Human perturbation on phosphorus cycles in one of China's most eutrophicated lakes

Lake Dianchi is one of the most phosphorus-polluted lakes in China. Understanding its phosphorus budget and the corresponding limitations for research are fundamental to lake management. We have therefore systematically analyzed publications and publicly available datasets regarding Dianchi P pollution since the year 1960 to explore the anthropogenic perturbation on phosphorus budget dynamics. Results show that current exogenous loading (> 600 tons ⋅a⁻¹) remains significant; inflow rivers and drainage pipes around the lakeshore are the main pathways (accounting for around 80% of loading). Besides shifts in the spatial–temporal pattern of exogenous loading, artificial treatments have also reshaped P removal patterns; consequently, only a small amount of P can be removed through lake outflows (<20%, around 140 tons ⋅a⁻¹). Although a reduction has been observed in exogenous P loading over the past decade, endogenous P loading has significantly increased, of which P flux via sediment release alone approximates more than 180 tons ⋅a⁻¹ in 2017. Generally, P budget improvements have been due to artificial treatments such as the Niulan River water diversion project, shifting directions on tail water emissions, and sediment dredging projects. However, these treatments also introduce negative consequences and therefore require further caution. The reuse of ”waste P” via treatments has especially become a challenge. To date, exogenous loading via atmospheric dry deposition, wintering birds, water diversion, degradation in geological phosphorus-rich regions and endogenous P cycling fluxes, etc., need further study. Here, suggestions for corresponding research are given. It is suggested that the restoration of spontaneous ”P balance” in the Dianchi system, and the modification of the regional urban development strategy should present issues facing plateau freshwater lakes in SW China which need to be addressed. Moreover, synthesized analyses and machine learning processes are new attempts worth promoting to other lakes where monitoring data is incomplete.