Investigation method
From March 2015 to December 2018, we surveyed 36 important tributaries of the TGR (Fig. 3) and conducted an investigation of macroinvertebrates. For the sake of convenience, we labeled tributaries from the reservoir dam to its tail area sequentially as R01–R36, i.e., R01 (Xiangxi River), R02 (Qinggan River), R03 (Shennong River), R04 (Baolong River), R05 (Guandu River), R06 (Daning River), R07 (Daxi-F River), R08 (Caotang River), R09 (Meixi River), R10 (Changtan River), R11 (Modao River), R12 (Tangxi River), R13 (Xiao River), R14 (Zhuxi River), R15 (Rangdu River), R16 (Ruxi River), R17 (Huangjin River), R18 (Dongxi River), R19 (Chixi River), R20 (Long River), R21 (Bixi River), R22 (Quxi River), R23 (Zhenxi River), R24 (Wu River), R25 (Lixiang River), R26 (Longxi River), R27 (Taohua River), R28 (Yulin River), R29 (Wubu River), R30 (Changtang River), R31 (Chaoyang River), R32 (Jialing River), R33 (Huaxi River), R34 (Yipin River), R35 (Daxi-J River), and R36 (Qi River).
Schematic diagram of important tributaries of TGR and sampling points (plotted by ArcGIS 10.5, https://www.32r.com/soft/16101.html).
This study was approved by the Environmental Protection Bureau of the Three Georges Reservoir.
A total of 175 sampling points were set up in all tributaries. Four parallel samples were taken from each sampling point. At least one sample was taken from each microhabitat (mainly including four microhabitats, i.e., shoal, deep pool, pebble and aquatic habitat). Parallel samples from the same sampling point were mixed together. The quantitative and qualitative sample collection methods were combined in this study. The quantitative collection was performed first, and then the qualitative collection for the same sampling point. The qualitative samples were collected by D-net. Quantitative samples of wadable sampling points were collected using a Surber net with an area of 0.3 m × 0.3 m. Quantitative samples of non-wadable sampling points were collected using a D-net with a bottom side length of 0.3 m. The collected samples were put into sample bottles (bags) and fixed with 5% formaldehyde solution. Then the samples were identified and classified under the laboratory conditions.
Selection of reference sites and impaired sites
A reference site refers to a sampling point with no or little anthropogenic disturbance, while a impaired site refers to a sampling point subject to obvious anthropogenic disturbance29. A total of 15 reference sites and 160 impaired sites were selected from 175 sampling points based on anthropogenic disturbance, vegetation coverage, population distribution, and the distribution of industry and agriculture in the vicinity of the sampling site6,8 (Table 7, Supplementary Fig. 1).
Creation and selection of the assessment metric index system
With reference to the river health assessment indexes in China13,15,18,24, North America6,24 and Europe5, and based on the ecological characteristics such as species composition and abundance, sensitivity, tolerance and functional feeding groups, we constructed 26 candidate metrics (Table 8) for B-IBI. These candidate metrics have significant or noticeable response to human activities, and normally, can be applied to relatively large geographic areas; therefore, they can be used to indicate the ecological quality of rivers6,23,24. Among these metrics, 17 were associated with species composition and abundance, which included the total number of taxon, the number of EPT taxa, the number of crustacean and mollusca taxa, the number of ephemerida taxa, the number of pteroptera taxa, the number of trichoptera taxa, the number of diptera taxa, the number of chironomidea taxa, the percentage of EPT, the percentage of crustacean and mollusca, the percentage of ephemerida, the percentage of pteroptera, the percentage of trichoptera, the percentage of dipteral, the percentage of chironomidea, the percentage of oligochaeta and the Shannon-Weiner diversity index. Species composition and abundance-related indexes reflect the diversity of macrobenthic communities. An increase in species diversity is associated with the improvement of community health, which indicates that the niche space and food sources are sufficient to support the survival and reproduction of multiple species. The candidate metrics related to sensitivity and tolerance in this study were the number of sensitive taxa, the number of tolerant taxa, the percentage of dominant species and the percentage of the top three dominant species. Different zoobenthos show different degrees of sensitivity and tolerance to the influencing factors in the river habitat, for which these characteristics can be used to assess the health status of the river. In addition, the taxa and percentage of functional feeding groups are closely associated with their living environment, and the parameters that were used to represent functional feeding in this study were the percentages of shredders, herbivores, filterers, scrapers and predators. Some of the representative images of the identified taxa were shown in Supplementary Fig. 1E,F.
The selection of core metrics for B-IBI mainly includes three steps: analysis of distribution range of candidate metrics, analysis of discriminant ability of candidate metrics and analysis of correlation between candidate metrics23.
Analysis of distribution range of candidate metrics
According to the numerical value of each biological metric in the reference site, an initial analysis was conducted to exclude the following two types of metrics: metrics with excessive nought values, which did not meet the requirement for a universal applicability; metrics with a scatter value distribution, and a standard deviation greater than or equal to the mean, indicating that the standard deviation of this value was relatively big and unstable, thereby unsuitable to be used as biological metrics6.
Analysis of discriminant ability of candidate metrics
After analyzing the distribution range of candidate metrics, those unsuitable for biological evaluation were eliminated. The distribution of the remaining eligible metrics for the reference site and the impaired site was analyzed using the box-plot, to mainly compare the distribution range of the 25th quantile to the 75th quantile of the reference site and the impaired site and the overlap of “box” InterQuartile Range (IQR), and judge which biological metrics could best distinguish between the reference site and impaired site. An IQ value ≥ 2 indicates a small overlapping part between the reference site and the impacted site, which means a significant difference in the related parameter between the reference site and the impacted site, suggesting a noticeable response to human activity6,24. The IQ scoring criteria were as follows6,24: 3 point, no overlapping between the two box bodies; 2 points, the box bodies have a small part of overlapping, but the median of neither body falls within the limits of its counterpart; 1 point, most parts of the box bodies overlap, and the median of at least one box body lies within the limits of its counterpart; 0 points, one box body falls within the limits of the other, or the medians of each body are within the other’s limits.
Correlation analysis of candidate metrics
Pearson correlation analysis was further performed of the metrics that met the preliminary conditions. If the correlation coefficient (left| {text{r}} right|) between two metrics is greater than 0.75, and they are intrinsically linked. Then most of the information reflected is overlapping. Therefore, it is OK to select one of them. If no intrinsic connection is found between two metrics, then both metrics can be selected even if the correlation coefficient is greater than 0.758.
After screening through the above three steps, core metrics of the B-IBI are finally determined.
Construction of B-IBI
The core biological metrics screened out by the above method were used as the metrics for final biological assessment. The metrics used for biological assessment were standardized using the ratio scoring method, to unify the evaluation metric23.
- (1)
For a metric that decreased with increasing interference, the metric was normalized by dividing the value of this metric at each sample point with the 95% quantile of all sample points:
$${text{V}}_{{text{i}}}^{prime } = {text{V}}_{{text{i}}} /{text{V}}_{{{95}% }} ;$$
- (2)
For a metric that increased with increasing interference, the metric was normalized by using the 5% quantile of this metric at all sample points as the reference object:
$${text{V}}_{{text{i}}}^{prime } = left( {{text{V}}_{{{text{MAX}}}} – {text{V}}_{{text{i}}} } right)/left( {{text{V}}_{{{text{MAX}}}} – {text{V}}_{{{5}% }} } right),$$
where Vi′ is the normalized value of the metric at the ith sampling point; Vi the actual value of the metric at the ith sampling point; V95% the 95% quantile of the metric; V5% is the 5% quantile of the metric; VMAX is the maximum value of this metric in all sampling points. The health thresholds of 5% quantile and 95% quantile can eliminate extreme abnormal values and retain most of biological information.
B-IBI assessment criteria
The 95% quantile of B-IBI distribution of all the sections/tributaries used for the health threshold can eliminate extreme abnormal values and retain most biological information. The distribution range lower than this value is divided into four portions, and the quartile close to the 95% quantile indicates a small disturbance. The biological integrity grade and the corresponding range of IBI6 are determined according to the 95% quantile and the quartile value, and the section/river health was classified into five grades, namely, excellent, good, fair, poor and very poor.
Source: Ecology - nature.com
