Driving forces of hydrological health and multifractal response of fish habitat in regulated rivers

Abstract

Climate change and anthropogenic activities are major influences on the hydrological cycle, further altering river hydrological health. However, the characteristics of the forces in driving the variations of hydrological health at long-short time scales (annual, seasonal, monthly), as well as the potential impacts of these variations on aquatic habitats, remain unclear. In this study, the flow threshold method was introduced to identify the inherent characteristics of river hydrological health degree (RHD) evolution in the upper reaches of the Yangtze River (URYR) through the extreme-point symmetric modal decomposition (ESMD) method and range of variation approach (RVA). The RHD under unregulated conditions was reconstructed to quantify the impacts of anthropogenic activities and climate change. Subsequently, a multifractal model was proposed to establish the relationship between RHD and habitat-weighted usage area (WUA) during the spawning period of the Four Famous Major Carps, aiming to analyze the response mechanisms of habitat conditions to RHD fluctuations. The results showed that the RHD in the URYR exhibited degradation characteristics, experiencing a moderate change with a value of 0.44. Climate change was identified as the dominant factor causing the annual-scale decline in RHD, with an average impact weight of 62.9%. At the annual scale, Anthropogenic activities exacerbate (-3.4), counteract (20.1), and counteract (20.5) the adverse climatic impacts at Yichang, Cuntan, and Zhutuo stations, respectively. Additionally, the effect of human activities during the flood season is slight, with the most favorable and unfavorable impacts occurring in December (50.7) at the Zhutuo station and in October (-27.2) at the Yichang station. Under the influence of driving forces, the multifractal correlation of the RHD-WUA system tended to homogenized as the time window increased, indicating the presence of potential nonlinear dependence, asymmetric fractal characteristics, and positive-to-negative persistence transitions. Therefore, modeling river health considering fish habitat cannot be limited to linear paradigms. The findings provide valuable insights for the management and restoration of aquatic ecosystems.

Keywords: Anthropogenic activities; Deep learning model; Fish habitat; Joint multifractal model; River hydrological health.