A hybrid PCA-GAM model for investigating the spatiotemporal impacts of water level fluctuations on the diversity of benthic macroinvertebrates in Baiyangdian Lake, North China

Abstract

Abstract To be effective, ecological restoration decisions for shallow lakes should consider the response of the biomes to water level fluctuation changes. A hybrid generalized additive model (GAM) based on principal component analysis (PCA), namely PCA-GAM model, was proposed to study the quantitative relationships between the water level fluctuation indicators and the diversity of benthic macroinvertebrates from short time (one month) to long time (twelve months) scales. For this model, the PCA method was applied to determining the main water level fluctuation indicators from the constructed indicator system, thereby reducing redundancy among the data variables; GAM was used to explore the quantitative relationships between the Margalef diversity index and the main water level fluctuation indicators after PCA screening. The results of the PCA-GAM model showed that the two-month water level amplitude value was the best water level fluctuation indicator for describing changes in the diversity of benthic macroinvertebrates in Baiyangdian Lake. The two-month water level amplitude value and the macroinvertebrate Margalef diversity index showed a negative correlation and an approximate one-dimensional quadratic function relationship. The macroinvertebrate diversity index in different regions responded differently to water level fluctuations, and the regions with low average water depth and close to the lake inlet were more sensitive to fluctuations than other regions. Our results suggested that shallow lake management activities aimed at improving biodiversity should focus on controlling water level fluctuations, and the amplitude of the artificial water level fluctuations from the single or multiple ecological water compensation should be reduced in order to promote lake ecological restoration in the future.