期刊
FRESHWATER BIOLOGY
卷 60, 期 9, 页码 1901-1916出版社
WILEY
DOI: 10.1111/fwb.12620
关键词
cyanobacteria; nutrients; hierarchical Bayesian models; classification
High densities of cyanobacteria can interfere with the use of lakes and reservoirs for recreation and as sources for drinking water, and one approach for reducing the amount of cyanobacteria is to reduce nutrient concentrations in the waterbody. An approach is described for deriving numeric targets for concentrations of total phosphorus (TP) and total nitrogen (TN) that are associated with a pre-specified probability of cyanobacterial biovolume that exceeds the recommended World Health Organization thresholds for recreation in the water. The analysis consisted of two phases. First, a divisive tree algorithm was used to identify groups of lakes in which the relationship between nutrients and cyanobacterial biovolume was similar. Second, hierarchical Bayesian models were used to estimate relationships between cyanobacterial biovolume, TP and TN, while partitioning the observed variance in biovolume into components associated with sampling variability, temporal variability, and among-lake differences. The final model accounted for 91% of the variance in cyanobacterial biovolume among different lakes and was used to identify nutrient concentrations that maintain a low probability of excessively high cyanobacterial biovolumes. When no classes of lakes were specified and the relationship between cyanobacterial biovolume and nutrient concentrations was modelled using a national data set, mean targets of 87 and 1100gL(-1) were derived for TP and TN, respectively, to maintain cyanobacterial biovolume below moderate risk levels as defined by the World Health Organization. After classification, mean nutrient targets in lakes that were found to be most susceptible to high biovolumes of cyanobacteria (i.e. deep lakes) were 61 and 800gL(-1) for TP and TN, while higher nutrient thresholds were observed for other classes of lakes.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据