3.1. Physico-Chemical and Microbiological Parameters
The results (range values) of physico-chemical and microbiological analyses from 25 sampling locations within Tiaoxi River across three seasons are summarized in
Table 2. The surface water temperature (WT) was in the range of 22.3–26.6 °C in autumn, 6–8.8 °C in winter, and 29–31.2 °C in summer. Water samples in all locations were within the pH range set by Ministry of Environmental Protection (MEP), People’s Republic of China (PRC) for surface water and also natural water’s pH limits (6.5–8.5) set for aquatic life and irrigation purposes [
42]. Most natural water systems require a pH range of 6.5–8 to support a diverse aquatic population [
43]. Significant statistical differences in pH were observed seasonally (
p < 0.05); however, no spatial variation was observed. EC values were within the range of 124–400 µS/cm in all the locations and the values varied significantly with seasons (
p < 0.05) (
Table 2). For most freshwaters, the EC ranges from 10 to 1000 µS/cm and elevated levels of above 1000 µS/cm can be seen in polluted water or water bodies that receive large quantities of land runoff [
42]. In streams and rivers, the conductivity is affected by various factors such as type of soils, bedrocks, and presence of inorganic dissolved solids. Sewage or wastewater could raise the conductivity due to the presence of chloride, phosphate, and nitrate [
44].
The seasonal and spatial variations in TN, TP, NO
3-N, NO
2-N, PO
4-P, NH
4-N, TOC, and Chl
a are shown in
Figure 2A–D and
Figure 3A–D. Among all these parameters tested, only TP, NH
4-N, and Chl
a showed significant (
p < 0.001) seasonal and spatial variations (
Table 2). The parameters TN, NO
3-N, NO
2-N, and TOC showed significant (
p < 0.05) spatial variations. Based on MEP guidelines, TN values were outside the acceptable limits (≥1 mg/L) in all locations for all seasons. The highest TN values (4.13 mg/L) were reported in samples collected from Location 18 in Autumn 2014 (
Figure 2A), and this location is near to a suburban/industrial area (junction of Changxing and Tiaoxi River) with ferry docking facilities. The sources for elevated levels of TN in water bodies include runoff from agricultural croplands and animal manure, discharge from wastewater treatment plants (WWTPs), and leakage from septic tanks [
45]. The presence of elevated levels of TN and ammonia in water is considered indicative of freshly polluted water by environmental management engineers [
46]. As per MEP guidelines, the acceptable TP levels for class III water bodies is <0.2 mg/L, but for lakes and reservoirs, the TP levels should be <0.05 mg/L. Here, TP levels were outside the acceptable range in Location 1 (Taihu Lake, ~1 km inside from Taihu Lake/Tiaoxi River junction) and the levels were closer to acceptable limits in Location 16 for the autumn (175.58 µg/L) and winter (187.42 µg/L) seasons (
Figure 2B). Location 16 is a suburban mixed residential and business area and the samples were collected at a junction between the main river and a canal that connects to Taihu Lake (
Table 2). Zheng et al. [
31] reported higher levels of TP and TN in Catchment 8 (urban land) of East Tiaoxi River; this area has two WWTPs, one of which is located near sampling locations 15 and 16 of the current study. The higher levels of TP observed at Location 16 could be due to effluents from the WWTP. As stated above, the concentrations of TP showed significant seasonal and spatial variation (
p < 0.001) and TP levels were comparatively high for most of the locations in winter 2015. Wang et al. [
27] reported similar TP levels for rivers surrounding Taihu Lake. Possible runoff from fertilized lawns and cropland, animal manure, and also domestic sewage entry into the water are likely causes [
47]. NO
3-N levels were within acceptable limits (<10 mg/L) as suggested by MEP but elevated levels of NO
2-N (>0.15 mg/L) were observed in Location 3 in autumn and in Locations 15 and 16 during the summer season (
Figure 2C,D). Sources of NO
2-N include human sewage, livestock manure, fertilizers, and erosion of natural deposits [
48]. Location 3 is a fishing village where people live on boats stationed at this location. Locations 15 and 16 are in a suburban area and the sampling was conducted in a junction between Tiaoxi River and a canal that connects to Taihu Lake where boats/ferries were docked. The higher levels of NO
2-N observed in these three locations may be due to the entry of human sewage into the water. A previous study showed that presence of higher concentrations of NO
2-N in water is a potential problem due to its toxicity to humans (more potential health effects are seen in infants) and livestock when consumed [
49].
NH
4-N levels were outside standard limits (>1 mg/L) in Location 16 in autumn and at Location 24 in winter (
Figure 3B). NH
4-N enters into water mostly from anthropogenic sources such as human sewage, municipal effluent discharges, livestock manure, and agricultural runoff. Elevated levels of NH4-N in surface water primarily exert toxic effects on the higher aquatic organisms such as fish and shrimps [
50]. Xu et al. [
51] reported similar results for surface water quality in the Taihu watershed. There are no specific standards for PO
4-P as per MEP, PR China, but a concentration of <20 µg/L is commonly present in streams and rivers. Elevated levels of >20 µg/L indicate pollution and can lead to excessive algal growth [
52]. In the present study, PO
4-P levels were high in all locations (
Figure 3A) on one or more occasions; however, no significant correlation between PO
4-P and Chl
a was observed (
Table 3). Similarly, for TOC there are no specific standards set by MEP, PRC. Both PO
4-P and TOC levels showed statistically significant (
p < 0.05) spatial variation (
Table 2,
Figure 3C,D). Most of the surface waters with low nutrient levels have Chl
a levels of <2.5 µg/L, but higher levels can be seen if there is high nutrient availability [
42]. In the current study, all locations showed higher Chl
a levels in all seasons, indicating high algal growth in Tiaoxi River water (
Figure 3D). Chlorophyll
a levels were high in the summer season followed by in autumn and winter, which can be correlated with warm temperatures in the summer and autumn seasons and the availability of nutrients. In general, the concentration of Chl
a was high if the location had a high TP concentration. Xu et al. (2010) [
53] reported similar results for Chl
a levels in Taihu Lake water. The prevalence of cyanobacteria and higher concentrations of Chl
a during the summer season in Taihu Lake has been reported previously [
23,
52]. The highest Chl
a levels were observed at Locations 1, 12, and 23 during the summer season, and Chl
a concentrations were statistically significant both spatially and seasonally (
p < 0.05) (
Table 2).
Microbiological Parameters
TVC was carried out to enumerate aerobic/facultative anaerobic mesophiles in the surface water, primarily to determine whether these counts showed any relationship with physico-chemical parameters and coliform counts. TVC values neither showed any seasonal or spatial significance statistically nor followed a similar trend to total coliform and fecal coliform numbers (
Table 2,
Figure 4A). As per MEP standards, the suggested standard limit for total/fecal coliforms for level III water bodies is <10,000/L (or <10/mL) but elevated levels of total coliforms were observed in all locations for all seasons, and much higher levels were observed in seven locations (Locations 2, 3, 5, 12, 15, 16, and 17) on one or more occasions with the highest at Location 16 (3.61 Log10 CFU/mL) during the summer season (
Figure 4B). The results correlate well with the land use pattern or possible mixing of waste in the above locations where either boats/ferries were docked or leakage of waste into the river through human activities was observed. Hagedorn and Liang [
28] also indicated a serious fecal contamination of Tiaoxi River and reported higher levels (2.54 log10 CFU/mL) of
E. coli for the water samples collected near Fengkou drinking water station. TC showed statistically significant (
p < 0.005) differences between locations (
Table 2,
Figure 4C). Previously, total coliforms were considered as bacterial water quality indicators to assess fecal contamination in recreational waters in the USA, as required by the Beaches Environmental Assessment and Coastal Health Act [
54] to reduce health risks. However, it was reported that some members of the coliform group live in the environment (i.e., outside of the gastrointestinal tract), which may show a false indication for fecal contamination in water [
55]. Therefore, TC counts are no longer used as an indicator for recreational waters as they are widespread in nature, but are still used to assess drinking water quality [
44] . Fecal coliform (FC) counts are used as guidelines for microbial water quality to assess fecal contamination. In the present study, most of the locations showed higher levels of FC (>250 CFU/100ml) in the winter and summer seasons (
Figure 4C) as compared to USEPA standards; however, no guidelines were suggested by MEP for FC in surface water in China. A high FC count was observed in five locations (Locations 2, 3, 5, 12, 16) on one or more occasions with the highest at Location 16 (3.62 log10 CFU/100mL) during the summer season. As indicated previously, these are the locations near residential areas where people either live on boats without adequate sanitation facilities, or urban residential areas with multiple waste inputs into the rivers such as leakage of waste from unknown sources. The higher levels of FC observed in these locations could be correlated with the discharge of effluent from a WWTP located near Locations 15 and 16. Only some fecal coliforms are pathogenic and a previous study showed that FC presence does not always correlate with pathogen presence [
56]. However, a high FC count implies impaired water quality and increased risk associated with the presence of pathogens [
57]. FC levels were comparatively higher in summer than in winter; this may be due to the runoff and heavy rainfall that occurred before summer sampling in 2015. The increased concentrations of fecal coliforms after rainfall events have been widely acknowledged in scientific literature [
4,
21,
58]. The higher levels of FC observed could also be due to warm temperatures, which can facilitate FC bacteria accustomed to such conditions [
59]. FC numbers showed significant spatial (
p < 0.05) variation.
3.3. Cluster Analysis for Spatial Grouping
Cluster analysis (CA) was applied to group sampling locations with similar water quality characteristics. A dendrogram generated by CA grouped the 19 locations into three clusters at (Dlink/Dmax) <60 (
Figure 5). The CA results are convincing, as the generated clusters share similar characteristic features and land use patterns. Based on the physico-chemical and microbiological results, each cluster was classified into respective pollution categories (
Supplementary Table S1). Cluster 1 includes eight locations (Locations 6–11, 13, and 17) and consists of mixed land use, either rural or urban/suburban residential areas with little industrial activity, corresponding to a relatively low level of pollution. Cluster 2 comprises four locations (4, 5, 15, and 16) which are mostly the junctions of East and West Tiaoxi River or other streams. These locations are predominantly close to urban and semi-urban residential areas with large-scale business, ferry transportation, and ferry docking activities, and in some of these locations, entry of wastewater to the river was noticed during sampling (
Figure 6). This cluster was classified as highly polluted based on the physico-chemical and microbiological results. Cluster 3 comprises seven locations (1, 2, 3, 12, 14, 20, and 21), and includes mixed land use and can be categorized as moderately polluted locations based on physico-chemical and microbiological analysis. Sampling locations 20 and 21 were close to sparse residential/industrial areas and Locations 2 and 3 are residential areas where a few people are living on boats (
Figure 6). The CA enabled us to categorize sampling locations based on water quality, so that in future studies, the number of sampling locations can be minimized for cost-effective monitoring of water quality in Tiaoxi River by choosing a few locations from each cluster based on the distance distribution and pollution levels in those locations. Previous studies have reported that a similar strategy has been successfully applied in water quality monitoring programs elsewhere [
12,
13,
61,
62], and the Tiaoxi River Taihu catchment is therefore similarly amenable to this rational approach.
3.4. Principal Component Analysis/Factor Analysis for Source Identification
Principal component analysis (PCA)/Factor analysis (FA) was performed using log transformed data to identify the factor(s) that influence the water quality during the entire sampling period and within the seasons (autumn, winter, and summer). Kaiser–Meyer–Olkin (KMO) and Bartlett’s tests were carried out to verify the suitability of data for PCA/FA. A KMO value of 0.5 or more is required to perform PCA and a lower KMO value indicates that the dataset is not suitable for PCA [
14].
In this study, the KMO value for the entire dataset was 0.53; however, the Bartlett’s test gave a
p value of <0.001, indicating the suitability of the data for PCA. The significance of the factor is evaluated by eigenvalue in PCA; the higher the eigenvalues, the higher the significance of factors, with 1.0 or greater eigenvalues considered significant [
12]. The PCA for the entire dataset yielded four PCs (with eigenvalues ≥1), which explained over 83% of the total variance in the dataset. The variable loadings on varimax-rotated PCs for the entire data set are provided in
Table 4. Variable loading is classified as “strong”, “moderate”, or “weak”, corresponding to their absolute loading values of >0.75, 0.75–0.50, and 0.50–0.30, respectively [
63].
The first component (VF1) accounted for 32.2% of the total variance and has strong positive loading for pH, EC, and TP, and strong negative loading for NO
3-N, indicating variability in physico-chemical sources (
Table 4). Normally, EC is used to indicate natural pollution and can be due to soil erosion or weathering effects on water quality during seasonal changes [
14]. This component also suggests that most of the variation is due to pH and EC changes. The second component (VF2) is responsible for 25.4% of the total variance and showed strong positive loading for TN, PO
4-P, and NH
4-N. This component also gave moderate negative loading to Chl
a, indicating nutrient pollution; this could be interpreted as influences from agricultural and domestic waste [
64]. The third component (VF3) explained 14.2% of the total variance and has strong positive loading for TVC and TC. This component also has moderate positive loading for Chl
a and represents influences of mainly microbial origin. The microbial factor TC can be associated with sewage pollution in the river. The fourth component (VF4), accounting for 12% of the total variance, has strong positive loading for WT and TOC. This component represents physico-chemical sources and could be interpreted as influences from organic pollution caused by domestic and industrial discharges. Similar results have been reported by other authors for water quality assessment by PCA/FA [
10,
64].