in central Kansas (Spruill nents (Figure 4); the first component effectively 1987), pore-water concentrations of sulfale were separated Neosho system sites from Spring River more than twofold greater in the Cottonwood River sites. Based on the same 22 sites used in MAN- 252 WILDHABER ET AL.

TABLE 2.—Means and one-way analysis of variance lesi results IP-values) for comparisons be\\veen the Nensho River TABL system and Spring River and between Spring River sites with and without Neosho madtoms. Abbreviations used are: ND =- none detected; NA = not applicable; DL = detection limit or detection limit range.

Wiihin-Spring River comparisons Between-river comparisons Neosho Neosho Neosho Spring madtoms madtoms system, River: lvalue present: absent-, />-value Dtcapi Measurement mean (N) mean {N} (f) mean (N) mean (W)

Community Pb (DL ' : Neosho rnadlom deniity (per 100 m ) 12.0000) 3.26(9) 0.042 (4.83) 3.2619) 0 111) NA Compn Density of potential benihic fish Decapi competitors (per 100 m-) 301.94(11) 120.89(20) 0.0045 (9.46> 195.49(9] 81.59 (11) i (8.46) Fish species rarefaction 6.18 (11) 8.10(20] 0.0053(9.0*1 8.1 I (9) 8.09 (II) 0.98 (0.00) Megal. Fish species richnesi 16.64(11) 16.35(20) 0.88 (0.02) 19.22(9) 14.00(11) 0.05 (4.42) Invertebrate tana richness 28.2 (5) 33.15(20) 0.17 (2.051 37.00(9) 30.00 (II) 0.021 (6.46] Emphcmeropiera. Plecopiera. Tricoptera (EPTi richness 1840(5) 17.65(20) 0.78 (0.08) 21.33(9) 14.64(11) 0.0038 III.02) HabiLal I.. Water depth (m) 0.33(11) '0.35(20) 0.56 (0.35) 0.37 (9.1 0.39(11) 0.54 (0.40) K Velocity al 60*. of deplh (m/sj 0.44 (III 0.47 (20) 0.56 (0.35 1 0.44 (9) 0.50 (II) 0.28 (1.231 Pore-water temperature (°C) 25.84 (6) 22.05 (20) 0.012 (7.4Oi 20.56 (9) 23.26(11) 0.055 14.23) Pore-waier turbidity (nephelometric units) 254.90(6) 59.27 (20) 0.0007(15.2li 86.83 (9) 43.37(11] ().04i (4.791 Substrate 238 mm (weight %) 12.64(6] 25.72 (20) 0026 (5.64i 22.050) 28.72(111 0.25 (1.42) Substrate <38 to S19 mm (weight %\ 26.67 (6) 35.24 (20) 0.030. (5.35i 36.85 (9i 33.93 (ID 0.46 (0.561 Substrate

ho River TABLE 2.—Cominued. iseil arc: Wilhin-Spnng River comparisons between-nver comparisons Neosho Neosho Neosho Spring madtomi oudtom^ iysiem: River: /•-value present: absent. /•-value Measurement mean (A/) mean (A/) in mean (N) mean I/V) in •.lue Decapoda 40.19(5) 94.10(18) 0.0085 (H.43) 110.89(9) "9.85 (9) 0.29 11.22) ) Meguloplera 54.51 (4) 207.01 (15) OAMX13 (20.25') 241.55(6) ISfe.SI (9) 0.42 (0.70) Pb i.DL = 0.38-2.70) Composite 0.73 (2) 2.01 115) 0.021 (6.61) 1.58(6) 2.36(9) 0.17 (2.15) Decapoda O.f>8 (1) 1.90(8) 0.15 (2.56) 1.46(2) 2.07 (6) 0.53 (0.45) (8.46) .'0.00) Megalopiera ND 2.71 ilO) NA i.Ol (2) 3.48 (8) 0.11 (3.26) '4.421 Zn (DL = 0.76-5.40) 6.461 Composite ::.64(6) 40.14 (20) 0.18 (1.92) 26.64 !9) 56.13(11) 0.005 (10.24)

.11.02) Decapoda 27.43 15) 36.02 (18) 0.32 (1.05) 27.35(9) 47.47(9) 0.032 (5.54) Meguloptera 24.24 (4) 55.34 (15) 0.023 (6.23) 32.64 (6) 78.65(9) 0.0043(11.871 iO.40) (1.23)

OVA, stepwise discriminant analysis produced a tween-river systems analyses, concentrations of Fe (4.79) and Mn were significantly higher in MegaJoptera ' (1.42) list of four significant variables: pore-water al- (0.56) kalinity, NH3, SOj, and temperature. The resulting than in Decapoda (respectively, F = 57.60 and (1.13) i discriminant function successfully categorized by 11.98; P = 0.0001 and 0.0019; /V = 28). Separate river system the 26 sites at which all variables were analyses of invertebrate taxonomic groups pro- (0.34) (0.81) measured (0% error rate). duced the same results as the composite analysis (0.15) and thus supported use of the composite analyses. Neosho Madtom versus No-Neosho Madtom Sites As noted earlier, Cd in pore water was only de- (0.93) in the Spring River tected at the mouth of Turkey Creek, where no (.3.15) Fish and invertebrates.—The following metrics madtoms were collected. Detectable concentra- (0.40) : ;s.5D were significantly greater at Spring River madtom tions of pore-water As, Fe. Hg, Mn, and Zn were (0.1 ft) sites than at no-madtom sites: potential competi- not significantly different between madtom and •(10.031 tors, fish species richness, benthic invertebrate no-madtom sites. Although its concentrations did (0.57) taxa richness, and EPT (Table 2). Fish rarefaction not differ significantly between madtom and no- (2.90) (0.10) did not differ between madtom and no-madtom madtom sites, Zn in pore water was elevated at the (1.04) sites (Table 2). mouths of Center Creek (116 p-g/L) and Turkey Physical habitat, water chemistry, and nutri- Creek (369 u,g/L) (Schrain et al. 1997). (7.64) ents.—Water chemistry and nutrient measure- Physical habitat, water chemistry, nutrients, and ments revealed a few differences between madtom metals combined.—The shortened list of variables (1.22) and no-madtom sites, and no differences in phys- used in multivariate analyses included pore-water (0.96). ical habitat were evident. Madtom sites had higher turbidity, alkalinity, NH:>. EPT (which paralleled NH3, alkalinity, and turbidity than no-madtom invertebrate taxa richness), and Cd and Zn-con- (1.33) sites (Table 2), but depth, water velocity, all sub- cemrations in composite Invertebrate samples. Re- strate size categories, temperature, pH, dissolved sults of MANOVA demonstrated a significant dif- oxygen, conductivity, hardness. NO: + NOV. SO.,, ference between madtom and no madtom sites (6.801 POj, and Cl in pore water did not differ signifi- (Wilks' lambda: P < 0.051; N = 16). Principal (8.4X) cantly between madtom and no-madtom sites. components analyses of the variables used in (5.30) Metals.—Only concentrations of meials in ben- MANOVA accounted for more than 84% of the thic invertebrates differed between madtom and variability in the data with only the first two prin- no-madtom sites. Cadmium and Zn concentrations cipal components: the first component effectively (3.21) in benlhic invertebrates were higher at no-madtom separated most Neosho madtom sites from no- i0.42i sites than at madtom sites, whereas detectable con- madtom sites (Figure 51. Based on the same 16 centrations of Fe, Mn, and Pb in benlhic inverte- sites used in MANOVA. invertebrate Zn concen- 10.74) brates did not differ significantly between madtom tration was the only significant variable in stepwise and no-madtom sites (Table 21. As with the be- discriminant analysis. The resulting discriminant 254 W1LDHABER ET AL.

smullei substrate; higher turbidity, allulinrry. hardneu. conductivity, un - ionized •mmoriit, lutlate, and manganese and lower nitrite.', nitrate if\ water; Um«r cadmium, greater Emphemeropier•, manganese, and zinc In invertebrates Plecoplen. «nd Tricopler htghet alkalinity and in water; Low«r cadmium and zinc in invertebrates

c ffl oc c c o o a a E N N o I 0 O o

5 S

-3-2-10 1 2 3 4 5 -3 -2-10 1 Principal Component 1 Principal Component 1

N Neosho/Cotlonwood Rivers No Neosho madtoms collected S Spring River Neosho madtoms collected

FIGURE 4.—Principal component (PCi analysis of FIGURE 5.—Principal components ('PC) analysis of combined Neosho. Cottonwood. and Spring river data Spring River sites with and without Neosho madtoms based on the shortened list of variables used for MAN- based on the shortened list of variables used for MAN- OVA. For ihe 22 sites at which all PC variables were OVA. For the 16 sites at which all PC variables were measured, detected, or both. PCI accounted for 43^ of measured. PCI accounted for 63% of the variability in the variability in the data and PC2 accounted for over the data and PC2 accounted for 19%. Correlations (or 19%. Correlations (or loadings) of the variables used for loadings) of the variables used for PCI were pore-water PCI were pore-water temperature (r = 0.17, P - 0.45), turbidity (r = 0.60. P = 0.01?). alkalinity (r = 0.84. P turbidity (r = 0.72. P - 0.0002). alkalinity (> = 0.85. = 0.0001). and NH, ir = 0.49. P = 0.056); EPT (r = P = 0.0001). hardness (r = 0.82. P = 0.0001). con- 0.92. P = 0.0001); and invertebrate Cd (r = -0.88. P ductivity (r = 0.71. P = 0.0002). NH, (r = 0.78. P = = 0.0001) and Zn (r = -0.95, P = 0.0001). 0.0001). NO; + NO, (r = -0.69. P = 0.0004). SO, (r = 0.61, P = 0.0024). Mn (r = 0.60. f = 0.003); fredlr index (r = -0.50, P ="0.01 7): and invertebrate Cd \r the Spring River, Neosho madtom populations may = -0.67. P = 0.0007), Mn (r = -0.57, P = 0.006). and Zn (r = -0.62, P = 0.002). also be directly limited by lower benthic inverte- brate abundance (i.e.. food) at sites where Neosho madioms were not collected, possibly as an indi- function based on invertebrate Zn successfully cai- rect result of contaminants. The Neosho madtom egorized as madiom or no-madiom sites 16 of the population numbers also appear limited by avail- 20 Spring River sites (20% error rate). able physical habitat, they may be affected by ba- sic water chemistry and nutrients in the Spring Discussion River. In contrast, our results suggest that com- Through this study, we have shown thai an in- petition between Neosho madtoms and other fishes tegrated approach is necessary to differentiate the is not limiting Spring River Neosho madtom pop- effects of natural and anthropogenic factors on fish ulations. populations and communities. Fishes of the Spring According to estimates from the mode! gener- Rivei. especially the Neosho madiom. may be di- ated for the Neosho system and based on habitat rectly limited by the presence of Pb, Zn, and Cd and water quality of the Spring River, observed in water and indirectly limited by the concentra- Neosho madiom densities in the Sprint; River tions of these metals in benthic invertebrate food above Center Creek were as expecied (i.e.. low), sources as a result of historic Pb-Zn mining. In whereas below Cenier Creek observed densities - INFLUENCES.ON .

were much lower than expected (Figure 3). The in pore water were sufficiently high to be toxic to Spring River below Center Creek appears to con- Spring River fishes. Based on the USEPA (1987) tain habitat that could support Neosho madtom chronic water quality criteria forZn, which is hard-

ochress; densities higher than was found, on average, in the ness-dependent, pore-water concentrations of Zn .uty Neosho system. Thus, Neosho madtom densities exceeded the chronic criterion by 78% at the in the upper portion of the Spring River appear mouths of Center and Turkey creeks. Furthermore, limited only by habitat, whereas densities below because toxicities of heavy metals may be cu- Center Creek appear limited not by physical hab- mulative (Sprague and Ramsay 1965; Wildhaber itat but by the presence of contaminants. Further- and Schmitt 1996). concentrations of Pb, Zn, and more, variation in the accuracy of predicted (rel- Cd that may not be individually toxic may be cu- ative.to observed) densities of Neosho madtoms mulatively toxic in the Spring River. in the Neosho system and Spring River above Cen- Higher metal concentrations in benthic inver- ter Creek (Figure 3) suggest !hat other environ- tebrates and lower densities of potential compet- mental factors not accounted for in the model also itors at sites where Neosho madtoms were not affect Neosho madtom densities. found suggest that Spring River fishes are exposed Highest concentrations of Pb, Zn, and Cd in the to metals indirectly via their food as well as di- Spring River (all media) occurred below the con- rectly via the water. As with waterborne metal con- fluence with Center Creek at sites where Neosho centrations. Pb. Zn, and Cd concentrations in ben- madtoms. were not found (Wildhaber et al. 1996, thic invertebrates were greatest at the mouths of 1997; Schrru'tt et al. 1997). In pore water, Pb was Center and Turkey creeks (Wildhaber et al. 1997). never detected and the only detectable Cd was at Working with laboratory rainbow trout Oncorhyn- the mouth of Turkey Creek: The two highest Zn chus mykiss fed a diet containing a mixture of Pb. levels in pore water occurred at the mouth of Tur- Zn, Cd, and copper (Cu), Farag et al. (1994) re- key Creek (highest) and at the mouth of Center ported scale loss and accumulation of metals in analysis of Creek. Average concentration of Zn in pore water pyloric caeca, and Woodward et al. (1994) dem- 10 madtoms at the mouth of Center Creek was 1.42 times great- onstrated reduced growth and tissue accumulation J for MAN- er than that of the next highest site. Concentrations of metals. The concentrations of Pb, Zn. and Cd "iables were of metals in benthic invertebrates paralleled those in invertebrates found by Farag et al. U994) to be ariability in in water, invertebrates having their highest Pb, Zn, detrimental to fish were less than those we ob- relations (or and Cd levels at the mouths of Turkey and Center served in benthic invertebrates at the mouth of ; pore-water r = 0.84. P creeks. During 1993, dissolved Zn concentrations Center and Turkey creeks; the concentration of Cu i: EPT (r = in the Spring Raver just below the confluence with was slightly higher (Table 3). The concentrations = -0.88. P its North Fork were 2.5-80 pig/L during low river of Pb and Cd in food found to be detrimental to I). flow and 50-80 iLgfL during high flow (Dames and fish by Woodward et al. (1994) were less than those Moore 1993). Dissolved Pb was never greater than we observed in benthic invertebrates at the mouth 1 u.g/L, and Cd never historically exceeded 0.2 of Turkey and Center creeks; the experimental con- lations may u,g/L, but higher concentrations occurred in Tur- centrations of Zn and Cu were 121% and 150% of hie mverte- key. Center, and Short creeks (Dames and Moore. the concentrations we observed at the mouth of icre Neosho Inc., Denver, Colorado, unpublished data). Al- Turkey Creek (Table 3). Other studies have dem- as an indi- though we did not detect either Pb or Cd by ICAP. onstrated detrimental effects of foodborne Pb ho madtorh others have documented elevated concentrations (Thomas and Juedes 1992) and Cd (Rhodes et al. ,d by 'avail- of these elements in the Spring River and its trib- 1985). We were not able to measure concentrations :cied by ba- utaries by more sensitive analytical methods. In of metals in fish and our invertebrate metal anal- the Spring the Spring River below Baxter Springs, dissolved yses were .done on taxa that may or may not be • t that corn- Pb averaged 70 u.g/L from 1974 to 1978. and dis- food of Neqsho madtoms. However, Czarneski other fishes solved Cd averaged about 2 u.g/L. From 1979 (1985) and Schmitt et al. (1993) reported elevated ladtom pop- to!991. dissolved Pb averaged 24 u.g/L and dis- concentrations of Pb. Zn. and Cd in black redhorse solved Cd averaged 3 u.g/L (Dames and Moore from Center Creek, and it is therefore likely that lodel gener- 1993). The dissolved Zn concentration in Center other benthic fishes are similarly contaminated. d on habitat Creek was 264 u.g/L in the summer of 1989 Consequently, our results and those of the studies ±r. observed (Schmitt et al. 1993). Zinc concentrations as great cited here suggest that dietary metals play a role pring River as 200.000 u,g/l_ have been reported in Short Creek in constrajning the Spring River Neosho madtom J (i.e.. 16-W), (Spruill 1987). population. id densities Of the mining-derived metals, Zn concentrations In the Spring River, depauperate invertebrate 256 WILDHABER ET AL.

TABLE 3.—Concentrations of .selected meials in invertebrate food source;, from this study. Farag et al. (1994). and Woodward ei al. (1994). The concentrations presented for Farag et al (1994) and Woodward et aJ. (1994) are the minimum levels at which a detrimental effect on hsh was observed. Concentrations are given as u,g;s wet weight (wet) or p.g/g dry weighl (dry).

Location or sludy: measurement lype Cadmium Copper - Lead Zinc Turkev Creek wei 0.7: 21.20 4.22 HM.29 Turkey Creek, dry 3.04 90.14 17.93 443.47 Center Creek: wei 1.15 18.77 606 126.85 Center Creek: drv 4.26 69.29 22.36 468.35 Farag et al. ( 1994): wet 0.24 26. 1 3 1.77 68.99 Woodward el al. (19941: dry 1.20 109 9.69 655

abundance may directly limit riffle-dwelling ben- Neosho system than in the Spring River parallels thic fishes, including vhe Neosho madiom. Most of previous observations of a preference for moder- the riffle-dwelling benthic fishes in Spring River ate- to fine-grained substrate by the Neosho mad- feed on benthic invertebrates, including the young, tom (Moss 1983; Fuselier and Edds 1994). The small instars of those used in the EPT index (Pfiie- larger average particle size in the Spring River and ger 1975; Mayden et al. 1980; Burr and Mayden the significant negative regression coefficient for 1982; Starnes and Starnes 1985). Our data illus- particle sizes larger than 38 mm in the model used lif! trate the greater numbers of EPT invertebrates at to predict Neosho madtom densities suggest sub- madtom sites than at no-madtom sites (Wildhaber strate limitations for the Neosho madtom and other et al. 1996). The similarity in habitat and the dif- riffle-dwelling benthic fishes in the Spring River, ferences in contaminant concentrations between especially above Center Creek. Perhaps the larger madiom and no-madtom sites suggest that ob- interstitial spaces in Spring River gravel do not served benthic invertebrate patterns resulted from afford as much protection from predators or as contaminants. Phipps et al. (1995.) demonstrated much food for Neosho madtoms as the Neosho sensitivity of aquatic invertebrates to waierborne system provides, but offers habitat and food for Pb, Zn, and Cd with Zn concentrations only 21 other species such as stonecats. and 35% of the surface water concentration we Basic water chemistry and nutrients, differed be- observed at the mouths of Turkey and Center tween river systems; most important are ihose that creeks, respectively (see SchmiU et al. 1997 for differed between madtom and no-madiom sites actual values). (alkalinity, NH3, and turbidity). The waier chem- Depth, velocity, and substrate are important to istry and nutrient patterns we observed in the two Neosho madtoms (Moss 1983;-Fuselier and Edds river systems parallel tho'se observed by Moss 1994). Our study showed no significant differences (1983). We know little aboul the importance of in either depth or velocity between river systems alkalinity and NH, to Neosho madtoms. The high or between Spring River madtom and no-madtom correlation found among the various water chem- sues. However, our study differs from previous istry and nutrient ^measurements makes any dis- investigations in that our analyses are based on cussion of the importance of alkalinity and NH? "W overall site means and not microhabitat values, by themselves highly speculative. However, the which may vary greatly within a site. The specific potential importance of basic water quality to Ne- substrate composition needs of fishes have been osho madtoms populations is suggested by the in- demonstrated by many researchers ('Moy)e and clusion of Cl, which was highly correlated with Vondracek 1985; Wood and Bain 1995) and some conductivity, hardness, and SO4. in the predictive studies have focused on threatened and endangered model for Neosho madtom densities, and by the species (Kessler and Thorp 1993. Freeman and significant differences between madtom and no- Freeman 1994), including other madtoms (Simon- madtom sites in alkalinity, NH,. and turbidity. son and Nevei 1992). Substrate panicle size in the Like other prairie stream fishes (Layher et al. Neosho system tended to be smaller than in the 1987), Neosho madtoms seem to prefer higher tur- Spring River, but there was no difference in par- bidities. Higher turbidities may afford Neosho ticle size distribution between madtom and no- madtoms more protection from predators and more madiom sites within the Spring River (Table 2). opportunity to capture prey with good visual acu- Our observation of smaller-substrate sizes in the ity. There was significantly hieher turbidity _ai_ INFLUENCES ON NEOSHO MADTOM DISTRIBUTION

Spring River madiom sites than at no-madiom sites In the Tn-State District, ground and surface waters but the difference was minimal compared to the in the Spring River drainage are affected by mining fourfold greater turbidity in the Neosho system to varying degrees, and SO4 is an indicator of min- ihan in the Spring River. ing-derived water pollution (Barks 1977; Spruill Many of the physical habitat, water chemistry, 1987). In these areas. SO4 results, from the oxi- nutrient, and community differences observed be- dation of pyrite as well as from the weathering of tween the Neosho system and the Spring River sulfide ore minerals (.sphalerite anJ galena). likely are due to the physiographic regions Contrary to what the USFWS (1991) suggested, drained. Although the main-stem reaches we sam- the observed fish community pattern suggests that pled in these three rivers are all found in the Prairie interspecific competition is not limiting Neosho ParkJand Province ecoregion, the upper reach and madtoms. Fish species richness of Spring River many of the tributaries of Spring River drain the madtom sites was higher than that of no-madtom very different Ozark Uplands Province (Bailey sites, which is likely due to the lower fish densities 1995). This ecoregional effect, which has been at no-madtom sites. After species richness was ad- documented by others (e.g.. Layher and Maughan justed for density by rarefaction, there was no dif- 1985; Rabeni and Sowa 1996; Leftwich et al. ference between Spring River madtom and no- 1997). is an important consideration in understand- madtom sites. The Spring River did have greater ing how Neosho madtom populations are being rarefaction than the Neosho system, as expected affected in the Spring River. The reach of the from descriptions by Cross and Collins (1995). Spring River supporting Neosho madtoms is the The significant positive correlation between Ne- most prairie-like because it is influenced by the osho madtom density and potential competitors as .-North Fork of the Spring River and Cow Creek, a group indicated (hat Neosho madtom densities which are prairie streams (Figure 1). The Ozark increase along with the density of other fishes. If. Uplands Province, part of which is drained by interspecific competition was a primary factor lim- some Spring River tributaries and the upper reach- iting Neosho madtom populations, Neosho mad- es of the main stem, has many spring-fed streams tom densities should have decreased as densities and is composed of limestone that contains large of potential competitors increased. Previous re- quantities of coarse chert and flint, unconsolidated search has supported (Gilliam et al. 1993: Winston chen acting as a water filter (Pflieger 1975). More 1995) and refuted (Angermeier 1982; Grossman than one-third of Missouri fishes have their dis- and Freeman 1987) interspecific competition as a "tribution centered in the Ozark Uplands ('Pflieger determinant of fish community structure. A likely 1975). The spring-fed nature, coarse substrate, scenario is one of alternating interspecific com- clear water, and high species diversity of the Ozark petition (density-dependent factors) and environ- Uplands are the likely reasons why Spring River mental impacts such as flooding or pollution (den- has lower temperature, larger substrate, lower tur- sity-independent factors) as determinants. Inter- bidity, and higher fish species rarefaction, respec- specific competition becomes important when den- tively, than the Neosho system. sity-independent factors are not limiting (Strange The Neosho and Spring River systems differ et al. 1992); currently. Neosho madtoms seem to substantially in soil types and land use (Moss be limited by density independent factors such as 1983), differences that are reflected in variables contaminants and habitat quality. More detailed such as conductivity, hardness, alkalinity. SO4.and studies and analyses of interspecific relationships metals such as Mn. The comparatively high con- between the. Neosho madiom and other species are centrations of dissolved constituents in the Cot- necessary to further define the role of competition tonwood River reflect the rocks and soil in its wa- in regulating Neosho madtom populations. tershed (Hem 1985) and the contribution of the Other factors not measured could affect Neosho Chase-Council Grove aquifer, the waters of which madtom populations. We focused on the benthic are characteristically high in SO4 and other ions aquatic communities of gravel bars where Neosho (Baker and Hansen 1988). Consequently, naturally madtom are found and did not attempt to assess high SO4 concentrations are typical of the Neosho communities in pools or other habitats. This de- River system (Kenny and Snethen 1993). In the cision was based on our primary focus of collect- carbonate-dominated Spring River, some elevation ing fishes with similar environmental preferences of SO4 occurs from the weathering of pyrite (iron and the scarcity of Neosho madtoms in any other sulfide) in the Pehnsylvanian-age shales that over- habitat (Fuselier and Edds 1994). Our focus on lie the western pan of its watershed (Spruill 1987). riffle-dwelling benthic fish specie:; precluded col- and £• lecting any data on fish predators. Predators (i.e., logical Services (ES) Field Office in Manhattan, chiru: black and temperate basses) in all three rivers are Kansas. Assistance with field collections and data Scien similar (Pfiieger 1975; Cross and Collins 1995), processing was provided by A. Bissing, D. Har- Burr. B. r- but we do not know if predator density differs desty. P. Heine, P. Lovely, B. Mueller, S. Olson. the bi between rivers. If it does, it could have influenced B. Poulton, B. Scharge, S. Russler. T. Thorn, R. lllino Natur some of the patterns we have documented. Walton, and D. Whiles of CERC; by M. Legg and Cross. F. 1 D. Munie, contracted through USEPA: by C. Char- 2nd e Conclusions bonneau from USFWS-ES Field Office in Colum- Czarneski When one evaluates limiting factors for rare bia. Missouri; by B. Wilkerson of Oklahoma De- Miss< fishes such as the Neosho madtom, it is important partment of Wildlife Conservation; and by D. of Ei 34:7/ to consider anthropogenic factors as well as phys- Wright of Missouri Southern State College ical habitat, basic water chemistry, and nutrients. Dames an (MSSC). P. J. Lamberson assisted compiling ref- forC Knowledge of either low habitat quality or envi- erences and reviewed the initial draft of this manu- Sprin ronmental contamination alone does not necessar- script. We thank J. Messick of MSSC for providing Den-. it i- I : ily lead to effective management decisions that laboratory space during the study. We gratefully Eisler. R. will stop suspected declines of fish populations. acknowledge the cooperation of the many private vertc Habitat improvement may not improve population landowners in Kansas, Missouri, and Oklahoma life S In i. status or community composition in a stream if the Eisler. R l\ who granted us permission to sample on their prop- of C. tr stream is also heavily contaminated. Conversely, erties. This manuscript was greatly improved by faun; removal of contaminants may also not affect spe- comments from D. F Woodward of the CERC and and II i • cies of concern because physical habitat or basic three anonymous reviewers. Farag. A water quality may be marginal for those popula- Ben- met;! Ui tions or communities. References Our-results suggest thai anthropogenic and nat- food ural factors limit Neosho madtom populations in Aadland. L. P. 1993. Stream habitat types: their fish Toxr Fausch. i- the Spring River. Where metals contamination is assemblages and relationship to flow. North Amer- ican Journal of Fisheries Managemem 13.790-806. iwe? minimal. Neosho madtom densities seem to be lim- irou; ited primarily by physical and chemical habitat Allen. G. T.. and S. H. Blackford. 1995. Contaminants evaluation of Neosho madtom habitats in ihc Ne- stre:> quality and availability. Where contamination has osho River drainage in Kansas. U.S. Fish and Wild- Scie occurred, Neosho madloms seem to be limited pri- life Service Comaminanis, Region 6 Contaminants Finger. 1 marily by the presence of contaminants acting di- Program. Contaminant Repon R6/513M/95, Man- spei rectly (via mortality or avoidance) or indirectly (by hattan. Kansas. Freeman by u suppressing, contaminating, or both the benthic Angcrmeier. P. L. 1982. Resource seasonably and fish dieis in an Illinois stream Environmental Biology spe. invertebrate food base). - 49- Future research into understanding the popula- of Fishes 7:25 1-264. Angermeier, P. L.. and J. R. Karr. 1984. Relationships Fuseliei tion dynamics of the Neosho madtom should in- between woody debris and fish habitat in a small in I: clude a more detailed look at regional and local uarmwater stream. Transactions of the American Ictu rail factors. A regional factor that may be important Fisheries Society .113:716-726. Gill, T. to Neosho madtom populations is the regulation APHA (American Public Health Association). American nep Water Works Association, and Water Pollution En- of water levels through impoundments on the Ne- ius osho and Cotton wood rivers. Local factors include vironment Federation. 1992. Standard methods for more comprehensive investigations of the effects the examination of water and wastewaier. 18ih edi- of microbabitat-scale environmental quality on uon. APHA, Washington. D.C. Bailey R. G. 1995. Description of the ecorepion!- of the Neosho madtom distribution across a gravel bar Uniied States. U.S. Departmem of Agriculture Mis- and an evaluation of fish communities, including cellaneous Publication 1391. predators, found in all habitats associated with Baker. C. H., Jr., and C. V. Hansen. !988. Kansas gravel bars where Neosho madloms are found. ground-water quality. U.S. Geological Survey Wa- ter-Supply Paper 2325:259-264. Acknowledgments Barki. J. H. 1977. Effects of abandoned lead and zinc This study was jointly funded and undertaken mines and tailings piles on water quality in trie Jop- lin area. Missouri. U.S. Geological Surve\ Water by the U.S. Environmental Protection Agency Resources, Investigations 77-75:1-49. Rolla. Mis- (USEPA). Region 7; the U.S. Geological Survey, souri through its Columbia Environmental Research Bryan. M. D.. G. J. Atchison. and M. B Sandheinrich .Center (CERC); and the USFWS, through its Eco- 1995. Effects of cadmium on the foraging behavior INFLUENCES ON NtOi'HO MAD I UM Uli i KIHTJ i ION fc! *

and growth of juvenile bluegill. Lc'pomis r>uur(,- ical characteristics ol' nacural water. 3rd edition. •lanhattan. ^•liirus. Canadian Journal of Fisheries and Aquatic U.S. Geological Survey Water-Supply Paper 2254. -> and data Sciences 52:1630-1638. Hurlben. S. H. 1971. The non-concept of .species di- . D. Har- Burr. B. M.. and R. L. Mayden. 1982. Life history of versity: a critique and alternative parameters. Ecol- 3. Olson, the brindled madiom Noiurus rniurus in Mill Creek. ogy 52:577-586. Thorn, R. Illinois (Pisces: Ictaluridae). American Vlidland James. F. C.. andS. Rathbun. 1981. Rarefaction, relative Naiurabsl 107:25-41. Legg and abundance, and diversity ol avian communities. Cross. F. B.. and J. T. Collins. 1995. Fiihes in Kansas. Auk 98:785-800. . C~Char- 2nd edition. University Press of Kansas. Lawrence. Johansson-Sjobeck. M. L.. and A. Larsson. 1979. Ef- n Colum- C/arneski. J. M. 1985. Accumulation ol'lead in fish from fects of inorganic lead on delta-aminoievulinic acid ,ioma De- Missouri streams impacted by lead mining. Bulletin dehydratase activity and hematological variables in id by D. of Environmental Contamination and Toxicology the rainbow trout. Saltnu gairdnerii. Archives of En- College 34:736-745. vironmental Contamination and Toxicology 8:419- Dames and Moore., 1993. Final remedial investigation •tling ref- 431. for Cherokee County. Kansas CERCLA site. Baxter Kenny. J. F., and D. H. Snethen. 19')3. Kansas stream his manu- Springs/Treece subfiles. Dames and Moore, Inc., water quality. L'.S. Geological. Survey Water-Sup- -jroviding Denver. ply Paper 2400:277-284. gratefully Eisler. R. 1988. Lead hazards to fish, wildlife, and in- Kerans, B. L.. and J. R. Karr. 1994. A benthic index of ny private vertebrates: a synoptic review. L'.S. Fish and Wild- biotic integrity (B-IBO for rivers of the Tennessee Oklahoma life Service Biological Report 85(1.14). valley. Ecological Applications 4:768-785. Eisler. R.. and R. J. Hennekey. 1977. Acute toxieities Kessler. R. K.. and J. H. Thorp. 1993. Microhabitat heir prop- of Cd-'. Ct**. Hg-' and Zn:' to estuarine macro- proved by segregation of the threatened spotted darter (Eth- fauna. Archives of Environmental Contamination eostoma maculalum} and closely related orangefin CERC and and Toxicology 6:315-323. darter (E. helium). Canadian Journal of Fisheries Farag, A. M.. C. J. Boese. D. F. Woodward, and H. L. and Aquatic Sciences 50:1084-1091. Bergman. 1994. Physiological changes and tissue Kiner, L. K.. C. Viiello. and K. Hash. 1997. Spring River metal accumulation in rainbow trout exposed to basin inventory and management plan. Missouri foodborne and waterborne metals. Environmental Department of Conservation. Jefferson City .-,: their tish Toxicology and Chemistry 13 2021-2029. Layhcr. W. G.. and E. Maushan. 1985. Relations be- • orth Amer- Fausch, K. D.. and R. J. White. 1981. Competition be- tween habitat variables and channel catfish popu- 13:790-806. tween brook trout (Salvelinus fontinalis) and brown lations in prairie streams. Transactions of the Amer- ontaminants troul (Saimo trutta) for positions in a Michigan ican Fisheries Society 114:771-781. - in the Ne- stream. Canadian Journal of Fisheries and Aquatic Layher. W. G.. E. Maughan, and W. D. Warde. 1987. Sciences 38:1220-1227. •-h and Wild- Spotted bass habitat suitability related to fish oc- Finger. T. R. 1982. Interactive segregation among three 'omaminums currence and biomass and measurements of physi- species of sculpins (Coitus). Copeia 1982.680-694. Vl/95. Man- ochemical variables. North American Journal of Freeman. B. J.. and M. C. Freeman. 1994. Habitat use Fisheries Management 7:238-251. by an endangered riverine rish and implications for ;|ity and fish Leftwich. K.N.. P. L. Angermeier. and C A. Dolloff. species protection. Ecology of Freshwater Fish 3: •ntal Biology 1997. Factors influencing behavior and iransfera- 49-58. bility of habitat models for a benthic stream fish. Fuselier. L.. and D. R. Edds. 1994. Seasonal variation Relationships Transactions of the American Fisheries Society 126: in habitat use by the Neosho madiom (Teleostei: a! in a small 725-734. he American Ictaluridae: Noiurus plncidus). Southwestern Natu- ralist 39:217-223. Lemaire-Gony, S.. P. Lemaire. and A. L. Pulsford. 1995. Gill. T. S.. J. C. Pant, and H. Tewari: 1989. Cadmium Effect of cadmium and benzoialpyrene on the im- m). American mune sysiem, gill ATPase and EROD activity of Pollution En- nephropathy in a freshwater fish. Puniius conchon- ius Hamilton. Ecotoxicology and Environment) European sea bass Dicunirarchus labrax. Aquatic I methods for" Safety 18:165-172. Toxicology 31:297-313. ier. 18th edi- Cilliam, J. F., D. F. Fraser. and M. Alkins-Koo. 1993. Ludwig, J. A., and J. F. Reynolds. 1988. Statistical ecol- Siructure of a tropical stream fish community: a role ogy: a primer on methods and computing. Wiley, regions of the for biotic interactions. Ecology 74:1856-1870. New York. riculture Mis- Gorman. O. T, and J. R. Karr. 1978. Habitat structure Luttrell. G. R., R. D. Larson. W. .1. Stark. N. A. Ash- and stream tish communities. Ecology 59:507-515. baugh. A. A.'Echelle. and A. V. Zale. 1992. Status Q88. Kansas Grossman. G. D.. and M. C. Freeman. 1987. Micro- and distribution of the Neosho madiom (Nuturus •.1 Survey Wa- habitat use in a stream fish assemblage. Journal of placidus) in Oklahoma. Procedures of the Oklahoma the Zoological Society of London 212:151-176. Academy of Science 72:5-6. lead and zinc Hall. L. W.. Jr.. M. C. Scon, W. D. Killen. Jr.. and R. Maret. T. R.. C. T. Robinson, and G. W. Minshall. 1997. ity in the Jop- D. Anderson. 1996. The effects of land-use char- Fish assemblages and environmental correlates in Survey Water acteristics and acid sensitivity on the ecological sta- least disturbed streams of iht: upper Snake River J. Rolla. Mis- tus of Maryland coastal plain streams. Environ- basin. Transactions of the American Fisheries So- mental Toxicology and Chemistry 15:384-394. ciety 126:200-216. SandheinrTch. Hem. J. D, 1985. Study and interpretation of the chem- Matthews. W. J.. and D. C. Heins. 1987. Community WILDHABER ET AL.

and evolutionary ecology of North American stream SAS Institute. 1990. SAS procedures guide, volume 6. fishes. University of Oklahoma. Norman. 3rd edition. SAS Institute, Cary. North Carolina. 4 Maydcn. R. L.. B. M. Burr, and S. L. Dewey. 1980. Schmitt. C. J., M. L. Wildhaber. J. B. Hunn. T. Nash. USF\ Aspect*, of the life history of the Ozark madtom. M. N. Tieger. and B. L. Steadman. 1993. Biomon- n fJoiurus alhaier.' in southeastern Missouri (Pisces itoring of lead-contaminated Missouri streams with Wagn Ictaluridaej. American Midland Naturalist 104: an assay for eryihrocyte d-aminolevulinic acid de- r 335-340. • hydratase (ALA-Dj in fish blood. Archives of En- 2 McCormick. F. H.. B. H. Hill. L. P. Parrish. and \V. T. vironmental Contamination and Toxicology 25: r Willingham. 1994. Mining impacts on fish assem- 464-475. Webe blages in the Eagle and Arkansas rivers, Colorado. Schmilt. C. J., M. L. Wildhaber. A. Allen, and B. C. i Journal of Freshwater Ecology 9:175-179. Poulton. 1997. The effects of historic zinc-lead i McMahon. T. E.. A. V Zale. and D. J. Orth. 1996. mining and related activities in the tri-states mining Aquatic habitat measurements. Pages 83-120 in B. district on aquatic ecosystems supporting the Ne- R Murphey and D. W. Willis, editors. Fisheries osho madlom. \arurus placidus. in Jasper County. techniques. 2nd edition. American Fisheries Soci- Missouri: Ottawa County. Oklahoma: and Cherokee ety. Bethesda. Maryland. County, Kansas. Final Report to U.S. Environmen- Merrill. R. W.. and K. W. Cummins. 1984. Introduction tal Protection Agency. Region 7. Kansas City. Kan- to the aquatic insects of North America, 2nd edition. sas. !»l|! Kendall/Hunt, Dubuque. Iowa. Scott. M. C., and L. W. Hall. Jr. 1997. Fish assemblages Milliken, G. G.. and D. E. Johnson. 1984. Analysis of as indicators of environmental degradation in Mary- messy data, volume 1: designed experiments. Wads- land coastal plains streams. Transactions of the worth. Belmonl, California. American Fisheries Society 126:349-360. Moss. R. E. 1983. Microhabitat selection in Neosho Simonson. T. D.. and R. J. Neves. 1992. Habitai suit- if River riffles. Doctoral dissertation. University of ability and reproductive traits on the orangefin mad- Kansas. Lawrence. lom Noiurus gilberti (Pisces: Ictalauridae). Ameri- Moyle. P. B.. and B. Vondracek. 1985. Persistence and can Midland Naturalist 127:115-124. structure of the fish assemblage in a small California Smith. B. J. 1988. Assessment of water quality in non- • stream. Ecology 66:1-13. coal mining areas of Missouri. U.S. Geological Sur- Neves. R. J.. and P. L. Angermeier. 1990. Habitat al- vey. Watei Resources Investigations 87-4286. Rol- la. Missouri. teration and its effecis on native fishes in the upper Tennessee River system, east-central U.S.A. Journal Somero. G. N., P. H. Yancey. T. J. Chow, and C. B. Snyder. 1977 Lead effects on tissue and whole or- of Fish Biology 37:45-52. ganism respiration of the estuarine teleosi fish. Cil- Omernik. J. M. 1987. Ecoregions of the conterminous lichthys mirabilis. Archives of Environmental Con- United Stales. Annals of the Association of .Amer- tamination and Toxicology 6:349-354. ican Geographers 77:1 18-125. Sprague, J. B., and B. A. Ramsay. 1965 Lethal levels Pflieger. W. L. 1975. The fishes of Missouri, 2nd edition. of mixed copper-zinc solutions for juvenile salmon Missouri Department of Conservation. Jefferson Journal of the Fisheries Research Board of Canada City. 22:425-432. Phipps, G. L.. V. R. Malison, and G. T. Ankelcy. 1995. Spruill. T. B: 1987. Assessment of waier resources in Relative sensitivity of threeJreshwaier benthic mac- lead-zinc mined areas in Cherokee County. Kansas roinvertebrates to ten contaminants. Archives of En- and adjacent areas. U.S. Geological Survey Water- vironmental Contamination and Toxicology 28: Supply Paper 2268. 281-286. Starnes. L. B.. and W. C. Starnes. 1985. Ecology and Plaits. W. S., W. F. Megahan. and G. W. Minshall. 1983. life history of the mountain madtom. Nmurus Methods for evaluating stream, riparian, and biotic eleutherus (Pisces Ictaluridae). American Midland conditions. U.S. Forest Service General Technical Naturalist I 14:331-341. Report INT-138. Strange. E. M., P. B. Moylc. and T. C. Foin. 1992. In- Raheni, C. F. and S. P. Sowa. 1996. Integrating bio- teractions between stochastic and deterministic pro- logical realism into,habitat restoration and conser- cesses in stream fisfi community assembly Envi- vation strategies for small streams. Canadian Jour- ronmental Biology of Fishes 36:1-15. nal of Fisheries and Aquatic Sciences 53:252-259. Taylor. W. R. 1969. A revision of the catfish genus Ni.- Rhodes. L.. and six coauthors. 1985. Interactive, effecis lurus Rafinesque. with an analysis of higher ver- of cadmium polychlorinated biphenyls. and fuel oil tebrate groups in the Iclaluridae. U.S. National Mu- on experimentally exposed English sole (Paruphrys seum Bulletin 282. veiulus}. Canadian Journal of Fisheries and Aquatic Thomas. P.. and M. J. Juedes. 1992. Influence of lead Sciences 42.1 870-1880. on the gluiaihione status of Atlantic croaker tissues Roark. S., and K. Brown. 1996. Effecis of metal con- Aquatic Toxicology 23:11-30. tamination from mine tailings on allozyme distri- USEPA (U.S. Environmental Prelection AgencyV 1986. butions of populations of greal plains fishes. En- Ambient waier quality criteria. USEPA. 440/ vironmental Toxicology and Chemistry 15:921- 5-86-001. Washington. D.C. 927. USEPA (U.S. Environmental Protection Agency). 1987. O!V HEOSHO MAUTOM DI5TKIBL I \ -Mi-

Ambient water quality criteria lur. iinc. USEPA. Elemental joiic---ncr'j!.:ons in benthic invertebrates 4405-37-003. Washington. D.C. collected from the Neosho. Coltonwood. and Spring USFWS (U.S. Fish and Wildlife Service!. 1991. Neosho rivers. Final Report io U.S. Fish and Wildlife Ser- madlom recovery plan. USFWS, Denver vice. Region 6. Manhattan. Kansas. Wagner. G. F. and B. A. McKeown. 1982. Changes in Wilkinson. C.. D. Edds. J. Dorlac. M. L. Wildhaber. C plasma insulin and carbohydrate metabolism of J. Schmitt. and A. Allen. 1996. Neosho madtom iinc-stressed rainbow irout. Sulmn \fairdnt;ri. Ca- distribution and abundance in the Spring River. nudian Journal of Zoology 60:2079-2084. Southwestern Naturalist 41:78-81. Weber. D. N. 1993. Exposure to sublethal levels of wa- Winston. M. R. 1995 Co-occurrence of morphologi- lerbornc lead alters reproductive behavior patterns cally similar species of stream fishes. American Nat- in fathead minnows ' Pime ptiales pi'tnneius). uralist 145:527-545. NeuroToxicology 14:347-358. Wood. B M.. and M. B. Bam. 1995. Morphology and Welsh. B. L. 1951. On companion of several mean val- microhabilat use in stream fish. Canadian Journal ues: an alternate approach. Biomeirica 38:330-336. of Fisheries and Aquatic Sciences 52:1487-1489. Wetzel. R. B. 1983. Limnology. Saunders. Philadelphia. Woodward. D. F. W. G. Brumbaugh. A J. DeLonay. E. E. Little, and C. E. Smith. 1994. Effects on rainbow Wildhaber. M. L.. and C. J Schmitt. 1996. Estimating trout fry of a mevals-comaminaied diet of benthic aquatic toxicitv as determined through laboratory invertebrates from ihe Clark Fork River. Montana. tests of Great Lakes sediments containing complex Transactions of the American Fisheries Society 123: mixtures of environmental contaminants. Environ- . 51-62. mental Monitoring and Assessment 4 1:255-289. Woodward. D. F. J N. Goldstein. A. M. Farag, and W. Wildhaber. M. L.. C. J. Schmitt. and A. L. Allen. 1996. G. Brumbaugh. 1997. Cutthroat trout avoidance of Historic zinc-lead mining and related activities and metals conditions characteristic of a mining water .their effects on aquatic ecosystems supporting the site: Coeur d'Alene River. Idaho. Transactions of Neosno madlom. Notums placidus. in Jasper Coun- the American Fisheries Society 126:699-706. : ty. Missouri: Ottawa County. Oklahoma: and Cher- Woodward. D. F. J. A. Hansen. H. L. Bergman. E. E. okee County. Kansas. Appendix A: tables A1-A35 Little, and A. J Delonay. 1995. Brown trout avoid- Draft Final Report to U.S. Environmental Protection ance of metals waier characteristic of the Clark Fork Agency. Region 7. Kansas City. Missouri River. Montana. Canadian Journal of Fisheries and Wildhaber. M. L., C. J. Schmitt. and A. L. Allen. 1997. Aquatic Sciences 52:2031-2037.