Identification of Regime Shifts in Shallow LakesUsing Remote Sensing: Analysis of MacrophyteDynamics and Management Implications

Abstract Shallow lakes are one of the richest freshwater ecosystems on Earth, with shallow aquatic vegetation being a fundamentalcomponent of wetland ecosystems, providing various services including water quality improvement, maintaining the balanceof aquatic ecosystems, and providing food and habitat for aquatic animals. Compared to deep lakes, shallow lakes aremore susceptible to external disturbances, leading to transitions from clear water states dominated by submerged plants toturbid water states dominated by algae. Identifying and exploring potential factors causing state transitions have significantimplications for the management of water environments in shallow lakes.Based on Landsat TM/ETM+/OLI remote sensingimages, this study constructed a decision tree classification model using maximum and minimum composite yearly NDVI valuesto remotely infer the evolution of aquatic plants in Qilu Lake from 1991 to 2020. The overall accuracy of the classificationreached 89.29%, with a Kappa coefficient of 0.7933 . Then, the Mann-Kendall trend test and mutation detection method wereused to conduct trend analysis and quantitative mutation identification of submerged plants in Lake Qilu from 1991 to 2020.The trend test showed that the lake ecosystem of Qilu Lake improved from 1991 to 2020 (β = 0.3582, P < 0.01), with the mostsignificant improvement occurring from 2004 to 2012 (β = 0.7872, P < 0.01). The mutation test indicated a significant statetransition in the lake ecosystem of Qilu Lake over the past 30 years, with the mutation point of state transition occurring in2006-2007. Based on state transition, the evolutionary process of the lake ecosystem of Qilu Lake can be divided into twostages: 1991-2006 and 2006-2020, with mean coverage rates of submerged plants in each stage being 8.57% and 15.29%,respectively