POPs in Baltic herring

This page is a knowledge crystal of subtype variable. The page identifier is Op_en2583
Moderator:Nobody (see all) Click here to sign up.

Upload data {{#opasnet_base_link:Op_en2583}}

Question

What are the concentrations of persistent organic pollutants (POPs) in Baltic herring.

Answer

Answer is under work and results are preliminary.

POP concentrations in Baltic sea fish have been measured from samples collected in EU-kalat project. The original data of individual fish samples is accessible through Opasnet base. This data is used here for a Bayesian model to calculate posterior concentration distributions (median and SD) for each congener. This data is then translated into TEQ, and can be used for health benefit assessment of Baltic herring.

Posterior congener median concentrations are presented below for each compound group (PCDD/F, PCB, BDE) analysed in EU-kalat.

Based on the mean posterior concentrations of individual congeners, TEQs are calculated for each congener by using TEF values by WHO and plotted below.

You can print out the numerical results of prior and posterior congener concentrations below. In addition the possible updated versions of the above figures are printed out.

+ Show code - Hide code


library(OpasnetUtils)
library(ggplot2)
library(reshape2)

objects.latest("Op_en2583", "pop_bayes")

oprint(dcast(resultsall@output, Congener~variable, fill=0, value.var = "resultsallResult"))

ggplot(subset(resultsall@output, grepl("PCB", Congener) & grepl("Medianpost", variable)), aes(x = Congener, y = resultsallResult, fill = Congener)) + geom_bar(stat = "identity") + 
  labs(x = "Congener", y = "Concentration (pg/g fat)") + coord_flip() + ggtitle("Median posterior concentrations of PCB's in Baltic herring")

ggplot(subset(resultsall@output, grepl("CD", Congener) & grepl("Medianpost", variable)), aes(x = Congener, y = resultsallResult, fill = Congener)) + geom_bar(stat = "identity") + 
  labs(x = "Congener", y = "Concentration (pg/g fat)") + coord_flip() + ggtitle("Median posterior concentrations of PCDD/F's in Baltic herring")

ggplot(subset(resultsall@output, grepl("BDE", Congener) & grepl("Medianpost", variable)), aes(x = Congener, y = resultsallResult, fill = Congener)) + geom_bar(stat = "identity") + 
  labs(x = "Congener", y = "Concentration (pg/g fat)") + coord_flip() + ggtitle("Median posterior concentrations of BDE's in Baltic herring")

ggplot(subset(teq@output, grepl("Medianpost", variable)), aes(x = Congener, y = Result, fill = Congener)) + geom_bar(stat = "identity") + 
  labs(x = "Congener", y = "TEQ (pg/g fat)") + coord_flip()  + ggtitle("Median posterior TEQ of POPs in Baltic herring")

Rationale

HELCOM has reported ^[1] that in the south western part of the Baltic and in Danish waters the average dioxin content in herring is 2-2.5 pg WHO-TEQ/g fresh weight. In comparison, levels are approximately double this figure in the Baltic Proper and the Gulf of Finland and four times higher in the Bothnian Sea and the southern part of the Bothnian Bay.

Northern Baltic sea

This model takes in measured congener concentrations of POPs in Baltic herring in northern part of Baltic sea (Bothnian Bay, Bothnian Sea, Åland Sea, Gulf of Finland). Measured data is used for Bayesian model to produce posterior medians and sds for each congener and also to calculate TEQ values. Numerical results are saved as variables to Opasnetbase and result figures are presented above in the Answer section.

+ Show code - Hide code

## This code is Op_en2583/pop_bayes on page [[POPs_in_Baltic_herring]]

library(OpasnetUtils)
library(ggplot2)
library(rjags)
library(reshape2)

dat <- opbase.data("Op_en3104", subset = "POPs")
dat <- dat[dat$Fish_species == "Baltic herring" , ]

dat <- subset(dat,!(is.na(dat["Result"])))
dat <- dropall(dat)
levels(dat$POP) <- gsub("HCDD", "HxCDD", levels(dat$POP))
levels(dat$POP) <- gsub("HCDF", "HxCDF", levels(dat$POP))
levels(dat$POP) <- gsub("CoPCB", "PCB", levels(dat$POP))

congeners <- levels(dat$POP) #names of different congeners in data

Y <- length(congeners) #number of congeners and j's in for loop

compdat <- dat[dat$POP %in% congeners[1:Y] , ] #data for current congener
Compound <- log10(compdat$Result+1E-2) #+1E-2 because zero concentrations are not allowed

mo <- textConnection("model{
                     for (j in 1 : Y ) {
                     tau1[j] ~ dunif(0.001, 1000)
                     muOfCompound[j] ~ dnorm(0, 0.001)
                     }
                     for( i in 1 : N ) {
                     Compound[i] ~ dnorm(muOfCompound[POP[i]], tau2[i])
                     tau2[i] <- tau1[POP[i]]*sqrt(n[i])
                     }
                     }
                     ")
dataList = list(
  Y = Y,
  Compound = Compound,
  POP = as.numeric(compdat$POP),
  N = length(Compound),
  n = compdat$N_individuals #number of fishes in sample
)

jags <- jags.model(mo, data = dataList, n.chains = 4, n.adapt = 1000)

close(mo)

#update(jags, 1000)
out <- coda.samples(jags, c('tau1', 'muOfCompound'), 500) # Stores a posterior sample
plot(out)

Meanlogpost = c()
for (j in 1 : Y) {
  logmean <- mean(out[[4]][,j]) #calculate mean of logmu for posterior (test 4)
  Meanlogpost = c(Meanlogpost, logmean)
}

Sdlogpost = c()
for (j in 1 : Y) {
  logsd <- sqrt(1/(mean(out[[4]][,j+Y])))
  Sdlogpost = c(Sdlogpost, logsd)
}

resultsall <- data.frame(
  Meanorig = aggregate(compdat$Result, compdat["POP"], mean), #calculate mean for original data
  Sdorig = aggregate(compdat$Result, compdat["POP"], sd)$x, #calcaulte sd of original data
  Medianorig = aggregate(compdat$Result, compdat["POP"], median)$x, #calculate median of original data
  Meanlog = aggregate(Compound, compdat["POP"], mean)$x, #calculate mean for logdata
  Sdlog = aggregate(Compound, compdat["POP"], sd)$x, #calculate sd for logdata
  Meanlogpost = Meanlogpost,
  Sdlogpost = Sdlogpost,
  Medianpost = 10^Meanlogpost-1E-02,
  Sdpost = 10^Sdlogpost-1E-02 #this might be incorrect
  )

oprint(resultsall)

tef <- Ovariable("tef", ddata = "Op_en4017", subset = "TEF values")
tef <- EvalOutput(tef)

colnames(resultsall)[1] <- "Congener"
resultsall <- melt(resultsall)
colnames(resultsall)[3] <- "Result"
resultsall <-  Ovariable("resultsall", data = resultsall)
resultsall <- EvalOutput(resultsall)
teq = resultsall * tef

objects.store(resultsall, teq)
cat("resultsall and teq stored for later use:\n", paste(ls(), collapse = ", "), "\n")

Southern Baltic sea

Pandelova et al. ^[2] have measured PCDD/Fs and PCB concentrations in Baltic herring taking samples from Gulf of Finland, Gulf of Riga and open Baltic sea. Results as TEQs (pg WHO-TEQ/g fw) are presented separately for herrings with different age.

PCDD/Fs(pg WHO-TEQ/g fw)
Obs	Area	Time	Age	Result
1	Gulf of Riga	2002	2.0–2.5	1.08
2	Gulf of Riga	2004	2.0–2.5	2.14
3	Gulf of Riga	2005	2.0–2.5	1.75
4	Gulf of Riga	2003	3.0–3.5	1.5
5	Gulf of Riga	2002	3.5–4.0	1.72
6	Gulf of Riga	2003	3.5–4.0	1.44
7	Gulf of Finland	2002	1.5–2.0	0.63
8	Gulf of Finland	2002	2.0–2.5	0.75
9	Gulf of Finland	2004	2.0–2.5	1.07
10	Gulf of Finland	2005	2.0–2.5	0.98
11	Gulf of Finland	2002	2.5–3.0	1.08
12	Gulf of Finland	2005	2.5–3.0	0.98
13	Gulf of Finland	2002	4.5–5.0	1.43
14	Gulf of Finland	2003	4.5–5.0	3.07
15	Open Baltic Sea	2004	1.5–2.0	1.25
16	Open Baltic Sea	2004	2.0–2.5	1.14
17	Open Baltic Sea	2002	3.5–4.0	2.18

PCBs(pg WHO-TEQ/g fw)
Obs	Area	Time	Age	Result
1	Gulf of Riga	2004	1.5–2.0	1.31
2	Gulf of Riga	2005	2.0–2.5	0.83
3	Gulf of Riga	2003	3.0–3.5	1.34
4	Gulf of Riga	2003	3.5–4.0	1.47
5	Gulf of Finland	2004	2.0–2.5	0.67
6	Gulf of Finland	2005	2.0–2.5	0.68
7	Gulf of Finland	2004	2.5–3.0	1.03
8	Gulf of Finland	2005	2.5–3.0	0.72
9	Gulf of Finland	2003	4.5–5.0	2.48
10	Open Baltic Sea	2004	1.5–2.0	1.21
11	Open Baltic Sea	2004	2.0–2.5	1.07
12	Open Baltic Sea	2003	4.0–4.5	2.09

References

↑ HELCOM 2004. Dioxins in the Baltic Sea [1]
↑ Pandelova et al. 2008. Levels of PCDD/F and dioxin-like PCB in Baltic fish of different age and gender, Chemosphere, 71, 369–378.

[1] HELCOM 2004. Dioxins in the Baltic Sea [1]

[2] Pandelova et al. 2008. Levels of PCDD/F and dioxin-like PCB in Baltic fish of different age and gender, Chemosphere, 71, 369–378.

[1]

[2]