This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Spatial patterns of phytoplankton across the southern Taihu basin of China were examined through five aspects: abundance, composition, richness, evenness and diversity. Data were collected from 33 sites in dry (April) and wet season (July), 2010. Global Moran's I statistics and Local Indicators of Spatial Association (LISA) were used to characterize the spatial autocorrelation for phytoplankton metrics at the whole area scale. It was found that the phytoplankton pattern presented significant spatial autocorrelation in both seasons. Specifically, the wet season witnessed more patterns of richness at local scale and more evenness patterns at regional scale. Spatial regression models were carried out to identify environmental factors that would simulate or limit of phytoplankton patterns. Results showed that diversity and composition of phytoplankton related significantly to nutrients. Also, phytoplankton richness could be predicted by potassium permanganate. No significant relationships were identified between environmental factors and phytoplankton evenness in both seasons. Our study also highlighted the importance of incorporating spatial dependence, when identifying the explanatory environmental factors for phytoplankton patterns.
Keywords: phytoplankton patterns, environmental factors, spatial analysis, seasonal variations
Great quantities of nutrients have been discharged into rivers worldwide, as a result of the rapid population growth, energy consumption, food production (Su et al. 2011; Zhang et al. 2012) and the subsequent ecological change during the past fifty years (Diodato and Ceccarelli 2005). Excessive nutrient loadings often cause eutrophication, of which an obvious symptom is the accelerated growth and accumulation of phytoplankton. Phytoplankton is comprised of diverse phyla and many freshwater genera can form blooms, which not only affect aquatic ecosystems , but also cause significant economic losses and finally threaten human sustainability . Given that algal blooms have become a critical global environmental issue , it is imperative to capture phytoplankton dynamics over time and their influential environmental factors.
Recent studies demonstrated that the composition and quantity of phytoplankton relied heavily on water environmental factors . Many studies used environmental factors to indicate the spatiotemporal variability in phytoplankton patterns . However, most previous studies ignored the issue of spatial autocorrelation. Spatial autocorrelation denotes the conditions where more similar attributes are found among neighboring observations than those far away . The phytoplankton patterns, to some extent, should exhibit spatial autocorrelation, given the comparable environmental characteristics of sites nearby . In addition, most recent studies were performed at a certain temporal dimension using a brief and static approach. The contributing environmental factors in different seasons have not been adequately discussed.
This paper addressed the above concerns, whereby an analysis was conducted using data from the case of southern Taihu basin, China. The objectives of this study are to: (1) compare the phytoplankton patterns in different seasons; (2) characterize the spatial autocorrelation of the phytoplankton patterns; and (3) analyze phytoplankton patterns in relation to environmental factors.
2. Materials and methods
2.1 Study area
Taihu Lake, located within Yangtze Delta, is the third largest freshwater lake in China (Fig. 1). It is 68.0 km in length and 35.7 km in width, with its basin covering an area of 36,500 km2. Southern Taihu basin has an area of 15,600 km2, with shoreline at the length of 60 km. The Taihu basin is one of the most developed regions in China, including the megacities of Shanghai, Wuxi, Suzhou, Huzhou and Jiaxing. This area belongs to subtropical monsoon climate, mainly with mountainous region and hills in the west, plain and river networks in the east. The region enjoys four distinct seasons with annual temperature averages 17.5â„ƒ, and rainfall averages 1,100 mm. Though it accounts for less than 0.4% of area in China, the Taihu basin area has a population of 3.6 million (2.9% of the total number in China), and contributes to 11% of the gross domestic product (GDP) of the nation. The river network covers 7% of the total basin with a total distributary length of about 120,000 km (Wang et al. 2009).
2.2 Sampling and chemical analysis
33 sites were sampled during the periods of April 10-12 and July 18-20, 2010. The average daily precipitation was 3.49 mm in April (dry season) and 6.56 mm in July (wet season). There was no rain on the days of sample collection. The 33 sampling sites were generally evenly distributed across the southern Taihu basin (Fig. 1). Phytoplankton samples were collected by filtering 10 L water samples through a 64 μm plankton net, and fixed with formalin. Phytoplankton samples (2 L of water) were collected at the time of water sampling from the same sites. At each site, water samples were taken using a 2.5 L PVC Go-flo sampler (General Oceanics, Miami, FL, USA).
All filtered water samples were placed in a deep freezer at -20°C before further processing in the laboratory. The analysis followed national quality standards for surface waters, China (GB3838-2002). Ammonium (NH4-N) and nitrate (NO3-N) were extracted by spectrophotometric method with salicylic acid. Total phosphorus (TP) and total nitrogen (TN) were obtained through ammonium molybdate spectrophotometric method. Water temperature, pH, and specific conductance were determined from vertical profiles taken on each sampling day using an YSI 6600 Multi-Parameter Water Quality Sonde. Potassium permanganate index (CODMn): acidic (alkaline) potassium permanganate method; dissolved oxygen (DO): electrochemical probe method; copper (Cu), zinc (Zn) and lead (Pb): atomic absorption spectrophotometry (chelating extraction). PO4, solid sediment (SS), inorganic matter (INOR), organic matter (OR) and chlorophyll a (chl-a) were analyzed according to standard methods (Jin and Tu 1990).
2.3 Metric analysis for phytoplankton patterns
Metric analysis provides a quantitative approach to depicting phytoplankton patterns. A wide variety of metrics have been developed in recent years, and they can be generally divided into five categories: abundance, composition, richness, evenness and diversity. In order to ensure comparability with previous studies and low redundancy among metrics, five metrics were selected from the five categories: total number of cells per sample, percentage of phyla, Shannon-Wiener index, Pielou index and Margalef index. Shannon-Wiener is a classical index reflecting biodiversity of any population in which each member belongs to a unique group, type or species . The Margalef index is a measure of species richness, which is calculated from the total number of species presented the abundance or total number of individuals . Pielou evenness index refers to how close in numbers each species in an environment are, and it is a measure of biodiversity which quantifies how equal the community is numerically .
2.4 Spatial autocorrelation
Moran's I index was applied to characterize the spatial autocorrelation of phytoplankton composition. Values of Moran's I range from -1 (indicating perfect dispersion) to 1 (perfect correlation). A zero value implies a random spatial pattern with no spatial autocorrelation. Positive values suggest spatially clustered patterns in adjacent sites and negative values indicate that samples reveal very different values from the neighboring ones . Local indicators of spatial association (LISA) were further applied, since Moran's I is not capable of identifying the location of clusters and their types of spatial autocorrelation . Four categories of LISA exist: high-high means high phytoplankton metric values are surrounded by high values; low-low means low phytoplankton metric values are surrounded by low values; high-low means high values surrounded by low values; and low-high means a low value surrounded by a high value neighborhood. All calculations were performed using GeoDa 0.9.5-i (Beta) software .
2.5 Spatial regression
Using simple univariate statistical method in the case of dependence among variables may cause severe under estimations (Nazemi and Elshorbagy 2012). Therefore in our study spatial regression was performed to determine the relationships between phytoplankton patterns (metrics in particular) and environmental factors. Spatial error and spatial lag are two primary types of spatial dependence. Therefore, regression models are employed to incorporate spatial dependence in the form of spatial error and spatial lag. Spatial error models imply that the random error terms are spatially correlated, while spatial lag models suggest the impacts from the adjacent sites. Independent variables obtained from stepwise regression were used in spatial regressions, in order to avoid the potential multicollinearity among environmental factors .
The equation of the spatial lag model is given by (Anselin 1995):
y =ρWy + Xβ +ε (1)
where y is a vector of observations on the dependent variable, Wy is a spatially lagged dependent variable for weights matrix W, X is a matrix of observations on the explanatory variables, ε is a vector of error terms, and ρ and β are parameters.
The equation of spatial error model is given by (Anselin 1995):
y =Xβ + ε (2)
with ε =λWε + μ
where y is a vector of observations on the dependent variable, W is the spatial weights matrix, X is a matrix of observations on the explanatory variables, ε is a vector of spatially auto-correlated error terms, μ is a vector of error terms, and λ and β are parameters. All spatial regression models were performed using GeoDa 0.9.5-i (Beta) software .
3.1 General descriptions of phytoplankton composition and water quality
A total of 47 species of phytoplankton were identified in dry season and 43 species were identified in wet season during the study period. These phytoplankton samples belonged to 7 genera of the Chlorophyta (18 species in dry season and 19 species in wet season), Bacillariophyta (17 species in dry season and 11 species in wet season), Cyanophyta (6 species in dry season and 7 species in wet season), Chrysophyta (2 species in dry season and 1 species in wet season), Dinophyta (2 species in dry season and 1 species in wet season), Cryptophyta (1 species both in dry and wet season), and Euglenophyta (1 species in dry season and 3 species in wet season). The most dominant genus was Bacillariophyta (39.23% and 25.41% of the total phytoplankton biomass in dry and wet season, respectively), followed by Cryptophyta (20.60% and 25.77% of the total phytoplankton in dry and wet season, respectively).
General data statistics were displayed in Table 1. Nine variables had significant differences between dry and wet seasons, in which seven were significant at 99% confidence level (p<0.01) and two were significant at 95% confidence level (p<0.05). Compounds of nitrogen and phosphorus all displayed significant changes. Mean value of temperature and chlorophyll-a increased from dry to wet season with significant changes. DO showed significant changes during the two seasons with mean value decreased. For metal, only Zn had significant changes with mean value decreased from dry to wet season.
3.2 Metric analysis
Fig. 2 and Fig. 3 showed the phytoplankton patterns, in terms of metrics values, for each sampling site in different seasons across the study area. Most minimum values of index were observed at Site 15, such as Shannon-Wiener and Margalef in both seasons, and Pielou in dry season. The other minimum values existed at Site 18 (abundance in dry season) and 20 (abundance in wet season). Moreover, the values of abundance, Shannon-Wiener, and Margalef index were generally smaller on the western side than on the eastern side of the southern Taihu basin. In terms of composition, Bacillariophyta was dominant in the whole area in both seasons; the biomass of Chlorophyta and Cryptophyta were high in both seasons, and Cyanophyta was high in wet season only.
The abundance and Pielou index showed no significant changes during the two seasons (Table 2). Conversely, the Shannon-Wiener and Margalef index of phytoplankton exhibited significant changes (p<0.01). Results about composition showed that Bacillariophyta, Euglenophyta and Cyanophyta displayed significant changes (p<0.05), while others showed no significant changes during the two seasons.
3.3 Spatial autocorrelation and seasonal changes
Great differences in spatial autocorrelation existed among the five metrics (Fig. 4 and Fig. 5). Specifically, Moran's I value was the highest for Pielou in wet season (0.928) and followed by Pielou in dry season (0.682). Moran's I values were the lowest for abundance in wet season (0.044). Differences in spatial autocorrelation were found between the two seasons. Moran's I values were lower in wet season than in dry season for Shannon-Wiener, Margalef and abundance, but higher for Pielou and composition. These results illustrated that richness of phytoplankton became more localized in wet season, while evenness distribution of phytoplankton became more regionalized in wet season.
As displayed in Fig. 4, high-high clusters for Shannon-Wiener were generally concentrated in upstream in dry season and in downstream near Taihu Lake in wet season. The locations of high-high clusters for Margalef changed from the eastern region to the downstream areas in the north. Such richness of phytoplankton would result from the assembled patterns of phytoplankton in rainy season. No significant changes for Pielou were detected between the two seasons. High-high clusters indicated the evenly distributed phytoplankton composition in upstream. Fig. 5 illustrated the composition of the three dominated phytoplankton. There were few high-high clusters identified between the two seasons.
3.4 Phytoplankton patterns in relation to environmental factors
Phytoplankton patterns in relation to environmental factors obtained from spatial regression were presented in Table 3. Most R2 values reached 0.6, denoting the predictive ability of spatial regression. Higher R2 in wet season implied that phytoplankton patterns were better explained by environmental factors in this season. In dry season, Shannon-Wiener was associated with total nitrogen (TN). Abundance and Margalef index had close relationship with CODMn, but no significant correlations were identified between Pielou index and environmental factors. In addition, changes of phytoplankton composition were linked with the changes of temperature (Bacillariophyta%), NO3-N (Chlorophyta% and Cyanophyta%), TP (Chlorophyta%), NH4-N (Dinophyta%), TN (Dinophyta%) and CODMn (Dinophyta%). In wet season, DO and chl-a presented close correlation with abundance, and only NH4-N exerted significant correlation with Shannon-Wiener. Similar to the dry season, no significant correlations were identified between Pielou and environmental factors. Margalef index was significantly associated with NH4-N, temperature and CODMn. Moreover, the phytoplankton composition were linked with chl-a (Bacillariophyta% and Dinophyta%), and NH4-N (Chrysophyta%).
4.1 Environmental factors associating with spatial patterns of phytoplankton
Our results revealed that nitrogen, water temperature and CODMn were the three most important environmental indicators accounting for the observed variations in phytoplankton patterns. Nitrogen and nitride compounds had been regarded as the main limiting nutrient to primary production in a variety of phytoplankton , and they significantly contributed to phytoplankton growth in summer and fall . Temperature and CODMn were proved to be critical environmental indictors of phytoplankton occurrence . Higher temperature contributed to the increased richness of species because it provided optimal factors for phytoplankton growth . However, Margalef index decreased obviously, even though temperature increased by twice from dry season to wet season in the study area. Such results implied that the impacts from temperature on species richness should be weak. Nitrogen concentrations were lowered by the increased rain in wet season, and the species richness declined to some extent. All these phenomena accounted for the positive association between the Margalef index and NH4-N in wet season. Moreover, the Margalef index in both seasons, as well as abundance in dry season and Dinophyta% in dry season, were positively correlated with CODMn. It had been reported that CODMn played a key role in regulating phytoplankton in different kinds of natural water systems, such as reservoirs , rivers and lakes , as well as the waters in Taihu basin . However, the previous studies showed that in Taihu Lake TP concentration had a significantly positive correlation with biomass of phytoplankton, especially the cyanobacteria and Microcystis , which was different from the case in our study area. This study illustrated that the influence of environmental factors on phytoplankton in the basin outside the Taihu Lake was different from the situation within the lake. In this study, of all nutrients registered, the concentrations of nitrogenous nutrients (TN in dry season and NH4-N in wet season) could be a possible reason for the general change of phytoplankton metrics observed within the whole study area.
4.2 Seasonal changes of phytoplankton communities
Phytoplankton patterns should vary during the two seasons, as stream flows associated with rainfall exert significant impacts on phytoplankton patterns through dispersion . In our study, phytoplankton metrics showed significant differences between the two seasons except abundance and Pielou evenness index, which meant that phytoplankton biomass and population evenness remained relatively constant. This behavior suggested that the distribution of phytoplankton had no significant changes between the two seasons, and the presence of phytoplankton had less variation in communities among the species. Movement of phytoplankton was accelerated by the activated stream flows on rainfall days, leading to decreased degree of diversity and evenness. Our research indicated that the diversity and evenness of phytoplankton was higher in dry season, which was similar to the case of Sepanggar Bay, Malaysia . Moreover, the environmental factors in different seasons can influence the composition of phytoplankton community. For example, diatoms and green algae can survive in rapidly flowing streams and consequently rely on physical disturbance to remain in suspension ; when the dominant species blooms, we would witness decreased diversity and low species richness . This trend was observed in our study, since the green algae became dominant in wet season.
The assemblage of several species of phytoplankton (Bacillariophyta, Chlorophyta, Cryptophyta, Cyanophyta) across the southern Taihu basin was similar to that in Taihu Lake in an earlier study . Bacillariophyta was often reported to develop in early spring and it could make a significant contribution to the biovolume of the phytoplankton community in rivers . This phenomenon was highlighted in our study that Bacillariophyta was dominant in dry season (Table 2). Rolland et al. (2009) reported that diatoms could live in strong-flowing waters that came from the rivers, while the present study indicated that Bacillariophyta decreased significantly in wet season. Cryptophyta was considered to be fast-growing and opportunistic species , and it benefited from high concentrations of nutrients . Our results showed that Cryptophyta was steady during the two seasons (Table 2), since the influence from the environmental factors was not significant on the Cryptophyta (Table 3). Blue-green algae (Chlorophyta and Cyanophyta) were found to be dominant in Taihu Lake and they also had a large biomass in southern Taihu basin, especially for wet season in which Cyanophyta increased significantly (Table 2). Furthermore, both Chlorophyta and Cyanophyta were influenced by NO3-N in dry season and when it came to wet season no environmental factors affected these two species.
4.3 Methodological aspects
Phytoplankton metrics were useful in quantitatively analyzing the spatiotemporal changes in phytoplankton patterns. Furthermore, the relationships between metrics and environmental factors could be used to test the existing theories and to develop models for environmental research. Therefore, metric analysis offered a useful framework for indirectly indicating the relationships between phytoplankton patterns and environmental factors. Our study also highlighted the importance of incorporating spatial dependence, when identifying the explanatory environmental factors for phytoplankton patterns, since significant spatial dependence in phytoplankton patterns was identified. This spatial dependence could help to understand the real effects of corresponding factors, because incorrect estimations might happen when applying traditional linear regression models, in which important impacts from neighboring sampling sites were ignored (Su et al. 2013). Specifically, the spatial lag model was suitable for all the phytoplankton metrics in both seasons, which suggested that spatial patterns of phytoplankton depended on not only local environmental factors, but also those observed in neighboring sites.
Despite the identification of relationships between phytoplankton patterns and environmental factors from this study, it still incorporated some limitations. First, the data set used in this research covered a very limited temporal dimension. Further studies need to be carried out regarding this aspect. Second, climate, land use, topography, hydrology and other socio-economic indicators were not discussed, which may account for the unexplained effects. Third, five metrics used in our study may not describe all the characteristics of phytoplankton patterns. More different kinds of metrics can be further applied and compared in the future. Fourth, we only identify the relationships at one spatial scale. More detailed comparison should be considered through different spatial scales.
This study characterized the spatial patterns of phytoplankton in relation to environmental factors across southern Taihu basin, China, and we have main conclusions as follows:
Spatial patterns of phytoplankton presented significant autocorrelation. Richness of phytoplankton became more localized in wet season, while evenness distribution of phytoplankton became more regionalized in wet season.
Shannon-Wiener, Margalef and composition of phytoplankton showed significant differences between the two seasons. Wet season showed more patterns of richness at local scale and more patterns of evenness at regional scale. For spatial regression, phytoplankton patterns were better explained by environmental factors in wet season.
Spatial lag model was suitable for all the phytoplankton metrics in both seasons. Nutrients played a major role in determining phytoplankton diversity and composition. CODMn could be used to predict phytoplankton richness in both seasons, as well as abundance in dry season.
Spatial regression is a promising tool for interpreting environmental factors of phytoplankton patterns in rivers. The application of spatial regression in analyzing phytoplankton dynamics can also be used for other phytoplankton research at other spatial scales.
We thank Editor-in-chief George Christakos for handling our manuscript, and two anonymous reviewers for providing useful suggestions. We also thank all the people involved in field work and sampling. This work was supported partly by the National Key Project of China (Grant No. 2011ZX07).