Material and Methods
Study area
The study was conducted in Lianhuan Lake. Lianhuan Lake, formed by tectonic slump, consists of 18 small lakes around, is located in the most low-lying center of the Songnen Plain in eastern China (Fig. 1). The lake has a mean depth of 2.14 m, a maximum depth of 4.6 m, and a surface area of 580 km2 (G. Yu, Harrison, & Xue, 2001). Lianhuan Lake become the first international waterfowl and hunting ground since 1985 in China. It lies in the cold temperate zone, characterized by semiarid climate. The average annual precipitation in the Lianhuan Lake catchment is 400 mm, 70 % of which occurs in the summer (June to August). During March to May it is generally dry with a highest frequency of dust storms. Apart from direct precipitation, the lake also receives water from Wuyuer and Shuangyang rivers through the southern part. Rapid economic development around the lake has led to large quantities of wastewater, from agricultural, industrial, and domestic sources, being released into the lake. In this study, 13 small lakes around the Lianhuan Lake were considered. The 13 lakes are different in natural forms and area and they exhibit different characters such as nutrients levels, fisheries community, and human disturbance patterns. Interestingly, these lakes are connected to each other. Among the 13 small lakes covered in this study, Habuta lake is geographically located far from the other lakes. Delong Lake, Yangcaohao Lake and Beiqin Lake are located in the upper reaches of Lianhuan lake. Durbote county is located to the east of these lakes (R. Wang et al., 2019).
Macroinvertebrate sampling
Sampling took place in spring (June), summer (August), and autumn (October) of 2021 at the at each site (Fig. 1). At each site, quantitative Petersen grab samples (0.025 m2) were collected in replicates and processed through a 300 μm mesh size sieve. Petersen grab and the mesh size sieve were visually inspected to ensure macroinvertebrates adhering to the grab and sieve were transferred to the composite sample. All materials were placed into a plastic jar and preserved in 80% alcohol. In the laboratory, all organisms in the samples were counted, weighed to the nearest 0.1 g and identified to genus or to the lowest taxonomic level possible. Species abundance for each sampling site was calculated as a density (individuals m-2) and biomass (gm-2) was calculated by adding the biomass of all species. The identification and classification of the macroinvertebrates was done according to (Dudgeon, 1999; Morse, Yang, & Tian, 1994; H. Wang, 2002)
Environmental variables
The Lake surface area was calculated using ArcGIS (ver. 10.7). At each sampling site, water temperature (WT,℃), conductivity (COND,μS/cm), pH, and dissolved oxygen (DO,mg/L) were measured in the field using a portable YSI Professional Plus instrument (made in the USA). Water depth (WD, m) was measured with a Speedtech handheld depth finder (made in the USA). Water was collected directly into 5-L polypropylene bottles at a depth of 0.5 m to quantify water chemistry factors. The water chemical including total phosphorus (TP, mg/L), total nitrogen (TN, mg/L), ammonium (NH4-N, mg/L), nitrate (NO3-N, mg/L), nitrite (NO2-N, mg/L), chemical oxygen demand (CODMn, mg/L), Chlorophyll a concentration (Chla, mg/L), and Suspended substance (SS, mg/L), were measured in the laboratory based on standard methods (Federation & Association, 2005). Chlorophyll a concentration (Chla, mg/L) was determined according to the protocols for standard observation and measurement in aquatic ecosystems by filtering 100 mL of the sampled water through GF/C whatman filter. Pigments extraction was done in 90% aqueous solution of acetone and Chla concentrations were measured spectrometrically
Classification of macroinvertebrate communities using Self-Organising Map
In order to analyze the classification macroinvertebrate communities, Self-organizing map (SOM) was applied for classifying sampling points according to the species’ abundance. SOM is an effective method for cluster analysis, which has a high explanatory ability in the study of ecological population(Giraudel & Lek, 2001). SOM consists of two layers of neurons, namely the input layer and the output layer connected by connection intensities (weights). Input layers acquire information from a data matrix, and output layers visualize the computational results (Song et al., 2007). In this study, the input layer was composed of species abundance and 74 sample points. The number of neurons in the output layer was determined in advance according as 5 ×\(\sqrt{\text{numberofsamplepoints}}\) ≈ 43 (Park, Céréghino, Compin, & Lek, 2003), and according to the minimum quantization error and the minimum topographic error (see Appendix S1 in Table S1). The optimal number of neurons in the output layer was determined to be 49 (Kohonen, 2001). Since SOM output layer has no distinct classification boundaries.K -means clustering analysis was performed on the SOM output layer neurons and classified the neurons into different groups. The Simple structure index (SSI) which is the value indicating the relative importance of each species in determining the distribution patterns of the samples in the SOM was then used to determine the optimal number of groups (Park et al., 2006). Therefore, the larger the values of SSI the higher the clustering quality (Dimitriadou, Dolničar, & Weingessel, 2002; Park et al., 2006).
Indicator species is a species closely related to environmental changes, used to measure the specificity and fidelity of a species to a certain environmental state (McGeoch, Van Rensburg, & Botes, 2002). Indicator values (IndVal) of all species in each group were calculated to determine the indicator species in each group. A 1000 permutations were performed to access the significance (P < 0.05) of the IndVal observed for each species (Arimoro & Keke, 2021). The calculation formula of IndVal was done as follows (Dufrêne & Legendre, 1997):