Concentrations of beneficial nutrients in fish

This page is a variable. The page identifier is Op_en1838
Moderator:Olli (see all)
Give your opinion to the peer rating of the content of this page. {{ #opasnet_rater: }} Citation of this page: Anna Karjalainen, Jouni T. Tuomisto, Patrycja Gradowska: Concentrations of beneficial nutrients in fish. Opasnet 2009. [2]. Accessed 19 Apr 2024.
Upload data Show results

Scope

What are the concentrations of beneficial nutrients in fish (such as omega-3 fatty acids (EPA + DHA + fatty acid 18:3 n-3)) in fish.R↻ Concentrations of sea and freshwater fish species are studied separately.

Definition

Data

⇤# : Omega-3_content_in_salmon check this --Arja (talk) 07:11, 21 September 2016 (UTC)

Concentrations of beneficial nutrients in fish extracted from ^[1], where they are indicated as mass units/100 g fish analysed.

Concentrations of omega-3 fatty acids in fish extracted from ^[1], where they are indicated as mass units/100 g fish analysed. Criteria for choosing the fish preparation method used prior the analyses:

one that is presumed as a common one used by the Finnish consumers (roasting), and
also as available for most of the fish in the database.

The data for omega-3 fatty acids was reported as mean values of the fish species^[2]. R↻

Omega-3 fatty acids, mg/g, as sum of n-3 PUFA (polyunsaturated fatty acids) precursors ALA (alphalinoleic acid), EPA (eicosapanthenic acid), and DHA (docosahexaenoic acid). DPA (docosapentaenoic acid) not available in the database.
Omega-3 fatty acids, mg/g, as DHA and EPA only.
- Vitamin D, ug/g
- Vitamin E, mg/g
- Selenium, ug/g
- Iodine, ug/g
- Group B vitamins, mg/g, as sum of B1 (tiamin), B2 (riboflavin), B6 (pyridoxin), B12 (cyanokobalamin) and niacin as its eqvivalents NE. R↻

The most commonly consumed fish species are described: Farmed salmon, wild salmon, herring, white fish, sprat, perch, flounder, pike-perch, bream, pike, vendace and burbot. Concentration data for imported fish in Finland is not available in many cases. R↻

Data 1

mass units/100 g fish analysed

Fish species	Omega-3-PUFA contents, mg/g	Vitamin D, ug/g	Vitamin E, mg/g	Selenium, ug/g	Iodine, ug/g	Group B vitamins, mg/g,
Baltic herring (Clupea harengus membras)	12.13	0.17	0.03	0.18	0.45	0.17
Herring (Clupea harengus)	12.13	0.17	0.03	0.18	0.45	0.17
Vendace (Coregonus albula)	12.77	0.13	0.03	0.22	0.76	0.00
Whitefish (Coregonus lavaretus)	10.71
Pike (Esox lucius)	6.75	0.03	0.02	0.22	0.34	0.08
Rainbow trout (Onchorhynchus mykiss)	20.52
Shrimp (Pandalus sp.)	2.49
Perch (Perca fluviatilis)	7.17
Saithe (Pollachius virens)	3.87
Atlantic salmon (Salmo salar)	17.42
Pike-perch (Sander lucioperca)	4.06
Tuna (Thunnus thynnus)	32.67

Data 2

There are no separate data for farmed and for wild salmon. Therefore they are assumed to be the same.
There are no data for vendace(sea), so assumption is that the parameters are the same as vendace(inland)
There are no data for bream(sea), so assumption is that the parameters are the same as bream(inland)
There are no data for wild salmon(inland), so assumption is that the parameters are the same as wildsalmon(sea)

First, the mercury data was used to form lognormal distributions for each species with parameters median and geometric standard deviation. Secondly, the proportion of methylmercury from total mercury was taken into account. Methylmercury proportion is assumed to follow triangular distribution (author judgement) with parameters (min=0.81, mode=0.93,max=0,98). R↻

The omega-3 distributions chosen by author judgement. Concentration distributions were formed according to amount of data. a) If 3 or more data points → triangular distribution, b) if 2 data points → uniform distribution, c) if 1 data point → normal distribution R↻

Sprat: omega-3 concentration(sprat) = omega-3 concentration(herring) closely related species, assumed to have similar composition

Units: mg/kg in fresh weight

Fish species	Omega-3 concentration
Farmed salmon (sea+inland)	1.553e+004
Wild salmon	1.516e+004
Herring(sea)	9790
White fish(sea)	7830
Sprat(sea)	9790
Perch(sea)	3030
Flounder(sea)	4220
Pike-perch(sea)	4060
Bream(sea)	4930
Pike(sea)	2920
Vendace(sea)	7510
Burbot(sea)	2010
Wild salmon(inland)	1.504e+004
White fish(inland)	7830
Perch(inland)	3030
Pike-perch(inland)	4060
Bream(inland)	4930
Pike(inland)	2920
Vendace(inland)	7510
Burbot(inland)	2010

Data 3

Since there are only mean concentration values available exponential distribution has been chosen to model the concentration of DHA and concentration of EPA in various sea and inland fish species.

Moreover, the Fineli database^[1] provides concentrations of nutrients in species caught mostly in the Baltic Sea (except vendace which was caught in inland waters). Thus, the concentration data for some of the species chosen is not available. Therefore, it has been assumed that the DHA (EPA) concentration in sea and freshwater species is the same.

Units: mg/g fish

Fish species	Mean DHA concentration	Mean EPA concentration
Baltic Herring	4.82	2.27
Vendace(inland)	4.09	2.06
Vendace(sea)	4.09	2.06
Whitefish(inland)	4.93	2.63
Whitefish(sea)	4.93	2.63
Pike(inland)	2.17	0.44
Pike(sea)	2.17	0.44
Perch(inland)	2.45	0.54
Perch(sea)	2.45	0.54
Atlantic Salmon	10.02	3.45
Pike-perch(inland)	3.01	0.92
Pike-perch(sea)	3.01	0.92

Dependencies

List of parents:

None

Unit

mg/g fish

Formula

Previous Analytica formulas can be found from a previous version of this page.

Bayes calculations

Model run 2.3.2017 [3]
Model run 3.3.2017 with multivariate normal and predictions [4]
Model run 19.5.2017 with only vit.param stored [5]

+ Show code - Hide code

# This is code Op_en1838/bayes on page [[Concentrations of beneficial nutrients in fish]]

library(OpasnetUtils)
library(rjags)
library(reshape2)
library(ggplot2)
library(car) # For scatterplotMatrix()

dat <- opbase.data("Op_en1838", subset = "Balticherring")
dat$Result <- as.numeric(gsub(",", ".", as.character(dat$Result)))
colnames(dat)

fisl <- sort(unique(as.character(dat$Fish_species)))
compl <- sort(unique(as.character(dat$variable)))
dat.wide <- reshape(
  dat, 
  idvar = "Sample_code", 
  v.names = "Result", 
  timevar = "variable", 
  direction = "wide"
)

mod <- textConnection("
  model{
    for(i in 1:N) { # Observation
      conc[i,1:C] ~ dmnorm(mu[], Omega[,])
    }
    mu[1:C] ~ dmnorm(mu0[1:C], S2[1:C,1:C])
    Omega[1:C,1:C] ~ dwish(S3[1:C,1:C],4)
    conc.pred[1:C] ~ dmnorm(mu[1:C], Omega[1:C,1:C])
  }
")

jags <- jags.model(
  mod,
  data = list(
    N = nrow(dat.wide),
    C = length(compl),
    conc = dat.wide[ncol(dat.wide)+(-3:0)],
    mu0 = rep(100,4),
    S2 = diag(4)/100000,
    S3 = diag(4)/10000
  ),
  n.chains = 4,
  n.adapt = 100
)

update(jags, 100)

samps.j <- jags.samples(
  jags, 
  c('mu', 'Omega', 'conc.pred'), 
  1000
)
js <- array(
  c(
    samps.j$mu[,,1],
    samps.j$Omega[,1,,1],
    samps.j$Omega[,2,,1],
    samps.j$Omega[,3,,1],
    samps.j$Omega[,4,,1],
    samps.j$conc.pred[,,1]
  ),
  dim = c(4,1000,6),
  dimnames = list(
    Exposure_agent = compl,
    Iter = 1:1000,
    Parameter = c("mu","Omega1", "Omega2", "Omega3", "Omega4", "conc.pred")
  )
)

# Mu for all questions
scatterplotMatrix(t(js[,,1]))
# All parameters for Exposure_agent 1 (DHA?)
scatterplotMatrix(js[1,,])

jsd <- melt(js)

coda.j <- coda.samples(
  jags, 
  c('mu', 'Omega', 'conc.pred'), 
  1000
)

plot(coda.j)

######## vit.param contains expected values of the distribution parameters from the model

vit.param <- list(
  mu = apply(js[,,1], MARGIN = 1, FUN = mean),
  Omega = solve(apply(js[,,2:5], MARGIN = c(1,3), FUN = mean))
    # solve matrix: precision->covariace
)

objects.store(vit.param)
cat("List vit.param stored.\n")

if(FALSE) {
concddeo.mu <- array(
  samps$mu,
  dim = c(length(compl), 1000, 4),
  dimnames = list(Compound = compl, Iter = 1:1000, Seed = 1:4)
)

concddeo.Omega <- array(
  samps$Omega,
  dim = c(length(compl), length(compl), 1000, 4),
  dimnames = list(Compound1 = compl, Compound2 = compl, Iter = 1:1000, Seed = 1:4)
)

concddeo.pred <- array(
  samps$mu,
  dim = c(length(compl), 1000, 4),
  dimnames = list(Compound = compl, Iter = 1:1000, Seed = 1:4)
)

concdf.mu <- melt(concddeo.mu)

scatterplotMatrix(t(concddeo.mu[,,1]))

concdf.pred <- melt(concddeo.pred)

scatterplotMatrix(t(concddeo.pred[,,1]))

ggplot(dat, aes(x = Result, colour = Fish_species))+geom_density()+
  facet_wrap(~ variable, scales = "free")

ggplot(concdf.pred, aes(x = value))+geom_density()+
  facet_wrap(~ Compound, scales = "free")

objects.store(concddeo.mu, concddeo.Omega, concddeo.pred)
cat("Arrays of concentrations of vitamin D, DHA, EPA, and Omega3) stored:\n")
cat("concddeo.mu (mean), concddeo.Omega (precision matrix), concddeo.pred (predicted distribution).\n")

} #if(FALSE)

Initiate ovariable conc_vit

Initiation 19.5.2017 [6]

+ Show code - Hide code

# This is code Op_en1838/initiate on page [[Concentrations of beneficial nutrients in fish]]

library(OpasnetUtils)

conc_vit <- Ovariable(
  "conc_vit",
  dependencies = data.frame(Name = "vit.param", Ident = "Op_en1838/bayes"),
  formula = function(...) {
    require(MASS)
    require(reshape2)
    vit <- mvrnorm(openv$N, vit.param$mu, vit.param$Omega)
    
    vit <- rbind(
      cbind(
        Fish = "Salmon",
        Iter = 1:nrow(vit),
        as.data.frame(vit) # Salmon results should be taken from Fineli
      ),
      cbind(
        Fish = "Herring",
        Iter = 1:nrow(vit),
        as.data.frame(vit)
      )
    )
    vit <- melt(vit, id.vars = c("Iter", "Fish"), variable.name = "Exposure_agent", value.name = "Result")
    vit <- Ovariable(output=vit, marginal = colnames(vit) %in% c("Iter", "Fish", "Exposure_agent"))
    return(vit)
  }
)

objects.store(conc_vit)
cat("Ovariable conc_vit stored.\n")

Result

Note! The first version of results only uses Data 3, for practical reasons (other nutrients than omega-3 are not yet needed).

Show results

Error creating thumbnail: Unable to save thumbnail to destination

References

↑ ^1.0 ^1.1 ^1.2 the Fineli database [1]
↑ Omega-3 fatty acids: database of fineli (www.ktlwww.ktl.fi/fineli)

[fineli-1] 1.0 ^1.1 ^1.2 the Fineli database [1]

[2] Omega-3 fatty acids: database of fineli (www.ktlwww.ktl.fi/fineli)

[1]

[2]

Concentrations of beneficial nutrients in fish

Contents

Scope

Definition

Data

Dependencies

Unit

Formula

Bayes calculations

Result

References

Navigation menu

Personal tools

Namespaces

Variants

Views

More

Search

Navigation

Page Tools

Tools

In other websites