In Korean streams and rivers, organic matter and nutrient pollution are nationwide problems, and are key issues in ecosystem health assessment and ecological management. This phenomenon is caused mainly due to rapid industrialization, urbanization, human population increase, and intensive farming (Limburg and Schmidt 1990, Ricciardi and Rasmussen 1999, Baer and Pringle 2000). Especially dredging, flow alteration, and channelization in the urban streams accelerated the degradation of physical habitat health (Paul and Meyer 2001, Allan 2004). Such anthropogenic disturbances influence biodiversity of various aquatic biota including fish, and disrupt food chains from the nutrient regime to top-carnivores in the lotic environment. For these reasons, the United States Environmental Protection Agency (Klemm et al. 1993) and the European Union (Didier et al. 1995, Oberdorff et al. 2002) developed an ecological health assessment system based on numerous multi-metric biological models and implemented the model for regional assessment. The fish models of the ecological multi-metric approach were originally based on the concept of “index of biological integrity” (IBI) to evaluate physical habitat, chemical, and biological conditions in lotic ecosystems (Karr 1981). This approach has been applied to streams and rivers of the U.S. (Karr et al. 1986), France (Oberdorff and Hughes 1992, Oberdorff et al. 2002) Australia (Harris 1995), India (Ganasan and Hughes 1998), Japan (Koizumi and Matsumiya 1997), and Korea (An et al. 2001, An et al. 2002, Choi et al. 2011, Lee and An 2014). One of the problems in the regional model applications, however, was the classification and class categorization of fish tolerance guilds and trophic guilds.

Fish are an ecological indicator to measure the chemical tolerance and assess habitat modifications in lotic ecosystems. The biological response or tolerance range differs depending on the magnitude of the pollution or ecological disturbances (United States Environmental Protection Agency 1988, Van Putten 1989). Especially, the presence or absence, and the abundance of regional fish fauna are frequently influenced by chemicals in the water. The tolerance based guild classification used in the biological integrity models is based only on fish dictionaries or experiences of fish ecologists or taxonomists. The classification of the tolerance guild into tolerant, intermediate or sensitive species (Klemm et al. 1993) is ambiguous, and an analysis based on quantitative data rather than simple field experiences is needed. In fact, the ecosystem health assessment model, which uses a metric of tolerance guild based on the subjective aspects and traditional experiences, were widely applied in national streams and rivers (Choi 1989, Kim 1997). The model values, thus, obtained from the multi-metric health assessments may be ambiguous in diagnosing the ecological health (Meador and Carlisle 2007). In spite of these facts, little is known about the tolerance ranges or magnitude of fish species in developed countries and Korea.

Each fish species has a set of optimal, sub-optimal, or lethal tolerance range in terms of temperature (thermal tolerance), physical habitat (habitat tolerance), water quality (chemical tolerance), and ecological relevance (ecological tolerance), and one or two factors in the tolerance level are critical to some fishes. Generally, physical hydrological factors such as water temperature (Beitinger et al. 2000), land use patterns (Wang et al. 2001), stream morphology (Wang et al. 2003), and flow regime (Bunn and Arthington 2002) are known to be important factors influencing the level of fish tolerance in aquatic ecosystems. Also, chemical parameters such as organic matter pollutant (Dyer et al. 2000), toxic chemicals (Brian et al. 2005), ionic content or salinity (Le François et al. 2004) or nutrients (N, P) are other key factors that control fish tolerance in the waterbodies of geographic regions. In addition, fish tolerance may also be associated with biological factors such as predation (Brown et al. 2005). The physical, hydrological and chemical preferences and needs for a certain type of stream condition greatly influence the regional distribution of fish species (Lee and An 2014). In real stream environments, multiple abiotic and biotic factors simultaneously influence any given fish species, which causes problems when quantifying the tolerance range of the fish.

The two target species, Carassius auratus and Cyprinus carpio, selected in this study are cyprinidae. They are widely distributed in Korean aquatic ecosystems, and well known as commercial fish species with high scientific value (Kim and Park 2002). These two species are categorized as omnivore species in the trophic guilds (Choi 1987, Kim et al. 2005, Lee and Noh 2006), and their major habitats are mainly stagnant regions of weirs, reservoirs, and rivers with slow current velocity and large amounts of sediments in the bottom (Kim et al. 2005). Little is still known about how these species tolerate various environmental stress caused by chemicals. Determination of tolerance ranges of different species for different environmental factors is a key issue for ecologists working in ecosystem health assessment. In this study, tolerance ranges of two fish populations (C_a-population and C_c-population) were determined on the basis of land use pattern, stream morphology, chemical quantity in water, habitat parameters, ecological indicators (trophic/tolerance compositions) of co-occurring fishes, and multi-metric assessment of stream health.

Fishes were collected by a modified wading method applied by An et al. (2006) based on the catch per unit effort (CPUE; Klemm et al. 1993). Fish sampling were conducted at 720 streams and rivers of the Han River watershed (river length (RL) = 514 km; basin area (BA) = 26,219 km²), the Nakdong River watershed (RL: 525 km, BA: 23,860 km²), Geum River watershed (RL: 414 km, BA: 9,886 km²), Yeongsan/Sumjin River watershed (RL: 348 km, BA: 8,267 km²) during 2008-2009 (Fig. 1). The time lapse for fish sampling was 60 minutes and all habitat types such as riffle, pool, and run area were included in 200m stream segments.

[Fig. 1.] Maps showing the sampling sites in four major Korean river watersheds of Han River (Region I), Nakdong River (Region II), Geum River (Region III) and Yeongsan/Sumjin River (Region IV).

Sampled fishes were identified by using an IBI-approach developed by An et al. (2006) and returned to their natural habitats. Some ambiguous specimens were preserved in 10% formalin to identify the taxa in the laboratory. All fishes were examined for external anomalies of DELT, namely deformities (D), erosion (E: skin, barbells), patlesion (L: open sores, ulcerations), and tumors (T)on the basis of the criteria developed by Sanders et al. (1999). Casting nets (mesh 7 × 7 mm) and kick nets (4 × 4 mm), which are the most common sampling gears in Korea, were used for fish sampling. The casting net was used for open water fish habitats without any obstructions like riffles, pools, and slow runs, whereas, the kick net was used at places with fast currents and many obstructions where one could hardly use a casting net.

In addition to regular sampling, thirty-four reference sites were chosen from the watershed to calculate maximum species richness lines (MSRLs) against stream orders by following the approach of Hornig et al. (1995). Each reference stream number from the 1st to 5th order streams in the survey of reference sites were 3, 4, 3, 11, and 13, respectively. A 1:15,000 map was used to select candidate reference locations and determine stream order based on the methodology of Strahler (1957). The MSRL was determined by empirical methods developed by Karr (1981) and 1st order regression analysis developed by Rankin and Yoder (1999).

A total of eight parameters that include biological oxygen demand (BOD), chemical oxygen demand (COD), total nitrogen (TN), nitrate-nitrogen (NO₃-N), total phosphorus (TP), phosphate-phosphorous (PO₄-P), suspended solids (SS), and electric conductivity were analyzed in this study. We analyzed monthly datasets that were obtained from the Ministry of Environment, Korea (MEK) from January 2008 to December 2009. The criteria of physico-chemical water quality were based on a 7-rank system set by the Ministry of Environment, Korea.

Fish samples were first analyzed according to the Index of Biological Integrity (IBI) developed by Karr (1981) and Barbour et al. (1999) and run through an ecological health assessment model. The samples were collected from four major watersheds of the Han River, Nakdong River, Geum River, and Yeongsan/Sumjin River during 2008-2009. The metrics (M) reflected three major ecological characteristics: species richness and composition, trophic composition, and fish abundance and health condition. The three groups each consisted of eight metrics: M₁, total number of native species; M₂, number of riffle-benthic species; M₃, number of sensitive species; M₄, proportion of individuals as tolerant species; M₅, proportion of individuals as omnivores; M₆, proportion of individuals as native insectivores; M₇, total number of individual native species; and M₈, proportion of abnormal individuals. The four metrics (M₁, M₂, M₃, and M₇) were evaluated using the maximum species richness line (MSRL) with stream orders. Each metric was given 1, 3, or 5 points respectively. The points were then summed up to determine 4 class criteria that ranged from excellent (A: 40-36), good (B: 35-26), fair (C: 25-16), and very poor (D: ≤ 15). Detailed descriptions with specific characteristics and scoring criteria of the model have been previously reported by An et al. (2006).

Habitat assessments were modified according to the Qualitative Habitat Evaluation Index (QHEI) described by Plafkin et al. (1989). Six habitat parameters chosen from widely used references (Hamilton and Bergersen 1984, Lafferty 1987, Bartholow 1989), were used for assessing QHEI: epifaunal substrate/available cover (Q₁), embeddedness (Q₂), channel flow status (Q₃), dam construction impact (Q₄), channel alteration (Q₅), and sediment deposition (Q₆). Each of the assessed habitat attributes was summarized and divided into 4 categories: “optimal (20, 18, 16),” “sub-optimal (15, 13, 11),” “marginal (10, 8, 6),” and “poor (5, 3, 1).”

SPSS ver. 18.0 (SPSS Inc., Chicago, IL, USA) was used to test significant differences between C_a-population (n = 479) and C_c-population (n = 145) by using datasets from four major rivers and streams. Two independent sample t-tests were conducted to compare water quality parameters, physical habitat variables, and ecological indicators between C_a-population (n = 479) and C_c-population (n = 145). In addition, SYSTAT 7.0 (Wilkinson 1997) was used for sample linear regression analyses and spearman-correlation analyses.

An analysis of stream morphology and land use patterns were carried out at sites where C_a-population and C_c-populations were sampled (Table 1). Stream width, stream order, and altitude were used as key variables for stream morphology. Three categories were set up to differentiate land use patterns, namely urban land, agricultural land, and forestry. The C_a-population occurred in habitats with a mean stream width of 73.3m (range: 1-550), mean stream order of 3.35 (1-6), and mean altitude of 38.2m (0-359), while the C_c-population occurred in habitats with a mean stream width of 89.6m (range: 1.1-625), mean stream order of 3.48 (1-6), and mean altitude of 30.9m (0-175; Table 1). Two statistically independent sample t-tests on stream morphology between C_a-population and C_c-population (Table 1) showed that there were no significant differences (p > 0.05) in stream width and stream order, but differences were found in altitude.

[Table 1.] Means, range (maximum-minimum) and quartile (1/4 quartile, median and 3/4 quartile) on stream morphologic parameters and land use pattern in streams and rivers where Ca-population and Cc-population was sampled

Means, range (maximum-minimum) and quartile (1/4 quartile, median and 3/4 quartile) on stream morphologic parameters and land use pattern in streams and rivers where Ca-population and Cc-population was sampled

Analysis of land use pattern showed that occurrence frequency of the C_a-population was 50.3% (1.5-100%) in the forestry region, 31.8% (1.0-86.8%) in the agricultural region, and 20.8% (0.5-100%) in the urban region. In the meantime, occurrence frequency of the C_c-population was 42.5% (7.1-88.5%) in the forestry region, 30.9% (2.0- 82.7%) in the agricultural region, and 31.2% (1.1-100%) in the urban region. Two statistically independent sample ttests on land use patterns between the two populations showed that there were significant differences (p < 0.001) in the urban and forestry regions, but not in the agricultural region. When compared to the C_a-population, the C_c-population was more widely distributed in the urban region and less widely distributed in the forestry region.

Statistical tests in the C_a-population and C_c-population indicated that stream morphology agreed with land use patterns in both populations. Both C_a-population and C_c-populations had a significant probability of p < 0.05 in the altitude (stream morphology), and fish occupancy of p < 0.001 in the urban and forestry region. The C_c-population was distributed in lower altitude regions (downstream) and urban regions, where point and non-point pollution sources are denser when compared to the C_a-population. These results indicate that the tolerance of C_a-population for stream morphology and land use pattern is greater than the C_c-population.

The chemical tolerance ranges of the C_c-population and C_a-population were determined after assessing water chemistry factors such as biological water quality (BOD), chemical oxygen demand (COD), total nitrogen (TN), nitrate-nitrogen (NO₃-N), total phosphorus (TP), phosphate- phosphorus (PO₄-P), total suspended solids (SS), and electric conductivity (Table 2; Figs. 2 and 3). Mean and maximum tolerance ranges for organic matters of BOD in the C_a-population were 3.71 mg L^-1 and 24.4 mg L^-1, respectively, and the tolerance values for COD were 7.14 mg L^-1 and 27.3 mg L^-1. Mean and maximum tolerance ranges for TP were 0.24 mg L^-1 and 2 mg L^-1 in the C_a-population, respectively. In the meantime, mean and maximum tolerance ranges for BOD in the C_c-population were 5.1 mg L^-1 and 24.4 mg L^-1, respectively, and the tolerance values for COD were 8.07 mg L^-1 and 27 mg L^-1. Mean and maximum tolerance ranges for TP were 0.39 mg L^-1 and 1.9 mg L^-1 in the C_c-population. Two statistically independent sample t-tests showed that fish tolerance for BOD, COD, and TP, except SS, were significantly (p < 0.05) greater in the C_c-population than the C_a-population (Table 2). These outcomes suggest that chemical tolerance of organic matter pollution and nutrient enrichment was greater in the C_c- population than the C_a-population.

[Table 2.] Means, range (maximum-minimum) and quartile (1/4 quartile, median and 3/4 quartile) on water quality in streams and rivers where Ca-population and Cc-population was sampled

Means, range (maximum-minimum) and quartile (1/4 quartile, median and 3/4 quartile) on water quality in streams and rivers where Ca-population and Cc-population was sampled

[Fig. 2.] Relative abundance (%) as a function of oxygen demand (organic matter), N nutrients in Ca-population (Carassius auratus) and Cc-population (Cyprinus carpio).

[Fig. 3.] Relative abundance (%) as a function of P nutrients, suspended solids and ionic content (Cond) in Ca-population (Carassius auratus) and Cc-population (Cyprinus carpio).

In addition, optimal peak ranges for two fish populations were determined by using quartile analysis (Table 2, Figs. 2 and 3). As shown in Figs. 2 and 3, the tolerance range for chemical parameters were wide in both the C_c-population and C_a-population, and the mean values for BOD, COD, and TP fell in the high chemical concentrations, indicating that both the C_c-population and the C_a-population are tolerant fish species along the chemical gradients. The chemical ranges of one-fourth and threefourths, based on the quartile, were a major distribution of the fish populations, and were also compared with criteria of stream water quality according to the Ministry of Environment, Korea (2006). Major distribution of the C_a-population on BOD was 1.5-4.5 mg L^-1, indicating that the stream chemical health, based on BOD, was judged as a little fair (III) to good health condition (Ib) according to the criteria of MEK (2006). Similarly, the major distribution range of the C_a-population on COD was 4.5-9 mg L^-1 and the distribution range on TP was 0.06-0.3 mg L^-1, indicating poor (IV) to good health (Ib) condition. Also, the C_a-population from 479 sampling sites were observed and 156 sites out of 479 sites turned out to be in very poor condition (BOD > 10 mg L^-1; COD > 11 mg L^-1; TP > 0.5 mg L^-1), and unfit for the survival of the C_a-population. Also, the C_a-population was observed in the clean environment (Ia), indicating that the fish can distribute in pristine conditions, even though the C_a-population is widely known as a tolerant fish population (Kim and Park 2002, Lee and Noh 2006).

Major distribution of the C_c-population on BOD was 2.3-6.4 mg L^-1, indicating that the stream chemical health, based on BOD, was judged as poor (IV) to somewhat good health condition (II) according to the criteria of MEK (2006). Similarly, the major distribution range of the C_c-population on COD was 5.8-10.2 mg L^-1, and the distribution range on TP was 0.1-0.5 mg L^-1, indicating a poor (V) to somewhat good health (II) condition. Also, the C_c-population from 145 sampling sites were observed and 53 sites out of 367 sites turned out to be in very poor condition (VI), unfit for survival of the C_c-population. Only 9% (13 of 145 sampling sites) of the C_c-population was observed in a clean environment (Ia), indicating that C_c-population was generally distributed in fairly poor quality water than in clean environments.

All analysis of the chemical tolerance range indicated that both fish populations had wide range of tolerance in major chemical parameters. However, the C_a-population was distributed in both polluted regions and clean regions, while the C_c-population was distributed more widely in severely polluted regions than the clean regions. These results suggest that the C_c-population has adapted more to the polluted region than the C_a-population, and that Carassius auratus was classified as a tolerant species as shown in conventional classifications of tolerance guild (An et al. 2006). In addition, the two populations were observed less frequently in clean environments as shown in the Figs. 2 and 3.

Regression analysis on relative abundance of the C_a-population and the C_c-population to the water chemical parameters are shown in Figs. 4 and 5. The data analysis was conducted by dividing the intervals of water quality data uniformly. The maximum relative abundance (Max R_a) values of the fish population were obtained from the intervals of water quality (Figs. 4 and 5). These two populations were statistically significant in all water chemical parameters in the level of p = 0.05, and especially statistically significant in the level of p < 0.001. The C_a-population had TN, NO₃-N, TN:TP ratio, and SS, and the C_c-population had BOD, TN, NO₃-N, and electric conductivity. Determinant coefficients of the C_a-population in the linear regression analysis were suspended solids (SS: R² = 0.543), NO₃-N (R² = 0.479), total nitrogen (TN: R² = 0.433), TN:TP ratio (R² = 0.412), total phosphorus (TP: R² = 0.34), BOD (R² = 0.264), electric conductivity (R² = 0.21) and PO₄-P (R² = 0.15) (Figs. 4 and 5). This outcome indicates that suspended solids, dissolved phosphate and nitrogen in water columns are key determinants that cause variation in abundance of the C_a-population, and these factors were closely associated with the habitat of Carassius auratus (Kim and Park 2002, Lee and Noh 2006). The determinant coefficients of the C_c-population were electric conductivity (R²= 0.456), NO₃-N (R² =0.402), TN (R² = 0.366), suspended solids (R² = 0.353), PO₄-P (R² = 0.344), TP (R² = 0.329), BOD (R² = 0.296) and TN:TP ratio (R² = 0.282) (Figs. 4 and 5). Regression analysis of the C_c-population suggests that electric conductivity, dissolved nitrogen and phosphorus and suspended solids in the water column are primary components that influence the abundance of the C_c-population. Overall, this data suggests that dissolved phosphate, and volatile and non-volatile suspended solids are key determinants for the abundance of the two populations.

[Fig. 4.] Linear regression analysis on water quality parameters for oxygen demand (organic matter), N nutrient and the maximum relative abundance in Ca-population (Carassius auratus) and Cc-population (Cyprinus carpio).

[Fig. 5.] Linear regression analysis on water quality parameters of P nutrient, suspended solids, ionic content (Cond) and the maximum relative abundance in Ca-population (Carassius auratus) and Cc-population (Cyprinus carpio).

Trophic preferences of two fish populations were analyzed by comparing insectivore and omnivore species, and also, tolerance preferences were analyzed by studying sensitive and tolerant species (Fig. 6). Barbour et al. (1999) pointed out that trophic guild and tolerance guild of fish are closely associated with water chemistry parameters such as organic matter and nutrient (N, P) pollution. Previous research on biological integrity of fish showed that omnivore species and tolerant species are frequently observed in the polluted regions, while insectivore species and sensitive species are frequently observed in the clean riffle regions (An et al. 2001, An et al. 2006, Choi et al. 2011, Lee and An 2014). These attributes of the fish compositions are usually used for classification of fish guilds (Karr 1981, Klemm et al. 1993). As shown in Fig. 6, the relative abundance of the C_a-population decreased with the percentage of insectivore and percentage of sensitive species and increased with the percentage of omnivores and the percentage of tolerant species. The relative abundance of the C_c-population also decreased with the percentage of insectivore and percentage of sensitive species and increased with the percentage of omnivores and percentage of tolerant species, even when the distribution range of the C_a-population was wider than the C_c-population (Fig. 6).

[Fig. 6.] Tolerance gradients of sensitive and tolerant species (%) and trophic gradients of insectivore and omnivore species (%) co-occurring with Carassius auratus and Cyprinus carpio.

According to the analysis of ecological indicators co-occurring in the two populations (Fig. 7), the relative abundance of tolerant species co-occurring with the C_a-population and C_c-population was 80.2% and 89.1%, respectively. In contrast, the relative abundance of sensitive species that co-occurred with the C_a-population and C_c-population was only 4.5% and 1.7%, respectively (Fig. 7). Numerous studies of tolerance guild metric (Karr 1981, Klemm et al. 1993, Barbour et al. 1999) showed that the tolerance guild of fish is closely associated with water chemistry parameters such as organic matter and nutrient (N, P) pollution. So, tolerant species tend to increase in polluted streams while sensitive species decrease in polluted streams.

[Fig. 7.] Comparison between streams with Ca-population (Carassius auratus) and Cc-population (Cyprinus carpio) on ecosystem health, tolerance guild and trophic composition (Number of sample in Ca=416, Cc=105).

Analysis of the trophic guild indicator co-occurring in the two populations showed that the relative abundance of omnivore and insectivore species that co-occurred with the C_a-population was 74.5% and 16.4%, respectively (Fig. 7). In contrast, the relative abundance of omnivore and insectivore species that co-occurred with the C_c-population was only 79.8% and 11.9%, respectively (Fig. 7). These outcomes suggest that omnivore species dominated the community in both fish populations and the magnitude of omnivore dominance was greater in the C_c-population than the C_a-population. Thus, based on the trophic indicator (omnivore, insectivore), biological tolerance was greater in the C_c-population than the C_a-population. Our results are supported by the fact that the predominance of omnivore species in the fish community is a typical example of water pollution (Klemm et al. 1993) and a good indicator for ecological disturbance in the stream environment (Shin et al. 2013). Both C_a-population and C_c-population groups were considered to be tolerant species in the stream regions with organic matter pollution.

The model value of the Index of Biological Integrity (IBI) in habitats co-occurring with C. auratus and C. carpio averaged 15.0 ± 4.3 and 12.9 ± 3.6, respectively (Table 3). This indicated a fairly healthy condition in the C_a-population and poor health condition in the C_c-population according to the modified criteria of US EPA (Klemm et al. 1993). Thus, the ecological health degraded in the streams of two fish populations, and the magnitude of degradation of ecological health was greater in the C_c-population than the C_a-population (Table 3). The multi-metric indicator analysis showed that metric values of seven (except M₄) out of eight metrics (M₁- M₈) were lower in the C_c-population than the C_a-population.

[Table 3.] Comparison of fish community metrics according to Index of Biological Integrity (IBI) between streams and rivers with Ca-population and Cc-population

Comparison of fish community metrics according to Index of Biological Integrity (IBI) between streams and rivers with Ca-population and Cc-population

The largest difference (≥ 0.3) in the metric values between the C_a-population and the C_c-population was seen in metrics M₆, M₇, and M₈. These metric values suggest that both species are categorized as highly tolerant species in the waterbodies, and the magnitude of tolerance (pollution) based on the multi-metric model, was greater in the C_c-population than the C_a-population. Statistically independent sample t-tests showed distinct differences in metric values between the C_a-population and C_c-populations at a probability level of 0.05 in the seven metrics, except M₅, and at the level of 0.01 in the six metrics.

Qualitative Habitat Evaluation Index (QHEI) in the C_a-population and C_c-population were evaluated as shown in Table 4. The model value of QHEI co-occurring with C. auratus and C. carpio averaged 64.7 and 67 respectively, (Table 4), indicating that the physical habitat health was in good condition according to the modified criteria of US EPA (Klemm et al. 1993; Table. 4, Fig. 7). The most significant difference in mean habitat metric values between the C_a-population and the C_c-population was seen for sediment deposition (Q₆), which was 10.3 and 12.5 respectively (Table 4). But, the remaining mean habitat metric values were similar for both species. Epifaunal substrate/available cover (Q₁), embeddedness (Q₂), and channel alteration (Q₅) ranged 10.1-10.3 and 10.1-10.6, respectively. The physical habitat of the C_a-population and C_c-population was assessed to be in fair condition (Table 4). In the meantime, mean habitat metric values of the C_a-population and C_c-population for channel flow status (Q₃) and dam construction impact (Q₄) ranged from 11.1-11.4 and 11.6-11.9 respectively, (Table 4). The physical habitat of C_a-population and C_c-populations was judged to be in good condition.

[Table 4.] Comparison of habitat parameters according to Qualitative Habitat Evaluation Index (QHEI) between streams and rivers with Ca-population and Cc-population.

Comparison of habitat parameters according to Qualitative Habitat Evaluation Index (QHEI) between streams and rivers with Ca-population and Cc-population.

To determine differences in physical habitats between the two populations, statistical tests were conducted under the assumption of equal variance on all metrics. According to the results of habitat health, values of probability (P) ranged from 0.19 to 0.96, indicating no significant difference in habitat conditions between the C_a-population and C_c-population. The probability (P) value in the epifaunal substrate/available cover (Q₁) was the highest among the six habitat metrics and the metric reflects the artificial habitat constructions and habitat simplification as well. These results indicate that the C_a-population and C_c-population reflected characteristics of tolerant species very well. Both populations occupied similar habitats with slow current velocity or dams and reservoirs with sediment deposition (Kim and Park 2002, Lee and Noh 2006).

Tests on overall tolerance range showed that both the C_a-population and C_c-population can be classified as tolerant groups after the analysis of land use-pattern, stream morphology, chemical water quality, habitat parameters, and ecological indicators (trophic compositions and tolerance guilds) of co-occurring fish along with the multimetric ecological health model. The magnitude of the tolerance range, however, was greater in the C_c-population than the C_a-population, and this was evident in chemical variables such as organic matter pollutants (BOD, COD), nutrients (TP, TN), suspended solids, and ionic contents. In the past, all research just categorized these two species as tolerant species using empirical methods such as simple descriptions of fish habitat, chemical conditions and fish dictionaries. International or domestic references with quantitative analysis based on land use pattern, stream morphology, chemical water quality, habitat parameters, ecological indicators, and ecological integrity model values did not exist. This research provides a clearcut evaluation of tolerance ranges or guild classification of the C_a-population and C_c-population in relation to physical, chemical, and ecological indicators, which are closely associated with external tolerance in the aquatic environment.