1. Introduction

1.1 Background

Lakes are very important property for all kinds of life. They provide water for consumption, fishing, irrigation, power generation, transportation, recreation, disposal of wastes, and a variety of other domestic, agricultural, and industrial purposes1). However, the current scientific evidence shows that the coastal, marine and freshwater lakes or reservoirs ecosystem are vulnerable to the impact of harmful algal blooms in world. Toxic algal or cyanobacteria blooms have been known to kill fishes, waterfowl and livestock, and dogs have died after eating mats of cyanobacteria or licking their fur after swimming in bloom-infested waters. In some cases, humans have also died after exposure to harmful algal toxins2). Unfortunately, not only Japan but also most of the developed and underdeveloped countries are not aware well enough about this issue. A very few studies have been conducted on algae transition behaviors3), their toxin production ability and mitigation strategies from a viewpoint of modeling.

1.2 Aims and objectives

The purpose of this study is modeling for predictive assessment and mitigation of cyanobacteria toxins in a eutrophic lake. The study area is the Lake Kitaura, Ibaraki Prefecture in Japan. Field samples were collected between July 2005 and September 2007.The major objectives of this study are: 1) to develop the vertical and horizontal transition of dominant algae model, 2) to develop toxin production model and their environmental mechanisms, and 3) to make a conceptual IMPACT (Integrating Mitigation Policies for Aquatic Cyanobacteria Toxins) model for controlling the nutrients that encouraged forming toxins.

2. Modeling of environment and algae transition

2.1 Hydrodynamic and ecosystem coupled model

The numerical model is based on the governing equations that are described in the Cartesian coordinate system3). The time variation in the fields of current velocity, water level, water temperature, density can be described by the equations of momentum, the continuity equation, the advection-diffusion equation of heat, and the equation of the state. Time variation of chemical materials and planktons can be described by the advection-diffusion equation. The ecosystem model is based on the observed features of a eutrophic food web in Lake Kitaura. The ecosystem model contains nine state variables such as: phytoplankton, cell quota of phosphorus and nitrogen, zooplankton, particulate and dissolved organic carbon, dissolved inorganic phosphorus and nitrogen, and dissolved oxygen.

Major improvement on algae transition model from the existing model3) are: 1) calibration of the horizontal eddy viscosity and diffusivity coefficients, 2) conservation of phosphorus and nitrogen, 3) visible light for photosynthesis, 4) consistency of algal flotation model with growth model, 5) finer mesh size (from 500 m to 200 m), and 6) calibration of many parameters and values.

2.2 Computational condition

The governing equations were solved numerically by finite difference scheme. Numerical simulation was carried out between June 2005 and March 2009. The observations were used for the boundary condition of meteorology and river.

2.3 Results

2.3.1 Water current velocity and temperature

The current velocity was reproduced well and strongly affected by the wind. Water temperature shows the longer stratification during June and July, and daily stratification in August. In terms of the annual variation in water temperature, the stratification was prolonged during June and July in 2008 since the atmospheric temperature increased rapidly during the same period.

2.3.2 Water quality

The simulation showed that concentration of dissolved oxygen was lower for a longer period in 2008 due to the stronger stratification. The concentration of dissolved inorganic phosphorus was usually exhausted, while the release rate of phosphorus increases in summer because anoxic conditions are sometimes formed by instantaneous stratification. The period of stratification and resulting anoxic condition were longer in the summer of 2008. Higher concentrations of dissolved inorganic phosphorus were both observed and predicted off Kamaya during the same period (Fig.1). The concentrations of dissolved inorganic phosphorus and nitrogen were higher in the middle and north of Lake Kitaura, respectively (Fig.2). This is because the phosphorus and nitrogen are supplied from the bottom in deep water and through rivers in north, respectively.

2.3.3 Algal transition and its mechanism

The numerical model could reproduce the seasonal variation in dominant algae; Planktothrix spp. in early summer, Microcystis spp. in late summer, and Cyclotella spp. between autumn and spring. The sudden increase in Planktothrix spp. in 2008 was also represented in the numerical simulation (Fig. 3). From several scenario simulations, it could be predicted that the exchange of water from other connecting lakes and rivers have the effect on algae transition in Lake Kitaura. The sudden increase in Planktothrix spp. is likely caused by the transfer of water containing Planktothrix spp. from Lake Nishiura via Wani River.

3. Modeling of toxic algae/cyanobacteria toxin production

3.1 Toxin production model

The timing and duration of the cyanobacteria blooms depends on the ability to make scum or colonies, light intensity, temperature, stratification, presence of nitrogen and phosphorus limited cyanobacteria. In the present study, toxin production is made by Microcystis spp., and is proportional to the growth of algae, while it depends on whether phosphorus or nitrogen limits the algal growth. The toxin remains in the cell for respiration. Toxin is released with extracellular release and mortality, and advects and diffuses with the surrounding current and turbulence. The degradation of toxin was taken into account by the decay coefficient which crosses the concentration of toxin. Numerical simulation was also tried under the assumption that phosphorus or nitrogen always limits the algal growth.

3.2 Results

The simulation results compare the toxin (the sum of MC-LR, MC-RR and MC-YR) production of three stations in the Lake Kitaura with the observational data in the month of July, August and September for 2005, 2006 and 2007. The simulation results show a good agreement with the observation and simulation all the stations and months in 2005 (Fig.4). If toxin production is assumed to be always limited by phosphorus, it is overestimated off Takei and off Kayama Stations where algal growth is limited by dissolved inorganic nitrogen. If toxin production is assumed to be always limited by nitrogen, it is underestimated off Tomoe Station where algal growth is limited by dissolved inorganic phosphorus. Consequently, toxin production strongly depends on what nutrient limits the algal growth. In 2006 and 2007 there is no toxin produced by cyanobacteria (dominant species by Microcystis ichthyoblabe is toxic/nontoxic) in the ecosystem of Lake Kitaura, however toxin is detected during this period in numerical simulation. This discrepancy may be attributed to the species of Microcystis.

4. Modeling of toxic algae mitigation strategies

The proposed IMPACT (Integrating Mitigation Policies for Aquatic Cyanobacteria Toxin) model assumes for preparing scenarios for alternative policy states like ideal state, moderate ideal state, poor state and very poor state (Fig.5). According to IMPACT model the scenario of Lake Kitaura is reflected the Scenario 2 (S-2). That means the nutrients management is not ideal.

If the national and prefectural government takes initiative for mitigation strategies with efficient nutrients management measures (Fig.6), and integrating lake basin management measures, then the actual condition of cyanobacteria mitigation policy will be 'optimum situation and ideal state' (Scenario 1). The finding of this study suggests that successful mitigation of cyanobacteria toxins is highly dependent on multi-functional, multi-stakeholder involvement, and relevant intergovernmental policy.

5. Conclusion

The environmental factors were not the key parameters explaining the transition of dominant algae in 2008, which was caused by the flux of algae due to water exchange between the lakes via Wani River. The toxin production behaviors are affected by the algae growth, nutrients limited condition and toxin decay coefficient. Toxin production in Lake Kitaura depends on whether phosphorus or nitrogen limits the algal growth. The ideal long term strategies for dealing with toxic algae are to prevent or reduce the occurrence of blooms. In this context, the proposed IMPACT model could be a decision framework for identifying suitable policies that mitigate cyanobacteria impacts.

Fig.1 Observed and simulated dissolved inorganic phosphorus off Kamaya station.

Fig.2 Simulated water quality and Microcystis concentration in August 2005

Fig.3 Simulation and observational transition of dominant species of algae off Kayama Station 2005-2009.

Fig.4 Simulation and observation toxin production are shown at the months of July, August, September in 2005 in off Tomoe, off Takei and off Kamaya stations in Lake Kitaura respectively.

Fig.5 Mitigation strategies IMPACT model

Fig.6 Nutrients management strategies flow chart.

湖は、飲料水、漁業、灌漑、輸送、リクリエーション、排水の自然浄化など、多くの機能を提供している。一方で、世界の湖では、富栄養化や貧酸素化が進行し、生態系の存続が脅かされている。特に、藍藻類による毒素の生産は、飲料水や食料の確保や生態系の保全に大きな影響を及ぼす。毒素の生産は、優占藻類の特性や栄養塩濃度、光量、水温等の周辺環境に依存するとされているが、その詳細なメカニズムはいまだ明らかにされていない。この要因の一つとして、物理学、化学、生物学的に複雑な現象を統合的に解析するツールがないことが挙げられる。

本研究は、流れ場-生態系結合数値モデルと毒素生産モデルを結合した数値モデルを構築し、時空間的に断片的な観測データを補間するとともに、毒素の生産に及ぼす影響因子を明らかにすることを目的としたものである。まず、流れ場や水質環境、藻類の種の変遷を再現できるように、流れ場-生態系結合数値モデルを高度化した。次に、毒素生産モデルを構築し、流れ場-生態系結合数値モデルに結合した後、毒素変動の数値シミュレーションを行い、毒素生産メカニズムを考察した。最後に、毒素生産を緩和するための概念的な環境システム解析モデル(IMPACTモデル:Integrating Mitigation Policies for Aquatic Cyanobacteria Toxins Model)を提案した。研究対象水域として、2005~2007年に藍藻類が増殖し、毒素が検出された北浦を選定した。

まず、既存の流れ場-生態系結合数値モデルでは、夏季に現れる間欠的な成層構造や、貧酸素水塊の発生によるリン酸態リンの溶出など、鉛直方向の諸過程については再現できていたが、水平方向の水質分布については観測結果との相違が認められていた。そこで、1) 水平渦動粘性係数、水平渦動拡散係数の再調整、2)水平解像度の向上、3)リンと窒素の全量の保存、4)光合成に用いられる光量として可視光を使用、5)藍藻類(Microcystis spp.)の生長モデルと浮上・沈下モデルの適合、6)生態学的パラメータの再調整により、既存の数値モデルを高度化した。その結果、夏季には、北浦北部で無機態窒素濃度が高く、中部でリン酸態リン濃度が高くなる様子など、水質環境の水平分布が再現された。また、数値モデルでは、Microcystis spp.、Planktothrix spp.、Cyclotella spp.の3種の藻類を状態変数としたが、2008年以降にPlanktothrix spp.が優占する様子が再現された。この主な原因として、シナリオ別の数値シミュレーションの結果、外浪逆浦を介して、西浦から北浦への流れが発生した際に、Planktothrix spp.が輸送され、北浦で生長したことが挙げられた。

藍藻類による毒素の生産は、コロニーの生成、光量、水温、成層、栄養塩濃度等に依存する。毒素生産モデルでは、毒素がMicrocystis spp.により生産され、Microcystis spp.の生長に比例して増加し、一定の半減期で減少するものとした。ただし、Microcystis spp.の生長がリン、窒素のどちらに制限されるかによって、異なる毒素の生産速度を与えた。Microcystis spp.の細胞外分泌、呼吸、枯死に対しては、毒素がMicrocystis spp.の体内または周囲の水中に保持され、移流・拡散するものと仮定した。動物プランクトンの摂食による毒素の蓄積や高次生態系への生物濃縮は、今後の課題とした。毒素生産の数値シミュレーションの結果、北浦北部では、Microcystis spp.の生物量が多いため、生産される毒素も多いが、それに加えて、Microcystis spp.の生長がリンによって制限されるため、毒素の生産量がさらに多いことが示された。Microcystis spp.の生長が常にリン、あるいは窒素によって制限されると仮定した場合は、北浦北部、中部の毒素生産量がそれぞれ過小評価、過大評価された。ただし、観測結果では、毒素は2005年に検出されたが、2006年以降は検出されなかったのに対し、数値シミュレーションの結果では、すべての年に毒素が検出された。今後は、Microcystis spp.の中での種別構成や動物プランクトンの摂食も考慮に入れた解析が必要である。

構築した数値モデルを政策決定に用いるため、毒素生産を緩和するための環境システム解析モデル(IMPACTモデル)を提案した。IMPACTモデルでは、湖を理想的な状態、適度に理想的な状態、貧弱な状態、非常に貧弱な状態に分類した。北浦は、湖の管理システムは存在するが、栄養塩類の管理システムが効果的でないため、適度に理想的な状態と診断された。政府が栄養塩類を効果的に管理する手法を用いて毒素生産の緩和戦略を主導することによって、湖が理想的な状態となる可能性が示された。

以上のように、本研究で構築された数値モデルは、富栄養化が進行した湖での水質環境、藻類の変遷、毒素生産の予測に有用である。また、環境システム解析モデルも含め、流域モデル、大気モデルとの結合によって、さらに統合的な解析システムの構築に発展することが期待される。

よって本論文は博士(工学)の学位請求論文として合格と認められる。