Fish assemblage collections were assessed in sub-watersheds containing various land use types to relate overall biotic condition to patterns of watershed disturbance and to the influence of introduced fishes. Four hundred historical and contemporary fish collections were analyzed in conjunction with land use and land cover data among 200, first- through fourth-order sub-watersheds across Pennsylvania. A high degree of collinearity among several land use types and the presence of non-indigenous fishes was determined, indicating multiple anthropogenic influences. Significant relationships were found between various land use types and the Jaccard coefficient of similarity, the primary index used to evaluate changes between fish collections, across all major drainages. In general, the Jaccard coefficient of similarity increased to a value of one as the percentage of forested land increased, while the Jaccard coefficient of similarity decreased as percentages of agricultural and urban land increased. Sub- watersheds that showed declines in species richness had significantly higher percentages of agricultural and developed land, while those sub-watersheds with higher amounts of forested land appeared relatively stable. Seventy-percent of those species that experienced declines were either insectivores or benthic insectivores. The largest increases among the fish guilds occurred within the insectivore-piscivore group. We conclude that land use influences at the sub-watershed scale and the introduction of non-indigenous fishes have had a significant influence on fish community composition.

Introduction

Historically, the distribution and abundance of fishes was influenced by a variety of geologic, physical, and biological factors. The formation of rivers and streams thousands of years ago initially guided the fish distribution patterns that we observe today (Hocutt and Wiley, 1986). Since that time, man has altered both the terrestrial and aquatic landscape to meet his needs: building roads, dredging waterways for travel and transport, and exploiting fisheries resources for food and recreation.

Left in a natural state, the individual components of the landscape change over time through the process of succession, but the overall structural and functional relationships among the components remains fundamentally the same (Schlosser, 1991). Alterations to these landscape components including agriculture, deforestation, and urbanization, shift these structural and functional relationships among landscape elements, thereby increasing sediment delivery to streams, altering hydrologic and thermal regimes (Fleischner, 1994), influencing water chemistry (Johnson et al., 1997), and modifying the aquatic food web (Klein, 1979; Schlosser, 1995). Such modifications have been cited as contributing to the extirpation of 73% of North American fish species (Miller et al., 1989) and to the decline of some eastern North American fishes (Fairchild et al., 1998).

Fish are useful for assessing the relative health of aquatic systems for several reasons. First, fish are sensitive to anthropogenic influences. We can monitor their movements from degraded to undegraded areas (Gagen et al., 1993), we can monitor their mortality in response to stress (Adams et al., 1996), and we can monitor them for disease and parasites (Moyle et al., 1986; Landsberg et al., 1998). Second, fish exhibit a variety of life history patterns. They occupy varied habitats (Aadland, 1993) and because some are quite long lived, their communities can show chronic effects of long-term environmental stress (Karr, 1981). Third, fish occupy a variety of trophic levels, they spawn on a variety of substrate types, and they respire through water. Thus, fish communities are structured by the direct and indirect effects of stress on the entire aquatic ecosystem and depend upon various environmental conditions over broad spatial areas for their survival (Plafkin et al., 1989; Fausch et al., 1990).

Documenting changes in fish assemblages among streams can provide important information regarding water resource quality and the biotic integrity of freshwater systems. The index of biotic integrity (IBI) is one such measure of anthropogenic stress (Karr, 1981). The IBI provides a framework for comparison among degraded stream systems and a reference condition using various fish community metrics to describe the aquatic environment from a one-time sampling period. The individual descriptors that comprise this index have been modified from the original and applied to regional analyses involving land use and land cover alterations, but with varied results (Lenat and Crawford, 1994; Roth et al., 1996; Lammert and Allan, 1999).

Prior studies investigating the influence of non-point source pollution, land-use impacts, and fish distribution have focused on stream bank and riparian areas (Armour et al., 1991; Wohl and Carline, 1996). Karr and Schlosser (1978) reviewed the critical role of riparian areas in determining stream habitat and the value of streamside buffers. As important as riparian areas are to the maintenance and integrity of streams (Wohl and Carline, 1996), most agricultural and urban land uses occur at a larger watershed scale and their impacts cannot be fully understood by looking only at adjacent riparian lands (Wang et al., 1997). Very few studies have investigated watershed-wide effects of land use on stream ecosystems (Wang et al., 1997), and many of those have focused on land-use relationships within a single watershed or groups of small neighboring watersheds (Lenat and Crawford, 1994; Weaver and Garman, 1994; Roth et al., 1996), making generalizations at the regional level difficult. With the advent of geographic information systems (GIS), studies of relationships among watershed attributes and stream ecosystem characteristics are more easily facilitated. In this study, we examine the relationships between watershed land use characteristics and changes in fish community structure among selected sub-watersheds across Pennsylvania.

The objectives of this study were (1) to determine if significant relationships exist among changes in fish community similarity and land use and land cover patterns, (2) to assess relationships between altered landscapes (i.e., urbanized and agriculture watersheds) and fish introductions, and (3) to determine if significant land use differences exist among watersheds with declining fish populations.

Methodology

Fish data and associated descriptors

Two separate data sets were used for analysis. The first was an historical data set, derived from Dr. Edwin Cooper's collections made between 1950 and 1974 (Cooper, 1983). The second data set was derived from contemporary collections made by the Pennsylvania Fish and Boat Commission (PFBC) and Dr. Jay R. Stauffer, Jr. between 1975 and 1995. These databases were selected because of their relatively widespread coverage throughout Pennsylvania, facilitating a large sample size for study. We scrutinized these databases for accuracy of species identification and site locality. We chose to remove or correct questionable data to ensure that collection events were accurately represented (Argent et al., 1997).

Data used for analysis were screened to be as similar as possible with respect to collection effort, sampling gear, season, and water chemistry (Table 1). No collections were included for analysis in which rotenone was used to sample fish populations because none of the contemporary data sets used this collection method. Surveys included for analysis contained stream reaches sampled using only electrofishing gear. Fishes collected by Cooper (1983) were generally sampled with two to three passes at each site, identified on site, and voucher specimens were retained. Stauffer collected fishes by sampling all available habitats until he thought that further effort would not produce additional species (J. Stauffer, Jr., The Pennsylvania State University, pers. comm.) and voucher collections were retained. The PFBC identified specimens on site to the species level and returned to them to the stream; therefore no voucher specimens exist for these sampling events.

Table 1.

Mean values for historic (Hist.) and contemporary (Cont.) fish collections. Parenthetic values denote total number of paired sites for which water chemistry data were available. An asterisk denotes a significant difference at the α = 0.05 level. No tests of significance were performed for the temperature variable in the Ohio River drainage because sample sizes were too low to perform a statistical comparison

 Environmental Variables 
 Alkalinity Mg l−1 CaCO3 pH Conductivity μs cm−1 Temperature C° 
Drainage Hist. Cont. Hist. Cont. Hist. Cont. Hist. Cont. 
Delaware 39.5 38.3 (24) 7.1 7.3 (25) 131.8 173.7 (24) 22.1 19.8 (13)* 
Susquehanna 33.8 36.2 (41) 7.3 7.2 (42) 110.5 124.1 (32) 17.5 16.2 (18) 
Ohio 45.8 50.1 (44) 7.3 7.3 (44) 241.4 261.3 (42) 22.3 17.4 (4) 
 Environmental Variables 
 Alkalinity Mg l−1 CaCO3 pH Conductivity μs cm−1 Temperature C° 
Drainage Hist. Cont. Hist. Cont. Hist. Cont. Hist. Cont. 
Delaware 39.5 38.3 (24) 7.1 7.3 (25) 131.8 173.7 (24) 22.1 19.8 (13)* 
Susquehanna 33.8 36.2 (41) 7.3 7.2 (42) 110.5 124.1 (32) 17.5 16.2 (18) 
Ohio 45.8 50.1 (44) 7.3 7.3 (44) 241.4 261.3 (42) 22.3 17.4 (4) 

Using these data sets, we calculated the Jaccard coefficient of similarity (JCS; where JCS = a/(a + b + c)) (Krebs, 1989) to compare the historical (1950–1974) collections to the contemporary (1975–1995) collections, where a = number of species found at both historical and contemporary sites, b = number of species found only at historical sites, and c = number of species found only at contemporary sites. Traditionally, this binary similarity coefficient is used to measure the similarity between two sites, usually a reference or unimpaired site and a site of interest or impaired site. But the JCS can also be used to assess changes in species composition between two sites by providing one metric that reflects changes in composition through time (Hansen and Ramm, 1994). Unlike other similarity coefficients, the JCS does not weight the presence or absence of a species and is therefore useful for the discrimination of rare occurrences (Jackson et al., 1989).

In addition to the JCS, we calculated measures of community and guild structure using the ‘Revision to Rapid Bioassessment Protocols for Use in Streams and Rivers’ guidelines (Barbour et al., 1997). The resulting classification provided a characterization of the regional biotic attributes of fish communities (Table 2). The ‘Protocol’ is based in part on the Index of Biological Integrity (IBI) (Karr, 1981; Karr et al., 1986), Gammon's (1980) Index of Well Being, standard fish taxonomy texts, and literature describing certain modifications for implementation in the northeast geographical region (Barbour et al., 1997).

Table 2.

Fish community descriptors selected to characterize community structure expressed as total number within each time period or the difference between each time period. Descriptors were derived from Barbour et al. (1997) 

Descriptors Interpretation 
Native fishes Higher species richness indicates higher water quality 
Introduced fishes Reflects direct and indirect human alteration 
Species lost Reflects decline in habitat quality or predation 
Intolerants Indicator of good water quality 
Omnivores Indicator of physical and chemical degradation 
Herbivores Indicator of aquatic vegetation community 
Filter-feeders Indicator of drift invertebrate community 
Insectivores Indicator of diverse macroinvertebrate communities 
Benthic insectivores Indicator of diverse benthic macroinvertebrate communities 
Insectivore-piscivores Indicator of high trophic level community stability 
Descriptors Interpretation 
Native fishes Higher species richness indicates higher water quality 
Introduced fishes Reflects direct and indirect human alteration 
Species lost Reflects decline in habitat quality or predation 
Intolerants Indicator of good water quality 
Omnivores Indicator of physical and chemical degradation 
Herbivores Indicator of aquatic vegetation community 
Filter-feeders Indicator of drift invertebrate community 
Insectivores Indicator of diverse macroinvertebrate communities 
Benthic insectivores Indicator of diverse benthic macroinvertebrate communities 
Insectivore-piscivores Indicator of high trophic level community stability 

Because the PFBC data set lacked abundance information, community descriptors had to be developed using presence data only. Therefore, a traditional IBI could not be developed, but many of its component descriptors were. These descriptors are intended to provide information regarding the quality of the aquatic environment (Karr et al., 1986).

Site selection

We used a watersheds GIS-data layer containing 9,854 individual polygons (sub-watersheds) to select appropriate areas for analysis. These sub-watersheds were derived from stream drainage boundaries present on United States Geological Survey (USGS)7.5″ quadrangle maps, that is, a 1:24,000 scale. Sub-watershed boundaries represent the surface drainage area for a particular named stream or stream section between tributaries that feed a larger stream. After the fish data had been assembled, sub-watersheds were selected in the areas that encompassed a sampling location. The initial selection yielded 200sub-watersheds of which there were 35 in the Delaware, 93 in the Susquehanna, and 72 in the Ohio river drainages. But this selection included only those areas containing a fish sampling location and did not include land upstream of a sampling location. Subsequently, additional sub-watershed boundaries were obtained to reflect the influence of up-stream processes. In the Delaware, Susquehanna, and Ohio river drainages, an additional 67, 373, and 257 sub-watersheds were selected from the watersheds GIS-data layer, respectively (Figure 1). Each sub-watershed where a sampling event occurred was classified by stream order (Strahler, 1957). The final data set included land use and land cover data from 897 sub-watershed sections, ranging in area from 7.7 to 103.7 km2 and in size from first through fourth order.

Figure 1.

Selected sub-watersheds used to relate land use with fish collections. Small map indicates location of Pennsylvania relative to other states in the northeastern United States.

Figure 1.

Selected sub-watersheds used to relate land use with fish collections. Small map indicates location of Pennsylvania relative to other states in the northeastern United States.

Land use and land cover data

We obtained land use and land cover data from the multi-resolution land characteristic (MRLC) data set (Table 3). The data set provides information at a 30-m spatial resolution using Landsat thematic mapper (TM) images stored in Arc/Info® GIS software. The MRLC consortium members used the USGS Anderson System, Gap Terrestrial Natural Vegetation classification, and the C-CAP land cover classification to develop this data set (Anderson et al., 1976; Jennings, 1993; Dobson et al., 1995). For Pennsylvania, the primary data source was leaves-on (summer) Landsat TM imagery acquired in 1991, 1992, and 1993 (Loveland and Shaw, 1996). Additionally, leaves-off TM data sets were acquired and referenced. We selected MRLC land use and land cover data using the sub-watershed boundaries we derived in Arc/Info® to obtain spatially explicit land use and land cover data about each fish collection site.

Table 3.

Land use and land cover classifications from the EPA Region III multi-resolution land characterization1 land cover data set

Land Cover Type2 Cover Type Description 
Water Areas of open water, with less than 30% vegetative cover 
Low intensity developed High percentage of residential development typifies this class 
High intensity developed High percentage of building materials and industrial development 
Hay/pasture/grass High percentages of grasses and vegetation that is regularly mowed for hay and/or grazed by livestock; predominantly hay fields and pastures, but also currently includes golf courses and city parks 
Row crops Areas regularly tilled and planted, often on an annual or biennial 
Probable row crops Grasslands that were not green during spring data acquisitions 
Conifer (evergreen) forest Areas comprised of 70% or higher coniferous species 
Mixed forest Areas comprised of both coniferous (< 70%) and deciduous (< 70%) trees 
Deciduous forest Areas comprised of 70% or higher deciduous species 
Woody wetlands (from NWI3Wetlands with substantial amount of woody vegetation present 
Emergent wetlands Wetlands without a substantial amount of woody vegetation 
Barren: quarry areas Includes all quarry areas (including sand/gravel operations) 
Barren: coal mines Region dominated by spectrally dark coal piles and strip mines 
Barren: beach areas Beaches 
Barren: transitional Areas likely to change to other land cover types, e.g., clear cuts 
 

2For data analysis % agriculture comprised hay/pasture/grass, row crops, and probable row crops; % developed comprised high and low intensity development; and % forested comprised coniferous forest, mixed forest, deciduous forest, and woody wetlands.

 

3NWI: National Wetlands Inventory.

 
Land Cover Type2 Cover Type Description 
Water Areas of open water, with less than 30% vegetative cover 
Low intensity developed High percentage of residential development typifies this class 
High intensity developed High percentage of building materials and industrial development 
Hay/pasture/grass High percentages of grasses and vegetation that is regularly mowed for hay and/or grazed by livestock; predominantly hay fields and pastures, but also currently includes golf courses and city parks 
Row crops Areas regularly tilled and planted, often on an annual or biennial 
Probable row crops Grasslands that were not green during spring data acquisitions 
Conifer (evergreen) forest Areas comprised of 70% or higher coniferous species 
Mixed forest Areas comprised of both coniferous (< 70%) and deciduous (< 70%) trees 
Deciduous forest Areas comprised of 70% or higher deciduous species 
Woody wetlands (from NWI3Wetlands with substantial amount of woody vegetation present 
Emergent wetlands Wetlands without a substantial amount of woody vegetation 
Barren: quarry areas Includes all quarry areas (including sand/gravel operations) 
Barren: coal mines Region dominated by spectrally dark coal piles and strip mines 
Barren: beach areas Beaches 
Barren: transitional Areas likely to change to other land cover types, e.g., clear cuts 
 

2For data analysis % agriculture comprised hay/pasture/grass, row crops, and probable row crops; % developed comprised high and low intensity development; and % forested comprised coniferous forest, mixed forest, deciduous forest, and woody wetlands.

 

3NWI: National Wetlands Inventory.

 

We determined land use and land cover values for each sampled sub-watershed and its associated up-stream area for each of the 14 distinct land use types, as the percentage of total catchment area. The area occupied by water, was not included in percentage calculations because it does not reflect a terrestrial landscape component. In addition, total percent of agriculture (hay/pasture/grass; row crops; and probable row crops), developed (high and low intensity), and forested (coniferous forest; mixed forest; deciduous forest; and woody wetlands) land was determined by summing discrete land use types together and calculating a percentage of total catchment area.

Data analyses

Regression analysis was used to determine if significant relationships existed between the JCS and each land use and land cover type (including composite land use and land cover types: total % agriculture, total % development, and total % forest). We used Pearson correlation analysis to determine if significant relationships existed among the various land use categories. Tests for homogeneity of variance and normality revealed that distribution-free statistics (non-parametric statistics) should be used for other statistical comparisons. For comparisons among land use and land cover variables and derived fish descriptors, the Kruskal-Wallis test was used. If a significant difference was found, the non-parametric multiple-comparison described by Dunn (1964) was performed. A Wilcoxon-signed test was used to determine if water quality variables were significantly different from one another. All statistical analyses were performed using JMP IN® software v. 3.1.2 (SAS, 1998) or MINITAB® Release 11 for Windows (Minitab, 1996) at a significance level of 0.05.

Results

Land-use characteristics

Among the sub-watersheds selected for study, land use and land cover characteristics seemed quite similar among Pennsylvania's major drainage basins with a few exceptions. Across all stream sizes, sub-watersheds within the Delaware River drainage had significantly more developed land, more wetlands, and less forested land than did either the Susquehanna or Ohio river drainage watersheds (p < 0.05). The Ohio and Susquehanna drainage watersheds contained similar proportions of all land use types and were not significantly different from one another (p > 0.05).

There was a high degree of collinearity among the different land use classification types within each major drainage basin (Argent, 2000). For example, the percentages of agricultural land were positively and significantly (p < 0.05) correlated with the percentages of developed land. In general, most of the forested land use types were negatively correlated with most of the altered (i.e., agricultural or developed) land use types. The only exception was the percent high-intensity development and the percent mixed forest within the Ohio River drainage. These results suggest that many watersheds contain land having both developed and agricultural areas, but not forested areas, and land that contains multiple forest types.

Land use values differed not only among major drainage basins but also across differing stream sizes. Overall, the highest percentages of forested lands were found in first order watersheds with declining amounts as stream sizes increased. On average, the percent of developed land was highest within second order watersheds followed by fourth order watersheds. With respect to agricultural land, the highest amounts were found in the third and fourth order watersheds. Within the Susquehanna River drainage there was a significant difference among the land use types across all stream sizes (p < 0.05). Within the Ohio River drainage, only developed land was significantly different among the four stream sizes (p < 0.05), with the highest proportion in second order streams. No tests of significance were performed for the Delaware River drainage because of low sample sizes of first order streams (n = 2).

Fish community structure

A total of 32, 46, and 46 fish species were included from the historical data set and 42, 53, and 56 fish species were included from the contemporary data set among sub-watersheds in the Delaware, Susquehanna, and Ohio river drainages, respectively (Appendix D in Argent, 2000). The least brook lamprey (Lampetra aepyptera), striped shiner (Luxilus chrysocephalus), Southern redbelly dace (Phoxinus erythrogaster), fathead minnow (Pimephales promelas), bigeye chub (Notropis amblops), sand shiner (Notropis ludabundus), black redhorse (Moxostoma duquesnei), and logperch (Percina caprodes) were all absent in the contemporary collections; with the exception of the striped shiner, these species occurred at less than five sites, historically. The JCS averaged 0.49, 0.66, and 0.52 among the Delaware, Susquehanna, and Ohio river drainages, respectively and these differences were significant from a JCS = 1.0 (p < 0.05). This indicates that all fish communities changed, but that fish communities in the Susquehanna drainage changed the least between historic and contemporary collections and those in the Delaware drainage changed the most.

A total of 7, 10, and 3 fishes were classified as introduced (non-indigenous or transplanted) from the historical data set and 9, 14, and 3 fishes were classified as introduced from the contemporary data set among the Delaware, Susquehanna, and Ohio river drainages, respectively (Argent, 2000). Fish introductions typically included salmonids stocked for recreational purposes and inter-drainage transfers such as smallmouth bass (Micropterus dolomieu). Comparing the JCS, calculated with and without introduced fishes, indicated that fish communities of the Delaware River drainage have been influenced more by fish introductions (R2 = 0.83) than either the Susquehanna (R2 = 0.86) or the Ohio river drainage (R2 = 0.90). A test of the null hypothesis that the slopes of the regression lines equal one yielded no significant differences (p > 0.05; Ott, 1988), indicating that exotic and transplanted fishes had not significantly altered fish community composition as measured with the JCS.

Overall, 72% of the selected watersheds had JCS values greater than 0.5, but only 5% of the watersheds had JCS values equal to one. The presence of species between historic and contemporary collections among the 200 watersheds was positively correlated (r = 0.94, p < 0.05), suggesting that species' distributions among all watersheds did not change substantially. But among 87 species collected within the three drainages, 45 experienced declines (i.e., occurred in fewer watersheds) while 42 experienced no change or an increase in distribution (Argent, 2000). Species which increased the most in distribution between the two time periods largely included insectivore-piscivore and introduced species (Figure 2). By contrast, 26 of the 45 fishes that experienced declines were insectivores, while eight were benthic insectivores. Significant differences existed between the number of declining insectivores and benthic insectivores observed and the expected numbers of both groups (Chi-square; p < 0.05). This result suggests that these feeding guilds declined non-randomly in response to something other than natural causes. Significant declines occurred between the historic and contemporary collections among all fish descriptors with the exception of introduced fishes and insectivore-piscivore fishes, which experienced significant increases (p > 0.05). Among the fish descriptors derived for each major drainage, there was a significant correlation between introduced and insectivore-piscivore fishes (r = 0.60, p > 0.05). Filter feeders, as represented by three lamprey species, were found only in the Ohio River drainage watersheds.

Figure 2.

Selected species whose distribution increased the most between historic and contemporary collections among the 200-selected study sub-watersheds.

Figure 2.

Selected species whose distribution increased the most between historic and contemporary collections among the 200-selected study sub-watersheds.

Forest land use and fish distribution relationships

Forested land uses upstream of sampled sub-watersheds ranged from 10 to 99%. The amount of forested land upstream of a sampling location was positively and significantly (p < 0.05) related to the JCS within each major drainage basin (Table 4). This positive association was driven largely by the relationship between the JCS and the percentage of deciduous forest (Table 5). The spread of JCS values associated with greater than 90% forested land can be attributed to the absence of one or more species in predominantly headwater streams that historically contain five or less species. The absence of only two species in a five species community for example, typical of many headwaters streams, results in a JCS of 0.6 instead of 1.0.

Table 4.

Analysis results comparing JCS with land use data. Table values are r-values reported to indicate the sign of the trend. The number of watersheds included in analyses is denoted in parenthesis

 Drainage Basin 
Land Use Delaware Susquehanna Ohio 
Type (35) (93) (72) 
Forested Land 0.68 0.70 0.71 
Agricultural Land −0.52 −0.70 −0.60 
Developed Land −0.74 −0.32 −0.30 
 (−0.79)1 

1Both linear and curvilinear (in parenthesis) models explained significant amounts of variation.

 
 Drainage Basin 
Land Use Delaware Susquehanna Ohio 
Type (35) (93) (72) 
Forested Land 0.68 0.70 0.71 
Agricultural Land −0.52 −0.70 −0.60 
Developed Land −0.74 −0.32 −0.30 
 (−0.79)1 

1Both linear and curvilinear (in parenthesis) models explained significant amounts of variation.

 
Table 5.

Correlation coefficients (r) obtained by comparing the Jaccard coefficient of similarity with each of the 14 land use types among sub-watersheds across all three major drainage basins

Land Use Category Delaware (35) Susquehanna (93) Ohio (72) 
% Low intensity development −0.744* −0.261 −0.259 
% High intensity development −0.710* −0.395* −0.164 
% Hay/pasture/grass −0.219 −0.617* −0.576* 
% Row crops −0.155 −0.562* −0.378* 
% Probable Row crops −0.646* −0.677* −0.498* 
% Conifer Forest 0.415* 0.000 0.182 
% Mixed Forest 0.228 0.192 0.071 
% Deciduous forest 0.671* 0.677* 0.773* 
% Woody wetland 0.192 0.110 0.130 
% Emergent Wetland 0.255 0.302 0.228 
% Barren:Quarry areas −0.367 0.155 0.063 
% Barren:Coal mines −0.210 −0.063 0.000 
% Barren:Beach Areas −0.245 0.000 0.000 
% Barren:Transitional −0.148 −0.110 −0.447* 

*Significant relationships at p-value < 0.05, using Pearson correlation coefficients.

 
Land Use Category Delaware (35) Susquehanna (93) Ohio (72) 
% Low intensity development −0.744* −0.261 −0.259 
% High intensity development −0.710* −0.395* −0.164 
% Hay/pasture/grass −0.219 −0.617* −0.576* 
% Row crops −0.155 −0.562* −0.378* 
% Probable Row crops −0.646* −0.677* −0.498* 
% Conifer Forest 0.415* 0.000 0.182 
% Mixed Forest 0.228 0.192 0.071 
% Deciduous forest 0.671* 0.677* 0.773* 
% Woody wetland 0.192 0.110 0.130 
% Emergent Wetland 0.255 0.302 0.228 
% Barren:Quarry areas −0.367 0.155 0.063 
% Barren:Coal mines −0.210 −0.063 0.000 
% Barren:Beach Areas −0.245 0.000 0.000 
% Barren:Transitional −0.148 −0.110 −0.447* 

*Significant relationships at p-value < 0.05, using Pearson correlation coefficients.

 

Within the Delaware, Susquehanna, and Ohio river drainages 23, 82, and 66 watersheds contained greater than 50% forested land use, respectively. All of these forested watersheds experienced declines among the various fish descriptors (Table 6). Declines were greatest within the insectivorous and benthic insectivorous fishes. No significant differences were detected among the remaining descriptors, including species richness and species loss, suggesting that these groups have remained relatively unchanged between historic and contemporary collections. There was no significant difference in the percentage of forested land between sites that lost species and sites that experienced an increase or no change in number of species (p > 0.05).

Table 6.

Percent decline among fish descriptors used to assess community change from sub-watersheds containing greater than 50% forested land use. Number of watersheds included that met the forested criteria are denoted in parenthesis. In remaining sub-watersheds, that is, 100 minus table values, species either increased or remained the same

 Drainage Basin 
 Delaware Susquehanna Ohio 
Fish Descriptor (23) (82) (66) 
Introduced 13 13 
Intolerant 26 18 41 
Insectivore 48 54 60 
Omnivore 26 24 41 
Benthic Insectivore 22 37 50 
Insectivore-piscivore 15 12 14 
Filter feeder N/A N/A 
Herbivore 11 24 
 Drainage Basin 
 Delaware Susquehanna Ohio 
Fish Descriptor (23) (82) (66) 
Introduced 13 13 
Intolerant 26 18 41 
Insectivore 48 54 60 
Omnivore 26 24 41 
Benthic Insectivore 22 37 50 
Insectivore-piscivore 15 12 14 
Filter feeder N/A N/A 
Herbivore 11 24 

Agricultural land use and fish distribution relationships

Agricultural watershed land uses upstream from collection sites ranged from 0 to 86%, with the highest values (those greater than 40%) occurring in the Susquehanna River drainage. The percentage of agricultural land use within the sub-watersheds was inversely and significantly (p < 0.05) related to the JCS within each of the major drainage basins (Table 4). This association was driven largely by the relationship between the JCS and the percentage of probable row crops (Table 5).

Within the Delaware, Susquehanna, and Ohio river drainages the JCS was less than 0.5 within 11, 18, and 12 watersheds that contained greater than 40% agricultural land use, respectively. Among these watersheds, declines were greatest among the insectivorous and benthic insectivorous fishes within the Susquehanna and Ohio river drainages (Table 7). No significant relationships were detected among agricultural land use practices, species richness, or species loss. There was no significant difference in the percentage of agricultural land among sites that lost species and sites that experienced an increase or no change (p > 0.05).

Table 7.

Percent decline among fish descriptors used to assess community change from sub-watersheds containing greater than 40% agricultural land use. Number of watersheds included that met the agricultural criteria are denoted in parenthesis. In remaining sub-watersheds, that is, 100 minus table values, species either increased or remained the same

 Drainage Basin 
 Delaware Susquehanna Ohio 
Fish Descriptor (11) (18) (12) 
Introduced 17 
Intolerant 20 44 58 
Insectivore 64 77 58 
Omnivore 27 50 33 
Benthic Insectivore 18 89 67 
Insectivore-piscivore 25 
Filter feeder NA NA 
Herbivore 83 50 
 Drainage Basin 
 Delaware Susquehanna Ohio 
Fish Descriptor (11) (18) (12) 
Introduced 17 
Intolerant 20 44 58 
Insectivore 64 77 58 
Omnivore 27 50 33 
Benthic Insectivore 18 89 67 
Insectivore-piscivore 25 
Filter feeder NA NA 
Herbivore 83 50 

Developed land use and fish distribution relationships

Developed watershed land uses upstream from collection sites ranged from 0 to 45%, with the highest values (those greater than 10%) occurring in the Delaware River drainage. Only within the Delaware River drainage could we detect a significant relationship between the JCS and developed land use (p < 0.05, Table 4). When an exponential regression equation was fit to these data, the amount of variation explained increased from 55 to 63%. Both high and low intensity developed areas were significantly correlated with the JCS (Table 6). But there was no relationship between the JCS and the percentage of developed land within the Ohio and Susquehanna river watersheds. In these watersheds, the maximum percentage of developed land was 4.5% in the Susquehanna drainage and 20% in the Ohio drainage.

Within the Delaware River drainage, only five watersheds contained developed land uses greater than 10%. Among intolerant, insectivore, omnivore, and benthic insectivore fishes, declines were found in one, four, two, and one watershed, respectively. Increases in both insectivore-piscivore and introduced fishes were found in these five watersheds. Only two watersheds in the Ohio River drainage contained developed land use greater than 10%. All watersheds within the Susquehanna contained less than 5% developed land. There was no significant difference in the percent-developed land among sites that lost species and sites that experienced an increase or no change (p > 0.05).

Selected species-specific responses to land use patterns

Among those fishes that experienced declines in distribution, increased amounts of agricultural land were observed, suggesting a negative influence of agricultural land use practices on fish distribution (Table 8). Those fishes that experienced the greatest declines in distribution occurred in watersheds containing large percentages of agricultural development. While watersheds containing developed land-uses were small in number, those in the Delaware River drainage containing greater than 10% developed land, experienced declines in species distribution among common shiner (Luxilus cornutus), blacknose dace (Rhinichthys atratulus), and spotfin shiner (Cyprinella spiloptera).

Table 8.

Species occurring at 10 or more sites in the historic collections that experienced declines in distribution of greater than 30%. The average % agricultural land is calculated for those watersheds in which a species was absent in the contemporary collection

 Number of Watersheds 
Species Historical Distribution Contemporary Distribution Avg. % Agric. Land 
Central stoneroller, Campostoma anomalum 82 56 31 
Redside dace, Clinostomus elongatus 25 14 15 
Spotfin shiner, Cyprinella spiloptera 19 34 
Silverjaw minnow, Ericymba buccata 18 39 
Satinfin shiner, Notropis analostanus 18 38 
Striped shiner, Notropis chrysocephalaus 15 36 
Swallowtail shiner, Notropis procne 13 54 
Mimic shiner, Notropis volucellus 10 35 
Bluntnose minnow, Pimephales notatus 41 26 33 
Pearl dace, Margariscus margarita 19 15 
Banded killifish, Fundulus diaphanus 10 34 
Fantail darter, Etheostoma flabellare 43 29 27 
Johnny darter, Etheostoma niger 39 27 24 
Tesselated darter, Etheostoma olmstedi 78 52 37 
Blackside darter, Percina maculata 14 28 
 Number of Watersheds 
Species Historical Distribution Contemporary Distribution Avg. % Agric. Land 
Central stoneroller, Campostoma anomalum 82 56 31 
Redside dace, Clinostomus elongatus 25 14 15 
Spotfin shiner, Cyprinella spiloptera 19 34 
Silverjaw minnow, Ericymba buccata 18 39 
Satinfin shiner, Notropis analostanus 18 38 
Striped shiner, Notropis chrysocephalaus 15 36 
Swallowtail shiner, Notropis procne 13 54 
Mimic shiner, Notropis volucellus 10 35 
Bluntnose minnow, Pimephales notatus 41 26 33 
Pearl dace, Margariscus margarita 19 15 
Banded killifish, Fundulus diaphanus 10 34 
Fantail darter, Etheostoma flabellare 43 29 27 
Johnny darter, Etheostoma niger 39 27 24 
Tesselated darter, Etheostoma olmstedi 78 52 37 
Blackside darter, Percina maculata 14 28 

Discussion

Our results illustrate the relationships between land use types and changes in fish community structure. Generally, forested watersheds experienced little or no change in fish communities between historical and contemporary collections, while fish communities in agricultural watersheds changed substantially. The influence of urbanization was not so clearly demonstrated and could be studied further if additional watersheds containing urban land use percentages greater than 10% were added to the analysis. Because the direct physical effects of forest, agricultural, and urban land uses could not be quantified in this study, causal relationships cannot be drawn from our results. However, the strong associations between the various land use types and the changes in fish community structure are consistent with the idea that watershed disturbance can negatively affect fish distribution and community structure.

Relationships between land use and fish community change

The physical effects that developed land use exerts on stream systems have been well documented. Developed land, including impermeable surfaces such as roads, parking lots, and rooftops, substantially increases runoff volume to streams (Gustav et al., 1994). Schueler (1994) calculated that the total runoff volume for a 0.4-hectare parking lot is nearly 16 times that produced by 0.4 hectares of undeveloped meadow. Urban development can result in habitat alteration, increased nutrient loading, and degradation of instream fish habitat (Scott et al., 1986). Urban runoff is often polluted with suspended solids, petroleum products, and other toxicants. Weaver and Garman (1994) demonstrated a strong negative relationship between the JCS and the percentage of urbanization in a Virginia watershed. In their study, the JCS was less than 0.5 at most sites and the percent-developed land approached 40%. Wang et al. (1997) found a similar trend between the IBI and developed land. Like Weaver and Garman (1994) and Wang et al., (1997), we found that as developed land use increased from 10 to 20% there was a marked change in fish communities. This decline between 10 and 20% is lower than the 25 to 50% reported by Steedman (1988), suggesting that urbanization effects may be more severe than previously believed. Additional watersheds need to be added to our analysis to confirm this trend, because only seven watersheds contained greater than 10% developed land. In Valley Creek, a southeastern Pennsylvania watershed, Kemp and Spotila (1997) documented the absence of several fishes and macroinvertebrates. Over a 20-year period, Valley Creek experienced increases in residential, industrial, and commercial urbanization of 22, 33, and 19%, respectively. Onorato et al. (1998) found that several species of cyprinids and percids declined in response to increasing urbanization, while centrarchids increased in distribution.

Unlike urban watersheds, agricultural watersheds experience perhaps the most disruptive land-use change because of the annual alteration of land (Singh, 1992). Each year, much agricultural land is tilled for crops or cut seasonally, thereby yielding fallow land, row crops, rotation meadows, or grazed meadows. These practices lead to changes in infiltration, increased erosion, and the increased potential for contaminant runoff (Lenat, 1984; Singh, 1992). Carline and Spotts (1998) report that agricultural watersheds experience higher levels of total suspended solids, declines in stream substrate permeability, and lower fish densities and biomass than forested watersheds. Row crop agriculture, in particular, can reduce the supply of coarse organic material, and destabilize flow, temperature, and channel morphology. Wang et al. (1997) demonstrated negative relationships between the IBI and percentages of agriculture among 103 first-through fifth-order Wisconsin streams; declines were greatest among sites where agricultural land exceeded 50%. We did not observe a similar trend. Rather we observed a consistent decline in the JCS as the percentage of agricultural land increased. Roth et al. (1996), investigating 23 Michigan stream sites, documented similar linear relationships between land use and the IBI and concluded that basin land use was the primary determinant of ecological stream conditions.

Forested watersheds are often viewed as critical to the biological integrity of stream systems, because they contribute organic litter and woody debris to streams, reduce stream bank erosion, and aid in the reduction of sediment transport to streams (Karr and Scholsser, 1978; Stevens and Cummins, 1999). Stream temperatures are also better regulated in forested watersheds than in open canopy watersheds, and stream flows are often more stable because forests resist the overland flow of water, thus reducing erosion (Gregory et al., 1991). These factors all contribute to stream stability, driving the addition of allochthonous inputs that facilitates the persistence of fish communities. Forested headwaters experienced few changes between historic and recent collections, even though they supported relatively simple fish communities. The only exceptions to this were the insectivores collected in the Ohio and Susquehanna river drainages. Reasons for their decline are unknown, but many of these sites did contain introduced insectivore-piscivore fishes. Within the agricultural watersheds, benthic insectivores and insectivores declined relative to the increased percentages of agricultural land use. Watersheds containing greater than 40% agriculture experienced the largest declines within these two feeding guilds and the intolerant fishes.

Among the 45 species that declined in overall distribution (Argent, 2000), many occurred in predominantly agricultural watersheds and were found at sites where insectivore-piscivore fishes were present. The absence of several species from contemporary collections does not necessarily indicate that these taxa are extirpated. Upon inspection of other contemporary data (Argent et al., 1997), only one species was absent, the Southern redbelly dace. The two watersheds that historically maintained this species contained greater than 90% forested land; hence, it is unlikely that their absence is directly related to land use alteration. It is more likely that such species had low population numbers in the historic collections and were never widely distributed. If fact, two of the eight species that were absent from contemporary collections are considered rare in Pennsylvania (Argent, 2000). Of the remaining six species only one, the striped shiner, seems to have declined in response to agricultural development increases.

Among the 42 species that increased in distribution between historic and contemporary collections, the brown trout (Salmo trutta) became the most widespread (Argent, 2000). The brown trout is both a widely stocked and naturalized sport fish (Beard and Carline, 1991; Kemp and Spotila, 1997) that occupies an insectivore-piscivore feeding guild within Pennsylvania. Although its direct influence on fish communities could not be assessed because abundance data were unavailable, many insectivore and benthic insectivore species were absent in contemporary collections where the brown trout was present. In Bottom Creek, Virginia, stomach content analysis of stocked brown trout longer than 280 mm in length revealed fish prey between 25 and 110 mm in length (Garman and Nielsen, 1982), a range that spans the total length of many insectivores, benthic insectivores, and juvenile fishes. Fish often comprised 80% or more of the consumption of brown trout longer than 280 mm (Garman and Nielsen, 1982). Penczak (1999), comparing fish communities before and after brown trout stocking in Poland's Pilica watershed, documented extirpations of native fishes and significant declines in other fishes where brown trout were present. In Alabama, the presence of largemouth bass (Micropterus salmoides) led to behavioral changes, emigration, and predation of central stoneroller (Power and Matthews, 1983; Power et al., 1985). Li and Moyle (1993) indicate that introduced species are most successful in streams that have been altered by human activity. While the addition of non-indigenous fishes did not have a direct significant influence on the structure of fish communities; the biotic and anthropogenic disturbances present in many of the sub-watersheds included in this study seem related to a combination of both widespread fish introductions and increased percentages of terrestrial disturbance.

The Jaccard coefficient of similarity

The use of the JCS to relate species change to land use change seems to show some promise as an appropriate response variable in the absence of abundance data. Previous studies have utilized the JCS (e.g., Weaver and Garman, 1994; Onorato et al., 1998), but many others have used the IBI as a measure of fish community integrity to compare with land use data. As Fausch et al. (1990) point out, the IBI is a method that relies on subjective scoring criteria, the descriptors provide redundant information in some cases (Angermeier and Karr, 1984), and comparisons between temporally isolated data sets cannot be performed. Even so, the results obtained are robust enough to provide insights into the biological integrity of different streams within a region and several modifications strengthen the validity of these regional metrics. The JCS on the other hand relies solely on the presence of a species between two collection periods and, in our opinion, when supported with other descriptors that reflect various feeding associations and community alterations can provide an equivalent assessment of biotic integrity.

Instream habitat has been correlated with stream fish assemblages (Matthews and Heins, 1987). Both Wang et al. (1997) and Roth et al. (1996) found strong associations between land use and their habitat index scores derived from measures of stream velocity, reach depth, riffle and pool structure, substrate, embeddedness, cover, and bank stability. Given the strong similarity between our JCS results and their IBI results, we believe that agricultural watersheds in this study have reduced instream habitat complexity. Forested watersheds on the other hand, can potentially provide instream habitat with additions of large woody debris and allochthonous material. With these additions, stream velocities are often more stable, pools deeper, and stream banks more stable.

Spatial scale

This investigation was performed at only one spatial scale, the sub-watershed. Other authors have addressed issues of spatial scale as they relate to fish and macroinvertebrate community structure using GIS. Roth et al. (1996) used both a 50-m buffer and a 30-m instream classification of local riparian vegetation, Richards et al. (1996) used a 100-m buffer, and Wang et al. (1997) used a 100-m buffer. In all cases, correlations between land use types and fish community indices were strongest at the watershed scale and tended to become weak and non-significant at local scales (Richards et al., 1996; Roth et al., 1996; Wang et al., 1997). Roth et al. (1996) reported that the extent of forested land within a 50-m riparian buffer was not an effective predictor of IBI values. This result suggests that processes affecting stream systems, particularly channel morphology, and their resident fish populations are operating at large spatial scales. In fact, most agricultural and urban land uses occur at the larger watershed scale and their impacts cannot be fully understood by looking at adjacent riparian lands (Wang et al., 1997). This does not preclude the importance of riparian areas for the maintenance of fish communities, but as Richards et al. (1996) demonstrated, stream buffers are better able to predict sediment-related habitat variables, (e.g., stream substrate characteristics and bank erosion) than are watershed characteristics.

One explanation for the lack of correspondence between fish community indices and riparian land use values is that most GIS land use data are derived at a resolution that does not incorporate the variability and spatial complexity of land use types common to most riparian areas. For example, in this study the MRLC land use data were derived at a scale of 30 m; however, most field-riparian studies quantify stream buffers in far greater detail than one or two homogenized grid cells within a GIS (Edwards and Huryn, 1996; Stevens and Cummins, 1999). Moreover, buffer size is an arbitrary assessment that may best be measured at scales greater than or less than 30 m, depending on stream size. Roth et al. (1996) reported a poor correspondence between their riparian field measurements and their riparian GIS measurements. Again this supports the contention that analyses must be performed at a spatial scale appropriate for the collected data and the research question(s) to be addressed.

This study serves as an important indicator of the relative persistence of Pennsylvania's fish communities in relation to environmental change. Future studies should address the specific role that introduced fishes may exert on native fishes, particularly the predation potential and the negative correlation between developed watersheds and native fishes. As Naiman et al. (1993) argue, to conserve the connectivity of functions within a watershed, the river corridor should be managed as an entire system, from well-buffered headwaters to downstream floodplains. To this end we should continue to investigate the processes affecting streams at the watershed level, because watersheds or sub-watersheds ultimately determine a stream's structure and function.

Acknowledgements

We thank the USGS National Fisheries Research and Development Laboratory (Research Work Order #47) and The Wild Resource Conservation Fund for providing financial support. We also thank the PFBC, Dr. Edwin Cooper, Dr. Jay R. Stauffer, and Penn State University for providing their data and the Office of Remote Sensing at Penn State University for the use of their GIS facilities.

References

Aadland, L. P.
1993
.
Stream habitat types: Their fish assemblages and relationship to flow
.
N. A. J. Fish. Manage
,
13
:
790
806
.
Adams, S. M., Ham, K. D., Greeley, M. S., Lehew, R. F., Hinton, D. E. and Saylor, C. F.
1996
.
Downstream gradients in bioindicator responses: point source contaminant effects on fish health
.
Can. J. Fish. Aquat. Sci.
,
53
:
2177
2187
.
Anderson, J. R., Hardy, E. E., Roach, J. T. and Witmer, R. E.
1976
.
A land use and land cover classification system for use with remote sensor data
,
U.S. Geological Survey Professional
.
Paper 964
Angermeier, P. L. and Karr, J. R.
1984
.
Relationships between woody debris and fish habitat in a small warmwater stream
.
Trans. Amer. Fish. Soc.
,
113
:
716
726
.
Argent, D. G., Carline, R. F. and Stauffer, J. R. Jr.
1997
. “
Historical and contemporary distribution of fishes in Pennsylvania. Final Report. Research Work Order No. 47
”. In
U.S. Geological Survey Biological Resources Division, The Pennsylvania Cooperative Fish and Wildlife Research Unit and The School of Forest Resources
,
Pennsylvania
:
The Pennsylvania State University
.
Argent, D. G.
2000
.
Fish distribution patterns in Pennsylvania: Anthropogenic influences and the status of and rare fishes
,
University Park, Pennsylvania
:
The Pennsylvania State University
.
Ph.D. Thesis
Armour, C. L., Duff, D. A. and Elmore, W.
1991
.
The effects of livestock grazing on riparian and stream ecosystems
.
Fisheries
,
16
:
7
11
.
Barbour, M. T., Gerritsen, J., Snyder, B. D. and Stribling, J. B.
1997
.
Rapid bioassessment protocols for use in streams and rivers: Periphyton, benthic, macroinvertebrates, and fish
,
Washington, D.C.
:
U.S. Environmental Protection Agency
.
EPA 841-D-97–002. Available
HTTP//:www.epa.gov/owow/monitoring/AWPD/RBP/bioasses.html
Beard, T. D. and Carline, R. F.
1991
.
Influence of spawning and other stream habitat features on spatial variability of wild brown trout
.
Trans. Amer. Fish. Soc.
,
120
:
711
722
.
Carline, R. F. and Spotts, D. E.
1998
. “
Early responses of stream communities to riparian restoration in agricultural watersheds, eastern USA
”. In
Headwaters: Water Resources and Soil Conservation
Edited by: Haigh, M. J., Křeček, J., Rajwar, G. S. and Kilmartin, M. P. pp.
167
173
. Proceedings of Headwater' 98, the Fourth International Conference on Headwater Control,
Merano, Italy
.
Cooper, E. L.
1983
.
Fishes of Pennsylvania and the Northeastern United States
,
University Park, PA.
:
The Pennsylvania University Press
.
Dobson, J. E., Bright, E. A., Ferguson, R. L., Field, D. W., Wood, L. L., Haddad, K. D., Iredale, H., Jensen, J. R., Klemas, V. V., Orth, R. J. and Thomas, J. P.
1995
.
NOAA Coastal Change Analysis Program (C-CAP): Guidance for Regional Implementation
NOAA Technical Report NMFS 123
Dunn, O. J.
1964
.
Multiple comparisons using rank sums
.
Technometrics
,
6
:
241
252
.
Edwards, E. D. and Huryn, A. D.
1996
.
Effect of riparian land use on contributions of terrestrial invertebrates to streams
.
Hydrobiologia
,
337
:
151
159
.
Fairchild, G. W., Horwitz, R. J., Nieman, D. A., Boyer, M. R. and Knorr, D. F.
1998
.
Spatial variation and historical change in fish communities of the Schuylkill River drainage, Southeast Pennsylvania
.
Am. Midl. Nat.
,
139
:
282
295
.
Fausch, K. D., Lyons, J., Karr, J. R. and Angermeier, P. L.
1990
.
Fish communities as indicators of environmental degradation
.
Am. Fish. Soc. Symp.
,
8
:
123
144
.
Fleischner, T. L.
1994
.
Ecological costs of livestock grazing in western North America
.
Conserv. Biol.
,
8
:
629
644
.
Gagen, C. J., Sharpe, W. E. and Carline, R. F.
1993
.
Mortality of brook trout, mottled sculpins, and slimy sculpins during acidic episodes
.
Trans. Am. Fish. Soc.
,
122
:
616
628
.
Gammon, J. R.
1980
.
The use of community parameters derived from electrofishing catches of river fish as indicators of environmental quality, in Seminar on Water Quality Management Tradeoffs
,
Washington, D.C.
:
U.S. Environmental Protection Agency
.
EPA-905/9–80-009
Garman, G. C. and Nielsen, L. A.
1982
.
Piscivority by stocked brown trout (Salmo trutta) and its impact on the Nongame fish community of Bottom Creek, Virginia
.
Can. J. Fish. Aquat. Sci.
,
39
:
862
869
.
Gregory, S. V., Swanson, F. J., McKee, W. A. and Cummins, K. W.
1991
.
An ecosystem perspective of riparian zones
.
BioScience
,
41
:
540
551
.
Gustav, R. S., Sovern, D. T. and Washington, P. M.
1994
.
Maintaining fish habitat in urban streams
.
Wat. Environ. Tech.
,
6
:
48
53
.
Hansen, M. J. and Ramm, C. W.
1994
.
Persistence and stability of fish community structure in a Southwest New York stream
.
Am. Midl. Nat.
,
132
:
52
67
.
Hocutt, C. H. and Wiley, E. O., eds.
1986
.
The Zoogeography of North American Freshwater Fishes
,
New York
:
John Wiley & Sons, Inc.
.
Jackson, D. A., Somers, K. M. and Harvey, H. H.
1989
.
Similarity coefficients: Measures of co-occurrence and association or simple measures of occurrence?
.
Am. Nat.
,
133
:
43
453
.
Jennings, M. D.
1993
. “
Natural Terrestrial Cover Classification: Assumptions and Definitions
”. In
Gap Analysis Technical Bulletin 2
,
Moscow, ID
:
U.S. Fish and Wildlife Service
.
Johnson, L. B., Richards, C., Host, G. E. and Arthur, J. W.
1997
.
Landscape influences on water chemistry in Midwestern stream ecosystems
.
Freshwat. Biol.
,
37
:
193
208
.
Karr, J. R.
1981
.
Assessment of biotic integrity using fish communities
.
Fisheries
,
66
:
21
27
.
Karr, J. R. and Schlosser, I. J.
1978
.
Water resources and the land-water interface
.
Science
,
201
:
229
234
.
Karr, J. R., Fausch, K. D., Angermeier, P. L., Yant, P. R. and Schlosser, I. J.
1986
.
Assessing biological integrity in running waters: A method and its rationale
,
Illinois Natural History Survey
.
Special publication 5
Kemp, S. J. and Spotila, J. R.
1997
.
Effects of urbanization on brown trout, Salmo trutta, other fishes and macroinvertebrates in Valley Creek, Valley Forge, Pennsylvania
.
Am. Midl. Nat.
,
138
:
55
68
.
Klein, R. D.
1979
.
Urbanization and stream quality impairment
.
Wat. Resources Bull.
,
15
:
948
963
.
Krebs, C. J.
1989
.
Ecological Methodology
,
New York
:
Harper & Row Publishers
.
Lammert, M. and Allan, J. D.
1999
.
Assessing biotic integrity of streams: Effects of scale in measuring the influence of land use/cover and habitat structure on fish and macroinvertebrates
.
Environm. Manage.
,
23
:
257
270
.
Landsberg, J. H., Blakesley, B. A., Reese, R. O., Mcrae, G. and Forstchen, R. R.
1998
.
Parasites of fish as indicators of environmental stress
.
Environm. Monitor. Assess.
,
51
:
211
232
.
Lenat, D. R.
1984
.
Agriculture and stream water quality, a biological evaluation of erosion control practices
.
Environm. Manage.
,
8
:
333
344
.
Lenat, D. R. and Crawford, J. K.
1994
.
Effects of land use on water quality and aquatic biota of three North Carolina Piedmont streams
.
Hydrobiologia
,
294
:
185
199
.
Li, H. W. and Moyle, P. B.
1993
. “
Management of Introduced Fishes
”. In
Inland Fisheries Management in North America
, Edited by: Kohler, C. and Hubert, W. pp.
287
307
.
Bethesda, Maryland
:
American Fisheries Society
.
Loveland, T. R. and Shaw, D.
1996
. “
Multi-resolution land characterization: Building collaborative partnerships
”. In
Gap Analysis: A Landscape Approach to Biodiversity Planning
, Edited by: Scott, J. M., Tear, T. H. and Davis, F. pp.
79
85
.
Bethesda, Maryland
:
American Society for Photogrammetry and Remote Sensing
.
Matthews, W. J. and Heins, D. C., eds.
1987
.
Community and evolutionary ecology of North American stream fishes
,
Norman, Oklahoma
:
University of Oklahoma Pressn
.
Miller, R. R., Williams, J. D. and Williams, J. E.
1989
.
Extinctions of North American fishes during the past century
.
Fisheries
,
14
:
22
36
.
Minitab
.
1996
.
Minitab reference manual for windows
,
Pennsylvania
:
State College
.
Release 11
Moyle, P. B., Li, H. W. and Barton, B. A.
1986
. “
The Frankenstein effect: Impact of introduced fishes on native fishes in North America
”. In
Fish Culture in Fisheries Management
, Edited by: Stroud, R. H. pp.
415
426
.
Bethesda, Maryland
:
American Fisheries Society
.
Naiman, R. J., Decamps, H. and Pollock, M.
1993
.
The role of riparian corridors in maintaining regional biodiversity
.
Ecol. Appl.
,
3
:
209
212
.
Onorato, D., Marion, K. R. and Angus, R. A.
1998
.
Longitudinal variations in the ichthyofaunal assemblages of the upper Cahaba River: Possible effects of urbanization in a watershed
.
J. Freshwat. Ecol.
,
13
:
139
154
.
Ott, L.
1988
.
An Introduction to Statistical Methods and Data Analysis
,
Boston, Massachusetts
:
PWS-Ken Publishing Company
.
Penczak, T.
1999
.
Impact of introduced brown trout on native fish communities in the Pilica River catchment (Poland)
.
Environm. Biol. Fishes
,
54
:
237
252
.
Plafkin, J. L., Barbour, M. T., Porter, K. D., Gross, S. K. and Hughes, R. M.
1989
.
Rapid bioassessment protocols for use in stream and rivers: Benthic macroinvertebrates and fish
,
Washington, D.C.
:
U.S. Environmental Protection Agency
.
EPA/444/4–89-011
Power, M. E. and Matthews, W. J.
1983
.
Algae-grazing minnows (Campostoma anomalum), piscivorous bass (Micropterus spp.), and the distribution of attached algae in a small prairie-margin stream
.
Oecologica
,
60
:
328
332
.
Power, M. E., Matthews, W. J. and Stewart, A. J.
1985
.
Grazing minnows, piscivorous bass, and stream algae: Dynamics of a strong interaction
.
Ecology
,
66
:
1448
1456
.
Richards, C., Johnson, L. B. and Host, G. E.
1996
.
Landscape-scale influences on stream habitats and biota
.
Can. J. Fish. Aquat. Sci.
,
53
:
295
311
.
Roth, N. E., Allan, J. D. and Erickson, D. L.
1996
.
Landscape influences on stream biotic integrity assessed at multiple spatial scales
.
Landscape Ecol.
,
11
:
141
156
.
Schlosser, I. J.
1991
.
Stream fish ecology: A landscape perspective
.
BioScience
,
41
:
704
712
.
Schlosser, I. J.
1995
.
Critical landscape attributes that influence fish population dynamics in headwater streams
.
Hydrobiologia
,
303
:
71
81
.
Schueler, T.
1994
.
The importance of imperviousness
.
Watershed Protection Techniques
,
1
:
110
111
.
Scott, J. B., Steward, C. R. and Stober, Q. J.
1986
.
Effects of urban development on fish population dynamics in Kelsey Creek, Washington
.
Trans. Am. Fish. Soc.
,
115
:
555
567
.
Singh, V. P.
1992
.
Elementary Hydrology
,
New Jersey
:
Prentice-Hall, Inc.
.
SAS (Statistical Analysis Software Institute)
.
1998
.
JMP'IN statistical discovery software
,
Cary, North Carolina
:
Duxbury Press
.
Version 3.2.
Steedman, R. J.
1988
.
Modification and assessment of an index of biotic integrity to quantify stream quality in Southern Ontario
.
Can. J. Fish. Aquat. Sci.
,
45
:
492
501
.
Stevens, M. H. and Cummins, K. W.
1999
.
Effects of long-term disturbance on riparian vegetation and in-stream characteristics
.
J. Freshwat. Ecol.
,
14
:
1
17
.
Strahler, A. N.
1957
.
Quantitative analysis of watershed geomorphology
.
Trans. Am. Geophys. Union
,
38
:
913
920
.
Wang, L., Lyons, J., Kanehl, O. and Gatti, R.
1997
.
Influences of watershed land use on habitat quality and biotic integrity in Wisconsin streams
.
Fisheries
,
22
:
6
12
.
Weaver, L. A. and Garman, G. C.
1994
.
Urbanization of a watershed and historical changes in a stream fish assemblage
.
Trans. Am. Fish. Soc.
,
123
:
162
172
.
Wohl, N. E. and Carline, R. F.
1996
.
Relations among riparian grazing, sediment loads, macroinvertebrates, and fishes in three central Pennsylvania streams
.
Can. J. Fish. Aquat. Sci.
,
53
:
260
266
.
Suppl. 1