nena masthead
NENA Home Staff & Editors For Readers For Authors

Influence of Different Habitat Factors on Creek Chub (Semotilus atromaculatus) within Channelized Agricultural Headwater Streams
Peter C. Smiley Jr., Kevin W. King, and Norman R. Fausey

Northeastern Naturalist,Volume 24, Special Issue 8 (2017): 18–44

Full-text pdf (Accessible only to subscribers.To subscribe click here.)

 



Access Journal Content

Open access browsing of table of contents and abstract pages. Full text pdfs available for download for subscribers.

Issue-in-Progress: Vol. 31 (1) ... early view

Current Issue: Vol. 30(4)
NENA 30(2)

Check out NENA's latest Monograph:

Monograph 23
NENA monograph 23

All Regular Issues

Monographs

Special Issues

 

submit

 

subscribe

 

JSTOR logoClarivate logoWeb of science logoBioOne logo EbscoHOST logoProQuest logo


Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 18 Vol. 24, Special Issue 8 Influence of Different Habitat Factors on Creek Chub (Semotilus atromaculatus) within Channelized Agricultural Headwater Streams Peter C. Smiley Jr.1,*, Kevin W. King1, and Norman R. Fausey1 Abstract - Natural history information on habitat relationships of Semotilus atromaculatus (Creek Chub) in channelized agricultural headwater streams in the northeastern region of North America is limited. We hypothesized that Creek Chubs within channelized agricultural headwater streams would be more strongly influenced by instream habitat than other physical and chemical habitat variables. We sampled Creek Chubs and measured watershed characteristics, riparian habitat characteristics, geomorphology, instream habitat characteristics, and water chemistry in 14 channelized agricultural headwater streams in central Ohio from 2006 to 2011. We found that the abundance, mean length, and biomass of Creek Chub were most strongly influenced by watershed characteristics (land use, soil type) and geomorphology (channel shape, channel size). Our results indicate that conservation and restoration practices designed to mitigate physical habitat degradation are most likely to benefit Creek Chub within channelized agricultural headwater str eams. Introduction Semotilus atromaculatus (Mitchill) (Creek Chub) is a North American fish species that is one of the most widespread and commonly occurring stream fishes in the eastern United States and southeastern Canada (Barber and Minckley 1971, McMahon 1982, Page and Burr 1991). The native distribution of this species encompasses most of eastern North America and extends westward from the Atlantic Coast to Wyoming and Montana in the United States and to Manitoba in Canada (Page and Burr 1991). Creek Chubs are found in a wide variety of stream types and sizes (Barber and Minckley 1971, Fitzgerald et al. 1999), but they exhibit a preference for small warmwater headwater streams (Copes 1978, Mahon et al. 1979, Shelford 1913, Starrett 1950, Thompson and Hunt 1930). Small Creek Chubs are primarily insectivorous, but larger individuals within small headwater streams are often the primary predator feeding on fishes, amphibians, and aquatic insects (Barber and Minckley 1971, Copes 1978, Quist et al. 2005, Storck and Momot 1989). Creek Chubs are classified by regulatory agencies in the United States as fish tolerant of natural and anthropogenic stressors (Barbour et al. 1999) because of their ability to persist within and adapt to variable hydrologic and chemical conditions (Blevins et al. 2013, Larimore et al. 1959, Nagrodski et al. 2013, Shelford 1913, Walker and Adams 2016). Yet, these fishes are often the most abundant fish species found in small, forested headwater streams (Heithaus and Grame 1997, Lotrich 1973). 1USDA Agricultural Research Service, Soil Drainage Research Unit, 590 Woody Hayes Drive, Columbus, OH 43210. *Corresponding author - rocky.smiley@ars.usda.gov. Manuscript Editor: Jay Stauffer Natural History of Agricultural Landscapes 2017 Northeastern Naturalist 24(Special Issue 8):18–44 Northeastern Naturalist 19 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 Creek Chubs also exhibit ecological and physiological responses to improvements in habitat quality and reduced concentrations of chemical pollutants (Dube et al. 2006, Fischer et al. 2010, Henshel et al. 2006, Lyons 2006, Portt et al. 1986, Stair et al. 1984, Thompson and Hunt 1930). Creek Chubs and other fishes are commonly found within channelized agricultural headwater streams (Table 1; Jordan et al. 2013, Smiley and Gillespie 2010). These small streams are common within agricultural watersheds throughout the northeastern portion of North America (i.e., the region extending west from Virginia to Missouri, north to Manitoba, and then back east to the Atlantic coast; Madramootoo et al. 2007, Needelman et al. 2007, Smiley and Gillespie 2010, Stammler et al. 2008). Channelized agricultural headwater streams (e.g., agricultural drainage ditches or open agricultural drains) are first-, second-, and third-order streams that have been modified or created for draining excess water from adjacent agricultural fields. Subsequently, these streams exhibit physical and chemical habitat degradation that includes loss of riparian habitat and vegetation, loss of instream habitat, altered hydrology, increased nutrient loadings, and pesticide contamination as a result of channelization, use of subsurface tile drains, and current nutrient and pesticide application practices. Developing conservation and restoration strategies for fishes within channelized agricultural headwater streams requires understanding the natural history of fishes within this stream type. Particularly, understanding the fish–habitat relationships within channelized agricultural headwater streams will provide predictions on what type of conservation and restoration practices will be most effective. Most agricultural conservation efforts focus on improving water quality by reducing nutrient, pesticide, and sediment inputs into agricultural watersheds. The current focus on water quality improvement might be an effective management approach if community and population structure of fishes are more strongly linked with water chemistry than physical habitat conditions. If community and population structure of fishes within these small streams are more strongly correlated with physical habitat conditions than water chemistry, then conservation efforts focusing on water chemistry improvements are less likely to benefit these fish compared to efforts targeting physical habitat improvements. Past research efforts within agricultural streams have focused on evaluating the impacts of channelization by comparing the differences in fishes and habitat between channelized and unchannelized streams and/or evaluating fish–habitat relationships across a continuum of channel modifications extending from recently channelized to recovering channelized streams to unchannelized streams (Smiley and Gillespie 2010). As a result of this emphasis on documenting channelization impacts, basic natural history information on fish–habitat relationships within channelized agricultural headwater streams is limited. Our previous research on fish–habitat relationships within channelized agricultural headwater streams in northeastern Indiana and central Ohio compared the degree of influence that riparian habitat, instream habitat, and water chemistry have on fish community structure. The initial results (Sanders 2012; Smiley et al. 2008, 2009) with the first 5 years Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 20 Vol. 24, Special Issue 8 of data from channelized agricultural headwater streams documented that fish community structure was most strongly influenced by instream habitat variables and only exhibited weak relationships with water chemistry (nutrients, pesticides, physicochemical variables) and riparian habitat variables. Most previous research in the United States and Canada has documented that the abundance (number of individuals), density (number of individuals per m2), or presence/absence of Creek Chubs is influenced by instream habitat (i.e., hydrology, substrate types, cover types) and watershed habitat variables (i.e., watershed size, land-use types) within unchannelized headwater streams (Table 1). Only a limited amount of information is available on Creek Chub–habitat relationships within channelized agricultural headwater streams and on the relationships of Creek Chub with riparian habitat, geomorphology, and water chemistry (Table 1). Additionally, only a few studies (Hubert and Rahel 1989, Magoulik 2000, Quist and Guy 2001, Wuellner et al. 2013) have compared the relative influence of different types of habitat variables on the population structure of Creek Chub. In this manuscript, we expand upon previous research findings (Sanders 2012; Smiley et al. 2008, 2009) by evaluating if the fish–habitat relationships we observed at the community level also occur at the population level within these small streams. We selected Creek Chub for this population-level assessment because they were the most abundant fish species captured within our study sites in Indiana and Ohio (Sanders 2012; Smiley et al. 2008, 2009). We also seek to increase the current understanding of Creek Chub–habitat relationships by comparing the relative influence of different habitat variables on the population structure of Creek Chub within channelized agricultural headwater streams. We sampled Creek Chub and measured watershed characteristics, riparian habitat, geomorphology, instream habitat, and water chemistry from 14 channelized agricultural headwater stream sites in central Ohio over a 6-year period to evaluate the relative influence of different habitat factors on the population structure of this common headwater fish. Specifically, our research hypothesis was that instream habitat will have the greatest influence on the population structure of Creek Chub in channelized agricultural headwater streams. Field-Site Description Upper Big Walnut Creek (UBWC) watershed is located in central Ohio (Fig. 1) and is part of the Scioto River watershed, which is one of the most biologically diverse watersheds in Ohio (Sanders 2001). This watershed is located in the humid continental climatic region of the United States. Daily temperatures range from an average minimum of -9.6 °C in January to an average maximum of 33.9 °C in July (King et al. 2008). Mean annual total precipitation within the watershed during our study varied from 903 mm to 1334 mm (P.C. Smiley Jr., USDA Agricultural Research Service, Columbus, Ohio, USA, unpubl. data). Thunderstorms during the spring and summer produce short-duration intense rainfalls. Moisture from December to March occurs primarily in the form of frozen precipitation or snow. Cropland consisting of corn and soybean is the dominant land use in the UBWC watershed Northeastern Naturalist 21 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 Table 1. Associations of Creek Chub with different types of habitat variables and individual habitat variables within channelized and unchannelized headwater streams in the United States and Canada. Abbreviations for types of habitat variables are: wate = watershed, riph = riparian habitat, geom = geomorphology, insh = instream habitat, watc = water chemistry. Individual habitat variable abbreviations are: wsize = watershed size, lu = land use, precip = precipitation, elev = elevation, ripht = riparian habitat type, ccov = canopy cover, wva = riparian woody vegetation amount, wd = water depth, mh = microhabitat type or microhabitat availability, ww = wet width, wv = water velocity, st = substrate type, pa = pool availability, pc = pool complexity, sa = surface area, disc = discharge, psize = pool size, ct = cover type, nutr = nutrients, herb = herbicides, wtemp = water temperature, cond = conductivity, do = dissolved oxygen, TSS = total suspended solids, pcb = polycholorinated biphenyl concentration. [Table continued on next page.] Type habitat Stream type/location Population variable variable Individual habitat variables Source Channelized headwater streams Iowa Length, age, growth riph, insh ripht, ccov, ct Fischer et al. 2010 Michigan Movement geom, insh culvert size, culvert type Briggs and Galarowicz 2013 Ohio Abundance riph wva Smiley et al. 2008 Ohio Abundance watc nutr, herb, wtemp, cond, pH Smiley et al. 2009 Channelized and unchannelized headwater streams Ontario Biomass, production insh wd Portt et al. 1986 Illinois Abundance insh mh Schlosser 1982 Indiana Growth, morphology watc cattle manure Leet et al. 2012 Iowa Abundance, biomass, length, presence insh ww, wv, st, ct Scarnecchia 1988 Ohio Abundance insh mh Trautman and Gartman 1974 Ontario Abundance wate lu Stammler et al. 2008 Unchannelized headwater streams Arkansas Density, age class insh ww, st Magoulick 2000 Arkansas Biomass insh ct Mitchell et al. 2012 Arkansas Movement insh pa, pc Walker and Adams 2016 Arkansas Density insh wd, pa Dekar and Magoulik 2007 Arkansas Survival, movement, abundance insh mh Hodges and Magoulick 2011 Illinois Abundance insh mh Schwartz and Herricks 2008 Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 22 Vol. 24, Special Issue 8 Table 1, continued. Type habitat Stream type/location Population variable variable Individual habitat variables Source Illinois Physiology wate, watc lu, wtemp, do Blevins et al. 2013 Kansas Growth, age classes insh st, ct Quist and Guy 2001 Kentucky Abundance, biomass wate wsize Lotrich 1973 Maryland Presence wate, insh, watc wsize, lu, elev, mh, st, pH Pinder and Morgan 1995 Minnesota Density wate, insh precip, wd, wv, ct Schlosser 1995 Minnesota Density insh wd, wv, ct Schlosser 1998 North Carolina Presence wate wsize Lemly 1985 Ohio Abundance, biomass wate, insh wsize, sa, mh Storck and Momot 1989 Ohio Abundance, age classes wate, insh wsize, wd, ct Storck and Momot 1981 Ohio Presence watc sewage Katz and Gaufin 1953 Ohio Size, growth wate, watc wsize, sewage Katz and Howard 1955 Pennsylvania Presence insh, watc disc, TSS, wtemp Wohl and Carline 1996 Pennsylvania Abundance wate, insh wsize, lu, psize Butler and Fairchild 2005 Penn. & New Jersey Presence riph ripht Ross et al. 2003 South Carolina Density, age classes insh ww, wd, wv Meffe and Sheldon 1988 Tennessee Abundance, age classes, size, growth insh disc Stair et al. 1984 Headwater stream – undocumented channelization status Indiana Size, age classes, growth, survivorship watc pcb Henshel et al. 2006 Ohio Abundance insh ct Gatz 2008 Northeastern Naturalist 23 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 Figure 1. Map depicting the locations of sampling sites (gray circles with black center dots) within the Upper Big Walnut Creek watershed, OH. The symbols have been sized to ensure their clarity and do not reflect actual site sizes or distances between sites. The inset map shows the location of the watershed within Ohio. Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 24 Vol. 24, Special Issue 8 (Smiley et al. 2014). The majority of headwater streams in the UBWC watershed are impaired by nutrient enrichment, pathogens, and habitat degradation stemming from current agricultural management practices (Ohio EPA 2005). We selected 14 sites in 7 channelized agricultural headwater streams in UBWC as our study sites (Fig. 1). Sampling sites located within the same stream were separated by a mean distance of 0.6 km (varying from 0.2 to 1.9 km). Each site was a 125-m-long reach near locations where we collected weekly grab samples of water for nutrient and pesticide measurements. Our established site lengths were on average 67 times the mean wet width (varying from 33 to 151 times the mean wet width) and frequently exceed the minimum site lengths recommended for collecting fishes via electrofishing within streams in the midwestern United States (i.e., 35 times mean wet width; Lyons 1992). Within each site, we established 6 permanent transects spaced 25 m apart in each site for sampling riparian habitat, geomorphology, and instream habitat. All sites possessed riparian habitats consisting mostly of herbaceous riparian vegetation and exhibited the straightened, over-enlarged, trapezoidal channel shape typical of channelized agricultural headwater streams in the region (Table 2; Smiley et al. 2011). We collected fishes and measured riparian habitat, geomorphology, instream habitat, and water chemistry from each site from May 2006 to November 2011. We also compiled information on watershed characteristics from each site. Methods Habitat sampling We obtained watershed characteristics by conducting geographic information system analyses with digital elevation models, orthophotos, and USDA Natural Resources Conservation Service (NRCS) soil survey results. Watershed size for each site was calculated as the total area encompassed by the watershed boundaries identified with digital elevation models. We determined the percent of 3 primary land-use types (i.e., agricultural, residential, forest/shrub) within each watershed from 2010 orthophotos generated by the Delaware County Auditor and based on 0.31-m resolution. We calculated the percent of 4 commonly occurring soil types (Bennington, Centerburg, Pewamo, Amanda) within the watersheds of our sites from the NRCS Soil Survey Geographic database. We established 12 quadrats (1 m x 10 m) in each site to characterize riparian habitat once annually in the fall (September to November). Two quadrats were located along each of the 6 permanent transects, with 1 quadrat placed on each streambank. The quadrats began at the waters edge and extended into the adjacent riparian habitat and sometimes into the agricultural fields. We identified and enumerated woody vegetation >1 m tall within each quadrat. The presence and absence of herbaceous vegetation within 3 height stratas (0–0.5 m, 0.5–2 m, 2–5 m) and the presence and absence of woody vegetation within 4 height stratas (0–0.5 m, 0.5–2 m, 2–5 m, and >5 m) were also noted in each quadrat. We used a spherical densionometer to obtain 3 measurements of riparian canopy cover at each transect (i.e., in the middle of the left quadrat, in the middle of the stream, and in the middle Northeastern Naturalist 25 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 of the right quadrat). We measured riparian widths with an electronic total station or a Real Time Kinematic global positioning system. Riparian widths were determined by calculating the straight-line distance between coordinate measurements obtained at the water’s edge and the edge of the agricultural fields. We calculated 15 riparian habitat variables from each site for each year. Herbaceous and woody structural richness is the number of height strata within a site containing herbaceous or woody vegetation. Percent frequency of herbaceous and woody vegetation in each height strata was calculated by dividing the occurrence of herbaceous and woody vegetation in each quadrat within each height strata by the total possible occurrence (12) of herbaceous and woody vegetation within all 12 quadrats in a site. The sum percent frequency of herbaceous and woody vegetation was the sum of the percent Table 2. Mean, minimum and maximum values of seasonal means of selected watershed, riparian, geomorphology, instream habitat, and water chemistry variables from channelized agricultural headwater streams in the Upper Big Walnut Creek watershed, OH, 2006 to 2011. Mean Minimum Maximum Watershed Watershed size (km2) 3.57 0.60 9.67 Percent agriculture 71.25 45.27 95.12 Percent bennington soil type 45.13 26.43 56.34 Percent centerburg soil type 18.19 0.93 56.15 Percent pewamo soil type 32.25 0.00 47.45 Riparian habitat Riparian width (m) 25.06 4.74 80.26 Woody vegetation density (#/m2) 0.15 0.00 0.80 Percent canopy cover 6.92 0.00 55.10 Sum frequency of woody vegetation 91.70 0.00 308.33 Sum frequency of herbaceous vegetation 214.69 133.33 283.33 Geomorphology Cross-section area (m2) 8.02 3.63 16.35 Top bank width (m) 8.54 5.95 12.09 Thalweg depth (m) 1.76 1.08 2.50 Sinuosity 1.03 0.99 1.29 Gradient (m/125 m) -0.19 0.01 -0.51 Instream habitat Water depth (m) 0.13 0.01 0.50 Water velocity (m/s) 0.02 -0.05 0.19 Wet width (m) 1.91 0.43 4.83 Percent clay 32.0 0.0 91.7 Percent gravel 22.0 0.0 71.0 Percent instream wood 1.0 0.0 13.0 Water chemistry Total nitrogen (mg/L) 4.66 0.36 22.88 Total phosphorus (mg/L) 0.13 0.00 0.75 Atrazine (μg/L) 2.64 0.00 85.26 Metolachlor (μg/L ) 0.73 0.00 15.65 Water temperature (°C) 17.79 3.60 36.21 Dissolved oxygen (mg/L) 87.05 9.70 315.20 Turbidity (NTU) 97.87 0.00 2816.00 Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 26 Vol. 24, Special Issue 8 frequency of herbaceous and woody vegetation in each height strata. The woody vegetation importance value was calculated as the sum of the relative frequency of woody vegetation (i.e., sum frequency of woody vegetation divided by the sum frequency of woody and herbaceous vegetation) and the mean woody vegetation density for each site. We also calculated mean percent canopy cover and mean riparian width. We conducted geomorphology surveys within each site in the fall of each year. Coordinate measurements for calculation of geomorphology variables were collected from a minimum of 9 points along each transect using either a Real Time Kinematic global positioning system or an electronic total station. We calculated 7 geomorphology variables from each site for each year. Specifically, we calculated the mean cross-section area, mean channel depth, mean thalweg depth, mean topbank width (i.e., the width of the channel at the bankfull elevation that is just below the elevation where the water would begin spilling over into the floodplains and agricultural fields), top-bank width to thalweg depth ratio, gra dient, and sinuosity. We took 1 measurement of wet width and 4 measurements of water velocity and depth along the 6 permanent transects within each site in the spring (May to June), summer (July to August), and fall of each year concurrently with fish sampling. One additional transect was established for the calculation of instantaneous discharge, and we made 10 equidistant measurements of water depth and velocity along this transect. We measured water depths with a top-setting wading rod, water velocity with an electromagnetic velocity meter, and wet widths with a tape measure. The dominant substrate type and instream cover types were visually identified at each point. For each site during each season, we calculated mean and SD of water depth, velocity, and wet width, as well as instantaneous discharge, the percent of each substrate type, the percent of instream cover types, and the number of substrate and cover types found. In total, we calculated 17 instream habitat variables from our seasonal instream habitat measurements. Our habitat assessment also included 18 water chemistry variables. We collected weekly grab samples of water from each site from April to December of each year for the measurement of nutrients, herbicides, and fungicides. Concentrations of nitrate+nitrite, ammonia, and dissolved reactive phosphorus were determined colorimetrically. We measured nitrate+nitrite and ammonia by application of the copperized-cadmium method and dissolved reactive phosphorus by the ascorbic reduction method (Parsons et al. 1984). We performed total nitrogen and total phosphorus analyses on unfiltered samples following alkaline persulfate oxidation (Koroleff 1983) with subsequent determination of nitrate+nitrite and dissolved reactive phosphorus. We used heated persulfate oxidation with a total organic carbon analyzer for measurements of dissolved organic carbon concentrations (Menzel and Vaccaro 1964). We determined concentrations of 4 herbicides (alachlor, atrazine, metolachlor, simazine), 1 herbicide metabolite (atrazine desethyl), and 2 fungicides (chlorothalonil, metalaxyl) using gas chromatography following standards protocols for pesticide analysis (US EPA 1995, Zaugg et al. 1995). We selected these 7 pesticides for measurement because of their known occurrence within agricultural Northeastern Naturalist 27 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 streams in the region (Battaglin et al. 2011, Gilliom 2007, Smiley et al. 2014) or our anticipation that they would be emerging contaminants as a result of known pest management issues within the UBWC watershed. We calculated mean nutrient and pesticide concentrations for each site from data collected during a 3-week period beginning 1 week before and ending 1 week after the week of fish sampling. Using a multiparameter meter, we obtained in situ measurements of dissolved oxygen, water temperature, pH, and specific conductivity from each site 3 times a year concurrently with fish sampling. Grab samples for turbidity were collected in conjunction with in situ physico-chemical measurements and measured in the lab with a turbidity meter. Fish sampling and population assessments At each site, we sampled fishes 3 times a year, once each in the spring, summer, and fall. Block nets were set at the upstream and downstream borders of the sites prior to fish sampling. We collected fishes with a backpack electrofisher (100 V, 60 Hz, DC current). Electrofishing began at the downstream border of a site and proceeded upstream. Care was taken to ensure that all microhabitat types within each site were sampled thoroughly during electrofishing. After electrofishing was completed, we also collected 5 samples with a seine (2 m x 4 m, 0.32-cm mesh) that were equally distributed throughout each site. Pools and slow-flowing areas were sampled with a seine haul, and fast-flowing riffle and run areas were sampled by kick-seining. We identified, measured, counted, and returned to the stream all Creek Chubs caught. Small fishes that could not be reliably identified in the field were anesthetized in MS-222, fixed in a 10% formalin solution, and taken to the lab for identification. We calculated the abundance (number of captures), mean length, and biomass of Creek Chub for each site during each seasonal collection. We determined the weight of each fish using published length–weight relationships of Creek Chub from an Illinois stream (Carlander 1969). Biomass of Creek Chubs at each site from each season was calculated as the product of the mean weight and abundance. Statistical analyses We conducted principal components analysis (PCA) separately for the 8 watershed variables, 15 riparian habitat variables, 7 geomorphology variables, 17 instream habitat variables, and 18 water chemistry variables to obtain the site scores from the first 2 PCA axes of each habitat factor that would be used as independent variables in our linear mixed-effect model analyses. Each PCA axis represents the underlying gradient in combinations of habitat variables that occurs among our study sites. Our use of PCA enabled us to objectively reduce the number of independent variables within our statistical analyses from 65 to 10. Variable-reduction methods to reduce multicollinearity prior to multiple regression analyses are necessary to avoid spurious results (Dormann et al. 2013). Principal components analyses with correlation cross-product matrixes were conducted with PC-ORD 4 for Windows (McCune and Mefford 1999). Examining the pairwise correlation coefficients among all 10 PCA axes, we found that the |r| values (absolute values of the correlation coefficients) were below Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 28 Vol. 24, Special Issue 8 the level that indicates the presence of strong multicollinearity (i.e., |r| = 0.7) among independent variables (Table 3; Dormann et al. 2013). We observed that 4 pairs of correlations (i.e., watershed PCA axis 1 and geomorphology PCA axis 1, watershed PCA axis 2 and riparian habitat PCA axis 2, watershed PCA axis 2 and instream habitat PCA axis 2, riparian habitat PCA axis 2 and geomorphology PCA axis 2) exhibited moderate levels of multicollinearity as indicated by |r| values between 0.52 and 0.62 (Table 3). We recognized that even moderate levels of multicollinearity may potentially influence the conclusions from the statistical analyses we planned to conduct as part of our mechanistic comparison of the influence of all habitat factors on the abundance, mean length, and biomass of Creek Chubs. To address this potential multicollinearity issue, we conducted multimodel inference analysis (Burnham and Anderson 2002) that involved determining the overall importance value of each independent variable based on 8 linear mixed-effect models with and without moderate levels of multicollinearity. The first step in our multimodel inference analysis was the development of 8 linear mixed-effect models that included at least 4 of 5 habitat factors (i.e., watershed characteristics, riparian habitat, geomorphology, instream habitat, water chemistry) in each model (Table 4). Model 1 is our global model that contains all 10 PCA axes. Models 2 to 5 represent 4 reduced versions of the global model (Model 1) that omit selected PCA axes to remove the moderate levels of multicollinearity present among 4 pairs of our independent variables (Table 2). Model 2 excludes riparian habitat PCA axis 2, geomorphology PCA axis 1, and instream habitat PCA axis 2. Model 3 excludes watershed PCA axis 1, riparian PCA axis 2, and instream habitat PCA axis 2. Model 4 excludes watershed PCA axis 2, geomorphology PCA axis 1, and geomorphology PCA axis 2. Model 5 excludes watershed PCA axis 1, watershed PCA axis 2, and geomorphology PCA axis 2. Model 6 is a model containing 7 PCA axes (watershed PCA axis 2, riparian habitat PCA axis 1, geomorphology PCA Table 3. Pearson correlation coefficients from pairwise correlations among the 10 principal components analysis (PCA) axes of watershed characteristics (WS), riparian habitat (RH), geomorphology (GE), instream habitat (IH), and water chemistry (WC) within channelized agricultural headwater streams in the Upper Big Walnut Creek, OH, 2006 to 2011. * indicate correlation coefficients are those pairs of principal components axes that exhibited a significant correlation (P < 0.05). WS PCA RH PCA GE PCA IH PCA WC PCA Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 WS PCA axis 1 - - - - - - - - - - WS PCA axis 2 0.00 - - - - - - - - - RH PCA axis 1 0.08 -0.11 - - - - - - - - RH PCA axis 2 -0.04 -0.54* 0.00 - - - - - - - GE PCA axis 1 -0.62* 0.10 0.05 -0.01 - - - - - - GE PCA axis 2 -0.16 0.14* -0.04 -0.56* 0.00 - - - - - IH PCA axis 1 -0.30* 0.15* 0.09 0.12 0.46* -0.31* - - - - IH PCA axis 2 -0.21* -0.52* -0.17* 0.10 0.08 0.29 0.00 - - - WC PCA axis 1 0.27* -0.03 0.25* -0.03 -0.05 -0.08 0.13* -0.09 - - WC PCA axis 2 0.12 -0.22 0.00 0.05 -0.02 -0.05 -0.39* -0.11 0.00 - Northeastern Naturalist 29 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 axis 1, instream habitat PCA axis 1, instream habitat PCA axis 2, water chemistry PCA axis 1, water chemistry PCA axis 2) (Table 4) that contain habitat variables that have been previously documented to influence the abundance, length, and biomass of Creek Chub within headwater streams (Table 1). Models 7 and 8 are reduced versions of Model 6 that exclude either geomorphology PCA axis 2 and instream habitat PCA axis 2 (Model 7) or watershed PCA axis 2 and geomorphology PCA axis 1 (Model 8) to remove moderate levels of multicollinearity (Table 4). The second step of the multimodel inference analysis consisted of conducting linear mixed-effect model analysis for each model. Specifically, for each model, we conducted linear mixed-effect model analysis with the 3 response variables as our dependent variables, the PCA axes as independent variables (fixed effects), and site as a random effect to address the issue of pseudoreplication that results from repeatedly sampling the same sites through time. Linear mixed-effect model analysis is an extension of the traditional linear regression analysis that incorporates fixed effects and random effects into the regression formula. We conducted linear mixed-effect model analyses with R (R Core Team 2014) using the lme function within nlme package (Pinheiro et al. 2014). Inspection of the residuals of the initial mixed-effect model analyses of the global model of abundance and biomass indicated the residuals were not normally distributed; therefore these 2 dependent variables were log (x + 1) transformed. We also accounted for the effect of heteroscedasticity in the variances among sites (abundance, mean length) or months (biomass) using the weights option (i.e., weights=varIdent(form=~1|site) or weights=varIdent(form=~1|months) within the lme function. The third step of the multimodel inference analysis involved obtaining the small sample Akaike information criterion (AICc ) score, ΔAICc (the difference in AICc between each model and the model with the minimum AICc), and Akaike weight (Wi) for each of the 8 models (Burnham and Anderson 2002, Johnson and Omland 2004). We based our multimodel inference analysis on the AICc because the ratio of n/K was less than 40 for all models (Burnham and Anderson 2002, Johnson and Omland Table 4. Summary of therincipal components analysis (PCA) axes of watershed characteristics (WS), riparian habitat (RH), geomorphology (GE), instream habitat (IH), and water chemistry (WC) that are present within the 8 models developed for multimodel inference analysis. A value of 1 indicates the variable is included in the model, and a value of 0 indicates the variable is not included in the model. WS PCA RH PCA GE PCA IH PCA WC PCA Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Global 1 1 1 1 1 1 1 1 1 1 model Model 2 1 1 1 0 0 1 1 0 1 1 Model 3 0 1 1 0 1 1 1 0 1 1 Model 4 1 0 1 1 0 0 1 1 1 1 Model 5 0 0 1 1 1 0 1 1 1 1 Model 6 0 1 1 0 1 0 1 1 1 1 Model 7 0 1 1 0 1 0 1 0 1 1 Model 8 0 0 1 0 0 1 1 1 1 1 Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 30 Vol. 24, Special Issue 8 2004). We also calculated the relative importance of each independent variable as the mean Wi over all models in which the independent variable occurs (Burnham and Anderson 2002, Johnson and Omland 2004, Kittle et al. 2008). Our importance value reports the mean Wi instead of the summed Wi because all independent variables did not occur in the same number of models (Kittle et al. 2008). AICc values were obtained with the AICc function within the AICcmodavg package (Mazerolle 2016) in R. We also reported the standardized regression coefficients from the best performing model (i.e., the model with the minimum AICc) for each response variable as an additional way to evaluate the relative importance of individual habitat variables in determining the population structure of Creek Chub and to identify the specific relationships that occurred between the abundance, mean length, and biomass of Creek Chub with the independent variables. Results We captured 10,401 Creek Chubs from our 14 study sites over a 6-year period. Creek Chub were captured at least once from every site during our study. The mean number of fish captured from each site per season was 45 and ranged from 0 to 809 fishes. Length of Creek Chub from each site per season averaged 8.4 cm and ranged from 1.6 to 18.6 cm. The largest fish captured was 22.0 cm in length. Biomass of Creek Chub from each site per season averaged 288 g and ranged from 0 to 2043 g. Our sampling occurred under a wide range of hydrological and chemical conditions that were expected to occur within these small streams (Table 2). Hydrologic conditions were quite variable, with some sites consisting of small fragmented pools with no water flow during the summer to some sites with elevated water depths and water velocities that follow precipitation events in the spring and late fall (Table 2). Clay and gravel were the most frequently occurring substrate types and within some sites the dominant substrate types (Table 2). Our sites typically contained minimal amounts of instream wood (Table 2), but some sites contained large amounts of aquatic plants that could serve as cover for fishes. The observed mean and range of nutrient and pesticide concentrations indicated that our sites exhibited wide variations in concentrations that are to be expected within agricultural streams that experience peaks in agricultural contaminants following storm events (Table 2; Smiley et al. 2014). Additionally, mean total nitrogen, maximum total nitrogen, and total phosphorus concentrations exceeded levels found to be capable of impacting stream communities (Table 2; Justus et al. 2010). With the exception of maximum atrazine concentrations, mean and maximum herbicide and fungicide concentrations were less than chronic and acute toxicity levels for fish (US EPA 2014). Average physicochemical conditions were within acceptable levels for aquatic life, yet our sites exhibited a wide range of water temperature, dissolved oxygen, and turbidity values (Table 2). Principal components analysis results The first 2 PCA axes of all habitat factors possessed eigenvalues that were greater than the broken-stick eigenvalues. Additionally, the cumulative percent of Northeastern Naturalist 31 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 the variance explained by the first 2 PCA axes ranged from 37% for water chemistry to 82% for watershed characteristics with a mean of 62% cumulative percent variance explained for all 5 habitat factors (Table 5). The first watershed PCA axis was a soil-type gradient where the percentages of Centerburg and Amanda soil types increased and percentages of Bennington and Pewamo soil types decreased with increasing site scores (Table 5). The second watershed PCA axis was a landuse gradient where percentages of forested and residential land use increased and Table 5. Loadings for each habitat variable from the first two principal components analysis (PCA) axes of watershed characteristics, riparian habitat, geomorphology, instream habitat, and water chemistry from channelized agricultural headwater streams in the Upper Big Walnut Creek watershed, OH, 2006 to 2011. [Table continued on next page.] Variables Axis 1 Axis 2 Watershed variables Watershed size 0.303 0.035 Percent agriculture 0.036 -0.605A, B Percent forested 0.081 0.560A, B Percent residential -0.153 0.538A, B Percent Bennington -0.492A, B 0.008 Percent Centerburg 0.486A, B -0.023 Percent Pewamo -0.461A, B 0.020 Percent Amanda 0.432A, B 0.167 % variance explained by axis 48.5 33.7 Riparian habitat variables Mean riparian width 0.103 -0.259 Percent canopy cover -0.306 0.165 % frequency herbaceous vegetation in height strata 1 0.029 -0.032 % frequency herbaceous vegetation in height strata 2 -0.172 -0.390A % frequency herbaceous vegetation in height strata 3 -0.030 -0.478A Sum frequency of herbaceous vegetation -0.101 -0.535A, B Herbaceous vegetation structural richness -0.137 -0.419A % frequency woody vegetation in height strata 1 -0.309 -0.007 % frequency woody vegetation in height strata 2 -0.328 -0.025 % frequency woody vegetation in height strata 3 -0.327 0.140 % frequency woody vegetation in height strata 4 -0.291 0.140 Sum frequency of woody vegetation -0.361A, B 0.062 Woody vegetation structural richness -0.300 -0.096 Mean woody vegetation density -0.312 0.079 Importance value of woody vegetation -0.352A, B 0.059 % variance explained by axis 48.7 19.8 Geomorphology variables Mean cross-sectionArea -0.514A, B 0.011 Mean top bank width -0.481A -0.204 Mean thalweg depth -0.504A, B -0.153 Mean channel depth -0.445A 0.320 Ratio of mean top bank width to mean thalweg depth 0.059 0.633A, B Sinuosity 0.044 0.519A, B Gradient -0.217 0.404A % variance explained by axis 52.3 29.3 Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 32 Vol. 24, Special Issue 8 percentages of agricultural land use decreased with increasing site scores (Table 5). The first riparian habitat PCA axis was a woody vegetation gradient where the sum frequency of woody vegetation and woody vegetation importance values decreased with increasing site scores (Table 5). The second riparian habitat PCA axis was a herbaceous vegetation gradient where the sum frequency of herbaceous vegetation decreased with increasing site scores (Table 5). The first geomorphology PCA axis Table 5, continued. Variables Axis 1 Axis 2 Instream habitat variables Mean water depth -0.426A, B 0.161 Standard deviation water depth -0.371A -0.136 Mean water velocity -0.352A -0.031 Standard deviation water velocity -0.338 0.001 Mean wet width -0.395A, B -0.104 Standard deviation wet width -0.108 0.322 Discharge -0.401A, B -0.088 Percent clay 0.027 -0.381A, B Percent sand -0.023 0.327 Percent gravel -0.135 0.487A, B Percent cobble -0.137 0.273 Percent terrestrial vegetation 0.087 -0.338 Percent leaf litter -0.038 -0.180 Percent instream wood -0.142 -0.088 Percent algae 0.055 0.187 Percent aquatic plants 0.111 0.000 Number of substrate types -0.179 -0.282 % variance explained by axis 23.5 15.9 Water chemistry variables Nitrate plus nitrite -0.395A 0.236 Ammonia -0.176 -0.329 Total nitrogen -0.428A, B 0.162 Dissolved reactive phosphorus -0.201 -0.374A Total phosphorus -0.200 -0.455A, B Dissolved organic carbon -0.191 -0.378A Alachlor -0.068 0.053 Atrazine -0.328 0.074 Atrazine desethyl -0.414A, B 0.085 Chorothalonil 0.009 0.053 Metalaxyl -0.013 0.044 Metolachlor -0.389A 0.075 Simazine -0.253 0.106 Water temperature -0.044 -0.171 Dissolved oxygen -0.067 0.204 Specific conductivity 0.034 -0.347 pH 0.080 0.055 Turbidity 0.001 -0.297 % variance explained by axis 19.7 17.7 ALoadings that were greater than 0.35. BLoadings that best characterized the underlying habitat gradients of each PCA axis. Northeastern Naturalist 33 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 was a gradient of channel size where channel cross-section area and thalweg depths decreased with increasing site scores (Table 5). The second geomorphology PCA axis was a gradient of channel shape where sinuosity and the ratio of mean top-bank width and mean thalweg depth increased with increasing site scores (Table 5). The first instream habitat PCA axis was a hydrology gradient where mean water depth, discharge, and mean wet width decreased with increasing site scores (Table 5). The second instream habitat PCA axis was a substrate gradient where percent gravel increased and percent clay decreased with increasing site scores (Table 5). The first water chemistry PCA axis was a nutrient-pesticide gradient where total nitrogen and atrazine desethyl concentrations decreased with increasing site scores (Table 5). The second water chemistry PCA axis was a nutrient gradient where total phosphorus concentrations decreased with increasing site scores (Table 5). Multimodel inference analysis results Our multimodel inference analysis indicated that all of the best models for each of the 3 response variables were reduced models that did not contain moderate levels of multicollinearity (Table 6). For abundance, model 3 had the least ΔAICc and the greatest Wi of all 8 models (Table 6). None of the other abundance models exhibited a ΔAICc < 2 that would indicate substantial support for another model (Burnham and Anderson 2002). Model 2 for mean length had the least ΔAICc and the greatest Wi of all models (Table 6). None of the other mean length models exhibited a ΔAICc < 2. For biomass, model 3 had the least ΔAICc and the greatest Wi of all models (Table 6). Model 7 for biomass exhibited a ΔAICc < 1 that indicates substantial support this model (Burnham and Anderson 2002) and its Wi was similar to the Wi of model 3. Model 7 differs only from model 3 in that it lacks the second geomorphology PCA axis. Given that the AICc penalizes for increasing number of parameters within a model we feel that identifying model 3 as the best model is justified because it has more parameters than model 7. Table 6. Summary of the number of parameters (k), small sample Akaike information criterion (AICc), difference in AICc between each model and the model with the minimum AICc (Δ AICc), and Akaike weight (Wi) from 8 models containing different combinations of independent variables to determine which habitat factor had the greatest influence on the abundance, mean length, and biomass of Creek Chub from channelized agricultural headwater streams within Upper Big Walnut Creek, OH, 2006 to 2011. Abundance Mean length Biomass k AICc ΔAICc Wi k AICc ΔAICc Wi k AICc ΔAICc Wi Global 26 372.43 6.8 0.02 26 761.67 5.8 0.05 30 301.31 6.6 0.02 model Model 2 23 369.19 3.6 0.11 23 755.87 0.0 0.84 27 299.59 4.9 0.04 Model 3 23 365.64 0.0 0.66 23 769.25 13.4 0.00 27 294.68 0.0 0.41 Model 4 23 379.47 13.8 0.00 23 760.16 4.3 0.10 27 303.96 9.3 0.00 Model 5 23 378.69 13.0 0.00 23 768.52 12.7 0.00 27 300.90 6.2 0.02 Model 6 23 370.79 5.1 0.05 23 768.73 12.9 0.00 27 297.65 3.0 0.09 Model 7 22 368.61 3.0 0.15 22 768.59 12.7 0.00 26 294.95 0.3 0.36 Model 8 22 376.51 10.9 0.00 22 766.28 10.4 0.00 26 298.58 3.9 0.06 Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 34 Vol. 24, Special Issue 8 We found that abundance was most strongly influenced by the channel-shape gradient (geomorphology PCA axis 2), with the land-use gradient (watershed PCA axis 2) and the channel size gradient (geomorphology PCA axis 1) being secondary and tertiary factors, respectively, of importance (Fig. 2). We observed that mean length was most strongly influenced by the soil-type gradient (watershed PCA axis 1), with the channel-shape gradient and the land-use gradient being secondary and tertiary factors, respectively, of importance (Fig. 2). We also documented that biomass was most strongly influenced by the land-use gradient, with the channelsize and -shape gradients being secondary and tertiary factors, respectively, of importance (Fig. 2). The standardized regression coefficients from the models with the lowest AICc scores also indicated that the abundance, mean length, and biomass of Creek Chub were most strongly influenced by watershed characteristics and geomorphology (Table 7). The standardized regression coefficients from the models with the lowest AICc scores also provided information on the types of relationships that occurred between the response variables and the independent variables (Table 7). Abundance and biomass exhibited negative correlations with the land-use gradient and decreased with increasing percentages of forested and urban land use and decreasing percentages of agriculture in the watershed (Table 7). Mean length was positively correlated with the soil-type gradient and increased with increasing percentages of Centerburg and Amanda soil types and decreasing percentages of Bennington and Pewamo soil types Table 7. Standardized regression coefficients from linear mixed-effect model analyses of abundance, mean length, and biomass of Creek Chub with habitat factors represented by the site scores from the first 2 axes from the principal component analysis (PCA) of watershed characteristics (WS), riparian habitat (RH), geomorphology (GE), instream habitat (IH), and water chemistry (WC) within channelized agricultural headwater streams in the Upper Big Walnut Creek, OH, 2006 to 2011. WS PCA RH PCA GE PCA IH PCA WC PCA Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Axis 1 Axis 2 Abundance -0.520A, B 0.079 -0.257B 0.252B -0.004 0.039 0.074 Mean length 0.542A, B 0.055 0.131 0.230B -0.158B -0.014 0.027 Biomass -0.382B -0.108 -0.423A, B 0.229 0.084 0.182B 0.166 AStandardized regression coefficients a having the greatest influence on each population-response variable. BStandardized regression coefficients of independent variables that were documented to have significant effect (P < 0.05) within the linear mixed-effect model analyses. Figure 2 [following page]. Mean importance values for each independent variable calculated based on sum of Akaike weights for each habitat factor over 8 models containing different combinations of independent variables to determine which habitat factor had the greatest influence on the abundance, mean length, and biomass of Creek Chub from channelized agricultural headwater streams within Upper Big Walnut Creek, OH, 2006 to 2011. Independent variable abbreviations are: WS = watershed characteristics, RH = riparian habitat, GE = geomorphology, IH = instream habitat, WC = water chemistry, a1 = principal components analysis axis 1, a2 = principal components analysis axis 2. Northeastern Naturalist 35 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 Figure 2. [Caption on previous page.] Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 36 Vol. 24, Special Issue 8 (Table 7). Abundance and mean length exhibited positive correlations with the channel- shape gradient and increased with increasing ratios of mean top-bank width and mean thalweg depth and increasing sinuosity (Table 7). Abundance and biomass were negatively correlated with the channel-size gradient and decreased with decreasing channel cross-section area and thalweg depth (Table 7). Discussion Our results did not support our hypothesis that instream habitat would have a greater effect on the population structure of Creek Chub in channelized agricultural headwater streams. We instead found that the abundance, mean length, and biomass of Creek Chub within these small streams were more strongly influenced by watershed characteristics and geomorphology than by riparian habitat, instream habitat, and water chemistry. Notably, instream habitat and water chemistry were not identified as being among the 3 most strongly influential habitat factors on Creek Chub (Fig. 2). Previous findings related to the habitat relationships of Creek Chub in headwater streams in North America have frequently documented the influence of instream habitat on Creek Chub (Table 1). Additionally, most studies (Hubert and Rahel 1989; Magoulik 2000; Quist and Guy 2001; Sanders 2012; Smiley et al. 2008, 2009) comparing the degree of influence of different habitat variables on the population structure of Creek Chub in North American streams have documented that instream habitat had a greater influence on the population structure of Creek Chub than other habitat factors. Our results related to mean length and biomass were consistent with those of Wuellner et al. (2013), who found the presence of Creek Chub was best explained by the watershed characteristic spatial position than by reach-scale measurements of riparian habitat, instream habitat, and water chemistry. Our results related to geomorphology were novel because the importance of channel size and shape on the population structure of Creek Chub had not been documented previously. Our results and those of Wuellner et al. (2013) highlight the importance of the larger-scale watershed and geomorphic characteristics in determining the population structure of Creek Chub in headwater streams in North America. Our results also represent the first documentation of the relationships of Creek Chub with land use, soil types, channel shape, and channel size within channelized agricultural headwater streams in North America (Table 1). Previous findings from channelized and unchannelized streams having a range of watershed sizes in North America indicate that Creek Chub exhibit either positive relationships with increasing amounts of forested land use (Bouska and Whitledge 2014, Nagrodski et al. 2013, Stammler et al. 2008), negative relationships with increasing amounts of urban land use (Fitzgerald et al. 1999, Horwitz et al. 2008), or positive relationships with increasing amounts of agriculture (Pinder and Morgan 1995). We documented that the abundance and biomass of Creek Chub decreased with joint increases in amounts of forested and urban land use and decreases in the amounts of agriculture in the watershed of channelized agricultural headwater streams. Based on previous findings (Bouska and Whitledge 2014, Nagrodski et al. 2013, Stammler et al. 2008), we expected that increases in abundance and biomass Northeastern Naturalist 37 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 of Creek Chub would occur with increasing amounts of forested land use. Our results instead suggest that the influence of increasing urban land use may override the potential benefits of increasing forested land use within the watersheds of channelized agricultural headwater streams. The variable relationships exhibited by Creek Chub with land use may be related to the ability of Creek Chub to physiologically adapt to a wide range of physical and chemical conditions (Blevins et al. 2013) or it could simply represent the site-specific conditions within a study area. Future research should quantify why Creek Chub exhibit variable relationships with land-use trends in headwater streams in North America. Soil type is one of several factors used to distinguish among ecoregions in the United States (Omerik 1987), and the composition of soil types within the watershed influences the hydrology, substrate characteristics, and chemical characteristics of streams (Maloney et al. 2013, Poff et al. 1997, Smith et al. 2003). Our understanding of the influence of soil types in the watershed on Creek Chub in headwater streams in North America is limited despite the potential importance of this watershed characteristic. Maloney et al. (2013) found that the percent of sand in the watershed was one of the most influential factors on the presence of Creek Chub in first- to fourth-order streams in Maryland. Specifically, the presence of Creek Chub decreased in sites within the Southeastern Plains and Mid-Atlantic Coastal Plains ecoregions as a result of increasing percentages of sand in the watershed that in turn corresponded with increased sand within the streams (Maloney et al. 2013). In contrast, we found mean length increased with increasing percentages of Centerburg and Amanda soil types and decreasing percentages of Bennington and Pewamo soil types. Centerburg and Amanda soil types are well-drained soils having more sand and less clay than the poorly drained Bennington and Pewamo soil types. Sites with the greatest percentage of Centerburg and Amanda soil types and having the greatest mean lengths were sites located in northern part of the watershed on the mainstem of the Upper Big Walnut Creek. Headwater sites located directly on the mainstem would be more likely to be colonized by larger-sized individuals than tributary sites located further away from the mainstem. Future research needs to examine the influence of soil type in the watershed on Creek Chub in headwater streams over a larger spatial scale to gain a greater understanding of this potentially important watershed characteristic. Watershed size is one watershed characteristic that was frequently documented to influence the abundance, density, length, biomass, and presence of Creek Chub within unchannelized headwater streams in the northeastern portion of North America (Table 1). Abundance, biomass, and occurrence of Creek Chub increases with decreasing watershed size (Butler and Fairchild 2005, Horwitz et al. 2008, Lemly 1985, Lotrich 1973, Maloney et al. 2013). These quantitative habitat relationships concur with the qualitative life-history descriptions of Creek Chub that indicate their preference for small streams (Copes 1978, Mahon et al. 1979, Shelford 1913, Starrett 1950, Thompson and Hunt 1930). In contrast, we did not observe any relationships of abundance, mean length, and biomass of Creek Chub with watershed size in channelized agricultural headwater streams. Perhaps the Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 38 Vol. 24, Special Issue 8 similarity in watershed sizes among our sites hindered detection of the influence of this watershed characteristic. Future research needs to quantify the influence of watershed size on Creek Chub in channelized agricultural headwater streams over a larger spatial scale and range of watershed sizes to gain a better understanding of the role of this watershed characteristic within this stream type. Little is known about the influence of geomorphology on abundance and success of Creek Chub in headwater streams in North America (Table 1). Creek Chub were associated with decreasing channel cross-section area in an agricultural watershed in Ohio (D’Ambrosio et al. 2009). We documented that abundance and biomass of Creek Chub increased with increasing channel cross-section area and thalweg depth within channelized agricultural headwater streams. We also documented that abundance and mean length increased with increasing ratios of mean top-bank width and mean thalweg depth and increasing sinuosity. These results suggest that stream channel alterations that modify the channel size and shape will influence success of Creek Chub within agricultural streams in North America. Our understanding of the responses of Creek Chub to stream channel alterations is limited despite the large amount of information available on the influence of stream channelization on fish community structure in agricultural streams (Smiley and Gillespie 2010). Increasing the length of stream channel enclosed by culverts installed at road crossings decreased the probability of occurrence and abundance of Creek Chub upstream of culverts within channelized agricultural headwater streams in Michigan (Briggs and Galarowicz 2013). Stream channelization has been found to reduce the number of larger-sized Creek Chub in a channelized stream site in Iowa (Scarnecchia 1988) or to have no effect on the occurrence of Creek Chub in agricultural headwater streams in Ontario (Stammler et al. 2008). Given the recent interest in the use of natural channel design, 2-stage channel design, and other novel stream channel alteration practices as part of stream restoration strategies in North America, more research is needed to identify which of these novel stream channel alteration practices positively influence Creek Chub in channelized agricultural headwater streams in the region. In conclusion, our results provided new information on the importance of watershed characteristics (land-use and soil-type gradients) and geomorphology (channel-shape and channel-size gradients) on the abundance, mean length, and biomass of Creek Chub in channelized agricultural headwater streams in the northeastern portion of North America. This information can be used to help develop conservation and restoration strategies for channelized and unchannelized headwater streams in agricultural watersheds in this region. Specifically, our results related to the population structure of Creek Chub and past results involving fish community structure (Sanders 2012; Smiley et al. 2008, 2009) indicate that conservation and restoration practices designed to mitigate physical habitat degradation within the watershed, within the channel, and within the streams are most likely to positively influence the population structure of Creek Chubs and other fishes within agricultural headwater streams. Our results also suggest that conservation and restoration practices intended to improve water quality without addressing physical habitat Northeastern Naturalist 39 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 degradation within channelized agricultural headwater streams are not likely to benefit the fishes within these streams. Acknowledgments We thank J. Bossley, S. Knight, R. Gillespie, and 2 anonymous reviewers for helpful comments on an earlier version of this manuscript. D. Gamble, E. Gates, S. Hess, A. Kemble, A. Rapp, G. Roberts, J. Risley, K. Seger, and R. Shaw assisted with field work. E. Fischer assisted with laboratory work and provided information on methods and QA/QC procedures for nutrient and pesticide measurements. K. Stillman assisted with preparing data summaries. M. Lauer helped with organizing and compiling the references section. We thank the numerous current and past USDA-ARS Soil Drainage Research Unit personnel for their assistance with field and laboratory work. B. Bacon provided information on watershed characteristics. Landowner and site information were provided by Soil and Water Conservation and NRCS districts in Delaware and Morrow Counties, OH. We are also grateful to the landowners who provided access to the sites. Literature Cited Barber, W.E., and W.L. Minckley. 1971. Summer foods of the cyprinid fish Semotilus atromaculatus. Transactions of the American Fisheries Society 2:283–289. Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. 1999. Rapid bioassessment protocols for use in streams and wadeable rivers: Periphyton, benthic macroinvertebrates and fish. Second Edition, EPA 841-B-99-002, US Environmental Protection Agency, Washington, DC. Battaglin, W.A., M.W. Sandstrom, K.M. Kuivila, D.W. Kolpin, and M.T. Meyer. 2011. Occurrence of azoxystrobin, propiconazole, and selected other fungicides in US streams, 2005–2006. Water, Air, and Soil Pollution 218:307–322. Blevins, Z.W., E.L. Effert, D.H. Wahl, and C.D. Suski. 2013. Land use drives the physiological properties of a stream fish. Ecological Indicators 24:2 24–235. Bouska, K.L., and G. Whitledge. 2014. Habitat associations of fish assemblages in the Cache River, Illinois. Environmental Biology of Fishes 97:27–42. Briggs, A.S., and T.L. Galarowicz. 2013. Fish passage through culverts in central Michigan warmwater streams. North American Journal of Fisheries Management 33:652–664. Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Springer, New York, NY. Butler, L.H., and G.W. Fairchild. 2005. Response of fish assemblages to winter in two adjacent warmwater streams. The American Midland Naturalist 154:152–165. Carlander, K.D. 1969. Handbook of Freshwater Fishery Biology. Volume One. Life History Data on Freshwater Fishes of the United States and Canada, exclusive of the Perciformes. Iowa State University Press, Ames, IA. Copes, F.C. 1978. Ecology of the Creek Chub Semotilus atromaculatus (Mitchill) in northern waters (Cyprinidae). Report # 12 on Fauna and Flora of Wisconsin, The Museum of Natural History, University of Wisconsin Stevens Point, Stevens Point, WI. D’Ambrosio, J.L., L.R. Williams, J.D. Witter, and A. Ward. 2009. Effects of geomorphology, habitat, and spatial location on fish assemblages in a watershed in Ohio, USA. Environmental Monitoring and Assessment 148:325–341. Dekar, M.P., and D.D. Magoulick. 2007. Factors affecting fish assemblage structure during seasonal stream drying. Ecology of Freshwater Fish 16:335–342. Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 40 Vol. 24, Special Issue 8 Dormann, C.F., J. Elith, S. Bacher, C. Buchmann, G. Cal, G. Carre, J.R.G. Marquez, B. Gruber, B. Lafourcade, P.J. Leitao, T. Munkemuller, C. McClean, P.E. Osborne, B. Reineking, B. Schroder, A.K. Skidmore, D. Zurrell, and S. Lautenbach. 2013. Collinearity: A review of methods to deal with it and a simulation study evaluating their performance. Ecography 36:27–46. Dube, M.G., D.L. MacLatchey, K.A. Hruska, and N.E. Glozier. 2006. Assessing the responses of Creek Chub (Semotilus atromaculatus) and Pearl Dace (Semotilus margarita) to metal mine effluents using in situ artificial streams in Sudbu ry, Ontario, Canada. Environmental Toxicology and Chemistry 25:18–28. Fischer, J.R., M.C. Quist, S.L. Wigen, A.J. Schaefer, T.W. Stewart, and T.M. Isenhart. 2010. Assemblage and population-level responses of stream fish to riparian buffers at multiple spatial scales. Transactions of the American Fisheries Society 139:185–200. Fitzgerald, D.G., R.P. Lanno, and D.G. Dixon. 1999. A comparison of a sentinel species evaluation using Creek Chub (Semotilus atromaculatus Mitchill) to a fish community evaluation for the initial identification of environmental stressors in small streams. Ecotoxicology 8:33–48. Gatz, A.J. 2008. The use of floating overhead cover by warmwater stream fishes. Hydrobiologia 600:307–310. Gilliom, R. J., 2007. Pesticides in US streams and groundwater. Environmental Science and Technology 41:3408–3414. Heithaus, M.R., and C. Grame. 1997. Fish communities of the Vermilion River watershed: Comparison of the main channel and tributaries. The Ohio Journal of Science 97:98–102. Henshel, D.S., D.W. Sparks, T.P. Simon, and M.J. Tosick. 2006. Age structure and growth of Semotilus atromaculatus (Mitchill) in PCB-contaminated streams. Journal of Fish Biology 68:44–62. Hodges, S.W., and D.D. Magoulick. 2011. Refuge habitats for fishes during seasonal drying in an intermittent stream: Movement, survival, and abundance of three minnow species. Aquatic Sciences 73:513–522. Horwitz, R.J., T.E. Johnson, P.F. Overbeck, T.K. O’Donnell, W.C. Hession, and B.W. Sweeney. 2008. Effects of riparian vegetation and watershed urbanization on fishes in streams of the Mid-Atlantic Piedmont (USA). Journal of the American Water Resources Association 44:724–741. Hubert, W.A., and F.J. Rahel. 1989. Relations of physical habitat to abundance of four nongame fishes in high-plains streams: A test of habitat suitability index models. North American Journal of Fisheries Management 9:332–340. Johnson, J. B., and K.S. Omland. 2004. Model selection in ecology and evolution. Trends in Ecology and Evolution 19:101–108. Jordan, M.A., D.L. Patel, K.E. Sanders, and R.B. Gillespie. 2013. The relative roles of contemporary and ancient processes in shaping genetic variation of a generalist fish in a catchment dominated by agriculture. Freshwater Biology 58:1660–1671. Justus, B.G., J.C. Petersen, S.R. Femmer, J.V. Davis, and J.E. Wallace. 2010. A comparison of algal, macroinvertebrate, and fish assemblage indices for assessing low-level nutrient enrichment in wadeable Ozark streams. Ecological Indicators 10:627–638. Katz, M., and A.R. Gaufin. 1953. The effects of sewage pollution on the fish population of a midwestern Stream. Transactions of the American Fisheries Society 82:156–165. Katz, M., and W.C. Howard. 1955. The length and growth of 0-year class Creek Chubs in relation to domestic pollution. Transactions of the American Fisheries Society 84:228–238. Northeastern Naturalist 41 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 King, K.W., P.C. Smiley Jr., B.J. Baker, and N.R. Fausey. 2008. Validation of paired watersheds for assessing conservation practices in Upper Big Walnut Creek watershed, Ohio. Journal of Soil and Water Conservation 63:380–395. Kittle, A.M., J. M. Fryxell, G.E. Desy, and J. Hamr. 2008. The scale-dependent impact of Wolf predation risk on resource selection by three sympatric ungulates. Oecologia 157:163–175. Koroleff, J. 1983. Determination of total phosphorus by alkaline persulphate oxidation. Pp. 136–138, In K.Grasshoff, M. Ehrhardt, and K. Kremling (Eds). Methods of Seawater Analysis. Verlag Chemie, Wienheim, Germany. Larimore, R.W., W.F. Childers, and C. Heckrotte. 1959. Destruction and reestablishment of stream fish and invertebrates affected by drought. Transactions of the American Fisheries Society 88:261–285. Leet, J.K., L.S. Lee, H.E. Gall, R.R. Goforth, S. Sassman, D.A. Gordon, J.M. Lazorchak, M.E. Smith, C.T. Jafvert, and M.S. Sepulveda. 2012. Assessing impacts of land-applied manure from concentrated animal feeding operations on fish populations and communities. Environmental Science and Technology 46:13440–13447. Lemly, A.D. 1985. Suppression of native fish populations by green sunfish in first-order streams of Piedmont North Carolina. Transactions of the American Fisheries Society 114:705–712. Lotrich, V.A. 1973. Growth, production, and community composition of fishes inhabiting a first-, second-, and third-order stream of eastern Kentucky. Ecological Monographs 43:377–397. Lyons, J. 1992. The length of stream to sample with a towed electrofishing unit when fish species richness is estimated. North American Journal of Fisheries Management 12:198–203. Lyons, J. 2006. A fish-based index of biotic integrity to assess intermittent headwater streams in Wisconsin, USA. Environmental Monitoring and Assessment 122:239–258. Madramootoo, C.A., W.R. Johnston, J.E. Ayars, R.O. Evans, and N.R. Fausey. 2007. Agricultural drainage management, quality, and disposal issues in North America. Irrigation and Drainage 56:S36–S45. Magoulik, D.D. 2000. Spatial and temporal variation in fish assemblages of drying stream pools: The role of abiotic and biotic factors. Aquatic Ecology 34:29–41. Mahon, R., E.K. Balon, and D.L.G. Noakes. 1979. Distribution, community structure, and production of fishes in the upper Speed River, Ontario: A preimpoundment study. Environmental Biology of Fishes 4:219–244. Maloney, K.O., D.E. Weller, D.E. Michaelson, and P.J. Ciccotto. 2013. Species distribution models of freshwater stream fishes in Maryland and their implications for management. Environmental Modeling and Assessment 18:1–12. Mazerolle, M.J. 2016. AICcmodavg: model selection and multimodel inference based on (Q)AIC(c). R package version 2.0-4. Available online at https://cran.r-project.org/ package=AICcmodavg. Accessed 27 July 2017. McCune, B., and M.J. Mefford. 1999. Multivariate Analysis of Ecological Data Version 4.01. MjM software, Gleneden Beach, OR. McMahon, T.E. 1982. Habitat suitability index models: Creek Chub. US Fish and Wildlife Service, Washington, DC. Meffe, G.K., and A.L. Sheldon. 1988. The influence of habitat structure on fish assemblage composition in southeastern blackwater streams. The American Midland Naturalist 120:225–240. Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 42 Vol. 24, Special Issue 8 Menzel D.W., and R.F.Vaccaro. 1964. The measurement of dissolved organic and particulate carbon in seawater. Limnology and Oceanography 9:138–142. Mitchell, D.M.W., S.A. Entrekin, and G.L. Adams. 2012. Structure and function of large wood in Ozark headwater streams and its relationship to fish community structure. Journal of Freshwater Ecology 27:335–349. Nagrodski, A., C.D. Suski, and S.J. Cooke. 2013. Health, condition, and survival of Creek Chub (Semotilus atromaculatus) across a gradient of stream habitat quality following an experimental cortisol challenge. Hydrobiologia 702:283–296. Needelman, B.A., P.J.A. Kleinman, J.S. Strock, and A.L. Allen. 2007. Improved management of agricultural drainage ditches for water quality protection: An overview. Journal of Soil and Water Conservation 62:171–177. Ohio Environmental Protection Agency (OH EPA). 2005. Total maximum daily loads for the Big Walnut Creek watershed. Final Report. Division of Surface Water, Columbus, OH. Omerik, J.M. 1987. Ecoregions of the conterminous United States. Annals of the Association of American Geographers 77:118–125. Page, L.M., and B.M. Burr. 1991. A Field Guide to Freshwater Fishes of North America North of Mexico. Houghton Mifflin Company, Boston, MA. Parsons, T.R, Y. Maita, and C.M. Lalli. 1984. A Manual of Chemical and Biological Methods for Seawater Analysis. Pergamon Press, Oxford, UK. Pinder, M.J., and R.P. Morgan II. 1995. Interactions of pH and habitat on cyprinid distributions in Appalachian streams of Maryland. Transactions of the American Fisheries Society 124: 94–102. Pinheiro J., D. Bates, S. DebRoy, D. Sarkar, and R Core Team. 2014. nlme: Linear and nonlinear mixed-effects models. R package version 3.1-117. Available online at http:// CRAN.R-project.org/package=nlme. Accessed 27 July 2017. Poff, N.L, J.D. Allan, M.B. Bain, J.R. Karr, K.L. Prestegaard, B.D. Richter, R.E. Sparks, and J.C. Stromberg. 1997. The natural flow regime: A paradigm for river conservation and restoration. Bioscience 47:769–784. Portt, C.B., E.K. Balon, and D.L.G. Noakes. 1986. Biomass and production of fishes in natural and channelized streams. Canadian Journal of Fisheries and Aquatic Sciences 43:1926–1934. Quist, M.C., and C.S. Guy. 2001. Growth and mortality of prairie stream fishes: Relations with fish community and instream habitat characteristics. Ecology of Freshwater Fish 10:88–96. Quist, M.C., F.J. Rahel, and W.A. Hubert. 2005. Hierarchical faunal filters: An approach to assessing effects of habitat and nonnative species on native fishes. Ecology of Freshwater Fish 14:24–39. R Core Team. 2014. R (version 3.03): A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. Vienna, Austria. Ross, R.M., R.M. Bennett, C.D. Snyder, J.A. Young, D.R. Smith, and D.P. Lemarie. 2003. Influence of Eastern Hemlock (Tsuga canadensis L.) on fish community structure and function in headwater streams of the Delaware River basin. Ecology of Freshwater Fish 12:60–65. Sanders, K.E. 2012. Relative importance of water quality and habitat to fish communities in streams influenced by agricultural land use in the Cedar Creek watershed, Indiana. M.Sc. Thesis. Indiana University-Purdue University Fort Wayne, Fort Wayne, IN. Northeastern Naturalist 43 P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 Vol. 24, Special Issue 8 Sanders, R.E. 2001. A Guide to Ohio Streams. Watkins Printing, Columbus, OH. Scarnecchia, D.L. 1988. The importance of streamlining in influencing fish community structure in channelized and unchannelized reaches of a prairie stream. Regulated Rivers: Research and Management 2:155–166. Schlosser, I.J. 1982. Fish community structure and function along two habitat gradients in a headwater stream. Ecological Monographs 52:395–414. Schlosser, I.J. 1995. Dispersal, boundary processes, and trophic level interactions in streams adjacent to beaver ponds. Ecology 76:908–925. Schlosser, I.J. 1998. Fish recruitment, dispersal, and trophic interactions in a heterogenous lotic environment. Oecologia 113:260–268. Schwartz, J.S., and E.E. Herricks. 2008. Fish use of ecohydraulic-based mesohabitat units in a low-gradient Illinois stream: Implications for stream restoration. Aquatic Conservation: Marine and Freshwater Ecosystems 18:852–866. Shelford, V.E. 1913. Animal Communities in Temperate America. The University of Chicago Press, Chicago, IL. Smiley, P.C., Jr., and R.B. Gillespie. 2010. Influence of physical habitat and agricultural contaminants on fishes within agricultural drainage ditches. Pp. 37–73, In M.T. Moore, and R. Kroger (Eds.). Agricultural Drainage Ditches: Mitigation Wetlands for the 21st Century. Research Signpost, Kerala, India. Smiley, P.C., Jr., R.B. Gillespie, K.W. King, and C. Huang. 2008. Contribution of habitat and water quality to the integrity of fish communities in agricultural drainage ditches. Journal of Soil and Water Conservation 63:218A. Smiley, P.C., Jr., R.B. Gillespie, K.W. King, and C. Huang. 2009. Management implications of the relationships between water chemistry and fishes within channelized headwater streams in the midwestern United States. Ecohydrology 2:294–302. Smiley, P.C., Jr., K.W. King, and N.R. Fausey. 2011. Influence of herbaceous riparian buffers on physical habitat, water chemistry, and stream communities within channelized agricultural headwater streams. Ecological Engineering 37:1314–1323. Smiley, P.C., Jr., K.W. King, and N.R. Fausey. 2014. Annual and seasonal differences in pesticide mixtures within channelized agricultural headwater streams in central Ohio. Agriculture, Ecosystems, and Environment 193:83–95. Smith, R.A., R.B., Alexander, and G.E. Scharz. 2003. Natural background concentrations of nutrients in streams and rivers in the conterminous United States. Environmental Science and Technology 37:3039–3047. Stair, D.M., V.R. Tolbert, and G.L. Vaughan. 1984. Comparison of growth, population structure, and food of the Creek Chub, Semotilus atromaculatus, in undisturbed and surface-mining-disturbed streams in Tennessee. Environmental Pollution 35:331–343. Stammler, K.L., R.L. McLaughlin, and N.E. Mandrak. 2008. Streams modified for drainage provide fish habitat in agricultural areas. Canadian Journal of Fisheries and Aquatic Sciences 65:509–522. Starrett, W.C. 1950. Distribution of the fishes of Boone County, Iowa, with special reference to the minnows and darters. The American Midland Naturalist 43:112–127. Storck, T.W., and W.T. Momot. 1981. Movements of the Creek Chub in a small Ohio stream. The Ohio Journal of Science 81:9–13. Storck, T.W., and W.T. Momot. 1989. Annual production of Creek Chub and Southern Redbelly Dace in a small woodland stream. The Ohio Journal of Science 89:55–61. Thompson, D.H., and F.D. Hunt. 1930. The fishes of Champaign County: A study of the distribution and abundance of fishes in small streams. Illinois Natural History Survey Bulletin 29:1–101. Northeastern Naturalist P.C. Smiley Jr., K.W. King, and N.R. Fausey 2017 44 Vol. 24, Special Issue 8 Trautman, M.B., and D.K. Gartman. 1974. Re-evaluation of the effects of man-made modifications on Gordon Creek between 1887 and 1973 and especially as regards its fish fauna. The Ohio Journal of Science 74:162–173. Walker, R.H., and G.L. Adams. 2016. Ecological factors influencing movement of Creek Chub in an intermittent stream of the Ozark Mountains, Arkansas. Ecology of Freshwater Fish 25:190–202. Wohl, N.E., and R.F. Carline. 1996. Relations among riparian grazing, sediment loads, macroinvertebrates, and fishes in three central Pennsylvania streams. Canadian Journal of Fisheries and Aquatic Sciences 53(Suppl. 1):260–266. Wuellner, M.R., R.G. Bramblett, C.S. Guy, A.V. Zale, D.R. Roberts, and J. Johnson. 2013. Reach- and catchment-scale characteristics are relatively uninfluential in explaining the occurrence of stream fish species. Journal of Fish Biology 82:14 97–1513. US Environmental Protection Agency (US EPA). 1995. Methods for the determination of organic compounds in drinking water. Supplement III, EPA-600/R-95/131, National Exposure Research Laboratory, Office of Research and Development, Cincinnati, OH. US EPA. 2014. Office of Pesticide Programs’ aquatic life benchmarks. Washington, DC. Available online at http://www.epa.gov/oppefed1/ecorisk_ders/aquatic_life_benchmark. htm#benchmarks. Accessed 17 March 2014. Zaugg, S.D., M.W. Sandstrom, S.G. Smith, and K.M. Fehlberg. 1995. Methods of analysis by the US Geological Survey National Water Quality Laboratory-determination of pesticides in water by C-18 solid-phase extraction and capillary-column gas chromatography/ mass spectrometry with selected-ion monitoring. Open-File Report 95-181, US Geological Survey, Denver, CO.