Conductivity elevation produces osmotic stress to aquatic biota and then alters biological communities. The responses of stream fish to conductivity remain unclear and strategies for protection are poorly developed. We collected data of fish and conductivity from sixty-two sites of the Taizi River to evaluate the changes to the fish community and species along the gradient of conductivity. Our results found that conductivity elevation was related to the regional development of urban and farmland and the local degradation of habitat quality. The community metrics of abundance and F-IBI, but not species richness and diversity, showed a significant linear correlation with conductivity. Conductivity of the top three F-IBI grades (excellent, good and normal) was significantly lower than those of the other two F-IBI grades (poor and bad). The boundary conductivity between normal grade and poor grade was approximately equal to 500 μS cm−1. We found different probability patterns for different species along the conductivity gradient; one capture probability pattern showed decline trend along the conductivity gradient. Except for two dominating and widespread species and one tolerant species, the remaining fish species of the first pattern should be designated as protection objects. In order to protect fish community integrity and sensitive species, sustainable land use management on the catchment scale and habitat quality improvement on the local scale should be given more attention by catchment managers.
The study of abnormal elevation in stream conductivity within a landscape has a long history (Hart et al., 1991), but has become an urgent ecological issue in recent years in light of global secondary salinization of freshwater environments (Cañedo-Argüelles et al., 2013) linked to urban development, agricultural production, industrial and mining activities, and the use of road salts (Daniel et al., 2002; Daley et al., 2009; Lindberg et al., 2011; Maloney and Weller, 2011). Increased salinity can affect a wide variety of stream life, including fish, by challenging inner osmotic balance, raising energetic cost, retarding growth performance, and altering community structure (Dunlop et al., 2005; Cañedo-Argüelles et al., 2012). Salinity is an indicative measure of the total ions concentrations that include cations (Na+, Ca2+, Mg2+, K+) and anions (Cl-, SO42-, CO32-, HCO3-), which can be surrogated as conductivity or total dissolved solids (Dunlop et al., 2005). For these reasons, conductivity is often used by managers to investigate and model the origin, transport and future trend, and evaluate the effects on aquatic organisms (Muschal 2006; Kimmel and Argent 2010).
The Taizi River catchment is located in the Liaoning Province of China, where there used to be an important production base for cereal crops and high-quality rice and a heavy industry base in the northeastern area. It is now undergoing a kind of complex development pattern including rapid urbanization, high-yield agricultural activity, and various resource extractions such as mining and oil wells (Ren et al., 2012; Ding et al., 2013; Zhang et al., 2015). These developments have been implicated in the elevation of conductivity that can be observed from historical data (Table 1). Detrimental effects of the increased conductivity on aquatic life of the Taizi River have not been well documented (Zhao et al., 2016).
The elevation of conductivity has significant effects on various aquatic organisms, from algae through to invertebrates and vertebrates (Dunlop et al., 2005). Physiologically, increased conductivity is linked to osmotic stress, which causes the loss of sensitive species or replacement by tolerant species which then change the community structure (Cañedo-Argüelles et al., 2013). Sensitive species (species objects) and community structure (community goals) of damaged streams are essential contents of biological conservation and catchment management (Simon et al., 2003). Fish is often used as a tool of bioassessment on water quality, particularly because of their responses to environmental pressure on long temporal scales (Karr 1981; Resh 2008). Increasingly, researchers have focused on the responses of fish to the gradient of conductivity (Weber-Scannell and Duffy, 2007; Kimmel and Argent, 2010). Several studies have found that conductivity is harmful to stream fish (Hitt et al., 2016), if regional conductivity continues to increase or remains high (Morgan II et al., 2012). As a result, comparison of the responses of fish community and individual species to the conductivity gradient can provide greater understanding of biotic degradation and effects of pollution, as well as facilitate management priorities aimed at protection and restoration of specific fish species and/or community in these damaged streams.
We chose the Taizi River as a study case, since the catchment has spatial differences in the degree and type of anthropogenic disturbance (Zhang et al., 2015). The Taizi River catchment is a sub-catchment of the Liaohe River basin, located in northeastern China. The length of the Taizi River is 413 km and the catchment area is approximately 13,880 km2. This catchment belongs to the humid and sub-humid climates of a warm temperate zone, and is characterized by regional variation of landscapes. The upper mountainous region is predominantly natural vegetation cover, and human activities are broadly restricted for the purpose of ecotourism. The middle hilly region includes large-scale agricultural activities, while the lower plain region is characterized by industrial development and urbanization. Although there was a lack of historical records on stream conductivity of the Taizi River, it was not hard to speculate that conductivity increased during the last four decades judging by the changing trends of the main ions (Table 1). The increased conductivity was related to the anthropogenic disturbance at the middle and lower regions.
The first objective of this study was to examine the relationship of conductivity to anthropogenic disturbance which might be the potential driver for the input of salinities. The second objective was to detect the effects of increased conductivity on fish community structure and capture probability of individual species that aided in the establishment of protection goals for catchment management.
Materials and Methods
A total of seventy study sites were sampled in the Taizi River catchment on Aug. 2009 and Oct. 2010 (Figure 1). For each study site, conductivity was measured by YSI Pro 2030 multi-parameter water quality analyzer in the field. 1 L of stream water sampled and taken back to laboratory with low temperature was used to measure the concentrations of the main ions (i.e. potassium, calcium, sodium, magnesium, chloride, sulfate, bicarbonate and carbonate) (see Supplementary materials).
Fish were collected by a backpack electro-fisher within a 300-m survey reach for wadeable streams. The electrofishing was conducted for 30 min following the zig-zag path within the survey reach. Gillnets were set up at both ends of the sampling reach in order to reduce the chance of fish being able to evade them as much as possible. In order to remedy the shortage of electro-fishing for non-wadeable streams, two types of gillnet (3 cm mesh and 6 cm mesh) were used. After the fish collection, species identification and biological information were recorded immediately; then all fishes were released except for unknown species, which were fixed in formaldehyde solution and taken back for further identification.
Fish community metrics
Four types of fish community metrics such as species richness, abundance, and Shannon diversity index (H’) and fish index of biotic integrity (F-IBI) were used in this study. The first two metrics were calculated from the field statistical records. H’ was calculated in Primer 5 software using ‘e’ as the base of the logarithmic calculation. F-IBI was calculated with the method by Karr (1981). Firstly, the sites with the least disturbance were selected as reference sites with the method by Blocksom et al., (2002). Secondly, four candidate variables (e.g. species richness, benthic fish proportion, tolerant fish proportion and viscid-egg fish proportion) screened by discriminant analysis and correlation analysis (Breine et al., 2004) were used for calculating F-IBI scores. Finally, F-IBI scores were classified into five healthy grades (excellent, good, normal, poor and bad) with the method by Baek et al., (2014).
Anthropogenic disturbances of regional and local scales
The significant effects of urban and farmland catchments on stream fish assemblage were well documented (e.g. Allan, 2004; Meyer et al., 2005), and so were chosen as anthropogenic disturbances of regional scale. SPOT5 (Satellite Positioning and Tracking 5) images with resolution of 2.5 m were used for catchment land use characterization. The multi-spectral and panchromatic bands were subjected to the geometric correction and Albers projection, and then were merged by ERDAS (Earth Resources Data Analysis System). Land use types were interpreted and obtained from the fused map. The unidentified land use information was confirmed by field survey. The data of land use extracted by ArcGIS were used to calculate the proportions of catchment urban and farmland for all sites.
Stream habitat quality assessment was conducted according to actual conditions in the field by the same person at local (reach) scale using the method by Barbour (1996) with appropriate modification (Zheng et al., 2007). This assessment had ten indices. The score of each individual index was between 0-20, while the total score was between 0-200. A score close to 200 indicates very good habitat quality, whereas a score close to zero means very poor habitat quality. In this study, the total score (Habitat Condition Score) and one index score indicating the intensity of human activities (Human Disturbance Score) were used for representing anthropogenic disturbances of local scale.
A weight-of-evidence method was used to evaluate the relationship between potential drivers (pH, water temperature, DO, TP, NH4+-N, fecal coliform and habitat quality) and conductivity before analysis. Only one potential driver, NH4+-N, was found to be biologically significant and eliminated by removing sites with NH4+-N > 2.0 mg l−1 (Jia et al., 2017). Finally, 62 of 70 sampling sites were used to explore the responses of steam fish to the gradient of conductivity in the Taizi River. A taxon was discarded from analysis if it was observed at < 5 sites to down weight the influence of opportunistic salt-tolerant and non-native organisms. The distance weighted least squares regression (Garson, 2013) was used to examine the responses of stream conductivity to anthropogenic disturbances at the local and regional scales, and the effects of conductivity on the metrics of fish community (STATISTICA 7.0). Mann-Whitney U test was used to assess the differences of conductivity levels between different grades of fish integrity (SPSS 18.0). The combination of generalized additive model (GAM) and logistic regression model (LRM) were used identifying the response features of individual fish species to increased conductivity in R.3.2.3 software (R_Core_Team, 2014). We used the 90% confidence bounds of GAM (3 degrees of freedom) to stimulate the fish potential distribution characteristics, and then LRM was used to calculate the capture probability along the gradient of stream conductivity for individual fish species (US Environmental Protection Agency, 2011). Statistical significance of all analysis was accepted at p < 0.05.
General fish assemblages
In this study, a total of 6704 individuals belong to 25 species were collected in the Taizi River. Based on the spatial heterogeneity of environmental quality and fish distribution, the conductivity ranges in which fish lived were different among the all species (Figure 2). Some species, e.g. Hypseleotris swinhonis, Huigobio Chinssuensis, Pungitius pungitius, Leuciscus waleckii, Phoxinus Lagowskii and Lampetra morii, lived in a narrow conductivity range. Conservely, other species, e.g. Rhodeus lighti, Gobio cynocephalus and Oryzias latipes, were distributed over a relative wide conductivity range.
Stream conductivity, land use and habitat quality conditions
The regression models certified the significant responses of stream conductivity to anthropogenic disturbances at both the regional and local scales (p < 0.05). At regional scale, the conductivity increased with the development of urban and farmland. Conductivity had a very strong positive correlation with proportions of urban area (r = 0.81) (Figure 3a) and a strong positive correlation with proportions of farmland area (r = 0.66) (Figure 3b). At the local scale, habitat condition score (r=−0.46) (Figure 3c) and human disturbance score (r=−0.45) (Figure 3d) had a medium negative correlation with conductivity, respectively. The response of conductivity to anthropogenic disturbances at the regional scale was stronger than to those at the local scale.
Stream conductivity and fish community metrics
The responses of fish community metrics to stream conductivity were different in that there were no significant responses of species richness and Shannon diversity to increasing conductivity (Figures 4a and 4c), whereas significant correlation between conductivity and abundance or F-IBI was found (p < 0.05) (Figures 4b and 4d). The number of fish species decreased gradually along the gradient of stream conductivity (Figure 4a). The similar trend was also found in the metrics of abundance and F-IBI (Figures 4b and 4d). The negative effect of the increasing conductivity on F-IBI (r=−0.65) was stronger than that on abundance (r=−0.28), indicating F-IBI was a better predictor for stream conductivity.
Conductivity exhibited obvious increasing trend from the excellent grade to bad grade of F-IBI (Figure 5), and Mann-Whitney U test showed that conductivity levels in the first three grades (excellent, good and normal) were significantly lower than those in the poor and bad grades (Table 2). If the stream conductivity exceeded 500 μS cm−1 approximately, F-IBI would decline to the poor grade (Figure 6), indicating more management needs and concerns should be paid on these damaged streams.
Fish species capture probability
The combination of GAM and LRM were used to calculate the species capture probability along the conductivity gradient. Three typical change types of species capture probability were shown in Figure 7, suggesting different species responding differently to stream conductivity. The first type showed that the capture probability decreased continuously along the conductivity gradient, such as P. lagowskii (Figure 7a). Fish belonging to the first type could be considered as sensitive indicator species of stream conductivity. The second type had an opposite change trend that revealed captured probability increasing along the gradient, such as Carassius auratus (Figure 7b), indicating a tolerant species of conductivity. The third change type of capture probability exhibited an optimum conductivity at 400 μS cm−1 and then a decline along the gradient, such as Zacco platypus (Figure 7c), indicating a medium sensitive species. Among the 25 fish species, 11 species exhibited the first change type of capture probability, while 6 species and 8 species belonged to the second and third change types, respectively (Supplementary materials).
Our results illustrate that stream conductivity is related to catchment land use (i.e. urban and farmland). The landscape characteristics, such as urbanization and agricultural production, can increase the stream conductivity level through surface runoff importing a large number of solutes (Cañedo-Argüelles et al., 2013; Hitt et al., 2016). Lerotholi et al., (2004) found that the crop productions needed much more groundwater irrigation, in which just a fraction of salts originated groundwater were absorbed by crops and a lot of salts were imported in the rivers through runoff, especially in the arid and semi-arid regions. The similar effects of agricultural activities on conductivity can also be found in other rivers (Isidoro et al., 2006; Koç, 2008). Additionally, the impacts of urbanization on conductivity have been well documented. In one hand, the stream conductivity is directly affected by the effluents from resident regions or sewage treatment plants (Cañedo-Argüelles et al., 2013; Chusov et al., 2014). The municipal effluents contain a large number of ions which could increase 1.3 to 10 times of stream conductivity after mixing with other drainage waters (Chusov et al., 2014). On the other hand, the impervious covers improve the hydrological connectivity between land and water body, which make it easier to input terrestrial solutes (e.g. road salts) into stream (Kaushal et al., 2014). A link has been found between stream chloride and road salt application (Shaw et al., 2012). Morgan II et al., (2012) also have reported that there is a strong correlation between stream chloride concentrations and impervious cover and/or road density. The above studies indicate that the stream conductivity is a good indicator to regional landscape pattern.
Our results have found that the degradations of riparian habitat contribute to the increasing of conductivity. The metric of human disturbance score used in this paper reflects the intensity of human activities influencing on the vegetation cover ratio in the riparian. In general, the streams characterized by high percentages of riparian vegetation cover have low levels of conductivity (Kasangaki et al., 2008), because a large amount of terrestrial solutes can be trapped by the riparian buffer, where is a key retention area of terrestrial ions (Madden et al., 2014). The serious human disturbances in riparian will decrease the proportion of vegetation cover and then improve conductivity level (Kasangaki et al., 2008; Madden et al., 2014). Collins et al., (2013) have reported that the restoration of riparian vegetation is not necessarily to reduce conductivity level in streams since the restoration occurring at insufficient width is not enough to resist the impacts of intensive human activities. It means that the riparian buffer could trap the terrestrial solutes when the anthropogenic disturbance intensity must decline to a certain level.
It has been suggested that the concentrations of dissolved solutes (i.e. conductivity, salinity) are affected by the local environmental process (flow regimes) and the regional environmental features (urbanization) simultaneously (Vogt et al., 2016). Our results also indicate that the increased conductivity is affected by catchment land use and local habitat quality. For the water quality management, nevertheless, the stream conductivity is a sensitive indicator of anthropogenic disturbances at both scales in the Taizi River catchment and thus we suggest that conductivity should be incorporated into the routine monitoring of surface water in China.
We used four fish community metrics for detecting the biological responses to conductivity, and found a similar changing trends for species richness and abundance, which showed firstly increased and then decreased along the conductivity gradient (Figure 3). This trend can be explained by the Subsidy-stress Gradient Model (SGM) developed by Odum et al., (1979). SGM is that some disturbances will provide the extra usable resources (i.e. organic matters, nutrients) that have a subsidizing effect on biological community if the inputs can be processed by the resident community, otherwise the inputs will have a detrimental effect and lead to negative biological response. A small amount of terrestrial solutes inputs may have an enriching effect on biological assemblage. Meanwhile, the low conductivity levels induced by terrestrial solutes inputs do not cause negative biological response or fish can resist this physiological disturbance.
The metrics of abundance and F-IBI showed significant responses to conductivity in this study. Compared to the metric of abundance, F-IBI has a relative higher correlation with the conductivity (Figure 3). F-IBI is a multimetric indices originally developed by Karr (1981), and is also an important component of ecological recovery endpoints since it can test whether a stream reach has been restored to similar integrity levels of fish community (Simon et al., 2003; Hall et al., 2014). Our results suggest that stream conductivity exceeding 500 μS cm−1 impair fish community integrity and the healthy conditions changed from normal grade to poor grade (Figure 6). The similar results reported by Morgan II et al., (2012), who found the conductivity values between 230-540 μS cm−1 caused the degradation of fish community integrity. Hitt and Chambers (2014) found that fish community composition would change when conductivity exceeded 600 μS cm−1. There was also a very different result that conductivity impairment to stream fish community was in the range of 3000-3500 μS cm−1 (Kimmel and Argent, 2010). We speculate that the major reason for different results may be related to the different fish community metric (the coefficient of community loss, Kimmel and Argent, 2010). According to the above results, we think that F-IBI is a better indicator for conductivity elevation since it can identify the fish community impairment at low conductivity ranges, and since it has relative well correlation with stream conductivity.
The response of macroinvertebrate community to conductivity has been well documented by the previous studies, which have found that increased conductivity caused a loss of sensitive taxa, especially in Ephemeroptera (Lindberg et al., 2011; US Environmental Protection Agency, 2011). The responses of fish species to conductivity are less understood (Weber-Scannell and Duffy, 2007; Hitt et al., 2016). Our results found there were three typical change types of species capture probability along the conductivity gradient. The first type reflects the decreasing of individual abundance with the increasing of conductivity. Fishes of the first type are sensitive indicator species of conductivity, and should be regarding as the potential protection objects in the Taizi River. Among the 11 fish species characterized by the first type of capture probability, three species of them are not suitable for potential protection objects, including P. lagowskii, Barbatula barbatula nuda and Misgurnus anguillicaudatus. The first two species are either widespread species or dominated species with the relative abundance of 37.6% and 16.8%, respectively, in the Taizi River, and thus they do not need too much conservation requirements in recent years. In addition, M. anguillicaudatus is a tolerant species which prefers silt bottom, and has successfully invaded many countries around the world (Urquhart and Koetsier, 2014). We speculated that our poor collection efficiencies in the lower reaches with silt bottom and deep water resulted in M. anguillicaudatus being wrongly classified in the first change type of capture probability. The remaining species might be regarded as potential protection objects in the Taizi River catchment. Thereinto, L. morri, Odontobutis yaluens and Squalidus chankaensis prefer to forest headwater reaches characterized by high dissolved oxygen, low turbidity and water temperature (Ding et al., 2013; Zhang et al., 2015). The population resources of these fishes have significantly reduced during the last several decades, L. morri in particular, which has been designated as a threatened species by the local government.
Although these fish species of the first type of species capture probability along the conductivity gradient should be regarded as the potential protection objects, it needs much more times and works on fish biology and ecology studies to make protection plans for each species that will be difficult to come true. Our results showed that the conductivity increasing was related to the catchment urbanization and local riparian development. The sustainable urbanization development on the catchment scale and the habitat quality improvement on the local scale are both effective ways to control conductivity level and reduce damages on sensitive fish species of conductivity. Specifically, rural sewage facilities construction and riparian naturalization should be considered as measures for entire catchment management.
Stream conductivity is a good indicator for land use development at the regional scale and habitat quality degradation at the local scale. The sensitive responses to anthropogenic disturbances suggest that stream conductivity could be play a bigger role in the future routine monitoring programs. The conductivity alters fish community, especially in fish integrity since the poor and bad grades can be separated from other grades along the stress gradient. Urbanization contributing to the increase of conductivity (Figure 3) should be paid more attentions for protection fish community integrity in this catchment. For example, rural sewage facilities construction and urban stream buffer zone naturalization should be considered as measures for improved land use management. Three typical change types of species capture probability have been identified in this study. The first type indicates that 8 fish species might be designated as potential protection objects in the Taizi River. Our results provide important management components of both species- and community-levels in the Taizi River catchment, and indicate that protection plans should be developed by the catchment managers shortly.
This study was supported by grants from the National Natural Science Foundation of China (41401066) and the National Natural Science Foundation of China (41571050).
Supplemental data for this article can be accessed on the publisher's website.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/uaem.