Institute of Ecological Science, Faculty of Biology, Vrije Universiteit, De Boelelaan 1087, 1081 HV Amsterdam, The Netherlands.
Environ Toxicol Pharmacol. 2002 Jul;11(3-4):167-72. doi: 10.1016/s1382-6689(01)00114-4.
Species sensitivity distributions (SSDs) are used in ecological risk assessment to derive maximum acceptable concentrations of toxicants in the environment from a limited set of ecotoxicity data obtained in the laboratory. Such distributions usually employ continuous bell-shaped functions such as the normal and the logistic distribution, which have the disadvantage that an arbitrary cut-off value must be chosen (usually the 5-percentile) to designate the concentration below which the fraction of species exposed above their no-effect level is considered acceptably small. In this paper the possibility is explored of introducing a true no-effect principle in the SSD framework by considering models with a finite lower threshold. Four of these distributions are elaborated, the uniform, triangular, exponential and Weibull distributions. The mathematical representations of these functions were re-parameterized allowing direct estimation of the threshold parameter by nonlinear regression. By way of example, a data set comprising chronic ecotoxicity of zinc to 21 different aquatic organisms was used. The exponential distribution did not describe the data well. The other distributions provided estimates for HC(0) (hazardous concentration for none of the species) between 1.66 and 7.83 μg/l. The triangular distribution fitted best to the data and was consistent with previous models. Since threshold-SSDs incorporate a true no-effect level they may be better communicable as a principle for environmental protection in comparison to the approach based on '95% protection'.
物种敏感度分布（SSD）被用于生态风险评估中，以便根据在实验室中获得的有限的生态毒性数据，从环境中推导岀毒剂的最大可接受浓度。此类分布通常釆用连续钟形函数，如正态分布和逻辑斯谛分布，这些分布的缺点是必须选择任意的截止值（通常是 5%分位数）来指定浓度，低于该浓度，暴露在无影响水平以上的物种比例被认为是可接受的小。在本文中，通过考虑具有有限下限的模型，探讨了在 SSD 框架中引入真正无影响原理的可能性。阐述了其中的四个分布，即均匀分布、三角形分布、指数分布和威布尔分布。这些函数的数学表示形式被重新参数化，允许通过非线性回归直接估计阈值参数。作为示例，使用了包含锌对 21 种不同水生生物慢性毒性的数据集。指数分布不能很好地描述数据。其他分布为 HC(0)（没有一种物种受到危害的浓度）提供了在 1.66 至 7.83μg/l 之间的估计值。三角形分布与数据拟合得最好，并且与以前的模型一致。由于阈值 SSD 包含真正的无影响水平，因此与基于“95%保护”的方法相比，它们作为环境保护的原则可能更容易被传达。
