Influence of environmental variables on macroinvertebrate community structure in Lianhuan Lake

Abstract The structural characteristics of the macroinvertebrate community can effectively reflect the health status of lake ecosystems and the quality of the lake ecological environment. It is therefore important to identify the limiting factors of macroinvertebrate community structure for the maintenance of lake ecosystem health. In this study, the community composition of macroinvertebrate assemblages and their relationships with environmental variables were investigated in 13 small lakes within Lianhuan Lake in northern China. A self‐organizing map and K‐means clustering analysis grouped the macroinvertebrate communities into five groups, and the indicator species reflected the environmental characteristics of each group. Principal component analysis indicated that the classification of the macroinvertebrate communities was affected by environmental variables. The Kruskal–Wallis test results showed that environmental variables (pH, total phosphorus, nitrate, water temperature, dissolved oxygen, conductivity, permanganate index, and ammonium) had a significant effect on the classification of the macroinvertebrate communities. Redundancy analysis showed that mollusks were significantly negatively correlated with pH and chlorophyll a, while annelids and aquatic insects were significantly positively correlated with chlorophyll a and dissolved oxygen. Spearman correlation analysis showed that the species richness and Shannon's diversity of macroinvertebrates were significantly negatively correlated with total phosphorus, while the biomass of macroinvertebrates was significantly negatively correlated with pH. High alkalinity and lake eutrophication have a serious impact on the macroinvertebrate community. Human disturbances, such as industrial and agricultural runoff, negatively impact the ecological environment and affect macroinvertebrate community structure. Thus, macroinvertebrate community structure should be improved by enhancing the ecological environment and controlling environmental pollution at a watershed scale.


| INTRODUC TI ON
Lake ecosystems are typical spatially heterogeneous ecosystems formed by interactions between the lake biological communities and their environments. This dynamic process can be illustrated using four-dimensional changes in biological and environmental elements (Protasov, 2008;Ward, 1989). A lake in a state of natural evolution, undisturbed by humans, possesses a reasonable structure and perfect function; such a lake is in a healthy state (Beck & Hatch, 2009;Cooke et al., 2016). Because of the increase in the human population and the rapid development of industrial production and urbanization, the human demand for water resources has been increasing in recent decades. Problems such as the overexploitation of water resources and environmental pollution have become increasingly prominent (Conor, 2015). As a significant component of the Earth's freshwater resources, lakes also face these problems. The structure and function of lake ecosystems have been seriously damaged, and their health is gradually deteriorating (Likens, 2010;Tilzer & Serruya, 2012).
Macroinvertebrates fulfill various important roles in lakes ecosystems, such as altering the geochemical condition of the sediment, promoting nutrient cycling, and facilitating the transfer of energy within food webs Odountan et al., 2019;Vaughn & Hakenkamp, 2001). Macroinvertebrates are also commonly used as indicators of aquatic ecosystem integrity because of their wide range of sensitivity to pollution and relative longevity (Du et al., 2021;Richman & Somers, 2010;Selvanayagam & Abril, 2016). The distribution of macroinvertebrate communities in aquatic systems is strongly affected by both natural factors and human activities (Yu et al., 2020). These natural factors mainly include water temperature, water depth, dissolved oxygen, pH, and the spatial heterogeneity of habitats (Bazzanti et al., 2009;Free et al., 2009;Shostell & Williams, 2007). Human activities, such as agricultural production, indirectly affect the community structure of macroinvertebrates by altering the nutrient levels within lake waterbodies (Wijesiri et al., 2018). Previous studies have shown that there is a strong correlation between macroinvertebrate communities and environmental factors, especially in streams, rivers, and lakes (Kłonowska-Olejnik & Skalski, 2014;Li et al., 2018;White & Irvine, 2003;Zhang et al., 2021). Thus, analyzing the response relationships between macroinvertebrates and environmental factors has guiding significance for improving the ecological environment of lakes.
Lianhuan Lake, located in Heilongjiang Province, is the largest alkaline lake in northern China. It was formed by a tectonic slump and consists of 18 small lakes (Wu et al., 2012). Lianhuan Lake plays important roles in climate regulation, groundwater replenishment, habitat biodiversity, and economic resource development in northern China (Ma et al., 2018;Wang et al., 2011). However, in recent years, problems such as biodiversity loss and habitat degradation have intensified owing to agricultural reclamation in the lake catchment area and tourism activities within the lake (Li et al., 2009;Sun et al., 2010). Li (2013) showed that Lianhuan Lake has the highest eutrophication level in northeast China. Chen et al. (2020) showed that the zooplankton community in Lianhuan Lake was in the degradation stage. At present, most of the studies conducted in Lianhuan Lake have focused on the assessment of heavy metals and nutrient elements in core sediments (Sun et al., 2010;. In this study, the community composition of macroinvertebrate assemblages and their relationships with environmental variables were investigated in 13 small lakes within Lianhuan Lake. The 13 small lakes were chosen because they belong to the same watershed but differ in their natural conditions (e.g., lake morphology and environmental characteristics; see Table 1), which could result in different relationships between their macroinvertebrate assemblages and environmental variables. The aims of this study were (a) to represent the spatial and temporal patterns of macroinvertebrates from 13 lakes in Lianhuan Lake across three seasons, to permit better understanding of the relationships between environmental variables and macroinvertebrates; (b) to identify the indicator taxa characterizing each community group; and (c) to determine the important environmental variables shaping the community structure of macroinvertebrates in Lianhuan Lake. We expected that withinlake environmental variables would explain a substantial amount of the variation observed in macroinvertebrates, providing a scientific basis for protecting the health and biodiversity of the ecosystems in Lianhuan Lake.

| Study area
The study was conducted in Lianhuan Lake, which was formed by a tectonic slump. It consists of 18 small lakes in the low-lying center of the Songnen Plain in eastern China. The lake has a mean depth of 2.14 m, a maximum depth of 4.6 m, and a surface area of 580 km 2 (Yu et al., 2001). Lianhuan Lake has become the first international waterfowl hunting ground in China since 1985. It lies in a cold temperate zone characterized by a semiarid climate. The average annual precipitation in the Lianhuan Lake catchment is 400 mm, 70% of which occurs during summer (June to August).
From March to May, it is generally dry, with frequent dust storms.
Apart from direct precipitation, the northern part of the lake also receives water from the Wuyuer and Shuangyang rivers. Rapid

| Macroinvertebrate sampling
Sampling was conducted at each site in spring (June), summer (August), and autumn (October) of 2020 ( Figure 1). The macroinvertebrate samples were collected using a modified Petersen grab (0.0625 m 2 ). At each site, replicate quantitative Petersen grab samples were collected and processed through a 500μm mesh sieve.
The sieve and Petersen grab were visually inspected to ensure that macroinvertebrates adhering to the grab and sieve were transferred to the composite sample. All collected materials were placed in a plastic jar and preserved in 80% alcohol. At the laboratory, all the organisms in each sample were counted, weighed to the nearest 0.1 g, and identified to genus or the lowest possible taxonomic level. The species abundance of each sampling site was calculated as the density (individuals/m 2 ), and biomass (g/m 2 ) was calculated by adding F I G U R E 1 Map showing the locations of the macroinvertebrates sampling sites in Lianhuan Lake the biomass of all species. Macroinvertebrates were identified and classified according to Liu et al. (1993), Morse et al. (1994), Tang (2006, and Wang (2002).

| Environmental variables
The lake surface areas and perimeters were calculated using ArcGIS

| Classification of macroinvertebrate communities using a self-organizing map
To classify the macroinvertebrate communities of the sampling sites according to species abundance, a self-organizing map (SOM) was applied. SOMs are an effective cluster analysis method with high explanatory ability in the study of ecological populations (Giraudel & Lek, 2001). The SOM consisted of two layers of neurons, the input layer and the output layer, connected using connection intensities (weighted connections). Input layers acquire information from a data matrix, while output layers visualize the computational results (Song et al., 2007). In this study, the input layer comprised species abundance data and 74 sample sites. The number of neurons in the output layer was determined in advance, according to 5 × √ number of sample sites ≈ 43 (Park et al., 2003) and the minimum quantization and topographical errors (see Appendix S1). Thus, the optimal number of neurons in the output layer was determined to be 49 (Kohonen, 2001). The SOM output layer had no distinct classification boundaries. K-means clustering analysis was performed on the SOM output layer neurons to classify the sites into different groups. The simple structure index (SSI), which indicates 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). The larger the SSI value, the higher the clustering quality (Dimitriadou et al., 2002;Park et al., 2006).
Indicator species are closely related to environmental changes and are used by measuring the specificity and fidelity of a species to a certain environmental state (McGeoch et al., 2002). The indicator values (IndVal) of all species in each group were calculated to determine the indicator species in each group. One thousand permutations were performed to assess the significance (p < .05) of the IndVal observed for each species (Arimoro & Keke, 2021). The calculation of the IndVal was performed as follows (Dufrêne & Legendre, 1997): where IndVal ij is the IndVal of species i in site cluster j. A ij is a measure of specificity; Nindividuals ij is the mean number of individuals of species i across the sites of group j, and Nindividuals i is the sum of the mean number of individuals of species i across all groups. B ij is a measure of fidelity; Nsites ij is the number of sites in cluster j where species i is present, and Nsites j is the total number of sites in that cluster. B ij is high when species i is present in all objects of cluster j. Indvals greater than 50% were regarded as indicator species. Five macroinvertebrate community indices (species richness, abundance, biomass, Shannon's diversity, and Pielou's evenness) were also calculated.

| Statistical analysis
Prior to analysis, the macroinvertebrate abundance data were logand Hellinger-transformed for SOM and redundancy analysis (RDA), respectively. With the exception of pH, all environmental variables were log-transformed to satisfy the normality and variance assumptions before performing RDA and principal component analysis (PCA). Before RDA, the gradient lengths were measured using detrended correspondence analysis. As the first gradient length was <4, a linear method was applied. The statistical significance of the species-environment correlations for each ordination axis was also determined based on 999 Monte Carlo permutations, and the eigenvalues of the first two axes were used to measure their importance (Ter Braak & Verdonschot, 1995).

| Environmental variables
With the exception of WT, the environmental variables showed significant differences (p < .05) among the lakes (Table 1). The PCA using 13 environmental variables explained 44.9% of the data variability along the first two axes (axis 1 = 23.8% of the total variance with an eigenvalue of 3.10, axis 2 = 21.1% of the total variance with an eigenvalue of 2.74). Axis 1 was positively correlated with NO 2 -N, TP, pH, SS, Chla, and DO, but negatively correlated with TN and WD.
Axis 2 was positively correlated with WT, NH 3 -N, and COD Mn , but negatively correlated with NO 3 -N and COND ( Figure 2).
The SSI showed that the clustering quality was the highest when the neurons in the SOM output layer were divided into five groups   The RDA ordination of macroinvertebrate composition with respect to environmental variables is presented in Figure 6. Stepwise forward selection yielded four environmental variables that were significant to the model. These variables were WT, pH, DO, and Chla ( Figure 6). These four variables accounted for 77% of the total variance in the macroinvertebrate species composition. The first RDA axis, which explained 45.3% of the total variability, was negatively correlated with WT, whereas the second axis, which explained 32.4% of the variability, was positively correlated with pH.

| Relationship between environmental variables and macroinvertebrate community structure
Among the strongest species-environmental associations, we found that mollusks such as G. albus, Radix pereger, and Stenothyra glabra were significantly positively correlated with WT and negatively cor- The Spearman correlation analysis (Table 3) showed that macroinvertebrate community indices were significantly affected by environmental variables. Macroinvertebrate abundance was most F I G U R E 3 Simple structure index (SSI) and optimal group number (a) Based on SOM neuron K-means cluster analysis.

| DISCUSS ION
Understanding the ecological status of lakes helps determine the ecosystem services provided by them (Grizzetti et al., 2016).
Therefore, it is important to assess these factors. The ecological status and water quality of a lake are affected by complex interactions between environmental variables. Understanding the relative effects of these environmental variables is a necessary step in determining the activities required for lake management. The macroinvertebrate community is an ideal indicator as they respond to a wide variety of physical, chemical, and biological factors (Rai et al., 2019). This study aimed to investigate the response of macroinvertebrate community structure to environmental variables in Lianhuan Lake.
The spatial patterns of the macroinvertebrate communities were fact that these sampling sites were located near the river mouth, which discharges water into the lakes from surrounding agricultural farms Xiao et al., 2014). The indicator species for Group I were Chironomidae and Corbicula fluminea, which are relatively tolerant to eutrophic conditions (Gong et al., 2001). C.
fluminea has also been shown to tolerate low temperatures (Gerard et al., 2009;Liu & Xiong, 2008). Group II, which mainly included sampling sites in Habuta Lake, was characterized by high COD Mn , TP, NH 3 -N, and NO 3 -N. Interestingly, the abundance of annelids such as B. sowerbyi and Herpobdella sp., which are associated with excess lake nutrients (Cai et al., 2017;Du et al., 2021), was remarkably high in Group II. This implies that, despite being located far away from the other lakes, the water sources of Habuta Lake are the same as those of the other lakes. Habuta Lake could be in a state of degradation owing to eutrophication.

Significant variability of environmental variables was recorded
in Group IV, which encompassed sampling sites from the eastern part of Lianhuan Lake. COD Mn , TP, and NH 3 -N were markedly high, while DO was notably low, likely because of the dominance of human activities such as crop farming east of the lake (Xiao et al., 2014 (Huang et al., 2014;Johnson et al., 2007;Mathur et al., 2008). Such increases have been shown to increase nutrients and change macroinvertebrate compositions (Johnson et al., 1993;Kubosova et al., 2010;Yu et al., 2016). G. pervia, Clinotarypus sp., and Tanypus sp.
were grouped into Group IV, possibly because they are tolerant of high levels of pollution (Wang, 2003). These species respond to organic pollution by increasing their abundance and their prevalence in Group IV supports this fact. They can live in extremely polluted waters with very low oxygen levels (Uwadiae, 2016).
This study also revealed that Delong Lake, Yangcaohao Lake, and Beiqin Lake, clustered in Group V, were characterized by good water quality, despite them being located in the upper reaches of Lianhuan Lake and sampled during summer when large amounts of surface runoff is discharged. This clearly demonstrates spatial variation in water quality. The fact that many indicator species were identified in Group V also implies that the conditions allowed many organisms to thrive.
Macroinvertebrates are an important part of the lake ecosystem, and their community structure characteristics are related to lake environmental variables. This study revealed that pH, TP, NO 3 -N, WT, DO, COND, COD Mn , and NH 3 -N had a significant effect on the classification of macroinvertebrate communities ( Figure 5). pH is known to influence the composition and abundance of macroinvertebrate communities. A study by Tamiru (2019) indicated that alkaline water reduces macroinvertebrate abundance, biomass, and diversity. Moreover, tolerance studies have revealed that tolerance to pH varies between macroinvertebrate species (Ormerod et al., 1987). Based on the RDA and Spearman correlation analysis, the biomass of the macroinvertebrates, especially that of mollusks, was significantly negatively correlated with pH in this lake. This could be attributed to the high pH values (8-10) which are experienced in the lake year-round (Li et al., 2009). Extreme pH environments can have direct toxic effects on mollusks and, under certain conditions, endanger the normal survival of various organisms . Peiffer et al. (1997) also noted that extreme pH conditions not only directly affect the birth rate of benthic invertebrates, reducing their biodiversity, but also cause benthic invertebrate poisoning by triggering the release of heavy metals. This could be another possible explanation for the negative influence of pH on macroinvertebrates in this study, as Li et al. (2009) revealed that the acidity and alkalinity of Lianhuan Lake changed with the differential enrichment of heavy metals caused by the discharge of industrial sewage.
Water temperature (WT) affects the physiological processes of organisms; thus, temperature dynamics may change their life cycle patterns and trophic interactions (Li et al., 2012). This may alter the community composition and biodiversity. Interestingly, there was no significant difference in the WT (p > .05) among the lakes (Table 1).
However, the RDA results indicated that WT had a significant influence (p < .05) on macroinvertebrate species composition. This indicates that the WT affected the macroinvertebrates of Lianhuan Lake seasonally, rather than spatially. According to the RDA results, most of the macroinvertebrates were significantly positively correlated with WT, which is in agreement with other studies (Buss et al., 2004).
Water temperature (WT) is an important factor in the embryonic development, larval growth, emergence, metabolism, and survival of macroinvertebrates (Haidekker & Hering, 2008). The fact that many indicator species were recorded in Group V further demonstrates that WT is an important factor, because these sites were sampled during summer and spring.
Nutrients are essential for maintaining the structure and function of lake ecosystems. However, excessive nutrients can reduce the water quality and deplete DO, leading to the death of aquatic organisms (Ouyang et al., 2018). On average, the macroinvertebrate species richness and abundance in this study exhibited a stress relationship with nutrients, mainly TP, Chla, and NO 2 -N.
This result is consistent with the conclusion that high nutrient concentrations negatively affect benthic invertebrate species richness and abundance. This conclusion has been previously drawn in manipulative experiments and observational studies (Dodson et al., 2000;Wang et al., 2007). When the nutrients in a lake are excessive, as observed in Group IV, the water quality deterioration and DO depletion caused by the decomposition of algal bloom biomass is likely to reduce species richness (Wang et al., 2007), leading to a negative relationship between nutrients and species richness. The high nutrient concentration recorded in this lake could be attributed to the use of fertilizers to increase agricultural production.

TA B L E 3 Spearman correlation analysis between macroinvertebrate community
indices and environmental variables in the Lianhuan Lake

| CON CLUS ION
In this study, the analysis of macroinvertebrate assemblages identified a gradient of macroinvertebrate diversity in Lianhuan Lake. It also captured the spatiotemporal variation in the macroinvertebrate community structure and identified the indicator species in the lake.
The SOM analysis of the macroinvertebrate communities revealed that eutrophication has a serious impact on them. The differences in the community structure and environmental variables between Groups I and V were remarkable, and the indicator species reflected the environmental characteristics of each group of communities.
The increasing alkalinity and eutrophication of the lake may have a serious impact on the macroinvertebrate community. This is clearly demonstrated by the significant negative correlation between macroinvertebrate biomass and pH, as well as the negative correlations between species richness and Shannon's diversity and TP. Highintensity human disturbances, such as industrial and agricultural runoff, negatively impact the ecological environment and affect macroinvertebrate community structure. Thus, the macroinvertebrate community structure in Lianhuan Lake should be strengthened by improving the ecological environment and controlling environmental pollution (nonpoint source pollution) in the watershed. The impact of environmental variables on macroinvertebrates in lakes is a long-term accumulating process. This study only spanned three seasons and did not cover all the lakes of Lianhuan Lake. Consequently, this study is subject to some time and space limitations. Watershed land-use intensity is also a key factor affecting macroinvertebrates.
Therefore, future research should focus on the impact of land use and other anthropogenic stressors on macroinvertebrates or compare future results to those of this study to explore the succession of the macroinvertebrate communities in Lianhuan Lake.

ACK N OWLED G M ENT
This work was financially supported by the National Key Research and Development Program of China (No. 2020YFD0900501).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.