A periphyton-based index for biological integrity was developed to assess the ecological status of lakes and reservoirs in Baiyangdian Watershed, China. Periphyton and environmental surveys were conducted at 20 sites during 2009-2010. A total of 22 surveyed metrics of periphyton, which were classified into 6 categories: biomass groups, community structure, pigment ratios, bacteria diversity, enzyme activities, and metabolism were evaluated in order to understand their responsiveness to environmental degradation. Out of these 22 metrics, six metrics were selected to constitute the periphyton-based index for biological integrity: chlorophyll b, chlorophyll c: chlorophyll a ratio, alkaline phosphatase activity, bacteria dominance, net daily metabolism, and autotrophic index by sensitivity and redundancy analysis. Then the periphyton-based index for biological integrity was verified by an independent validation-site data set. The test results indicated the index could not only discriminate disturbed sites from reference sites, but also be triggered by land use types. The scores of periphyton-based index for biological integrity were significantly correlated with the environmental parameters, especially ammonia nitrogen and total phosphorus. These results suggest that this index may be a potential tool for assessing the ecological status of lakes and reservoirs, polluted mainly by organic contaminants in Baiyangdian Watershed.
Assessment of the influence of human disturbance on aquatic ecosystems, based on the characteristics of resident biota, has been an established tool since the 1900s. Many studies have reported correlations between variables or a set of variables describing environmental degradation and shifts in algae, macrophytes, invertebrates, or fish assemblages (Kerans and Karr, 1994; Buffagni et al., 2004; Johnson and Buchanan, 2014; Dupler et al., 2015; Petesse et al., 2016). The index of biological integrity (IBI), which was originally based on fish communities (Karr, 1981), is a comprehensive, multi-metric index that represents the ecological status of aquatic ecosystems. In recent years, more IBIs were developed based amongst others on macrophytes (Umetsu et al., 2018), plankton (Pomari et al., 2019), macroinvertebrates (Nestlerode et al., 2019) and periphyton (Tan et al., 2015). Periphyton has some potential advantages as an indicator: Periphyton occurs in a variety of habitats and accounts for the majority of primary production in shallow lakes and rivers (Burkholder and Wetzel, 1990). It is also easy to collect and handle. Moreover, periphyton sampling usually causes minimal impact on the resident biota (Omar, 2010). Periphytic algae are sensitive to change in water quality and, in particular, respond rapidly to nutrient enrichment in lakes. Their taxonomy, pigments or photosynthetic capacity are sensitive to subtle changes in environmental conditions and thus may respond to disturbances that may not visibly affect other communities (Burns and Ryder, 2001; Allan and Castillo, 2007; Sabater et al., 2007; Hambrook, 2010). The short generation times of periphyton allow them to respond more rapidly changes in water quality than macrophytes or fauna. Thus, periphyton has increasingly been used as an integrated bioindicator for the evaluation of the aquatic ecosystem health (Artigas et al., 2010; Ma et al., 2011).
As periphyton is responsive to many environmental factors (Stevenson, 1996), it could be useful for assessing the effects of various human alterations on aquatic ecosystems. The primary pollutants of Baiyangdian Lake are organic pollution and nutrients, because of their relation to duck breeding, aquaculture and domestic wastewater. In this study, we investigated the periphyton of reservoirs and lakes in the Baiyangdian Watershed during 2009–2010, with the goal of developing a periphyton-based index of biological integrity (PER-IBI) as an alternative index which could evaluate the ecological status of Baiyangdian Lake, particularly in respect of organic pollution.
Materials and methods
The Baiyangdian Watershed (37.8°–40.4°N; 113.3°–116.6°E; Figure 1) is experiencing extreme water shortage. Four rivers (Ping River, Zhulong River, Pu River and Tang River) that flow into Baiyangdian Lake have completely dried up recently (Long et al., 2010). Only Fu River is perennial, whereas the remaining three inflow rivers (Cao River, Daqing River and Xiaoyi River) have become to be seasonal. To alleviate the shrinking of Baiyangdian Lake, several intra-basin and inter-basin water transfer projects have been undertaken; both Xidayang Reservoir and Wangkuai Reservoir are supporting to Baiyangdian Lake. Xidayang Reservoir, located in Tang River, has a drainage area of 4420 km2, with total storage capacities of 12.58 × 108 m3, Wangkuai Reservoir is located in Daqing River, with 13.89 × 108 m3 total storage capacities. Moreover, the only perennial river, Fu River, is seriously polluted. Large amounts of untreated industrial and municipal wastewater is discharged directly into Fu River and eventually into Baiyangdian Lake. This amounts to a CODMn and BOD5 discharge varied from 1630 to 2243 t and from 997 to 1583 t per year, respectively (Qiu et al., 2009). The average concentrations of Zn, Cu, Pb and Cd in sediments from Fu River were 1147.1, 190.5, 109.7 and 3.3 mg kg−1 dry weight, respectively (Hu et al., 2011).
We investigated 20 sites and collected 360 samples (20 sites × 3 replicates × 6 times) in 3 sub-watersheds of the Baiyangdian Watershed during 2009–2010; each site was designated as either reference or impaired based on its water quality and land use type (Table 1). Both reservoirs are important drinking water source of Hebei province and drinking water conservation areas. All activities that will destroy the water environment or its surrounding forests are forbidden, pollutants are prohibited to discharge, chemical fertilizers are not allowed to abuse and aquatic organisms are protected in conservation areas. 12 sites in the reservoirs (Figure 1A,B) were designated as reference sites, and 8 sites in Baiyangdian Lake (Figure 1C) were designated as disturbed. The collected samples were randomly divided into either development-site (development of PER-IBI) or validation-site data set (testing of PER-IBI), and each dataset included samples from 3 sub-watersheds. Ten sites (6 reference; 4 disturbed) were designated as the development-site dataset, while the other 10 sites (6 reference; 4 disturbed) were used as the validation-site dataset.
Periphyton sampling and processing
Periphyton was sampled during August and November in 2009, and April, June, August, and November in 2010. Periphyton samples were collected using activated carbon fiber (2 × 10 cm, TK-1600, specific area 1450–1550 m2 g−1, pore size 18–21 A; Jiangsu Tongkang Activated Carbon Fiber Co., Ltd. China) as substratum (Wang et al., 2013). The carriers with 10 activated carbon fiber coupons were placed horizontally and positioned at a depth of 20 cm. The fibers were placed vertically and exposed to water for 15 days. Periphyton on the fibers was scraped off, and the fibers were repeatedly washed with sterile water to ensure that no residues remained on the fibers. Then, the samples were divided into two parts, P1 and P2. 5% formalin solution in filtered river/lake water was added to P1, which was used to study algal composition. P2 was used for other analysis, including chlorophyll a (chla), chlorophyll b (chlb), chlorophyll c (chlc), ash-free dry weight (AFDW), polysaccharide content (PSC), algal density, extracellular enzyme activities of β-glucosidase (GLU), leucine-aminopeptidase (LEU), and alkaline phosphatase (ALP), measurements according to Ma et al. (2011), and the samples stored at -20 °C until analysis. The autotrophic index (AI), which is an indicator of the relative importance of autotrophs vs. heterotrophs and the detritus in the periphyton community, was calculated as the ratio of AFDW and chla (Weber, 1973). Bacillariophyta (BAC), Cyanophyta (CYA), and Chlorophyta (CHL) were identified at the phylum level using a microscope (OLYMPUS BX41) at 400× magnification, and the proportions of cell number of BAC, CYA, and CHL to the sum of three algae were calculated.
Bacterial community structure was characterized using a DNA fingerprinting method. DNA extraction was performed according to (Dumestre-Toulet and Kintz, 2001). Following extraction, DNA concentration was determined by fluorimetry using the DNA Quantitation Kit Fluorescence Assay (Genomic DNA Purification Kit, #K0512, Pure Extreme®, Fermentas, CAN). Thereafter, the periphyton bacterial diversity indices were calculated following PCR-DGGE analysis (Muyzer and Smalla, 1998). The gel image was captured using a CCD camera and image analysis was performed using Gelquest 2.7.3 software (Sequentix, GER), which allows fragment detection and quantification. The values of peak density and relative peak height of each sample were used to assess diversity (Shannon-Wiener) (H), richness (Margalef) (Dmg), evenness (Shannon-Wiener) (E) and dominance (Berger-Parker) (d) of periphytic bacteria communities (Bunn et al., 2010).
Periphyton metabolism was measured over 24 h using a pair of enclosed chambers which contained the artificial substrata (Bunn et al., 2010). Estimates of daily respiration (R24) were obtained by determining the change in dissolved oxygen (DO) over 24h in darkness. GPP was determined by integrating changes in DO during daylight hours and adding R24 for the same period. Net daily metabolism (NDM) was calculated as the difference between GPP and R24. The results were represented as the change in DO concentration over time per square meter of subsample area per hour (mg O2 m−2 h−1) (Fellows et al., 2006).
Measurement of physicochemical factors
Secchi depth (SD) was measured by Secchi disk, water temperature (T), pH, electricity conductivity (EC), and dissolved oxygen (DO) of water were directly determined in the field using a YSI 6920 Multi-parameter Water Quality Probe. Water samples for chemical analysis were collected in triplicates to be filtered and frozen in the laboratory until analysis. Total nitrogen (TN), total phosphorus (TP), nitrate (NO3-N), ammonium (NH4-N), chemical oxygen demand (CODMn), biochemical oxygen demand (BOD5), fecal coliform bacteria (FC), oil, mercury (Hg), cadmium (Cd), chromium (Cr), lead (Pb), and anionic surfactants (AS) were determined according to standard protocols (Gilcreas, 1985).
Development of periphyton-based index for biological integrity
A total of 180 samples (development data set) was used to develop the PER-IBI. We started with developing initial attributes (total 22 metrics), which were classified into 7 categories: biomass groups (AFDW, chla, chlb, chlc and PSC), periphytic algae density and relative abundance (AD, BAC, CHL, CYA), pigment ratios (chlb: chla, chlc: chla), bacteria diversity (H, Dmg, E and d), enzyme activities (ALP, GLU and LEU), metabolism (GPP, R24 and NDM), and community structure (AI). Later on, a careful selection of the metrics was performed aiming to assure that various aspects of the periphyton communities were assessed. Sensitivity and redundancy tests were performed to select the metrics for PER-IBI.
Firstly, potential metrics were identified when the test showed significant differences (p < 0.05) between reference and disturbed sites using the non-parametric Mann-Whitney test during both summer and winter. The separation power (SP) and the seasonal stability index (SSI) were used to select the final metrics from the set of potential metrics. SP is defined as the degree of overlap between boxes (25th and 75th percentiles) in box plots of the metric values for reference and disturbed sites. Metric sensitivity was evaluated according to box and whisker plots (Barbour et al., 1996). We assigned SP = 3 when boxes did not overlap between the reference and disturbed site groups, SP = 2 when interquartile ranges overlapped but did not reach medians, SP = 1 when only one median was within the interquartile range of the other box, and SP= 0 when both medians were within the range of the other box.
Secondly, a redundancy analysis was performed by the Spearman correlation test, with pairs of metrics regarded as sensitivity, in order to simplify the index, reduce the cost of analysis, and avoid redundant information. When one or more metrics showed redundancy by being highly correlated (Spearman r > 0.80, p < 0.05), only one metric was chosen from each category into the index, and its correlation coefficients should be higher with water quality parameters (Casatti et al., 2009).
Index development and validation
Then a 0–10 scaling system was used to standardize the ranges of the selected metrics, as recommended by Hill et al. (2000). Metrics that decreased with impairment were scored from 0 to 10 by dividing the metric value by the 90th percentile of all sites, and multiplied by 10. For metrics that increased with impairment, the metric value was divided by the 90th percentile of all sites and subtracted from 1, and multiplied by 10. Similar percentiles have been used in other studies as benchmarks for ecological assessment (Wu et al., 2012). The total PER-IBI values were the sum of the selected metrics’ scores based on the 0–10 scaling system.
The PER-IBI ability to discriminate between reference and impaired was tested using a validation dataset. The calculation and scaling systems of the final metrics for PER-IBI were similar as those used for the development dataset. Besides SP, two other methods were used to evaluate the PER-IBI at the validation. The PER-IBI scores should be significantly correlated with the environmental variables. Spearman rank analysis and linear model were used to examine the correlations of PER-IBI and the metrics that constituted PER-IBI with water quality parameters for all sample sites, including the development and validation data sets. Nonparametric Mann-Whitney test and Spearman rank correlation test were performed by SPSS 20.0. Relationships between Periphyton metrics and environmental variables were examined by the CANOCO software version 4.5. Metrics data from periphyton samples were log10(x +1) transformed to normalize data before being included in the analysis. We carried out redundancy analysis (RDA) on periphyton metrics data which was constrained by environmental variables. Forward selection was used to reduce the environmental variables that significantly explained the distribution pattern of periphyton at a cut-off point of p = 0.1. The significance of the RDA axes was assessed using the Monte Carlo permutation test (999 unrestricted permutations). Bonferroni correction was used for correcting probabilities for multiple comparisons. To clarify the response of periphyton to environmental change, RDA of the attributes matrix constrained by environmental variables were performed.
The environmental variables varied widely across sites (Table 2). Reference sites showed higher pH, DO, and SD, but lower EC, NH4-N, TP, COD, BOD5 and heavy metals than those from disturbed sites.
Development of periphyton-based index for biological integrity
17 metrics were able to discriminate among reference and degraded sites in both summer and winter according to Box-and -Whisker plots, with a SP ≥ 2 between reference and disturbed sites, and statistically different according to the Mann-whitey U-test (p < 0.05), except PSC, chlb：chla, GLU, R24 and NDM (Table 3).
With the redundancy test results considered, six metrics that had less occurrence of obtaining redundancy value (r > 0.80, p < 0.05) were considered for the index development—chlb, chlc: chla, ALP, d, NDM, and AI. These metrics represented different characteristics of periphyton communities and showed less variation among seasons. Box plots of PER-IBI scores showed good separation between reference and impaired sites (Figure 3).
Validating periphyton-based index for biological integrity
PER-IBI and its metrics were tested with 10 sites by using the validation-site dataset. The application of PER-IBI, calculated by the same criteria and scaling systems as for the development-site dataset, indicated significant differences between reference and disturbed sites for both summer and winter (Mann-Whitney test, p < 0.05), except for AI in winter, SP ≥ 2 (Figure 4). 6 metrics and PER-IBI were highly correlated with environmental variables (Table 4). PER-IBI differentiates between reference and disturbed sites indicated the suitability of the developed PER-IBI for the context of the present work (Figure 5).
Assessing the results of the study area
The final PER-IBI scores for the study areas showed a wide range of values from 1.51 to 53.27 in 2009 and from 1.69 to 51.77 in 2010 (maximum 60.00, Table 5). Overall, the final PER-IBI scores were classified into 4 categories which were >90th, 50th-90th, 25th-50th and <25th percentiles of the site values, respectively (Zalack et al., 2010), “good” (49.80-60.00), “fair” (46.60-49.80), “poor” (32.30-46.60), and “very poor” (0-32.30). In table 5, the data was better in 2009 than in 2010. Generally, the ecological status of Baiyangdian Lake was “poor” or “very poor”.
In our study, PER-IBI was highly correlated with environmental variables (Table 4). RDA analysis showed that axis1 explained 89.8% of the total variance (eigenvalue of 0.698) and axis 2 explained 5.6% (eigenvalue of 0.043, Figure 6A). The first axis was mainly related to SD, DO, CODMn, NH4-N and TP, which were reflecting the trophic state, axis 2 was related to heavy metals (Cd and Pb) and anionic surfactants. These environmental variables related to trophic state have very important effects on PER-IBI and its metrics. In our study, chlb, ALP and d increased with nutrients, while NDM and PER-IBI decreased with trophic states. And PER-IBI was positively correlated with SD and DO, and highly negatively correlated with CODMn, NH4-N and TP (Figure 6A). PER-IBI on the other hand had less relation to Pb and negative correlation with Cr. This indicated that nutrients played a greater role on periphyton than heavy metals in Baiyangdian Lake. Figure 6B showed that PER-IBI was capable of indicating the influences from anthropogenic activities. Less disturbed regions distributed to the left of axis 1, associated with high chlc：chla and NDM, and less chlb, d, AI and ALP. The PER-IBI scores of reservoirs (i.e. Protection areas of drinking water source) were higher than other land use types, whereas the PER-IBI scores of cropland and residential areas were lowest.
Multi-metric IBIs based on epilithic algae have been used as tools for monitoring stream/lake health for a long time in the USA and European countries (Griffith et al., 2005; Hill et al., 2000). Because of the requirement of high expertise and the lack of relevant experts on algae identification, PER-IBI is rarely applied in monitoring and assessment in China. We are trying to construct an alternative PER-IBI, which is more dependent on instruments than on professional skills. In this study, PER-IBI included 6 individual metrics (chlb, chlc:chla, ALP, d, NDM, and AI) covering different attributes, which allowed responses to different types of pressures. For example, chlb is positively correlated with nutrients; ALP is useful as an indicator of phosphorus limitation in freshwater ecosystems (Mulholland and Rosemond, 1992; Scott et al., 2008). The dominance of periphytic bacteria is a good metric for ecosystem assessment, because bacterial dominance means the presence of bacterial taxa showing varying tolerance to a wide range of stresses, including resistance to extreme temperatures, pH, and concentrations of heavy metals (Satyanarayana et al., 2005). AI is positively correlated with organic contamination (Biggs and Close, 1989); NDM of periphyton is in response to abundant nutrient levels in a freshwater ecosystem (McCormick et al., 2001). In general, the PER-IBI separated reference sites from impaired sites effectively, while it also had strong relationships with water quality parameters.
PER-IBI is more correlated to the trophic state of Baiyangdian Lake, than the presence and concentration of heavy metals. These observations were consistent with the studies by Köhler et al. (2010) and Montuelle et al. (2010). In Lake Baiyangdian, the main pollutants are nutrients and organic matter. There are more than 100,000 people living around Lake Baiyangdian and nutrient concentrations are higher due to the domestic effluents discharged into the lake from residential zones, and development of fisheries and duck feeding also add nutrients. Although heavy metal would reduce periphyton’s biomass, enzyme activities and metabolic, higher nutrient level could mitigate toxic effects of metallic pollutants (Serra et al., 2010; Guasch et al., 2004). In this study, there are significant differences of PER-IBI among different land use types, indicating the PER-IBI could reflect the disturbance degree to some extent.
PER-IBI was significant positive correlated with other index, which based on phytoplankton, and zoobenthos biodiversity, used in Baiyangdian Lake (Table 6). And PER-IBI classifications were concordant with the assessments of Baiyangdian Lake using other index (Wang et al., 2011; Xie et al. 2010; Xu et al. 2012a; Xu et al. 2012b). Both Wangkuai Reservoir and Xidayang Reservoir were in “good” state, and Baiyangdian Lake was in “poor” state, especially in sampling site 13, 14, 15 were in “very poor” state.
In this study, relatively small number of sampling sites were used for developing PER-IBI. To our knowledge, there is no exact definition on the minimal sampling size for developing PER-IBI. Yoder and Rankin (1995) suggested that more than 40 sites would need to be sampled to obtain the adequate dataset for developing a reliable index. Our index was developed on 20 sampling sites only, therefore, our number of sampling sites might be insufficient. Although PER-IBI had good performance in Baiyangdian Lake, more sampling sites and long-term time series of sampling would improve the overall performance, as a potential alternative index of the index. Thus, further work is required to test its suitability in a wider statistical series in more sampling sites.
Our constructed periphyton-based index of biotic integrity (PER-IBI) could be an alternative tool to assess ecological status of Baiyangdian Lake owing to its satisfactory performance on organic pollutants and/or trophic status. Furthermore, we propose to test and extend the tested area to other aquatic ecosystems in this and other regions. The index could be applied to the whole watershed without basic conceptual changes by adjusting the metrics of periphyton attributes. And finally, PER-IBI could be a generalized tool for water assessment for organic pollutants at regional or national level after application to additional sites that vary in pollutant levels.
This study was financially supported by the National Natural Science Foundation of China (Grant No. 41401615) and the National Nonprofit Research Institutes Fundamental Research funding (CAFINT2013K06)