Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Use of a Generalized Additive Model to Investigate Key Abiotic Factors Affecting Microcystin Cellular Quotas in Heavy Bloom Areas of Lake Taihu

  • Min Tao,

    Affiliation Donghu Experimental Station of Lake Ecosystems, State Key Laboratory of Freshwater Ecology and Biotechnology of China, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, People's Republic of China

  • Ping Xie ,

    xieping@ihb.ac.cn

    Affiliation Donghu Experimental Station of Lake Ecosystems, State Key Laboratory of Freshwater Ecology and Biotechnology of China, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, People's Republic of China

  • Jun Chen,

    Affiliation Donghu Experimental Station of Lake Ecosystems, State Key Laboratory of Freshwater Ecology and Biotechnology of China, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, People's Republic of China

  • Boqiang Qin,

    Affiliation Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing, People's Republic of China

  • Dawen Zhang,

    Affiliation Donghu Experimental Station of Lake Ecosystems, State Key Laboratory of Freshwater Ecology and Biotechnology of China, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, People's Republic of China

  • Yuan Niu,

    Affiliation Donghu Experimental Station of Lake Ecosystems, State Key Laboratory of Freshwater Ecology and Biotechnology of China, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, People's Republic of China

  • Meng Zhang,

    Affiliation Donghu Experimental Station of Lake Ecosystems, State Key Laboratory of Freshwater Ecology and Biotechnology of China, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, People's Republic of China

  • Qing Wang,

    Affiliation Donghu Experimental Station of Lake Ecosystems, State Key Laboratory of Freshwater Ecology and Biotechnology of China, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, People's Republic of China

  • Laiyan Wu

    Affiliation Donghu Experimental Station of Lake Ecosystems, State Key Laboratory of Freshwater Ecology and Biotechnology of China, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, People's Republic of China

Use of a Generalized Additive Model to Investigate Key Abiotic Factors Affecting Microcystin Cellular Quotas in Heavy Bloom Areas of Lake Taihu

  • Min Tao, 
  • Ping Xie, 
  • Jun Chen, 
  • Boqiang Qin, 
  • Dawen Zhang, 
  • Yuan Niu, 
  • Meng Zhang, 
  • Qing Wang, 
  • Laiyan Wu
PLOS
x

Abstract

Lake Taihu is the third largest freshwater lake in China and is suffering from serious cyanobacterial blooms with the associated drinking water contamination by microcystin (MC) for millions of citizens. So far, most studies on MCs have been limited to two small bays, while systematic research on the whole lake is lacking. To explain the variations in MC concentrations during cyanobacterial bloom, a large-scale survey at 30 sites across the lake was conducted monthly in 2008. The health risks of MC exposure were high, especially in the northern area. Both Microcystis abundance and MC cellular quotas presented positive correlations with MC concentration in the bloom seasons, suggesting that the toxic risks during Microcystis proliferations were affected by variations in both Microcystis density and MC production per Microcystis cell. Use of a powerful predictive modeling tool named generalized additive model (GAM) helped visualize significant effects of abiotic factors related to carbon fixation and proliferation of Microcystis (conductivity, dissolved inorganic carbon (DIC), water temperature and pH) on MC cellular quotas from recruitment period of Microcystis to the bloom seasons, suggesting the possible use of these factors, in addition to Microcystis abundance, as warning signs to predict toxic events in the future. The interesting relationship between macrophytes and MC cellular quotas of Microcystis (i.e., high MC cellular quotas in the presence of macrophytes) needs further investigation.

Introduction

Toxic cyanobacterial blooms in eutrophic lakes, rivers and reservoirs are encountered worldwide [1][3]. Microcystins (MCs) produced by some species of freshwater cyanobacteria are potent hepatotoxins and tumor promoters by inhibiting protein phosphatase types 1 and 2A [4], [5]. They can transfer via the food chain and accumulate in organisms [6], [7], causing poisoning even death of plants, invertebrates, fish, birds and mammals [8][11] in addition to effects on human health through chronic exposure [12], [13].

The MC toxic risks during cyanobacterial proliferations are determined by variations in both the abundance of toxic cyanobacterial strains and the production of MC by the toxic cells [14], [15]. The environment influences MCs indirectly by affecting the above two aspects. There have been many experimental and field studies to document the impact on MC production of various factors such as temperature, nutrients [16][19], light [17], [20], pH [21], [22], iron [23], xenobiotics [24], and predators [25][27], but the conclusions are sometimes different or even contradictory perhaps due to rather complex interactions of these factors in the field. It remains a great challenge to investigate how environmental factors interactively affect the toxicity of cyanobacteria. Thus, intensive and large-scale field surveys based on an effective model for data analysis are badly needed.

Generalized additive model (GAM) [28] is an extension of the generalized linear model. The advantage of the GAM is the adaptability for non-normally distributed variables. It is a flexible and effective technique for dealing with non-linear relationships between the response and the set of explanatory variables, and it is non-parametric generalization of multiple linear regression that is less restrictive in assumptions about the underlying distribution of data. The model assumes that the dependent variable is dependent on the univariate smooth terms of independent variables rather than independent variables themselves. The basic GAM model used took the following form:

E(Y|X1,X2,…,Xp) = B0+S1(X1)+S2(X2)+…+Sp(Xp) where Si(Xi), i = 1,2,…,p are nonparametric smooth functions (smoothing spline) for independent variable Xi. The function Si is estimated in a flexible manner and does not have to be nonlinear for all independent variables in GAM. The model is a useful and scientific tool applied in many scientific aspects [15], [29], [30].

Lake Taihu is the third largest freshwater lake in China, which historically has been beset by occurrences of cyanobacterial blooms dominated by Microcystis in warm seasons each year [31]. The coverage area of cyanobacterial blooms increased rapidly in recent years, posing serious threat to water supply for millions of inhabitants around the lake [32]. Water works located in the northern area supply drinking water to millions of residents of Wuxi city. Several field studies on MCs have been executed in Lake Taihu in recent years [33][37], but most of these studies were focused on two bays (Meiliang and Gonghu Bays) with simple description of seasonal changes of MCs, while systematic research on the whole-lake was still absent.

Mainly for these reasons, a systematic survey at 30 sites across the whole areas of Lake Taihu was conducted, and the spatiotemporal dynamics of MC concentrations, abundance and composition of major phytoplankton groups and various physicochemical parameters were monitored monthly from January to December 2008. MC cellular quotas which was calculated as the quotient obtained by dividing intracellular MCs concentration by Microcystis density presented MC-producing capability of Microcystis population. The main purpose of this study was to use GAM to investigate quantitative relationships between various environmental factors and MC cellular quotas from recruitment period of Microcystis to cyanobacterial bloom seasons, so as to clarify the possible mechanisms of environmental factors affecting MC-producing capability of Microcystis in the lake.

Materials and Methods

Ethics Statement

No specific permits were required for the described field studies. The location studied is not privately-owned or protected in any way and the field studies did not involve endangered or protected species.

Study area

Lake Taihu (119°54′–120°36′N, 30°56′–31°33′E) in Jiangsu Province, is a subtropical, shallow, highly eutrophic freshwater lake with a surface area of 2338 km2, a mean depth of 1.89 m. It serves as an important resource for drinking water, irrigation, aquaculture, and industrial waters, in addition to being a popular recreational and tourist attraction. The occurrence of heavy cyanobacterial blooms in warm seasons has increased in frequency and intensity in recent years, which damages the function of the lake as a drinking water supply, posing a risk to public health [31].

Sampling and analyzing

The lake was sampled at 30 sites (Fig. 1) from January to December, 2008. Sites 1–14 located in the northern area of the lake, where water was seriously polluted by human activities and was used as drinking water source, were sampled monthly. Water samples of other sites (sites 15–30) were collected quarterly. Each sample was a mixture collected from the top (0–0.5 m, surface water) and the bottom (0–0.5 m over sediment) of the water column with a 5 L Schindler sampler [38].

Values of water transparency and water depth were obtained in situ. Water temperature, pH, dissolved oxygen (DO) and conductivity were measured in situ with YSI Environmental Monitoring System 6600 (YSI Incorporated, Yellow Springs, OH, USA). Dissolved inorganic carbon (DIC) and dissolved organic carbon (DOC) were measured using a TOC Analyser (OI-1020A, OI Analytical, College Station, TX, USA), and some metal ions such as Na+ and K+ were analysed by ion chromatography (Dionex DX-100, Dionex Corporation, Sunnyvale, CA, USA).

Chemical parameters, including total nitrogen (TN), ammonia nitrogen (NH4-N), nitrate nitrogen (NO3-N), nitrite nitrogen (NO2-N), total phosphorus (TP), phosphate phosphorus (PO4-P), and chlorophyll a were measured for each sample according to the methods described by Greenberg et al. [39].

Water samples for identification of phytoplankton (1 L) were fixed in situ with acetic Lugol's solution [40]. In laboratory, each sample was concentrated to 50 ml after sedimentation for 48 h. Then 0.1 ml concentrated samples were counted using an Olympus microscope (BX50, Olympus, Tokyo, Japan) under magnification of ×400 after complete mixing. Colonial Microcystis cells were separated by using an ultrasonic crusher (JY88-II, Scientiz, Ningbo, Zhejiang, China), and then the single cells were counted. Phytoplankton species were identified with reference to the methods detailed by Hu and Wei [41] and John et al. [42].

MCs in lake water (1 L) were separated into extracellular MCs (toxins dissolved in water) and intracellular MCs (toxins in particulate) through filtering with a filter (Waterman GF/C, Whatman, Maidstone, Kent, UK). Filter films were extracted thrice in methanol (75%). The suspensions were centrifuged at a relative centrifugal force (RCF) of 24475× g (30 min at 4°C, Jouan KR22i, Jouan, Saint-Herblain, France) and the supernatant was diluted 1∶5 with distilled water. The MCs in distilled supernatant were directly concentrated on phase extraction cartridges (10 ml, C18, 500 mg), which were previously activated with 10 ml methanol (100%) and 10 ml distilled water. MCs were eluted from the cartridges with 10 ml methanol (100%) and then evaporated to dryness. The residue was dissolved in 100 µl distilled water and used for the qualitative and quantitative analysis of MCs. MC concentration was measured by using a Finnigan LC-MS system (Thermo Electron Corporation, San Jose, CA, USA) according to the methods described by Wang et al. [36].

Statistical analyses

Regression analysis was performed using GAM, provided by PROC GAM procedure of the SAS software (release 9.1.3, SAS Institute incorporated, Cary, NC, USA) to assess the effects of environmental factors on MC-producing capability of Microcystis spp. in the recruitment, growth and proliferation phases of Microcystis bloom-forming (March to November). In order to better understand the underlying trend of any given factor, PROC GAM separates the linear trend from any general nonparametric trend during the fitting as well as in the final report. This makes it easy to determine whether the significance of a smoothing variable is associated with a simple linear trend or a more complicated pattern [43][45].

The model used the amount of intracellular MC concentration in each Microcystis cell (MC cellular quotas) as the dependent variable and abiotic factors such as temperature, pH, water depth, conductivity and nutrients as the independent variables. Zero values in MC cellular quotas were identified as outliers and excluded from the analysis. The “spline” function was used in MODEL statement to request an additive model using a cubic smoothing spline with four degrees of freedom by default for each environmental factor [43], [45]. Using conservative degrees of freedom in GAM is of benefit of avoiding over-fitting and lowers the computing cost. The F-statistic calculated from GAM vaguely indicated the relative strength of effect of an independent factor on dependent variable in the model. F-statistics were standardized to sum up to 100 within model. The product of standardized F-statistics (%) of each parameter and R-squares of the whole model presented the contribution of each parameter to MC production [15]. Factors with high significance levels (P<0.01) and accounting for the majority of the variations in MC production in the model were identified as key factors that have strong effects on MC production and discussed in detail. The combined effect of the linear and nonparametric contributions for each key factor was plotted using ODS Graphics statement [43], [45].

To test the conclusions of this study, the simplified model based on key factors was also applied to the data from previous studies in Lake Taihu, and the results were compared with results generated from the present in situ observations.

Other statistic analyses including Independent-samples T test and Spearman's correlation were carried out with SPSS version 13.0 for Windows (SPSS incorporated, Chicago, IL, USA).

Results

Environmental parameters

Annual mean and ranges of the physical and chemical variables for the Lake Taihu in 2008 are presented in Table 1. Lake Taihu is an alkaline system, with pH values above 7.5 during the experimental period. Water temperature varied from 3.9 to 32.4°C and monthly means of water temperature in the northern area peaked in July and August (Table 1, Fig. 2A). Conductivity demonstrated an adverse seasonal variation trend to DIC: reached peaks in April before bloom broke out and experienced persistent decline until October except for June (Fig. 2 B).

thumbnail
Figure 2. Seasonal variations of A) Water temperature, Water depth and Secchi depth; B) NO3-N, Na+, dissolved inorganic carbon (DIC) and conductivity and C) Microcystis density, MC concentration and MC cellular quotas in the northern area of Lake Taihu (values for each month are the mean value of fourteen sites).

https://doi.org/10.1371/journal.pone.0032020.g002

thumbnail
Table 1. Mean and ranges of the environmental parameters during the study period of Lake Taihu.

https://doi.org/10.1371/journal.pone.0032020.t001

A total of 87 phytoplankton taxa were recorded, with Microcystis spp. being the absolute dominant species in most months of the year. Temporal variation in phytoplankton abundance of various groups in the northern area was shown in Figure 3. Diatoms (Bacillariophyceae, mainly Cyclotella spp.), Cryptophyta and Chrysophyta jointly prevailed over the other groups only in winter and early spring. As the flourish of non-N-fixing Microcystis spp. in May, cyanobacteria became the absolutely advantageous taxa and maintained the superiority status in the remaining seasons. Microcystis spp. mainly contained M. aeruginosa, M. flos-aquae, M. viridis and M. wesenbergii in the study period. Spatial distribution of Microcystis biomass of Sites 1–30 from spring to autumn were demonstrated in Figure 4. Microcystis was abundant in the three northern bays and west littoral zones. Of which Meiliang bay was the most severely polluted area in Lake Taihu. Since spatial distribution of Microcystis revealed high risks in the northern area of Lake Taihu (Fig. 4), the present study is mainly focused on this area. Other potential MC-producing cyanobacteria (such as Anabaena spp. and Oscillatoria spp.) also multiplied during periods of Microcystis spp. dominance but accounted for a marginal part of cyanobacteria biomass (Fig. 3). Seasonal changes of Microcystis abundance in the northern area were shown in Figure 2C. Dramatic increase of Microcystis spp. gave rise to explosion of cyanobacteria density in May.

thumbnail
Figure 3. Temporal variation of phytoplankton density composition in the northern area of Lake Taihu.

https://doi.org/10.1371/journal.pone.0032020.g003

thumbnail
Figure 4. Spatial distribution of Microcystis density, MC concentration and MC cellular quotas in A) spring (May), B) summer (August) and C) autumn (November) of Lake Taihu.

https://doi.org/10.1371/journal.pone.0032020.g004

Dynamics of MC concentration and MC cellular quotas

Spatial distribution of MCs in bloom seasons was shown in Figure 4, revealing a high health risk of MC exposure in the northern area, especially in Meiliang Bay (sites 1–7, 4.82 µg L−1 as a mean) where the MC concentrations were up to almost 14 times (Independent-samples T test; P<0.0001) higher than those in East Taihu (sites 28–30, 0.35 µg L−1 as mean). In the southern area, MC cellular quotas presented an adverse spatial distribution pattern to Microcystis abundance and MC concentration, which disclosed higher MC-producing capability of Microcystis (Fig. 4).

Seasonal variation of MCs in the northern area was shown in Figure 2C. MC concentration was much higher in summer and autumn than in the other seasons, and was at a low level in the first five months of 2008, but increased quickly from June to October when water temperature was above 20°C. It was obvious that variations of MC concentration did not always coincide with that of Microcystis abundance (Fig. 2C). Microcystis abundance and MC cellular quotas both presented positive correlations to MC concentration (Spearman's R = 0.46 and 0.63 respectively; P<0.0001) from recruitment period of Microcystis to bloom seasons (March to November), which indicated the importance of these two factors in prediction of toxic events. MC cellular quotas provide an estimate of mean MC-producing capability of Microcystis cells. The MC cellular quotas were higher in months when Microcystis cell abundance was relatively low (for example, in April and October) than in summer when Microcystis spp. bloomed (Fig. 2C).

Results of GAM and test on previous data

Lake Taihu was an ideal system to study the complex mechanisms of various environmental parameters affecting MC-producing capability of Microcystis spp. for the naturally high levels of MCs observed in water and the high abundance of Microcystis (absolutely dominating the phytoplankton community) in the bloom seasons.

From the results described above, the northern area was seriously polluted by toxic Microcystis. In consideration of the high risks and high Microcystis (absolutely dominating the phytoplankton community) abundance with high MC concentration, the present study was focused on the northern area (site 1–17) and GAM was used to investigate the key abiotic factors affecting MC-producing capability of Microcystis. Totally, all the abiotic environmental factors included in GAM could explain about 78% of the variations in MC cellular quotas. From the various factors, the highest weighted (P<0.01) four (conductivity, dissolved inorganic carbon, water temperature and pH), which accounted for the majority of the variations (54%) in MC cellular quotas in the study period (Fig. 5A), were finally selected to simplify our model (taking the costs and timeliness of monitoring into consideration). Nitrogen and phosphorus concentration had little effects on MC production of Microcystis spp. (data not shown).

thumbnail
Figure 5. Plots showing the combined effect of the linear and nonparametric contributions for each important environmental factor on MC production by Microcystis spp. from recruitment period of Microcystis to bloom seasons of the GAMs run for A) the present study (R2 is the product of standardized F-statistics of each factor and R-squares of the whole model) and B) previous data.

https://doi.org/10.1371/journal.pone.0032020.g005

In most field studies, complete data of conductivity, water temperature, pH and dissolved inorganic carbon (the key abiotic environmental factors confirmed in the present study) were usually lacking. Only two studies [36], [37] which offered relatively more information (three of the four factors mentioned here) were picked out to draw a comparison to the results obtained from the present study. The results showed that the three abiotic factors (water temperature, pH and conductivity) were all significant at 1% level in the GAM and could explain 61% of the variations in MC cellular quotas. Plots from the multivariate model showed that the overall trends of MC cellular quotas with the three factors in previous studies were similar to that in the present research (Fig. 5B), in spite of the differences in the details (Table 2).

thumbnail
Table 2. A comparison between the results of GAMs generated from present study and previous studies (Wang et al., 2010; Wilhelm et al., 2011).

https://doi.org/10.1371/journal.pone.0032020.t002

Discussion

Temporal and spatial distribution of Microcystis spp. and MCs

The changes in abundance of Microcystis could not completely explain the fluctuations in MC concentration. Potentially MC-producing and non-MC-producing cells can coexist in natural cyanobacterial populations and the proportion of toxic cells can differ considerably over time during bloom season [46][48]. Many species of Microcystis spp. in Lake Taihu were potential MC producer like M. aeruginosa, M. flos-aquae and M. viridis, while another common species M. wesenbergii was reported to be non-toxic [49]. If a genotype can produce MCs, it should contain intact genes from the microcystin synthetase (mcy) gene cluster [50], [51]. Several studies have targeted the mcy gene cluster for the determination of relative abundance of MC-producing Microcystis cells in the total Microcystis population [50][52]. A study conducted in Meiliang Bay of Lake Taihu during the same period as the present research investigated the proportion of toxic Microcystis based on mcy gene (mcyA) and partial Microcystis-specific 16S rDNA sequence using real-time PCR. This research revealed shifts from non-toxic to toxic Microcystis strains from June to October, 2008 [53], which supported the result of increasing MC-producing capability in the northern area during this period in the present study. It seemed that the proportion of potentially toxic Microcystis cells increased with the development of bloom [35] when water temperature was above 20°C in Lake Taihu. It might be assumed that high MC cellular quotas in early spring are a result of the recruitment of highly toxic cells surviving the winter which preserved the mcy genotype composition from one year to the next [46].

The maximum mean concentration (15.2 µg L−1) and the maximum concentration (78.0 µg L−1) of MCs were both detected in October, significantly higher than those of Microcystis blooms in other regions of the world [54][56], revealed the severe contamination by MCs in Lake Taihu. In the present study, Microcystis abundance and MC concentration reached their peaks in different months, although they presented positive correlations. The possible explanation for this might be that compared to Microcystis density (Spearman's R = 0.46; P<0.0001), MC cellular quotas which represented the proportion of toxic cells had a closer relationship (Spearman's R = 0.63; P<0.0001) with MC concentration, so the maximum MC concentrations did not occur in the period of the heaviest algal blooms, but appeared in October when MC cellular quotas reached peak value simultaneously. Previous studies reported similar results [57], [58]. Seasonal variation of MC concentration indicated that the potential MC threat is present both during both bloom and non-bloom seasons in Lake Taihu, thus water safety in non-bloom seasons should also make an appeal.

Spatial distribution of Microcystis density and MC concentration in the bloom season warned of the high risks posed by MCs in the northern area. Meanwhile, it should be noticed that MC concentration in some sites of the eastern and western areas was also at a danger level in summer and autumn, despite the low density of Microcystis cells. The MC-producing capability of Microcystis in these areas was quite high. Interestingly, aquatic macrophytes could always be found in these sites during the bloom season. Phytoplankton and aquatic macrophytes are the primary producers in aquatic ecosystem; they compete intensely for various resources [59], and inhibit each other through secreting allelochemical such as microcystins and phenolic compounds [11], [60][62]. High MC production in this condition might be due to allelopathy and response of Microcystis to suboptimal conditions for growth resulted from competition with macrophytes.

Abiotic factors influencing MC cellular quotas and test on previous studies

Environmental factors may affect the mean MC-producing capability of Microcystis spp. (MC cellular quotas) through the following two aspects, the proportion of potential MC-producing cells [63], [64] which possess the mcy gene cluster encoding MC synthesis [51] and expression level of mcy gene cluster [65], [66]. They may influence competitive ability of toxic and nontoxic Microcystis strains and the physiological condition of the toxic cells which are associated with MC production through their direct impact on cell division rate [14], [18], [20]. Various parameters including biotic and abiotic factors were reported to be regulating factors of MC production of cyanobacteria in field and laboratory (Table 3). The different results from previous studies and the complexity of interactions occurring among environmental factors in nature confirm the importance of very local conditions in the present research. In this study, we aimed to investigate the abiotic factors affecting MC cellular quotas and took no account of any biotic ones such as predators [25][27] in the model. All the parameters included in GAM accounted for 78% of the variations in MC cellular quotas. Theoretically, the biotic factors might account for the remaining 22% which could not be explained by the abiotic parameters included in the model.

thumbnail
Table 3. A comparison of the environmental factors affecting MC production of cyanobacteria from literatures and the present study.

https://doi.org/10.1371/journal.pone.0032020.t003

Nitrogen and phosphorus concentration had little influence on MC production while factors related to carbon fixation and proliferation of Microcystis presented significant effects on MC cellular quotas. Perhaps nutrient loading in Lake Taihu (annual means of TN and TP were 2.98 and 0.123 mg L−1, respectively) was at a saturation level to MC production of Microcystis.

Conductivity which explained 21% of the variation of MC cellular quotas was the highest weighted parameters in the statistical model (Fig. 5A). It is a parameter related to the ability of electric conduction of water, and can indicate the ion concentration. Microcystis utilize various inorganic ions such as macronutrients and trace metal for growth. With the development of the population, available ions for growth decline and may become insufficient. Toxic cells have a competitive advantage over nontoxic ones under suboptimal conditions for growth [47], [67]. When environment conditions (for example, nutrients) were no longer well appropriate for growth, the proportion of toxic cells in Microcystis spp. increased and resulted in rise of MC production and MC concentration. The shifts from non-toxic to toxic Microcystis strains with the development of bloom when temperature was above 20°C might be due to deterioration of growth environment. Similarly, a test on previous data (Fig. 5B, Table 2) also showed a significant decrease in MC cellular quotas with conductivity (P = 0.0008).

Dissolved inorganic carbon (DIC) and pH could explain 12% and 10% of the changes in MC production, respectively (Fig. 5A). They were important parameters related to carbon fixation and proliferation of Microcystis. The photosynthesis of phytoplankton depletes dissolved carbon dioxide and increases pH and the concentration of dissolved oxygen in water. In such alkaline environments (low CO2/O2 ratios), cyanobacteria enable themselves to overwhelm other phytoplankton through establishing a carbon-concentrating mechanism (CCM) which adapts them to fluctuating inorganic carbon (Ci) and O2 conditions to concentrate Ci more than 1,000-fold inside the cell [68]. HCO3 transport system is one of the two functional elements composing CCM. Most commonly, HCO3 is transported by an HCO3 ATP binding cassette (ABC) transporter and two Na+-dependent HCO3 transporters [69]. Consequently, Na+ is required for the active transport of Ci and can not be replaced by other monovalent metal ion such as K+ [70]. A negative correlation between Na+ concentration and DIC in Microcystis recruitment period and bloom seasons (Spearman's R = −0.237 P = 0.006) was found. The inscrutable decrease of Na+ limited uptake of Ci by Microcystis, caused Ci accumulation in water and posed relative deficiency of intracellular inorganic carbon (Ci,i). MCs might be produced in response to a relative deficiency of Ci,i to enhance the efficiency of the adaptation of the photosynthetic apparatus to fluctuating inorganic carbon conditions in cyanobacterial cells [71]. This may help to explain the positive relationship between DIC and MC cellular quotas. Effects of pH presented a curve pattern with the lowest MC cellular quotas occurred in a moderate level (pH = 8.3 in the present study and pH = 8.5 in test on previous data, Fig. 5, Table 2). Because of the recruitment of highly toxic cells surviving the winter in early spring when pH level was low, MC cellular quotas were quite high [46], [47], [67]. It might be assumed that decreasing MC cellular quotas at increasing pH are due to the suitable conditions in company with multiplication and dominance in phytoplankton of Microcystis favored nontoxic cells over toxic ones. As the development of bloom, a gradually aggravating lack of free CO2 and decline in HCO3 at increasing pH level when pH exceeded the value of 8.3 led to the enhance of MC production [22].

Temperature explained 11% of the variation of MC cellular quotas (Fig. 5A). MC cellular quotas changed with water temperature in a unimodal pattern with the maximum value occurred at 21.5°C. Test on previous studies showed a similar pattern with the maximum value at 24.5°C (Fig. 5B, Table 2). These results gave support to previous studies in both experiments [72] and field [73], suggesting that the optimal temperature for MC production by M. aeruginosa was between 20 and 25°C. As shown in Figure 2C, MC cellular quotas peaked in October when the mean water temperature was 21.8°C and changed with temperature as the predict pattern (Fig. 5A, Table 2) from June to November. However, MC-producing capability was quite low in May when the water temperature was theoretically optimal for MC production, which might be because of the small proportion of toxic cells.

The test on previous data shows that the overall trends of MC cellular quotas with the three factors (water temperature, conductivity and pH) in studies of Wang et al. [36] and Wilhelm et al. [37] were similar to that in the present research, in spite of the differences in details (Table 2). It should be noted that most of previous data applied in the model was from a study conducted in Gonghu Bay of Lake Taihu [36]. Compared with the northern area on which the present research focused, environmental condition in littoral Gonghu Bay with less water exchange was relatively stable. The different hydrographic conditions might be the cause of some differences between the two models. In spite of lack of DIC data, the test results from previous data reconfirmed the important roles of the other key factors obtained from the present study in regulation of MC production in Lake Taihu.

Variations of MCs are directly related to population dynamics of cyanobacteria [22] which include cell abundance, proportion and the physiological conditions of toxic cells. Although abundance of Microcystis cells are a traditionally indicator of toxic risks posed by MCs in many circumstances, the present study indicates that the changes in Microcystis abundance can not completely explain the fluctuations in MC concentration in Lake Taihu, and that water temperature, DIC, conductivity and pH are also important regulating factors.

Conclusions

The health risks of MC exposure in Lake Taihu were high, especially in the northern area. Microcystis density and parameters affecting MC-producing capability of Microcystis were both important in predicting MC variation. As a powerful and scientific predictive modeling tool to discover the hidden pattern of predictors and improves the predictive performance, generalized additive model (GAM) was used to investigate quantitative relationships between abiotic environmental factors and MC cellular quotas from recruitment period of Microcystis to bloom seasons. The results of the model together with a test on previous data indicated that factors related to carbon fixation and proliferation of Microcystis (conductivity, DIC, water temperature and pH) presented significant correlations with MC cellular quotas, suggesting their possible use, in addition to Microcystis abundance, as warning signs to predict toxic events. The interesting relationship between macrophytes and MC cellular quotas of Microcystis needs further investigation.

Acknowledgments

We greatly appreciate the useful comments and suggestions by Dr. Brett Neilan. The authors are also grateful to the staff of Taihu Ecosystem Research Station for field and laboratory assistance.

Author Contributions

Conceived and designed the experiments: MT PX JC. Performed the experiments: MT BQ DZ YN MZ QW LW. Analyzed the data: MT. Contributed reagents/materials/analysis tools: PX JC BQ. Wrote the paper: MT.

References

  1. 1. Paerl HW, Fulton RS III, Moisander PH, Dyble J (2001) Harmful freshwater algal blooms, with an emphasis on cyanobacteria. Scientific World Journal 1: 76–113.
  2. 2. Havens KE, James T, East TL, Smith VH (2003) N∶P ratios, light limitation, and cyanobacterial dominance in a subtropical lake impacted by non-point source nutrient pollution. Environmental Pollution 122: 379–390.
  3. 3. Te SH, Gin KYH (2011) The dynamics of cyanobacteria and microcystin production in a tropical reservoir of Singapore. Harmful Algae 10: 319–329.
  4. 4. Carmichael WW (1992) Cyanobacteria secondary metabolites-the cyanotoxins. Journal of Applied Bacteriology 72: 445–459.
  5. 5. Nishiwaki-Matsushima R, Ohta T, Nishiwaki S (1992) Liver tumor promotion by the cyanobacterial cyclic peptide toxin microcystin-LR. Journal of Cancer Research and Clinical Oncology 118: 420–424.
  6. 6. Gérard C, Poullain V, Lance E, Acou A, Brient L, et al. (2009) Influence of toxic cyanobacteria on community structure and microcystin accumulation of freshwater mollusks. Environmental Pollution 157: 609–617.
  7. 7. Lance E, Neffling MR, Gérard C, Meriluoto J, Bormans M (2010) Accumulation of free and covalently bound microcystins in tissues of Lymnaea stagnalis (Gastropoda) following toxic cyanobacteria or dissolved microcystin-LR exposure. Environmental Pollution 158: 674–680.
  8. 8. Carmichael WW (2001) Health effects of toxin-producing cyanobacteria: “the CyanoHABs”. Human and Ecological Risk Assessment 7: 1393–1407.
  9. 9. Chen J, Zhang DW, Xie P, Wang Q, Ma ZM (2009) Simultaneous determination of microcystin contaminations in various vertebrates (fish, turtle, duck and water bird) from a large eutrophic Chinese lake, Lake Taihu, with toxic Microcystis blooms. Science of the Total Environment 407: 3317–3322.
  10. 10. Chen JZ, Ye JY, Zhang HY, Jiang XJ, Zhang YX, et al. (2011) Freshwater toxic cyanobacteria induced DNA damage in apple (Malus pumila), rape (Brassica napus) and rice (Oryza sativa). Journal of Hazardous Materials 190: 240–244.
  11. 11. Jiang JL, Gu XY, Song R, Wang XR, Yang LY (2011) Microcystin-LR induced oxidative stress and ultrastructural alterations in mesophyll cells of submerged macrophyte Vallisneria natans (Lour.) Hara. Journal of Hazardous Materials 190: 188–196.
  12. 12. Azevedo SMFO, Carmichael WW, Jochimsen EM, Rinehart KL, Lau S, et al. (2002) Human intoxication by microcystins during renal dialysis treatment in Caruaru-Brazil. Toxicology 181: 441–446.
  13. 13. Chen J, Xie P, Li L, Xu J (2009) First identification of the hepatotoxic microcystins in the serum of a chronically exposed human population together with indication of hepatocellular damage. Toxicological Sciences 108: 81–89.
  14. 14. Orr PT, Jones GJ (1998) Relationship between microcystin production and cell division rates in nitrogen-limited Microcystis aeruginosa cultures. Limnology and Oceanography 43: 1604–1614.
  15. 15. Halstvedt CB, Rohrlack T, Ptacnik R, Edvardsen B (2008) On the effect of abiotic environmental factors on production of bioactive oligopeptides in field populations of Planktothrix spp. (Cyanobacteria). Journal of Plankton Research 30: 607–617.
  16. 16. Watanabe MF, Oishi S (1985) Effect of environmental factors on toxicity of a cyanobacterium (Microcystis aeruginosa) under culture conditions. Applied and Environmental Microbiology 49: 1342–1344.
  17. 17. Sivonen K (1990) Effects of light, temperature, nitrate, orthophosphate, and bacteria on growth of and hepatotoxin production by Oscillatoria agardhii strains. Applied and Environmental Microbiology 56: 2658–2666.
  18. 18. Vézie C, Rapala J, Vaitomaa J, Seitsonen J, Sivonen K (2002) Effect of nitrogen and phosphorus on growth of toxic and nontoxic Microcystis strains and on intracellular microcystins concentrations. Microbial Ecology 43: 443–454.
  19. 19. Dai RH, Liu HJ, Qu JH, Zhao X, Hou YN (2009) Effects of amino acids on microcystin production of the Microcystis aeruginosa. Journal of Hazardous Materials 161: 730–736.
  20. 20. Wiedner C, Visser PM, Fastner J, Metcalf JS, Codd GA, et al. (2003) Effects of light on the microcystin content of Microcystis strain PCC 7806. Applied and Environmental Microbiology 69: 1475–1481.
  21. 21. Song LR, Sano T, Li RH, Watanabe M, Liu YD, et al. (1998) Microcystin production of Microcystis viridis (cyanobacteria) under different culture conditions. Phycological Research 46: Suppl. 219–23.
  22. 22. Jähnichen S, Petzoldt T, Benndorf J (2001) Evidence for control of microcystin dynamics in Bautzen reservoir (Germany) by cyanobacterial population growth rates and dissolved inorganic carbon. Archiv fuer Hydrobiologie 150: 177–196.
  23. 23. Utkilen H, Gjølme N (1995) Iron-stimulated toxin production in Microcystis aeruginosa. Applied and Environmental Microbiology 61: 797–800.
  24. 24. Wang JX, Xie P, Guo N (2007) Effects of nonylphenol on the growthand microcystin production of Microcystis strains. Environmental Research 103: 70–78.
  25. 25. Jang MH, Ha K, Lucas MC, Joo GJ, Takamura N (2004) Changes in microcystin production by Microcystis aeruginosa exposed to phytoplanktivorous and omnivorous fish. Aquatic Toxicology 68: 51–59.
  26. 26. Yang H, Xie P, Ke ZX, Liu YQ, Wu SK, et al. (2006) The impact of planktivorous fishes on microcystin concentrations in Meiliang Bay, Lake Taihu. China. J Freshw Ecol 21: 721–723.
  27. 27. Jang MH, Ha K, Takamura N (2008) Microcystin production by Microcystis aeruginosa exposed to different stages of herbivorous zooplankton. Toxicon 51: 882–889.
  28. 28. Hastie TJ, Tibshirani RJ (1990) Generalized Additive Models. Chapman and Hall, London, UK.
  29. 29. Yee TW, Mitchell ND (1991) Generalized additive models in plant ecology. Journal of Vegetation Science 2: 587–602.
  30. 30. Pope CA III, Rodermund DL, Gee MM (2007) Mortality Effects of a Copper Smelter Strike and Reduced Ambient Sulfate articulate Matter Air Pollution. Environmental Health Perspectives 115: 679–683.
  31. 31. Xie P (2008) Historical Development of Cyanobacteria with Bloom Disaster in Lake Taihu, Science Press, Beijing, PRC (in Chinese).
  32. 32. Ma RH, Kong FX, Duan HT, Zhang SX, Kong WJ, et al. (2008) Spatio-temporal distribution of cyanobacteria blooms based on satellite imageries in Lake Taihu, China. Journal of Lake Science 20: 687–694. (in Chinese).
  33. 33. Shen PP, Shi Q, Hua ZC, Kong FX, Wang ZG, et al. (2003) Analysis of microcystins in cyanobacteria blooms and surface water samples from Meiliang Bay, Taihu Lake, China. Environment International 29: 641–647.
  34. 34. Liu YQ, Xie P, Zhang DW, Wen ZR (2008) Seasonal dynamics of microcystins with associated biotic and abiotic parameters in two bays of Lake Taihu, the third largest freshwater lake in China. Bulletin of Environment Contamination and Toxicology 80: 24–29.
  35. 35. Ye WJ, Liu XL, Tan J, Li DT, Yang H (2009) Diversity and dynamics of microcystin-Producing cyanobacteria in China's third largest lake, Lake Taihu. Harmful Algae 8: 637–644.
  36. 36. Wang Q, Niu YA, Xie P, Chen J, Ma ZM, et al. (2010) Factors affecting temporal and spatial variations of microcystins in Gonghu Bay of Lake Taihu, with potential risk of microcystin contamination to human health. Scientific World Journal 10: 1795–1809.
  37. 37. Wilhelm SW, Farnsley SE, LeCleir GR, Layton AC, Satchwell MF, et al. (2011) The relationships between nutrients, cyanobacterial toxins and the microbial community in Taihu (Lake Tai), China. Harmful Algae 207–215.
  38. 38. Niu Y, Shen H, Chen J, Xie P, Yang X, et al. (2011) Phytoplankton community succession shaping bacterioplankton community composition in Lake Taihu, China. Water Res 45: 4169–4182.
  39. 39. Greenberg AE, Clesceri LS, Eaton AD (1992) Standard methods for the examination of water and wastewater. American Public Health Association, Washington, D.C.
  40. 40. Parsons TR, Maita Y, Lalli CM (1984) A Manual of Chemical and Biological Methods for Sea Water Analysis. Pergamon, Elmsford, New York, USA.
  41. 41. Hu HJ, Wei YX (2006) Freshwater Algae in China, Science Press, Beijing, PRC (in Chinese).
  42. 42. John DM, Whitton BA, Brook AJ (2002) The freshwater algal flora of the British Isles-an identification guide to freshwater and terrestrial algae. Cambridge University Press, UK.
  43. 43. SAS Institute Inc (1999) SAS Procedures Guide, Version 8. Cary, NC: SAS Institute Inc.
  44. 44. Dong X (2001) Fitting Generalized Additive Models with the GAM Procedure. SUGI Proceedings.
  45. 45. SAS Institute Inc (2004) SAS/STAT 9.1 User's Guide. Cary, NC: SAS Institute Inc.
  46. 46. Kurmayer R, Christiansen G, Chorus I (2003) The abundance of microcystin-producing genotypes correlates positively with colony size in Microcystis sp. and determines its microcystin net production in Lake Wannsee. Appl Environmental Microbiology 69: 787–795.
  47. 47. Sabart M, Pobel D, Briand E, Combourieu B, Salençon MJ, et al. (2010) Spatiotemporal Variations in Microcystin Concentrations and in the Proportions of Microcystin-Producing Cells in Several Microcystis aeruginosa Populations. Applied and Environmental Microbiology 76: 4750–4759.
  48. 48. Joung SH, Oh HM, Ko SR, Ahn CY (2011) Correlations between environmental factors and toxic and non-toxic Microcystis dynamics during bloom in Daechung Reservoir, Korea. Harmful Algae 10: 188–193.
  49. 49. Xu Y, Wu ZX, Yu BS, Peng XG, Yu LZ, et al. (2008) Non-microcystin producing Microcystis wesenbergii (Komárek) Komárek (Cyanobacteria) representing a main waterbloom-forming species in Chinese waters. Environmental Pollution 156: 162–167.
  50. 50. Dittmann E, Neilan BA, Erhard M, von Döhren H, Börner T (1997) Insertional mutagenesis of a peptide synthetase gene that is responsible for hepatotoxin production in the cyanobacterium Microcystis aeruginosa PCC 7806. Mol Microbiol 26: 779–787.
  51. 51. Tillett D, Dittmann E, Erhard M, von Döhren H, Börner T, et al. (2000) Structural organization of microcystin biosynthesis in Microcystis aeruginosa PCC7806: an integrated peptide-polyketide synthetase system. Chemistry & Biology 7: 753–64.
  52. 52. Furukawa N, Noda S, Tsuneda T (2006) Highly sensitive real-time PCR assay for quantification of toxic cyanobacteria based on microcystin synthetase A gene. Journal of Bioscience and Bioengineering 102: 90–96.
  53. 53. Shi LM, Cai YF, Yang HL, Li PF, Kong FX, et al. (2009) Dynamics of composition of different Microcystis spp. genotypes and abundance of toxic Microcystis in Meiliang Bay of Lake Taihu during bloom. Journal of Lake Science 21: 801–805. (in Chinese).
  54. 54. Kotak BG, Lam AKY, Prepas EE (1995) Variability of hepatotoxin microcystin-LR in hypereutrophic drinking water lakes. Journal of Phycology 31: 248–263.
  55. 55. Kotak BG, Lam AKY, Prepas EE, Hrudey SE (2000) Role of chemical and physical variables in regulating microcystin-LR concentration in phytoplankton of eutrophic lakes. Canadian Journal of Fisheries and Aquatic Sciences 57: 1584–1593.
  56. 56. Haney JF, Ikawa M (2000) Final Report: A Survey of 50 NH Lakes for Microcystins (MCs), Report 65. New Hampshire Water Resources Research Center, Durham.
  57. 57. Ozawa K, Fujioka H, Muranaka M, Yokoyama A, Katagami Y, et al. (2005) Spatial distribution and temporal variation of Microcystis concentration in Lake Biwa. Environmental Toxicology 20: 270–276.
  58. 58. Zhang DW, Xie P, Liu YQ, Chen J, Wen ZR (2009) Spatial and temporal variations of microcystins in hepatopancreas of a freshwater snail from Lake Taihu. Ecotoxicology and Environmental Safety 72: 466–472.
  59. 59. Casanova MT, Burch MD, Brock MA, Bond PM (1999) Does toxic Microcystis aeruginosa affect aquatic plant establishment? Environmental Toxicology 14: 97–109.
  60. 60. LeBlanc S, Pick FR, Aranda-Rodriguez R (2005) Allelopathic effects of the toxic cyanobacterium Microcystis aeruginosa on duckweed, Lemna gibba L. Environmental Toxicology 20: 67–73.
  61. 61. Zhang M, Wang ZQ, Xu J, Liu YQ, Ni LY, et al. (2010) Ammonium, microcystins, and hypoxia of blooms in eutrophic water cause oxidative stress and C-N imbalance in submersed and floating-leaved aquatic plants in Lake Taihu, China. Chemosphere 82: 329–339.
  62. 62. Zhu JY, Liu BY, Wang J, Gao YN, Wu ZB (2010) Study on the mechanism of allelopathic influence on cyanobacteria and chlorophytes by submerged macrophyte (Myriophyllum spicatum) and its secretion. Aquatic Toxicology 98: 196–203.
  63. 63. Yoshida M, Yoshida T, Takashima Y, Hosoda N, Hiroishi S (2007) Dynamics of microcystin-producing and non-microcystin-producing Microcystis populations is correlated with nitrate concentration in a Japanese lake. FEMS Microbiol Lett 266: 49–53.
  64. 64. Davis TW, Berry DL, Boyer GL, Gobler CJ (2009) The effects of temperature and nutrients on the growth and dynamics of toxic and non-toxic strains of Microcystis during cyanobacteria blooms. Harmful Algae 8: 715–725.
  65. 65. Kaebernick M, Neilan BA, Börner T, Dittmann E (2000) Light and the transcriptional response of the microcystin biosynthesis gene cluster. Appl Environ Microbiol 66: 3387–3392.
  66. 66. Sevilla E, Martin-Luna B, Vela L, Bes MT, Fillat MF, et al. (2008) Iron availability affects mcyD expression and microcystin-LR synthesis in Microcystis aeruginosa PCC7806. Environmental Microbiology 10: 2476–2483.
  67. 67. Janse I, Kardinaal WEA, Meima M, Fastner J, Visser PM, et al. (2004) Toxic and nontoxic Microcystis colonies in natural populations can be differentiated on the basis of rRNA gene internal transcribed spacer diversity. Applied and Environmental Microbiology 70: 3979–3987.
  68. 68. Giordano M, Beardall J, Raven JA (2005) CO2 concentrating mechanisms in algae: mechanisms, environmental modulation, and evolution. Annual Review of Plant Biology 56: 99–131.
  69. 69. Kaplan A, Schwarz R, Lieman-Hurwitz J, Reinhold L (1991) Physiological and molecular aspects of the inorganic carbon-concentrating mechanism in cyanobacteria. Plant Physiol 97: 851–855.
  70. 70. Miller AG, Turpin DH, Canvin DT (1984) Na+ requirement for growth, photosynthesis, and pH regulation in the alkalotolerant cyanobacterium Synechococcus leopoliensis. Journal of Bacteriology 159: 100–106.
  71. 71. Jähnichen S, Ihle T, Petzoldt T, Benndorf J (2007) Impact of inorganic carbon availability on microcystin production by Microcystis aeruginosa PCC 7806. Applied and Environmental Microbiology 73: 6994–7002.
  72. 72. Van der Westhuizen AJ, Eloff JN (1985) Effects of temperature and light on the toxicity and growth of the blue-green alga Microcystis aeguginosa (UV–006). Planta 163: 55–59.
  73. 73. Li SX, Xie P, Xu J, Zhang X, Qin J, et al. (2007) Factors shaping the pattern of seasonal variations of microcystins in Lake Xingyun, a subtropical plateau lake in China. Bulletin of Environment Contamination and Toxicology 78: 226–230.