3.1. Macroinvertebrate community composition
In 2023, a total of 270 species of macroinvertebrates were found in the study area, belonging to 5 phyla, 30 orders and 127 families. Among them, there are 194 species in plain terrain, 130 species in hilly terrain and 171 species in mountainous terrain. Chironomidae had the highest number of species among all sites, with 64 species accounting for 23.7% of the total. Palaemonidae, Atyidae, Polypedilum, and Corophiidae were the dominant species during the survey period.
The macroinvertebrate community composition differed significantly among mountains, hills, and plains (Stress = 0.1945, ANOSIM_R = 0.1094, p = 0.001) (Figure 2), and therefore the topography-based grouping effectively attenuated the effect of natural conditions on community composition. Specifically, macroinvertebrate community composition at mountain sites was dominated by Insecta (71%), followed by Gastropoda (12%), Oligochaeta (6%), with Heptageniidae, Cricotopus sp, Baetidae, Hydropsychidae, and Polypedilum sp as dominant taxa. The macroinvertebrate community composition at the hill sites was dominated by Insecta (63%), followed by Gastropoda (12%), Oligochaeta (67%), with Palaemonidae, Atyidae, Polypedilum sp, and Semisulcospira sp as dominant taxa. The macroinvertebrate community composition at the plain sites was dominated by Insecta (63%), followed by Gastropoda (12%), Bivalvia (67%), with Palaemonidae, Atyidae, Polypedilum sp, Corbicula sp, and Corophiidae as the dominant taxa.
3.2. Differences in land use variables
The study area exhibits a high land-use gradient, as evidenced by the significant variation in land-use composition across different sites (Table 1), with agricultural land (51.87 ± 25.24) being the most common land use type in all sites, followed by forest (23.6 ± 27.62) and grassland (2.80 ± 11.16) being the least common. In terms of different terrains, forest (59.08 ± 22.28) was the most common land use type in hills, agricultural land (65.57 ± 21.08) in the hills and agricultural land (58.20 ± 19.96) in plains. In terms of different land use types, agricultural land (65.57 ± 21.08) was the most common in the hills, forest (59.08 ± 22.28) in the mountains, grassland (11.55 ± 20.81) in the mountains and water bodies (17.12 ± 15.57) and urban areas (15.40 ± 13.52) in the plains.
Table 1 Differences in land-use composition of sites with different topography, give the p-value for the Kruskal-Wallis H-test
Land use type
|
All sites
|
Mountain sites
|
Hill sites
|
Plain sites
|
kruskal.test
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Mean
|
SD
|
Farmland %
|
51.87
|
25.24
|
24.64
|
19.28
|
65.57
|
21.08
|
58.20
|
19.96
|
p < 0.001
|
Forest land %
|
23.36
|
27.62
|
59.08
|
22.28
|
19.89
|
22.59
|
9.22
|
15.05
|
p < 0.001
|
Grassland %
|
2.80
|
11.16
|
11.55
|
20.81
|
0.21
|
0.68
|
0.04
|
0.20
|
p < 0.001
|
Water area %
|
11.01
|
13.42
|
2.37
|
2.65
|
5.18
|
2.61
|
17.12
|
15.57
|
p < 0.001
|
Urban land %
|
10.76
|
12.63
|
1.64
|
2.56
|
9.09
|
11.20
|
15.40
|
13.52
|
p < 0.001
|
3.3 MMI results for macroinvertebrates in different terrains
3.3.1. Mountain results
Of all 40 mountain sites, 29 were classified as impaired and 11 as reference sites. We began with 87 indicators and reduced the number to 67 after removing those with small distributions. Only 2 indicators, M59 (% number of filter feeders) and M79 (Dispersal), met the screening criteria for discriminability and the correlation between them did not exceed 0.75, so M59 and M79 were the final indicators retained for the mountain MMI calculation (Figure 3). Finally, the MMI score was calculated for each point and it was found that the score was not significantly correlated with the CDI (R = -0.2, p = 0.21) (Appendix Figure 2).
3.3.2. Hilly results
These points were categorised in the same way as above. Out of a total of 37 mound points, 29 are impaired and 8 are reference points. We started with 87 indicators and reduced this to 60 after removing indicators with small distributions. Only four indicators met the screening criteria for discriminability, so these were tested for correlation and, as before, variables with a final correlation of no more than 0.75 were retained. The final retained metrics were M17 (number of bivalve taxonomic units), M25 (percentage of individuals in dominant taxonomic units), M68 (Maglev species richness index) and M77 (Rao quadratic entropy index) (Figure 4). The final MMI score was significantly correlated with the CDI (R = -0.4, p = 0.015) (Appendix Figure 3).
3.3.3. Plain results
Of all 92 plain points, 69 are impaired and 23 are reference points. We started with 87 indicators and reduced them to 57 after removing indicators with small distributions. Nine indicators met the screening criteria for discriminatory ability, so these were tested for correlation, and as before, variables with a final correlation of no more than 0.75 were retained. As there were multiple indicator combination scenarios, the indicator combination scenario with the highest correlation coefficient was selected based on the correlation of MMI scores with CDI for each scenario, i.e. scenario b (R = -0.4, p < 0.001) (Appendix Figure 4). The final metrics selected for final retention were M22 (number of taxonomic units of telopods and molluscs), M23 (number of total taxonomic units), M53 (percentage of the number of individuals of telopods and molluscs), M59 (percentage of the number of individuals of filter feeders), M60 (percentage of the number of individuals of collectors), and M71 (BI) (Figure 5).
3.3.4. All point results
When the points were grouped without the use of topography, out of a total of 169 points, 135 were classified as impaired and 34 as reference points. We started with 87 indicators and reduced this to 60 after removing indicators with small distributions. Fifteen indicators met the screening criteria for discriminatory ability, so these were tested for correlation and, as before, variables with a final correlation of no more than 0.75 were retained. As there were multiple indicator combination scenarios, the indicator combination scenario with the highest correlation coefficient, scenario j, was selected based on the correlation of MMI scores with CDI for each scenario (R = -0.34, p < 0.001) (Appendix Figure 5). The final metrics selected for final retention were M10 (number of taxonomic units of aquatic insects), M19 (number of taxonomic units of crustaceans), M21 (number of taxonomic units of crustaceans and molluscs), M24 (percentage of individuals in EPT), M25 (percentage of individuals in dominant taxonomic units), M63 (percentage of individuals in trappers), M71 (BI), M74 (FEve), M83 (Rheophily), M86 (Functional Feeding Group) (Figure 6).