2006 SOUTHEASTERN NATURALIST 5(3):393–412 Influence of Catchment Disturbance on Pteronotropis euryzonus (Broadstripe Shiner) and Semotilus thoreauianus (Dixie Chub) Kelly O. Maloney1,2,*, Richard M. Mitchell1, and Jack W. Feminella1 Abstract -We examined relationships between catchment-scale disturbance from military training and two dominant fish species, Pteronotropis euryzonus (broadstripe shiner) and Semotilus thoreauianus (Dixie chub) in headwater streams at the Fort Benning Military Installation (FBMI), GA. Disturbance was estimated as the percent of the catchment that was bare ground and unpaved road cover. Relative abundance of broadstripe shiners and Dixie chubs were negatively and positively related to disturbance, respectively. This complementarity likely resulted from contrasting life histories, feeding behaviors, and habitat preferences between the two species. Absolute abundance of broadstripe shiners increased, whereas relative abundance of Dixie chubs decreased, with stream discharge, suggesting that both species were affected by local habitat conditions. Additionally, the average body size of both species was lower in high-disturbance streams, signifying that both species were affected by disturbance. Results also suggest a disturbance threshold, where streams with disturbance levels of 􀂧 5–8.1% of the catchment had broadstripe shiner proportions below those in low-disturbance streams. About 71–88% of second-order catchments on FBMI lie below this threshold level, suggesting that many streams on FBMI are potentially suitable for the broadstripe shiner. Introduction Stream fish assemblages are governed by biotic (e.g., competition, predation), abiotic (e.g., local and reach habitat), and spatial (e.g., geographic position, longitude) factors (Jackson et al. 2001, Marsh-Matthews and Matthews 2000, Matthews and Robison 1998, Schlosser 1987). Anthropogenic actions may disrupt native fish assemblages by destabilizing one or more of these factors. For example, human introductions of nonnative fish that circumvent natural physical barriers to migration may disrupt biotic controls on native assemblages by altering trophic structure (Mills et al. 1994, Moyle 1999, Rahel 2000). Further, construction of flow obstructions such as impoundments and culverts may restrict migration and thus impact assemblages, particularly in headwaters (Baxter 1977, Winston et al. 1991). Arguably, in small streams, the most pervasive and thus significant anthropogenic disruption occurs from altered stream physicochemical (habitat) conditions associated with disturbance from land use within the surrounding catchment (Scott and Helfman 2001). 1Department of Biological Sciences, 331 Funchess Hall, Auburn University, Auburn AL, 36849-5407. 2Current address - Illinois Natural History Survey, Kaskaskia Biological Field Station, RR#1, Box 157, Sullivan, IL 61951. *Corresponding author - kom@uiuc.edu. 394 Southeastern Naturalist Vol. 5, No. 3 Fish abundance and diversity typically decrease with increasing urban and agricultural land use within catchments (Lenat and Crawford 1994, Snyder et al. 2003, Wang et al. 2001). Most often, such reductions occur from degraded streamwater quality and/or loss of instream habitat. Land-use changes resulting in forestland conversion and soil disturbance often increase erosion and subsequent sedimentation within receiving streams. Direct impacts of sedimentation on fish may range from increased emigration and/or decreased immigration to sediment-induced mortality (Auld and Schubel 1978, Bergstedt and Bergersen 1997, Cordone and Kelley 1961, Ritchie 1972), whereas indirect impacts of sedimentation may be manifested in decreased fish habitat and food quality or quantity (Bergstedt and Bergersen 1997, Berkman and Rabeni 1987, Cordone and Kelley 1961, Sutherland et al. 2002). Effects of increased sedimentation from catchment and soil disturbance on fish are well known; however, a scarcity of research exists describing responses in naturally sandy-bottomed streams where populations may be naturally adapted to high sedimentation. Furthermore, many studies have been conducted within catchments with significant urban and/or agricultural land use, areas where degraded stream water often includes chemical pollutants (e.g., pesticides, heavy metals) that may mask impacts on fish populations from sediment alone. Military installations provide a unique opportunity to study the effects of land use on stream fish assemblages for several reasons. First, military bases often are large experimental units that have minimal agricultural or urban influences. Such conditions can facilitate studies of landscape-level disturbance without the confounding factors associated with agricultural or urban sprawl. Second, bases may provide refuges for imperiled species (Cohn 1996, Goodmann 1996), allowing study of sensitive species that may not be possible elsewhere. Last, because of the need for different landscape conditions in training exercises, land use within military bases often varies, thereby providing instructive ecological contrasts that are manifested at the catchment scale. Thus, military bases may provide unique landscape conditions that are conducive to examining the effects of anthropogenic land use on ecosystem structure and function at a variety of spatial scales. Pteronotropis euryzonus Suttkus (broadstripe shiner) and Semotilus thoreauianus Jordan (Dixie chub) are two common headwater species in the Southeast that show overlapping distributions (Boschung and Mayden 2004, Mettee et al. 1996). Broadstripe shiners are restricted to the Chattahoochee Basin, whereas Dixie chubs occupy a broader geographic range, from the Tombigbee Basin, AL, east to the Ochlockonee River, GA (Boschung and Mayden 2004, Johnston and Ramsey 1990, Suttkus 1955). Broadstripe shiners are primarily drift feeders that consume mostly aquatic insects and detritus (Suttkus 1955), and are associated with coarse woody debris in swift, deep water (Boschung and Mayden 2004, Suttkus 1955). Dixie chubs are trophic generalists that consume a variety of aquatic and terrestrial 2006 K.O. Maloney, R.M. Mitchell, and J.W. Feminella 395 insects, worms, small fish, mollusks, crayfish, and plant detritus, and prefer small, clear streams (Boschung and Mayden 2004, Mettee et al. 1996). Broadstripe shiners broadcast eggs over vegetation without much parental care (Katula 1993; C.E. Johnston, Auburn University, Auburn, AL, pers. comm.), whereas Dixie chub males excavate nests where females deposit eggs, which are subsequently covered with coarse sediment and defended by males (Boschung and Mayden 2004, Johnston and Ramsey 1990, Maurakis et al. 1993). The broadstripe shiner is classified as a vulnerable species and considered rare in Georgia, whereas the Dixie chub is currently stable (Warren et al. 2000). The contrasting feeding and life-history traits between these two species may make them differentially susceptible to catchment disturbance; however, no study has reported patterns of these two species with respect to catchment disturbance. The objectives of our study were to 1) examine relationships between catchment disturbance and relative and absolute abundances of the broadstripe shiner and Dixie chub, and 2) compare size structure between broadstripe shiners and Dixie chubs in streams draining low- vs. highdisturbance catchments. Methods Study site We studied seven small tributaries of Upatoi Creek, a 6th-order stream in the Chattahoochee River Basin, on the Fort Benning Military Installation (FBMI), GA (Fig. 1). Fort Benning occurs within the Southeastern Plains ecoregion, and contains mainly oak-hickory-pine and southern mixed forests with underlying sandy or sandy clay-loam soils (Griffith et al. 2001, Omernik 1987). Study streams were small and perennial (1st- to 2nd-order) with narrow, low-gradient (range of channel slope 0.8–2.7%), sandy-bottom channels (range of mean particle size 0.56–0.89 mm; Maloney et al. 2005), and intact riparian zones. Land use within study catchments was patchy, ranging from almost entirely forested to catchments used extensively for military training and silviculture. Military land-use practices (i.e., mechanized training by tanks and armored personnel carriers) and general use of unpaved roads resulted in the creation of bare ground, which increased the influx of sediment from upland disturbance to streams through numerous ephemeral channels. The degree of this disturbance varied among the catchments studied (Maloney et al. 2005). Land cover/instream physicochemical variables We quantified spatial and land-use/land-cover data with Arcview® 3.2 GIS (Environmental Systems Research Institute, Redlands, CA) using coverages from the SERDP Ecosystem Management Project (SEMP) data repository (http://sempdata.wes.army.mil/). We estimated catchment area (Area) above sampling locations using a 1993 digital elevation model 396 Southeastern Naturalist Vol. 5, No. 3 (DEM, 10-m resolution) and obtained grid coordinates of sampling sites from global positioning system (GPS) units. For our land-use component, we defined disturbance level (%BGRD) as the percent of bare ground on slopes > 5% summed with the percent of unpaved road cover within a catchment, calculated from a 1999 Landsat image (30-m resolution), 1995 road coverage (10 m), and the 1993 DEM (Maloney et al. 2005). As an index of landscape-scale factors, we calculated distance from each site to the mainstem Upatoi Creek (L_Upatoi), which we considered a potential colonization source for fish, using a 1999 digitized streams-coverage map (1:24,000) and summing the distances between each study stream and Upatoi Creek. In addition, prior to sampling fish at each site, we quantified stream discharge (Q, incremental method; see Gore 1996) using a Marsh-McBirney Model 2000® flowmeter and streamwater pH using a Beckman model 200® pH meter about bimonthly from January 2000 to April 2003 at each site, and also estimated relative abundance of coarse woody debris (CWD) in the stream channel during April 2003. We defined CWD as pieces of wood and live roots at least partially submerged and > 2.5 cm in diameter, and used a modified transect method (Wallace and Benke 1984) to quantify CWD along 15 transects (spaced 5 m apart) per stream, expressing CWD relative abundance as percentage of coverage of the stream bed. Figure 1. Locations of study catchments (polygons) within Fort Benning Military Installation, GA. Dotted line in middle figure represents the Chattahoochee River, which separates Alabama (AL) and Georgia (GA) (modified from Maloney et al. 2005). Numbers in the right figure identify watersheds on the same stream (e.g., 2 and 3 on the Sally Branch represent Sally Branch Tributaries 2 and 3, SB2 and SB3, respectively). 2006 K.O. Maloney, R.M. Mitchell, and J.W. Feminella 397 Fish and habitat sampling We sampled fish assemblages at three random locations along a 100-m long representative reach in each stream in March (spring), July (summer), and December (winter) 2003, using the 2-pass removal-depletion method (Seber 1982) with a backpack electroshocker (Smith-Root LR-24®) and block seines. Each location consisted of an adjacent pool and run mesohabitat; riffles were not present. We identified and recorded all fish collected for each mesohabitat and, except for voucher specimens used for taxonomic confirmation, we returned all individuals to the stream. We also recorded standard length (SL) to the nearest mm for all fish collected. For each mesohabitat, we measured wetted width at five equally spaced crossstream transects (n = 5), and recorded depths at five locations along each transect (n = 25). We then averaged width and depths values for each mesohabitat. Young-of-year fishes were excluded from analysis due to difficulty in capture and identification. Statistical analyses Preliminary observations indicated that the broadstripe shiner and Dixie chub were the dominant species collected and in some streams, constituted 100% of the collected individuals; therefore, we restricted our analyses to these species. We quantified absolute and relative abundance of both species for each stream and season. We normalized the proportional data using the arcsine-square-root transformation and the absolute abundance data using the square-root transformation (Zar 1999). We then used regression analysis to determine relationships between proportions of each of these species and %BGRD, CWD, Q, and L_Upatoi. Analysis of collinearity using variation inflation factors (VIFs) revealed no high collinearity within this set of explanatory variables (i.e., all VIFs < 10; Myers 1990); however, pH was highly correlated with %BGRD, CWD, and Area (all r > 0.6) and thus not analyzed further, and CWD was highly correlated (r > 0.6) with %BGRD and thus was excluded from multiple regression models. Model selection was performed using adjusted R2 (R2 adj), Akaike Information Criteria corrected for small sample size (AICc), and Akaike weights (wi; Burnham and Anderson 2002). Best models were considered to have the smallest AICc and largest wi and R2 adj; however, models that deviated < 2 AICc from the best model (􀂨AICc) were highly supported. We calculated 95% confidence intervals of broadstripe shiner and Dixie chub relative abundance for the three least-disturbed streams and used these intervals as a measure of low-disturbance variation, which enabled us to identify potential disturbance thresholds. We chose the three least-disturbed streams because each stream showed less than 􀂧 5% catchment disturbance and had higher amounts of and more stable benthic habitat than the four other study streams (see Maloney et al. 2005). We also calculated the %BGRD for all 2nd-order catchments (n = 249) within the Fort Benning boundary to approximate the number of potential “refuge” sites for broadstripe shiners. 398 Southeastern Naturalist Vol. 5, No. 3 We also generated size-frequency distributions for each species in the study streams. However, as a result of low collections in some streams, we pooled specimens collected over the entire study from the three lowest and from the three highest disturbed catchments, and then tested for differences in size frequencies between these high- and low-disturbance categories using a nonparametric analysis of variance by ranks (Kruskal-Wallis test; Zar 1999). Results Land cover/instream physicochemical variables The proportion of bare ground on slopes > 5% and unpaved road cover (%BGRD) ranged from 3.15 to 13.65% (Table 1). Catchment area ranged from 0.72 km2 (Site SB3) to 3.35 km2 (KM1), distance to the Upatoi Creek ranged from 1.73 km (Site LC) to 26.15 km (HB), and instream CWD relative abundance ranged from 3.3 (Site SB3) to 12.4 (LC) percent of streambed, respectively (Table 1). Average discharge among all study streams was highest in spring (0.019 m3/s), intermediate in winter (0.014 m3/ s), and lowest in summer (0.009 m3/s). Wetted width ranged from 1.0 m (BC2) to 2.1 m (KM1), and streamwater pH ranged from 4.9 to 6.5 (Table 1). Pool volume was highest in winter (mean = 1.17 m3), intermediate in spring (1.02 m3), and lowest in summer (0.87 m3), whereas run volume was largest in spring (mean = 0.85 m3), followed by summer (0.76 m3) and winter (0.69 m3) (Table 1). Fish assemblage We collected 10 fish species over the study (Table 2). Broadstripe shiners and Dixie chubs each composed > 30% of total fish collected in every season (Table 2), and together they composed 48–100% of the total fish collected in each stream in each season (Table 3). The remaining eight species each composed 􀂔 22% of total fish collected in each season (Table 2). Season-specific total richness ranged from 1 to 7, with the fewest species collected in SB4 (one [Dixie chub] in each season) and the most species in LC (5–7 per season; Table 3). Absolute abundances of broadstripe shiners and Dixie chubs Absolute abundance of both species exhibited a seasonal response to catchment disturbance. In spring, broadstripe shiner absolute abundance was best modeled by a negative relationship with %BGRD (R2 adj = 0.42; Table 4). In summer, absolute abundance of broadstripe shiners was best modeled by a positive relationship with Q (R2 adj = 0.58); however, a two-variable model including a negative relationship with %BGRD and positive relationship with Q also had high support (􀂨AICc = 0.44, R2 adj = 0.79). In winter, a positive relationship between Q and absolute abundance of broadstripe shiners was the best model (R2 adj = 0.56,), although two univariate models also had support (CWD: 􀂨AICc = 2.11, R2 adj = 0.41; %BGRD: 􀂨AICc = 2.28, 2006 K.O. Maloney, R.M. Mitchell, and J.W. Feminella 399 Table 1. Summary of land use, reach, and habitat scale variables. %BGRD = catchment disturbance (see text for further explanation), Area = catchment area (km2), L_Upatoi = distance to Upatoi Creek (km), CWD = coarse woody debris relative abundance (% stream bottom coverage). Data for wetted width, CWD, pH, pool volume, and run volume are season means (SE). Site Discharge Wetted Pool Run Stream abbrev. %BGRD Area L_Upatoi CWD Season (m3/sec) width (m) pH volume (m3) volume (m3) Bonham BC2 3.15 0.75 2.05 10.1 (2.2) Spring 0.005 1.2 4.89 (0.05) 0.65 (0.07) 0.30 (0.08) Tributary Summer 0.001 1.1 4.92 (0.03) 0.50 (0.13) 0.55 (0.31) Winter 0.005 1.0 5.24 (0.19) 0.67 (0.04) 0.31 (0.07) Sally Branch SB2 8.12 1.23 11.92 8.7 (1.6) Spring 0.027 1.4 5.92 (0.13) 0.62 (0.10) 0.40 (0.13) Tributary Summer 0.009 1.4 6.24 (0.03) 0.38 (0.02) 0.61 (0.05) Winter 0.016 1.5 6.03 (0.06) 0.86 (0.39) 0.29 (0.02) Sally Branch SB3 10.49 0.72 12.80 3.3 (0.9) Spring 0.007 1.3 6.06 (0.13) 0.28 (0.06) 0.14 (0.02) Tributary Summer 0.004 1.3 6.52 (0.06) 0.20 (0.01) 0.22 (0.08) Winter 0.008 1.3 6.13 (0.06) 0.34 (0.02) 0.31 (0.04) Sally Branch SB4 13.65 1.00 12.86 3.6 (1.5) Spring 0.012 1.5 5.45 (0.19) 0.32 (0.07) 0.31 (0.04) Tributary Summer 0.006 1.4 5.78 (0.15) 0.35 (0.06) 0.24 (0.03) Winter 0.009 1.5 5.54 (0.19) 0.38 (0.08) 0.29 (0.04) Hollis Branch HB 6.62 2.15 26.15 6.5 (2.2) Spring 0.018 2.0 5.13 (0.08) 1.65 (0.46) 0.76 (0.10) Summer 0.013 1.8 5.43 (0.04) 0.74 (0.06) 0.93 (0.21) Winter 0.018 1.9 5.08 (0.04) 2.24 (0.83) 1.02 (0.09) Kings Mill Creek KM1 5.01 3.35 3.11 7.5 (1.1) Spring 0.037 2.1 4.95 (0.06) 1.35 (0.34) 1.12 (0.17) Tributary Summer 0.020 1.9 5.07 (0.03) 2.14 (0.58) 1.56 (0.22) Winter 0.029 1.8 4.98 (0.05) 1.94 (0.17) 0.98 (0.25) Lois Creek LC 3.67 3.32 1.73 12.4 (2.1) Spring 0.044 2.0 4.87 (0.02) 2.49 (0.61) 3.47 (0.95) Summer 0.013 2.0 4.89 (0.07) 1.97 (0.21) 1.43 (0.46) Winter 0.022 1.9 5.04 (0.07) 1.68 (0.19) 1.85 (0.69) 400 Southeastern Naturalist Vol. 5, No. 3 Table 2. Absolute and relative abundance (in parentheses) of fish species collected during the study. Number collected (% of total) Family Species Common name Spring Summer Winter Aphredoderidae Aphredoderus sayanus (Gilliams) Pirate perch 8 (3.7) 3 (1.8) 5 (2.1) Centrarchidae Lepomis gulosus Cuvier Warmouth 0 (0) 0 (0) 1 (0.4) Lepomis miniatus Jordan Redspotted sunfish 3 (1.4) 1 (0.6) 1 (0.4) Cyprinidae Notemigonus crysoleucas (Mitchill) Golden shiner 0 (0) 1 (0.6) 1 (0.4) Pteronotropis euryzonus (Suttkus) Broadstripe shiner 67 (30.7) 69 (41.8) 117 (50) Semotilus thoreauianus Jordan Dixie chub 90 (41.3) 67 (40.6) 81 (34.6) Esocidae Esox americanus Gmelin Redfin pickerel 1 (0.5) 5 (3) 2 (0.9) Ictaluridae Ameiurus natalis (Lesueur) Yellow bullhead 0 (0) 4 (2.4) 2 (0.9) Percidae Percina nigrofasciata (Agassiz) Blackbanded darter 1 (0.5) 1 (0.6) 2 (0.9) Petromyzontidae Ichthyomyzon gagei Hubs and Trautman Southern brook lamprey 48 (22) 14 (8.5) 22 (9.4) Total 218 165 234 Table 3. Fish species richness and absolute and relative abundance (in parentheses) of the broadstripe shiner and the Dixie chub by stream and season. Stream abbreviations defined in Table 1. Spring Summer Winter Stream Broadstripe Broadstripe Broadstripe abbreviation Richness shiner Dixie chub Richness shiner Dixie chub Richness shiner Dixie chub BC2 3 7 (50) 6 (43) 2 4 (80) 1 (20) 2 4 (67) 2 (33) SB2 2 0 (0) 13 (81) 5 4 (13) 23 (72) 4 18 (51) 13 (37) SB3 1 0 (0) 17 (100) 4 5 (28) 11 (61) 4 2 (6) 30 (88) SB4 1 0 (0) 12 (100) 1 0 (0) 22 (100) 1 0 (0) 23 (100) HB 5 18 (27) 38 (57) 6 9 (41) 5 (23) 5 21 (60) 8 (23) KM1 3 31 (46) 3 (4) 4 31 (76) 3 (7) 6 23 (52) 4 (9) LC 7 11 (44) 1 (4) 7 16 (64) 2 (8) 5 49 (86) 1 (2) 2006 K.O. Maloney, R.M. Mitchell, and J.W. Feminella 401 R2 adj = 0.39; Table 4). In spring, absolute abundance of Dixie chubs was best modeled by a two-variable model including a negative relationship with Q and a positive relationship with L_Upatoi (R2 adj = 0.97; Table 4); however, the simple model indicating a positive relationship with L_Upatoi also had strong support (􀂨AICc = 1.51, R2 adj = 0.91). In summer and winter, %BGRD best explained variation in absolute abundance of Dixie chubs (R2 adj = 0.69 and 0.84, respectively); however, for winter, a negative relationship with CWD was supported (􀂨AICc = 1.69, R2 adj = 0.79; Table 4). Relative abundances of broadstripe shiners and Dixie chubs In all seasons, the proportion of the total assemblage as broadstripe shiners was strongly negatively related to %BGRD, whereas proportion of the assemblage as Dixie chubs was strongly positively related to %BGRD (Table 5, Fig. 2). In spring, summer, and winter the best model for the broadstripe shiners was a negative relationship with %BGRD (R2 adj = 0.76, 0.84, and 0.86, respectively; Table 5); however, a positive relationship with CWD was supported in winter (􀂨AICc = 2.76, R2 adj = 0.79). Variation in relative abundance of Dixie chubs was best modeled by a two-variable model, which included a positive relationship with %BGRD and negative relationship with Q in spring (R2 adj = 0.90) and winter (R2 adj = 0.94); however, the simple model containing a positive relationship with %BGRD also Table 4. Best two models of multiple regression analysis on the absolute abundance of broadstripe shiners and Dixie chubs (best models had smallest 􀂨AICc and largest wi). Abbreviations: CWD = relative abundance of coarse woody debris, Q = discharge, %BGRD = catchment disturbance (see text), L_Upatoi = linear distance to the Upatoi Creek main stem, k = number of parameters in model, AICc = Akaike Information Criterion corrected for small sample size, 􀂨AICc = deviation in AICc from best model, wi = Akaike weights, SSE = sum of squares error. Numbers in parentheses denote standardized regression coefficients. CWD was not included in multiple regression models due to collinearity with other variables. * = second and third best model were both included because both were equally supported. Adjusted Variable Parameters in model k AICc 􀂨AICc wi SSE R2 Number of broadstripe shiners Spring %BGRD (-0.72) 2 12.39 0.00 0.65 15.11 0.42 Q (0.49) 2 15.58 3.19 0.13 23.84 0.09 Summer Q (0.81) 2 6.46 0.00 0.45 6.48 0.58 %BGRD (-0.49), Q (0.65) 3 6.90 0.44 0.36 2.54 0.79 Winter Q (0.80) 2 11.14 0.00 0.50 12.64 0.56 CWD (0.71)* 2 13.25 2.11 0.17 17.09 0.41 %BGRD (-0.44)* 2 13.42 2.28 0.16 17.51 0.39 Number of Dixie chubs Spring Q (-0.24), L_Upatoi (0.88) 3 –6.63 0.00 0.67 0.37 0.97 L_Upatoi (0.96) 2 –5.12 1.51 0.31 1.24 0.91 Summer %BGRD (0.86) 2 2.52 0.00 0.88 3.69 0.69 CWD (-0.60) 2 8.88 6.36 0.04 9.16 0.24 Winter %BGRD (0.93) 2 –0.58 0.00 0.66 2.37 0.84 CWD (-0.91) 2 1.11 1.69 0.28 3.02 0.79 402 Southeastern Naturalist Vol. 5, No. 3 explained a high amount of variation in both seasons (spring: 􀂨AICc = 1.53, R2 adj = 0.73; winter: 􀂨AICc = 3.62, R2 adj = 0.80; Table 5). In summer, variation in Dixie chub relative abundance was best modeled by a positive relationship with %BGRD (R2 adj = 0.83; Table 5). For spring and summer, at 5.0% catchment disturbance, the proportion of broadstripe shiners fell below the 95% confidence limit for the three least-disturbed streams, whereas this threshold occurred at 8.1% catchment disturbance for winter (Fig. 2, top 3 panels). Dixie chub relative abundance showed an opposite pattern, being above this threshold at 8.1% catchment disturbance for spring and summer and at 10.5% for winter (Fig. 2, bottom 3 panels) . Body size of broadstripe shiners and Dixie chubs Sizes of both broadstripe shiners and Dixie chubs were significantly different between streams in high- versus low-disturbance categories (Fig. 3). Mean SL of broadstripe shiners was smaller in high-disturbance streams (26.0 ± 1.35 mm) than low-disturbance streams (SL=37.9 ± 0.87 mm; 􀁲2 = 26.50, P < 0.0001), and mean SL of Dixie chubs followed the same trend (i.e., SL = 42.9 ± 1.32 mm vs. 77.7 ± 4.9 mm in high- vs. lowdisturbance streams, respectively; 􀁲2 = 35.18, P < 0.0001). Potential refuge areas for broadstripe shiners Study-site catchment-disturbance levels spanned a large portion of the range of disturbance in all 2nd-order catchments on FBMI (Fig. 4). Using Table 5. Best two models of multiple regression analysis on the relative abundance of broadstripe shiners and Dixie chubs (best models had smallest 􀂨AICc and largest wi). Abbreviations: CWD = relative abundance of coarse woody debris, Q = discharge, %BGRD = catchment disturbance (see text), L_Upatoi = linear distance to the Upatoi Creek mainstem, k = number of parameters in model, AICc = Akaike Information Criterion corrected for small sample size, 􀂨AICc = deviation in AICc from best model, wi = Akaike weights, SSE = sum of squares error. Numbers in parentheses denote standardized regression coefficients. CWD was not included in multiple regression models due to collinearity with other variables. Adjusted Variable Parameters in model k AICc 􀂨AICc wi SSE R2 % broadstripe shiner of total Spring %BGRD (-0.89) 2 -18.79 0.00 0.24 0.176 0.76 CWD (0.70) 2 -12.25 6.54 0.03 0.447 0.39 Summer %BGRD (-0.93) 2 -20.82 0.00 0.92 0.131 0.84 %BGRD (-0.89), 3 -14.08 6.75 0.03 0.127 0.80 L_Upatoi (-0.08) Winter %BGRD (-0.94) 2 -21.25 0.00 0.73 0.124 0.86 CWD (0.91) 2 -18.49 2.76 0.18 0.183 0.79 % Dixie chub of total Spring %BGRD (0.70), Q (-0.44) 3 -14.01 0.00 0.59 0.128 0.90 %BGRD (0.88) 2 -12.48 1.53 0.27 0.433 0.73 Summer %BGRD (0.93) 2 -18.13 0.00 0.84 0.193 0.83 %BGRD (0.85), Q (-0.23) 3 -13.96 4.17 0.10 0.129 0.86 Winter %BGRD (0.75), Q (-0.40) 3 -19.53 0.00 0.84 0.058 0.94 %BGRD (0.91) 2 -15.91 3.62 0.14 0.265 0.80 2006 K.O. Maloney, R.M. Mitchell, and J.W. Feminella 403 inferred disturbance-threshold levels from above, 177 (71%) catchments within Fort Benning fell below the lower threshold of 5% catchment disturbance, whereas 219 (88%) catchments were below the upper threshold of 8.1% catchment disturbance (Figs. 4 and 5). Discussion Catchment land use is often among the best predictors of fish-assemblage integrity, and measures such as the Index of Biotic Integrity (IBI), a multimetric index of assemblage structure based on relative abundance and functional group composition, often signal impairment from landscape disturbance (Allan et al. 1997, Karr 1991). Unfortunately, the low number of species we collected (10, mean 􀂧 3 species/stream) and high relative abundance of the broadstripe shiner and Dixie chub precluded our use of an IBI. However, our results support IBI predictions that percent tolerant individuals increases with increasing disturbance (Karr 1981, Schleiger 2000). Disturbance effects on broadstripe shiner and Dixie chub abundances Our findings suggest that the broadstripe shiner was negatively affected by catchment-scale disturbance, whereas the Dixie chub apparently Figure 2. Proportions of the broadstripe shiner (top 3 panels) and Dixie chub (bottom 3 panels) of total individuals collected plotted against catchment disturbance (%BGRD) for the seven study streams during Spring, Summer, and Winter 2003. Curved lines are 95% confidence intervals. Solid lines represent trends using means, dashed lines represent 95% upper and lower confidence limits of the three leastdisturbed study streams in the data set (BC2, KM1, LC), which was used to define a disturbance threshold (see Statistical Analyses subsection). 404 Southeastern Naturalist Vol. 5, No. 3 benefited from disturbance. Additionally, our results suggest that absolute abundance of both species was affected by local habitat and distance to the main-stem colonization source. We suggest that different life-history traits, feeding behaviors, and/or habitat requirements of the two species account for their opposite responses to catchment disturbance. For example, the requirement of vegetation and lack of parental care in the spawning behavior of the broadstripe shiner is a disadvantage when spawning over highly mobile streambeds, where risk of egg burial and associated mortality is high. However, the spawning behavior of the Dixie chub (i.e., greater degree of parental care and use of coarse sediments), may be less at risk in highly mobile stream beds. Hence, the reduced bed stability in more disturbed streams (Maloney et al. 2005) likely negatively affected the broadstripe shiner to a greater degree than the Dixie chub. Figure 3. Size-class frequency distributions of the Dixie chub (left 2 panels) and the broadstripe shiner (right 2 panels) between the three least-disturbed (top panels) and three most-disturbed (bottom panels) study streams. 2006 K.O. Maloney, R.M. Mitchell, and J.W. Feminella 405 Figure 4. Catchment disturbance (%BGRD) for all 2nd-order catchments within the Fort Benning Military Installation boundary (solid circles), ranked in order of increasing %BGRD. Study sites (open circles) are superimposed on the range of 2ndorder catchments. Shaded bar indicates range of potential disturbance thresholds, dashed lines indicate number of catchments below lower threshold level (5.0, n = 177 catchments) and the upper threshold level (8.1, n = 219). Disparate relationships between land use and the two species also may have resulted from dissimilar feeding behavior. In studies from other catchments strongly influenced by military training, fish assemblages were composed mostly of trophic generalists (Quist et al. 2003), patterns that appear to apply to fish assemblages found in Ft. Benning streams. The more selective drift-feeding behavior of broadstripe shiners may be more of a disadvantage than the generalist-feeding behavior of Dixie chubs in streams with highly mobile stream beds because of reduced foraging efficiency associated with increased suspended and deposited sediment (but see Gardner 1981, Ryan 1991). Total suspended solids (TSS) increases with catchment disturbance in these streams (Houser et al. 2006); however, how TSS affects foraging efficiency of both species needs further investigation. In terms of habitat, broadstripe shiners may be at a disadvantage in highdisturbed catchments because of reduced CWD and mesohabitat volumes in associated streams. Over the entire study, broadstripe shiner relative abundance was positively related to CWD (regression model, R2 = 0.67, F = 9.96, P = 0.03, 􀁠 = 0.82), suggesting that CWD is required for this species. Furthermore, absolute abundance of broadstripe shiners increased with in406 Southeastern Naturalist Vol. 5, No. 3 creasing discharge in each season. Relative to undisturbed sites, stream channels draining disturbed catchments at FBMI showed lower CWD abundance and bed stability (Maloney et al. 2005), which may have decreased pool depth while increasing current velocity (Angermeier and Karr 1984). In this context, fewer deep-pool/run habitats may constitute reduced habitat quantity and quality for broadstripe shiners. In contrast, Dixie chubs not only appeared to be less affected by reduced available habitat in disturbed streams, but their relative abundance decreased with increasing discharge in all seasons, and Dixie chub absolute abundance decreased with increasing discharge and proximity to Upatoi Creek in spring; taken together, these patterns suggest that Dixie chubs prefer smaller headwater streams (see also Schleiger 2000). Most streams on FBMI have intact riparian zones; however, we found that instream CWD is lower in highly disturbed streams (Maloney et al. 2005), possibly a result of historic land-use practices. One potential conservation strategy for broadstripe shiners would be to increase instream CWD to levels comparable to amounts in streams draining lowdisturbance catchments. Figure 5. Map of Fort Benning Military Installation showing the spatial arrangement of 2nd-order catchments classified as below (< 5% BGRD, n = 177), within (5–8.12%, n = 42), and above (> 8.12%, n = 30) identified disturbance thresholds. Dotted line represents the Chattahoochee River. 2006 K.O. Maloney, R.M. Mitchell, and J.W. Feminella 407 Surprisingly, distance to a potential colonizing source, Upatoi Creek, was only included in a weakly supported two-variable model explaining summer relative abundance of broadstripe shiners and the best model (together with discharge) for spring absolute abundance of Dixie chubs. Proximity to a larger system is often an important component in explaining variation in fish-assemblage structure (Gorman 1986, Osborne and Wiley 1992), which may explain the strong relationship with spring absolute abundance of Dixie chubs. Proximity to a colonization source may not have affected these two species to any great extent, but other species likely were affected. For example, 69% of southern brook lamprey (58 of 84), 55% of pirate perch (11 of 20), and 78% of redfin pickerel (seven of nine) were collected in the two streams (KM1 and LC) that drained directly into Upatoi Creek. The limited explanatory power of distance to Upatoi main stem that we observed also may be a result of position within the Upatoi drainage rather than just the nearest longitudinal distance to Upatoi Creek. Osborne and Wiley (1992) argued that position within the drainage is as important as distance from the main stem, with higher species richness occurring within streams located lower in the drainage. Our small sample size precluded any rigorous analysis on basin position, but the stream located second furthest upstream (LC) in our study had the highest species richness, whereas the one furthest downstream (HB) had the second highest richness for spring and summer, suggesting that position in drainage may be less important in structuring fish assemblages in this system. Dissimilar habitat requirements and life-history traits may account for lower relative and absolute abundances of broadstripe shiners and the higher relative and absolute abundances of Dixie chubs in high-disturbance streams, but they do not account for low relative and absolute abundances of Dixie chubs in low-disturbance streams. Juvenile Dixie chubs (20–40 mm SL) may consume similar prey as adult broadstripe shiners and both prefer similar habitats (Barber and Minckley 1971, Ross et al. 2001), so it is possible that juvenile Dixie chubs are competitively inferior to adult broadstripe shiners, and are displaced from sections where broadstripe shiners occur (i.e., low-disturbance streams). The absence of many competitors in high-disturbance streams may benefit tolerant Dixie chubs, a pattern observed in other tolerant/pioneer species (Byers 2002, McAuliffe 1984, Resh et al. 1988). Disturbance effects on broadstripe shiner and Dixie chub body size Differences in size distributions between the most- and least-disturbed streams for both species, with significantly smaller individuals in the mostdisturbed streams, provides some evidence that both species were negatively affected by catchment disturbance at FBMI. The presence of juveniles of both broadstripe shiners and Dixie chubs suggests that both species were capable of reproducing and recruiting in disturbed streams. However, relative to the least-disturbed streams, both species had high mortality, and/or emigration. Either or both mechanisms could result from decreased adult 408 Southeastern Naturalist Vol. 5, No. 3 habitat (cover, spawning habitat) and/or adult food availability attributable to increased sediment disturbance (Angermeier and Karr 1984, Ritchie 1972, Ryan 1991, Sutherland et al. 2002). Decreased habitat availability appears particularly likely in this regard, as available habitat in terms of pool size and abundance of CWD (resting habitat, potential refugia from predation) decreased with increasing catchment disturbance. Moreover, in our sites, stream flashiness (i.e., rapid fluctuations in stream stage in response to storms) increased with catchment disturbance (Maloney et al. 2005), which may have further decreased available habitat in the most-disturbed streams. Low numbers of small individuals of both species in the least-disturbed streams also may have skewed both species to a smaller average size in most-disturbed streams. One reason for low abundance of small individuals in these streams may have been sampling inefficiency; however, this factor is unlikely because numerous small individuals were collected in the mostdisturbed streams. A second reason for this pattern is that habitat for small fish may have been limited in the least-disturbed streams. This explanation is unlikely because these streams had higher amounts of CWD, mesohabitat volumes, and more stable beds than the most-disturbed streams. A third reason may be predation by larger fish. The three least-disturbed streams had higher abundances of larger fish predators (two centrarchids and seven redfin pickerel) than the three most-disturbed streams (no centrarchids, one pickerel). Therefore, it is possible that the least-disturbed streams had higher predation on smaller individuals than the most-disturbed streams; however, predation has not been quantified in these systems. Disturbance thresholds and potential refuge areas at Fort Benning We also observed a potential disturbance threshold at 􀂧 5–8.1% of the catchment area as bare ground on slopes > 5% and unpaved roads. Of the 249 second-order catchments on FBMI, 177 (71%) have disturbance levels below the 5% disturbance threshold, and 219 (88%) have levels below the 8.1% BGRD threshold, suggesting a potential for many suitable locations for broadstripe shiners. A potential conservation strategy in streams at FBMI that may also be applicable to other low-gradient southeastern streams, is to limit the amount of catchment disturbance to levels that remain below apparent thresholds. Our GIS-based predictive-modeling approach enabled a rapid assessment of potential suitable catchments for broadstripe shiners; however, such an approach may fail to identify already occupied systems or additional potential catchments if models are incorrectly parameterized. As such, we caution sole use of GIS-based predictive-models in developing conservation plans. These approaches should be used in tandem with in-depth quantitative surveys of target populations to fully protect sensitive species. The conservation of southeastern fishes is an important issue because of the high degree of diversity and endemism and high amount of impaired streams from extensive historic and contemporary catchment disturbance in the Southeast. As human demands on private lands increase, the role of 2006 K.O. Maloney, R.M. Mitchell, and J.W. Feminella 409 public lands, such as military bases, as reserves will become increasingly important. Our study of the rare broadstripe shiner at FBMI suggests that the installation may be a refuge for this species because it is found in a large proportion of minimally disturbed catchments (Fig. 5). However, successful conservation of sensitive metapopulations requires colonized patches to be connected by habitat corridors (Gonzalez et al. 1998); therefore, we recommend that FBMI maintain high streamwater quality in Upatoi Creek to allow passage of broadstripe shiners to small headwater streams. Additional conservation strategies for the broadstripe shiner, which may apply to other vulnerable headwater-stream species that require further study are to: 1) limit catchment-scale disturbance to levels below identified thresholds; 2) reduce incoming sediment from ephemeral channels; and 3) restore limiting habitat resources (e.g., bed stability, CWD) necessary for population sustainability. Acknowledgments We thank personnel at the Fort Benning Military Installation for access to the study sites, particularly Hugh Westbury, SEMP Host Site Coordinator. We also thank Lisa Olsen and Virginia Dale for initial classification of Landsat imagery; Michael Buntin, Brian Helms, Steve Herrington, and Adriene Burnette for field assistance; and Brian Helms, Dennis DeVries, Carol Johnston, and two anonymous reviewers for comments on the manuscript. The project was supported by the US Department of Defense Strategic Environmental Research and Development Program (SERDP) Ecosystem Management Program (SEMP), projects CS-1114C and CS-1186 to Oak Ridge National Laboratory, and by the Auburn University Center for Forest Sustainability Peaks of Excellence Program. Oak Ridge National Laboratory is managed by the University of Tennessee-Battelle, LLC for the US Department of Energy under contract DE-AC05-00OR22725. Literature Cited Allan, J.D., D.L. Erickson, and J. Fay. 1997. The influence of catchment land use on stream integrity across multiple spatial scales. Freshwater Biology 37:149–161. Angermeier, P.L., and J.R. Karr. 1984. Relationships between woody debris and fish habitat in a small warmwater stream. Transactions of the American Fisheries Society 113:716–726. Auld, A.H., and J.R. Schubel. 1978. Effects of suspended sediment on fish eggs and larvae: A laboratory assessment. Estuarine and Coastal Marine Science 6:153–164. Barber, W.E., and W.L. Minckley. 1971. Summer foods of the cyprinid fish Semotilus atromaculatus. Transactions of the American Fisheries Society 100:283–289. Baxter, R.M. 1977. Environmental effects of dams and impoundments. Annual Review of Ecology and Systematics 8:255–283. Bergstedt, L.C., and E.P. Bergersen. 1997. Health and movements of fish in response to sediment sluicing in the Wind River, Wyoming. Canadian Journal of Fisheries and Aquatic Sciences 54:312–319. Berkman, H.E., and C.F. Rabeni. 1987. Effect of siltation on stream fish communities. Environmental Biology of Fishes 18:285–294. 410 Southeastern Naturalist Vol. 5, No. 3 Boschung, Jr., H.T., and R.L. Mayden. 2004. Fishes of Alabama. Smithsonian Books, Washington, DC. 736 pp. Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. 2nd Edition. Springer Verlag, New York, NY. 488 pp. Byers, J.E. 2002. Impact of non-indigenous species on natives enhanced by anthropogenic alteration of selection regimes. Oikos 97:449–458. Cohn, J.P. 1996. New defenders of wildlife. BioScience 46:11–14. Cordone, A.J., and D.W. Kelley. 1961. The influences of inorganic sediment on the aquatic life of streams. California Fish and Game 47:189–228. Gardner, M.B. 1981. Effects of turbidity on feeding rates and selectivity of bluegills. Transactions of the American Fisheries Society 110:446–450. Gonzalez, A., J.H. Lawton, F.S. Gilbert, T.M. Blackburn, and I. Evans-Freke. 1998. Metapopulation dynamics, abundance, and distribution in a microecosystem. Science 281:2045–2047. Goodmann, S.W. 1996. Ecosystem management at the Department of Defense. Ecological Applications 6:706–707. Gore, J.A. 1996. Discharge measurements and streamflow analysis. Pp. 53–75, In F.R. Hauer and G.A. Lamberti (Eds.). Methods in Stream Ecology. Academic Press, New York, NY. 674 pp. Gorman, O.T. 1986. Assemblage organization of stream fishes: The effect of rivers on adventitious streams. American Naturalist 128:611–616. Griffith, G.E., J.M. Omernik, J.A. Comstock, S. Lawrence, G. Martin, A. Goddard, V.J. Hulcher, and T. Foster. 2001. Ecoregions of Alabama and Georgia, (color poster with map, descriptive text, summary tables, and photographs). US Geological Survey (map scale 1:1,700,000), Reston, VA. Houser, J.N., P.J. Mulholland, and K.O. Maloney. 2006. Upland disturbance affects headwater-stream nutrients and suspended sediments during baseflow and stormflow. Journal of Environmental Quality 35:352–365. Jackson, D.A., P.R. Peres-Neto, and J.D. Olden. 2001. What controls who is where in freshwater fish communities: The roles of biotic, abiotic, and spatial factors. Canadian Journal of Fisheries and Aquatic Sciences 58:157–170. Johnston, C.E., and J.S. Ramsey. 1990. Redescription of Semotilus thoreauianus Jordan, 1877, a cyprinid fish of the southeastern United States. Copeia 1:119–130. Karr, J.R. 1981. Assessment of biotic integrity using fish communities. Fisheries 6:21–27. Karr, J.R. 1991. Biological integrity: A long-neglected aspect of water resource management. Ecological Applications 1:66–84. Katula, B. 1993. Spawning a “winged minnow”—the broadstripe shiner. American Currents (Spring):20–21, 30. Lenat, D.R., and J.K. Crawford. 1994. Effects of land use on water quality and aquatic biota of three North Carolina Piedmont streams. Hydrobiologia 294:185–199. Maloney, K.O., P.J. Mulholland, and J.W. Feminella. 2005. Influence of catchmentscale military land use on physical and organic-matter conditions in small Southeastern Plains streams (USA). Environmental Management 35:677–691. Marsh-Matthews, E., and W.J. Matthews. 2000. Geographic, terrestrial, and aquatic factors: Which most influence the structure of stream fish assemblages in the midwestern United States? Ecology of Freshwater Fish 9:9–21. 2006 K.O. Maloney, R.M. Mitchell, and J.W. Feminella 411 Matthews, W.J., and H.W. Robison. 1998. Influence of drainage connectivity, drainage area, and regional species richness on fishes of the Interior Highlands in Arkansas. American Midland Naturalist 139:1–19. Maurakis, E.G., M.H. Sabaj, and W.S. Woolcott. 1993. Pebble-nest construction and spawning behaviors in Semotilus thoreauianus (Pisces: Cyprinidae). Association of Southeastern Biologist Bulletin 40:27–30. McAuliffe, J.R. 1984. Competition for space, disturbance, and the structure of a benthic stream community. Ecology 65:894–908. Mettee, M.F., P.E. O’Neil, and J.M. Pierson. 1996. Fishes of Alabama and the Mobile Basin. Oxmoor House, Birmingham, AL. 820 pp. Mills, E.L., J.H. Leach, J.T. Carlton, and C.L. Secor. 1994. Exotic species and the integrity of the Great Lakes. BioScience 44:666–676. Moyle, P.B. 1999. Effects of invading species on freshwater and estuarine ecosystems. Pp. 177–191, In O.T. Sandlund, P.J. Schei, and Å. Viken (Eds.). Invasive Species and Biodiversity Management. Kluwer Academic, Boston, MA. 448 pp. Myers, R.H. 1990. Classical and Modern Regression with Applications. 2nd Edition. Duxbury Press, Belmont, CA. 488 pp. Omernik, J.M. 1987. Ecoregions of the conterminous United States. Annals of the Association of American Geographers 77:118–125. Osborne, L.L., and M.J. Wiley. 1992. Influence of tributary spatial position on the structure of warmwater fish communities. Canadian Journal of Fisheries and Aquatic Sciences 49:671–681. Quist, M.C., P.A. Fay, C.S. Guy, A.K. Knapp, and B.N. Rubenstein. 2003. Military training effects on terrestrial and aquatic communities on a grassland military installation. Ecological Applications 13:432–442. Rahel, F.J. 2000. Homogenization of fish faunas across the United States. Science 288:854–856. Resh, V.H., A.V. Brown, A.P. Covich, M.E. Gurtz, H.W. Li, G.W. Minshall, S.R. Reice, A.L. Sheldon, J.B. Wallace, and R.C. Wissmar. 1988. The role of disturbance in stream ecology. Journal of the North American Benthological Society 7:433–455. Ritchie, J.C. 1972. Sediment, fish, and fish habitat. Journal of Soil and Water Conservation 27:124–125. Ross, S.T., W.M. Brenneman, W.T. Slack, M.T. O’Connell, and T.L. Peterson. 2001. The Inland Fishes of Mississippi. University Press of Mississippi. Jackson, MS. 624 pp. Ryan, P.A. 1991. Environmental effects of sediment on New Zealand streams. New Zealand Journal of Marine and Freshwater Research 25:207–221. Schleiger, S.L. 2000. Use of an index of biotic integrity to detect effects of land uses on stream fish communities in west-central Georgia. Transactions of the American Fisheries Society 129:1118–1133. Schlosser, I.J. 1987. The role of predation in age- and size-related habitat use by stream fishes. Ecology 68:651–659. Scott, M.C., and G.S. Helfman. 2001. Native invasions, homogenization, and the mismeasure of integrity of fish assemblages. Fisheries 26:6–15. Seber, G.A.F. 1982. The Estimation of Animal Abundance. 2nd Edition. MacMillan Publications, New York, NY. 654 pp. Snyder, C.D., J.A. Young, R. Villella, and D.P. Lemarié. 2003. Influences of upland and riparian land-use patterns on stream biotic integrity. Landscape Ecology 18:647–664. 412 Southeastern Naturalist Vol. 5, No. 3 Sutherland, A.B., J.L. Meyer, and E.P. Gardiner. 2002. Effects of land cover on sediment regime and fish-assemblage structure in four southern Appalachian streams. Freshwater Biology 47:1791–1805. Suttkus, R.D. 1955. Notropis euryzonus, a new cyprinid fish from the Chattahoochee River system of Georgia and Alabama. Tulane Studies in Zoology 3:85–100. Wallace, J.B., and A.C. Benke. 1984. Quantification of wood habitat in subtropical Coastal Plain streams. Canadian Journal of Fisheries and Aquatic Sciences 41:1643–1652. Wang, L., J. Lyons, P. Kanehl, and R. Bannerman. 2001. Impacts of urbanization on stream habitat and fish across multiple spatial scales. Environmental Management 28:255–266. Warren, M.L., B.M. Burr, S.J. Walsh, H.L. Bart, R.C. Cashner, D.A. Etnier, B.J. Freeman, B.R. Kuhajda, R.L. Mayden, H.W. Robison, S.T. Ross, and W.C. Starnes. 2000. Diversity, distribution, and conservation status of the native freshwater fishes of the southern United States. Fisheries 25:7–31. Winston, M.R., C.M. Taylor, and J. Pigg. 1991. Upstream extirpation of four minnow species due to damming of a prairie stream. Transactions of the American Fisheries Society 120:98–105. Zar, J.H. 1999. Biostatistical Analysis. 4th Edition. Prentice-Hall, Upper Saddle River, NJ. 663 pp.