Difference between revisions of "Concentrations of PCBs in fish"

From Testiwiki
Jump to: navigation, search
m (fish category added)
Line 4: Line 4:
 
== Scope  ==
 
== Scope  ==
  
'''Concentrations of PCBs in fish '''describes concentrations of polychlorinated biphenyls (PCBs) in fish. Concentrations of marine and freshwater fish species are studied separately.  
+
'''Concentrations of PCBs in fish '''describes average concentrations of polychlorinated biphenyls (PCBs) in fish. Concentrations of marine and freshwater fish species are studied separately.  
  
 
== Definition  ==
 
== Definition  ==
Line 10: Line 10:
 
=== Data ===
 
=== Data ===
  
Probability distributions of total PCB concentration (sum of 37 PCB congeners) in various fish species were extracted from <ref>http://www.pyrkilo.fi/beneris/index.php/Image:Analytical_data_of_foods-Finland_for_BENERIS_hk_ak.xls</ref> where concentrations of 37 PCB congeners found in marine and freshwater fish samples were collected. Three probability distributions (Weibull, lognormal and gamma) were fitted to the total PCB concentration data. In several cases the number of data points was too small to give reliable fit results. Therefore, the best fitting distribution was chosen based on the statistical data analysis and goodness-of-fit evaluation performed on a largest set of data points (BS Herring, 47 samples) and then fitted to data of the rest of fish species. It turned out that gamma is the best fitting distribution.
+
Probability distributions of average PCB concentration (dioxin-like PCB congeners) in various fish species were extracted from <ref>http://www.pyrkilo.fi/beneris/index.php/Image:Analytical_data_of_foods-Finland_for_BENERIS_hk_ak.xls</ref> where concentrations of 37 PCB congeners found in marine and freshwater fish samples were collected. Average PCB concentration was estimated by the arithmetic mean. Because of small number of measurements per species the non-parametric bootstrap procedure was used to create the probability distribution of the arithmetic mean.
 +
For fish species with n concentration measurements the procedure employed can be summarized as follows:
  
 +
# Draw m samples of size n with replacement from the original sample. These samples are called bootstrap samples.
 +
# Calculate the sample mean for each of the bootstrap samples.
 +
# Make the density histogram of bootstrap means. 
 +
# Find a parametric probability density function that matches the histogram best.
 +
 +
As a result a set of parametric distributions was obtained that characterize the uncertainty in the average concentration of PCBs in various fish species.
 +
The parameters of these distributions are summarized below in results part.   
 +
 +
The number of bootstrap replicates m was chosen to be n^n or 2000 (if the numerical evaluation of all n^n samples was impossible).
 
<br>  
 
<br>  
  
Line 18: Line 28:
 
=== Unit  ===
 
=== Unit  ===
  
ng/g fw<br>  
+
pg/g fw WHO-TEQ <br>  
  
 
=== Formula  ===
 
=== Formula  ===
Line 26: Line 36:
 
== Result  ==
 
== Result  ==
  
<br>
+
Distributions used:
 +
Normal(mean,standard deviation)
 +
Beta(alfa,beta)
 +
Gamma(alfa,beta)
 +
Lognormal(mu,sigma)
  
 
{| border="1"
 
{| border="1"
 
|-
 
|-
 
| '''Fish species'''  
 
| '''Fish species'''  
| '''Mean total PCB concnetration'''
+
| '''Distribution(parameters)'''
 
|-
 
|-
 
| Baltic Herring  
 
| Baltic Herring  
| 52.503
+
| Normal(2.8797,0.2818)
 
|-
 
|-
 
| Vendace(inland)  
 
| Vendace(inland)  
| 9.677
+
| Normal(0.8767,0.1559)
 
|-
 
|-
 
| Vendace(sea)  
 
| Vendace(sea)  
| 8.658
+
| Normal(0.4468,0.0816)
 
|-
 
|-
 
| Whitefish(inland)  
 
| Whitefish(inland)  
| 10.204
+
| Normal(0.9605,0.271)
 
|-
 
|-
 
| Whitefish(sea)  
 
| Whitefish(sea)  
| 37.74
+
| Lognormal(0.7341,0.097)
 
|-
 
|-
 
| Pike(inland)  
 
| Pike(inland)  
| 6.724
+
| Gamma(0.1081,5.1975)
 
|-
 
|-
 
| Pike(sea)  
 
| Pike(sea)  
| 17.69
+
| Normal(1.0544,0.1604)
 
|-
 
|-
 
| Perch(inland)  
 
| Perch(inland)  
| 5.439
+
| Beta(7.9494,8.298)
 
|-
 
|-
 
| Perch(sea)  
 
| Perch(sea)  
| 39.143
+
| Normal(2.3615,0.3956)
 
|-
 
|-
 
| Atlantic Salmon  
 
| Atlantic Salmon  
| 194.386
+
| Normal(11.2524,1.515)
 
|-
 
|-
 
| Pike-perch(inland)  
 
| Pike-perch(inland)  
| 10.986
+
| Normal(0.6929,0.1551)
 
|-
 
|-
 
| Pike-perch(sea)  
 
| Pike-perch(sea)  
| 27.8
+
| Normal(1.4828,0.2169)
 
|}
 
|}
  

Revision as of 13:39, 26 May 2009



Scope

Concentrations of PCBs in fish describes average concentrations of polychlorinated biphenyls (PCBs) in fish. Concentrations of marine and freshwater fish species are studied separately.

Definition

Data

Probability distributions of average PCB concentration (dioxin-like PCB congeners) in various fish species were extracted from [1] where concentrations of 37 PCB congeners found in marine and freshwater fish samples were collected. Average PCB concentration was estimated by the arithmetic mean. Because of small number of measurements per species the non-parametric bootstrap procedure was used to create the probability distribution of the arithmetic mean. For fish species with n concentration measurements the procedure employed can be summarized as follows:

  1. Draw m samples of size n with replacement from the original sample. These samples are called bootstrap samples.
  2. Calculate the sample mean for each of the bootstrap samples.
  3. Make the density histogram of bootstrap means.
  4. Find a parametric probability density function that matches the histogram best.

As a result a set of parametric distributions was obtained that characterize the uncertainty in the average concentration of PCBs in various fish species. The parameters of these distributions are summarized below in results part.

The number of bootstrap replicates m was chosen to be n^n or 2000 (if the numerical evaluation of all n^n samples was impossible).

Causality

Unit

pg/g fw WHO-TEQ

Formula


Result

Distributions used: Normal(mean,standard deviation) Beta(alfa,beta) Gamma(alfa,beta) Lognormal(mu,sigma)

Fish species Distribution(parameters)
Baltic Herring Normal(2.8797,0.2818)
Vendace(inland) Normal(0.8767,0.1559)
Vendace(sea) Normal(0.4468,0.0816)
Whitefish(inland) Normal(0.9605,0.271)
Whitefish(sea) Lognormal(0.7341,0.097)
Pike(inland) Gamma(0.1081,5.1975)
Pike(sea) Normal(1.0544,0.1604)
Perch(inland) Beta(7.9494,8.298)
Perch(sea) Normal(2.3615,0.3956)
Atlantic Salmon Normal(11.2524,1.515)
Pike-perch(inland) Normal(0.6929,0.1551)
Pike-perch(sea) Normal(1.4828,0.2169)

See also


References