Research on river health evaluation and response mechanism of eco-hydrology in mountainous rivers : a case study of Lai Chi Wo River in Hong Kong, China

Abstract

Ecosystem problems, especially biodiversity and river health, have been concerned worldwide due to impacts of rapid socioeconomic developments and climate change in past several decades. However, due to shortage of field observations, our understanding of ecosystem and river health in mountainous rivers is still vague. In order to explore the biological integrity and eco-hydrological interactions, this study used fish data, macroinvertebrates data and hydrologic data collected at a mountainous river, the Lai Chi Wo (LCW) River, in Hong Kong, China, for evaluating its ecological health and relationship between aquatic organisms and environmental factors. Five stations along the river were selected for collecting fish data, macroinvertebrates data and river water data in this study. Fish-based biotic integrity index (F-IBI) was adopted to evaluate river health. Partial least square structural equation model (PLS-SEM) was used to assess the interactions between fish, macroinvertebrates, and environmental factors. Using the initial selection, distributing range, discriminatory power and Pearson's correlation analysis, this study adopted F-IBI to evaluate mountainous river ecological health. With assignment method and ratio method, the study evaluated the ecological health at the five stations along the river. Analysis of F-IBI values revealed a spatial variation from the river downstream to upstream, and impacts of human interference on the river resulting in a fair habitat to fish. The study disclosed that different relationships between water levels and fish abundance at five stations at monthly scale. The results also divulged the appropriate range of environmental factors (water temperature, dissolved oxygen, turbidity, and pH) in terms of F-IBI values at five stations. Further, this study used PLS-SEM to analyze the complex interactions of hydrometeorology, water temperature, nutrient, substrate, fish and macroinvertebrates. The model results showed different driving modes at five sampling stations. Station D, locating in downstream and close to the sea, was vulnerable to seawater intrusion during the dry season. Freshwater macroinvertebrates were intolerant to seawater, so station D becomes a hydrometeorological driven model. Stations A and C were narrow with rocks, which increased the chance of interaction with water and air or rocks, forming a nutrient-driven mode. Both stations B and Z were blocked by the weirs and form two small ponds. The accumulated sands and stones were conducive to macroinvertebrates and formed a substrate-driven model. With further analysis of relationship between fish and environmental factors, and relationship between fish and macroinvertebrates, this study disclosed the chain of causality of environmental factors and aquatic animals at different stations. Compared with the result of linear fitting or redundancy analysis, PLS-SEM model shows more precise interaction between environmental factors and the direct and indirect effects on macroinvertebrates and fish.