Determining Long-Term Trends of Four Fast-Eutrophicated Lakes in China and Finland

Lake eutrophication has increased in pace in recent decades and has caused serious environmental problems However, the development trends have not been fully determined as it is difficult to recognize complex effects emanating from both climate and human mechanisms. China has many lakes in different trophic stages, which represent three developing stages from forestto agriculture-, and then to urban-lake, typically in Lakes Lugu, Taibai, and Taihu. To determine long-term water quality trends, the three lakes were chosen for statistic analysis on dominant effects on the diatom-inferred nutrient changes, and to undertake dynamic modelling regarding climate-controlled nutrient changes. The results indicate the significant turning points of water quality in Lakes Lugu, Taibai and Taihu occurring in the 1990s, 1950s and 1940s respectively, which were effected from human activities by increases in tourism, farming and urbanization respectively. Water quality changes in Lakes Lugu, Taibai and Taihu captured 68.4%, 54.9%, and 86.0% of the temperature variations before the turning points. The anthropogenic impacts explained 84.0%, 96.4% and 96.0% of the water quality variations after the turning points, where the sharp change of water quality by human activity has played an accelerated effect on the gentle change of temperature. Compared with the 4 phases of water quality development in Pyhäjärvi Lake (SW Finland), Lakes Lugu and Taibai have experienced the 1 and 2 phases, and Taihu has experienced from the 2 to 3 phases during the last 150 years. Phase 4 has not occurred in the three lakes, but it is a key period during the eutropication we need to pay attentions.


Introduction
In natural systems, eutrophication is a nutrient-enhanced process in a timescale of several hundred years (Kalff, 2002).However, the process can be dramatically accelerated by intensive human disturbances, which will endanger the structures and functions of lake ecosystems (Carpenter et al., 1995;Howarth et al., 2000;Qin, Xu, & Dong, 2011).Nowadays, many lakes in China have been eutrophicated or reveal euthrophic trends due to excessively development and using lake and catchment resources (e.g.Dong et al., 2008;Liu et al., 2007;Wu, Huang, Zeng, Schleser, & Battarbee, 2007).Changes in lake water quality influenced by human activity have been recognized as a progressive process from the original forest-type lake with primary vegetation types to the agriculture-type lake strongly influenced by tillage practices, and finally to the urban-type lake characterized by industrialization and urbanization (e.g.Fritz, 1989;Rasmussen, 2005;Yu, Xue, Lai, Gui, & Liu, 2007).To analyze the processes of lake water quality within different developmental stages, such as changes in forest-agricultural-urban lakes, can provide a more comprehensive perspective to understand lake development but there have been few study in China (Yang et al., 2010).lake.Lugu Lake is a remote and isolated area in southwestern China with a low-level economy and weak human activity, and the forest covers 68.9% of the catchment area (Dong et al., 2008).The water was measured in an oligotrophic stage and the water quality is very good (Zhang, 2014).Tillage practice has become the major economic activity in eastern China, such as in the catchment of Taibai Lake.71.4% of the area is farming land (Liu et al., 2007), which represents an agriculture-lake type.Northern Taihu Lake catchment, located in the east of China is a densely populated and highly economically developed area.It covers 36,500 km 2 in the catchment, only 0.38% of the national territory area, but holds 4.2% of the total population and contributes 11% of the Gross Domestic Product in China (Xie, Yu, & Zhang, 2001).Thus, it is a typical urban-lake type.In comparison to the three Chinese lakes mentioned above, the water quality of Pyhäjärvi Lake in SW Finland not only had experienced changes from oligotrophic condition to eutrohication, but also experienced water quality recovering (Ventelä et al., 2015).Therefore the history of water quality changes in Pyhäjärvi Lake can provide a reference for comparing the developments of Chinese lakes' water quality.Considering Chinese lakes in different trophic stages are differentially influenced by climatic and anthropogenic impacts, three Chinese lakes (Lakes Lugu, Taibai and Taihu) were chosen to study the forest-agricultural-urban lakes in eutrophication processes.
A comprehensive understanding of lake water quality changes must rely on long-term monitoring (Degobbis et al., 2000;Sayer, Davidson, Jones, & Langdon, 2010).However, it is almost impossible to find records longer than 50 year records in China, and this is so for the three lakes.For this reason, a paleolimnological approach using diatom fossils was introduced in this study, which has proved useful for analyzing changes of lake water quality history (e.g.Harris & Vollenweider, 1982;Levine et al., 2012;Luoto & Ojala, 2014;Smol & Douglas, 1996), because among paleolimnological multi-proxies the diatom fossil is a practical and workable indicator in reconstructing the past lake water quality (e.g.Anderson, 1993;Chen, Yang, Dong, & Liu, 2011;Dong et al., 2008;Hall, Leavitt, Smol, & Zirnhelt, 1997).However, the reconstruction based on diatom analysis can hardly distinguish anthropogenic impact from natural influence and is unable to estimate the contributions of different factors.For this reason, a nutrient dynamic model was constructed in the present study to simulate lake nutrient changes under natural conditions, by setting climate boundaries and initial human-influence conditions.
In this paper, based on the four-lake water quality changes analyzed by diatom-inferred nutrient reconstructions, we used Principal Component Analysis (PCA) to diagnose the major change signals of lake water quality controlled by both climate and human activity.Meanwhile, a lake nutrient dynamic model was constructed to simulate nutrient changes under the climate conditions.By comparing the two-approach results, we attempted to distinguish the impacts of climate only and overlaying human activity.Finally, to compare with Pyhäjärvi Lake trends, we figured out respective stages of the three Chinese lakes, and analyzed the potential trends of their water quality.

Study Site
In this study, three lakes of Lugu, Taibai and Taihu (Figure 1) were chosen respectively as typical cases of forest-, agriculture-and urban-type lakes according to the developing stages and the strengths of human activities in the catchments.
(1) Lugu Lake (27.27°E, 100.78°N, 2690.75 m a.s.l) is located in the upper reach of the Yangtze River, in Yunnan Province, southwest China.The water surface area is 48.25 km 2 within 246.26 km 2 of the catchment area.The average water depth is 40.3 m with a maximum of 93.5 m, and the water volume is 1.953×10 9 m 3 (Wang & Dou, 1998).The multi-year mean of the total annual precipitation is about 920 mm, and the mean annual temperature is 12.7 °C with a minimum of -10.3 °C and a maximum of 31.4 °C (Wang & Dou, 1998).This lake is supplied mainly by surface runoff from the catchment.As a deep-water lake, the seasonal stratification is significant and the thermocline generally develops at a depth of 14.4 m (Wang & Dou, 1998).The lake catchment is an under-populated region, and the forest land, farming land and residential land occupies about 68.9%, 11.77% and 1.56%, respectively of the catchment area (Cai, 2014).Lugu Lake is considered to be an oligotrophic lake with a mean TP concentration of 11.7 μg/L (Chen et al., 2014).
(2) Taibai Lake (29.13°E, 115.8°N, 16 m a.s.l) is located in the middle reach of the Yangtze River, Hubei Province, eastern China.The water surface area is 25.1 km 2 within 960 km 2 of the catchment area.The average water depth is 3.2 m with a maximum of 3.9 m, and the water volume is about 8×10 7 m 3 (Wang & Dou, 1998).The multi-year mean of the total annual precipitation is about 1,272.5 mm, and the mean annual temperature is 16.7 °C with a minimum of -13.8 °C and the maximum of 39.8 °C (Wang & Dou, 1998).Water supply derives primarily from surface runoff (Wang & Dou, 1998).The lake catchment is a typical agriculture area, and the forest land, farming land and residential land occupies about 16.55%, 71.4% and 4.37%, respectively, of the total area (Liu et al., 2007).According to the investigation in 2002 AD, the mean TP concentration is about 125.5   (Dong et al., 2008), and Pyhäjärvi (Weckström et al., 2015).We therefore used the diatom assemblage variations in lake sediments to reflect nutrient changes during historical periods.In this paper, the method of Principle Component Analysis (PCA) was applied to extract the principal component (PC) of diatom assemblages.PCA is an approach aimed at extracting a few PCs to replace numerous original variables by using the dimension reduction method, and should reflect original information as much as possible.Therefore, the PC can capture the major signals of diatom assemblage variations, which was used as a lake nutrient proxy in this study.

Region
The PCA was performed in C2 program Version 1.5 (Juggins, 2003), when only diatoms with abundances >2% were included in the dataset.Before running the PCA, we constructed a data matrix.The columns of the matrix were the abundances (%) of different diatom species, and the rows corresponded to the top-down samples from the lake sediment cores.In the dataset, the sample numbers of Lakes Lugu, Taibai, Taihu and Pyhäjärvi were 16, 24, 28 and 103 respectively, and the species numbers were 50, 46, 35 and 48 respectively.

Lake Climate-Nutrient Dynamic Model
Although diatom assemblages and the extracted signals from lake sediments can reflect the combined effect of climatic and anthropogenic impacts on water quality changes, and also can infer the potential main-control forces, it can hardly distinguish climatic influence from anthropogenic impacts.Nutrients are important for water quality changes (Kalff, 2002), among which total phosphorus (TP) is one of the most important component to evaluate lake nutrient levels (Vollenweider, 1968).This is also the case with most Chinese lakes (Wang & Dou, 1998).Therefore, a lake climate-nutrient dynamic model was constructed to simulate TP changes controlled by natural conditions such as precipitation, evaporation, runoff, regional hydrology and the geomorphology, and so on.
Lake nutrient accumulation and discharge are significantly affected by lake morphology and hydrology, and the nutrient equilibrium is greatly influenced by water exchange speed (Vollenweider, 1968).Based on the analysis for the morphological characteristics and hydrological data from 30 lakes in Europe and North America, Vollenweider (1968) proposed a relationship of lake nutrient concentration (C l ), catchment nutrient concentration (C r ) and hydraulic residence time (τ).Larsen and Mercier (1976) then improved Vollenweider's mode and proposed the relationship described as: Hydraulic residence time is defined as the time length lake water needs to change once, and it can be expressed as: Where V is the lake volume, Q is the net inflow.By introducing the concept of hydrologic budget (P+R-E-Q=0), Equation (2) can be expressed as: Wherein P, R, E represent lake surface precipitation, catchment-derived runoff and lake surface evaporation.
Water phosphorus load (P l ) is the function of discharge (Q) and concentration (C l ): P l = C l QP l can be calculated according to the formula below: In this paper, Equation ( 4) was used to calculate TP content with lake morphology and hydrology as boundary conditions, and was applied to estimate changes of lake water quality.For comparison purpose, the TP load was divided by the lake volume to convert to TP concentration.Model validation has been done by a control test published in Yu, Liao, and Li (2013) and Guo, Yu, and Qin (2015).

Total Variance Explained
To further explain the variations of climatic and anthropogenic impacts on the water quality change, we introduced an index of relative variance in this study.For comparison purpose, time series of the PCs, temperature and human activity index were normalized and calculated the standard deviations.Finally, we used a ratio of standard deviation of temperature (or human activity index) to that of PCs, in order to describe the contribution weighting of temperature (or human activity).In this paper, the ratio can reflect the fitting degree between two compared series.Namely, the higher ratio indicates better fitting, while lower ratio indicates poor fitting. www.ccsen

PCA R
The diatom results, the Lakes Lug The four la a certain ra rapidly wi before ca. in North T then it dec AD and th Lake incre increased r    Taihu and Pyhäjärvi, respectively, during the post-turning point periods.These suggest that human activity strongly influenced lake water quality changes during the recent decades.

Trends of Lake Water Quality
According to the PC1 series during the last 150 years, water quality changes in Pyhäjärvi Lake are shown by a four-phase model (Figure 7a): phase 1 is in a stage of oligothrophic condition before the 1930s, phase 2 is in a stage of water quality deterioration during the 1930s-1990s, phase 3 represents a stage of accelerated deterioration during the 1990s-2000s, and phase 4 is in a stage of water recovering since the 2000s.Comparing with Pyhäjärvi Lake, the three Chinese lakes can be inferred to show similar stages.Water quality in Lakes Lugu and Taibai has experienced the processes from phases 1 to 2 during the last 150 years (Figures 7b and 7c), and Taihu Lake has experienced the process from phases 2 to 3 (Figure 7d).Phase 4, however, has not yet occurred in the Chinese lakes.Based on Pyhäjärvi experiences and trend analysis, we estimated that water quality of Taihu Lake, now in phase 3, will likely step into phase 4 when the lake is protected and restored by appropriate measures likely in the Pyhäjärvi Lake.The water quality of Lakes Lugu and Taibai, in phase 2 now, will step into phase 3 if they are persistently affected by intensive human activities like the 1990s' situation in the Pyhäjärvi, but it will probably step directly into phase 4 if they are treated by effective protecting measures and management.

Conclusions
In this study, through combined paleolimnological and simulated approaches, we attempted to diagnose the lake trophic processes and turning points and to estimate the trends of water quality on the reference of Pyhäjärvi Lake.The comparisons between the simulated TPs and the PCs as inferred from the diatom assemlages from lake sediments showed that during the last 150 years, there were significant turning points in water quality changes occurring in the three lakes of China.It occurred in the 1990s in Lugu Lake, resulting primarily from catchment tourism development.The turning point of Taibai Lake occurred in the 1950s, causing by agricultural reclamation.The turning point of Taihu Lake occurred in the 1940s, mainly due to the urbanization in the lake basin.When during the last 150 years, the turning point in Pyhäjärvi Lake occurred in the 1930s, a major result also from increasing influence of human activity in the catchment.According to the calculation results of relative variance, before the turning points of each lake, temperature played a leading role in water quality changes, while during the periods of the post-turning points, human activity was a major force to control water quality change in the eutrophication process.
The change process of water quality in Pyhäjärvi Lake during the last 150 years show four-phase developments (phase 1-4).Comparing with Pyhäjärvi Lake, each of the three Chinese lakes also shows some similarities in the change process of water quality.Water quality in Lakes Lugu and Taibai has experienced the processes from phases 1 to 2 during the last 150 years, and Taihu Lake has experienced the process from phases 2 to 3, while phase 4 has not occurred in the three Chinese lakes yet.Based on Pyhäjärvi experiences and our trend analysis, we also predicted that the water quality of Taihu Lake will likely step into phase 4 if protection and restoration measures for water quality are taken into practice.Water qualities in Lakes Lugu and Taibai will step into phase 3 if they are persistently affected by intensive human activities, but it will likely step directly into phase 4 if they are protected by effective measures and managements.

Figure
Figure 4. C Figure 5 were used to calculate relative variance.The calculation results show that changes of PC1s in Lakes Lugu, Taibai, Taihu and Pyhäjärvi captured 68.4%, 54.9%, 86.0% and 56.6% of the temperature variations respectively, during the pre-turning point periods of each lake.These indicate that temperature was the main force controlling lake water quality changes during these periods.The results also show that the human activity indices explained 84.0%, 96.4%, 96.0% and 73.6% of the PC1s (or PC2) change in Lakes Lugu, Taibai,