ENVIRONMENTAL

Selenastrum capricornutum EC50 72h

 

Endpoint

 

The acute aquatic toxicity model for algae predicts the concentration of a substance that inhibits 50% of the growth (EC₅₀) of the test alga Selenastrum capricornutum within a designated period.

 

 

Data

 

The training set consists of EC₅₀ values for 278 chemicals [1]:

 

• Test duration - 72 hours,
• Test species - Selenastrum capricornutum (unicellular green alga).

 

The tested chemicals belong to the following categories:

 

Baseline narcotics - 100 chemicals,
Narcotic amines - 14 chemicals,
Anilines - 20 chemicals,
Acrylates - 5 chemicals,
Reactive unspecified - 90 chemicals.

•  

 

Model

 

The acute aquatic toxicity model for algae is based on the existing profile for acute aquatic toxicity OASIS (AAT OASIS) developed in LMC. New rules associated with algae specification were added and used to classify chemicals according their mode of action. In the contrast of AAT OASIS profile, chemicals acting as basesurface narcotics, phenols, esters and methacrylates, are grouped in one category - Baseline narcotics. The following models were developed based on regression analysis of the data:

 

Narcotic toxicants

log 1/EC50 = 1.8(±0.2)+ 0.81(±0.04)logKow

 

Narcotic amines

log 1/EC50 = 3.4(±0.1)+ 0.34(±0.1)logKow

 

Acrylates

log 1/EC50 = 4.3(±0.03)+ 0.26(±0.02)logKow

 

Anilines

log 1/EC50 = 3.3(±0.3)+ 0.37(±0.1)logKow

 

 

Domain

 

The stepwise approach [2] was used to define the applicability domain of the model.  It consists of the following sub-domain levels:

 

• General parametric requirements - includes ranges of variation of log K_OW and MW,
• Structural domain - based on atom-centered fragments (ACFs).

 

A chemical is considered In Domain if its log K_OW and MW are within the specified ranges and its ACFs are presented in the training chemicals.  The information implemented in the applicability domain is extracted from the correctly predicted training chemicals used to build the model and in this respect the applicability domain determines practically the interpolation space of the model.

 

 

Statistics

 

The precision of the regression models is characterized by the following estimates: the 95% confidence intervals of model parameters, coefficient of determination (R²), mean squared error (estimate of error variance, s²), and F value:

 

Narcotic toxicants

 

• Coefficient of determination R² = 0.78,
• Mean squared error s² = 0.22,
• F value = 338,
• Number of chemicals, n = 100.

 

Narcotic amines

 

• Coefficient of determination R² = 0.64,
• Mean squared error s² = 0.14,
• F value = 21,
• Number of chemicals, n = 14.

 

Acrylates

 

• Coefficient of determination R² = 0.99,
• Mean squared error s² = 0.002,
• F value = 206,
• Number of chemicals, n = 4.

 

Anilines

 

• Coefficient of determination R² = 0.32,
• Mean squared error s² = 0.22,
• F value = 8.4,
• Number of chemicals, n = 20.

 

 

References

 

1. National Institute of Technology and Evaluation. CHRIP (Chemical Risk Information Platform) databse: http://www.safe.nite.go.jp/english/db.html (last visited on January 21, 2013)

2. S. Dimitrov, G. Dimitrova, T. Pavlov, D. Dimitrova, G. Patlevisz, J. Niemela and O. Mekenyan, J. Chem. Inf. Model. 45 (2005)  839-849.