Charles Santerre posted a message on the Listserve on 4 September
reproducing and article printed in the New York Post of 29 August by Dr.
Elizabeth Whelan. Charles has previously, 20 March 2002, drawn attention in
the Listserve to an article about PCBs (polychlorinated biphenyls) in farmed
salmon. I pointed to the possible hazards of polychlorinated hydrocarbons in
farmed fish in a review published in 1998 (International Journal of Food
Science and Technology, 33, 99-125) so have some interest in the topic. I
wrote to Charles at that time but did not post my comments on the Listserve.
A little late following up the latest posting, but I would like to comment
on the subjects PCBs in fish generally and salmon in particular.
Peter Howgate
(note new e-mail address)
********************************
Not being an expert in the subject of highly chlorinated hydrocarbons (HCHs)
in fish and their possible hazards I find some difficulty in evaluating the
literature on PCBs in fish. PCBs are synthetic chemicals not found in nature
by natural processes other than in trace amounts made by chlorinating
biphenyl. The are some 200 possible combinations, (congeners), of number and
positions of chlorine atoms in the molecule, but those of concern have a
structure similar to that of dioxins. They are referred to as dioxin-like
PCBs and there are 11-14 of them depending on which publication you are
consulting. These dioxin-like PCBs are not all equally toxic in animal
studies and individual congeners are given weighting factors, TEFs, Toxic
Equivalency Factors, according to their potencies relative to the most toxic
of the dioxin congeners. The toxicity of mixtures can then be compared by
summing the concentrations of the individual dioxin-like PCBs times their
TEF to give the Toxic Equivalent (TEQ). Many sites on the Internet explain
this concept more fully and provide details.
The Whelan piece refers to a report of the 'Environmental Working Group'.
Its Internet site describes itself as: 'The Environmental Working Group
(EWG) is a not_for_profit environmental research organization dedicated to
improving public health and protecting the environment by reducing pollution
in air, water and food.'. Its report on PCBs in farmed salmon is available
at: www.ewg.org/reports/farmedPCBs. The summary page refers to 10 samples of
farmed salmon purchased at grocery stores in Washington DC, San Francisco,
and Portland, Oregon, though the table of data shows analyses for 11
samples. The survey did not apparently include any samples of wild salmon.
The second paragraph of the summary claims that: 'On average the farmed
salmon have 16 times the dioxin-like PCBs found in wild salmon, .... '.
However, another page in the report claims that 'Farmed salmon contains five
to 10 times the PCBs of wild salmon.', and that 'The average level of total
PCBs in EWG's ten farmed salmon samples was 27.3 ppb, or 5.2 times higher
than the average PCB level of 5.3 ppb in four wild salmon tested by Canadian
scientists (Easton et al. 2002).'. There is also a claim that 'On average,
farmed salmon from EWG's supermarket study contained ... at least 3.4 times
the PCB level of other commercial seafood.'. Let me examine these claims.
The EWS site shows the sources of the 11 salmon: 5 from Canada, 2 from
Chile, 1 each from Iceland, USA and Scotland, and 1 unknown. A table shows
the concentrations of 155 congeners or groups of 2-5 congeners, (presumable
not in resolved in the analysis). The concentrations or mean concentrations
in picog/g for the separate sources are:
Canada 25 940
Chile 7 755
USA 25 100
Iceland 32 600
Scotland 67 800
Unknown 25 300
All 26 910
The Chilean, Scottish and Icelandic are almost certain to be Atlantic
salmon, the Canadian and USA to be a species of Pacific salmon, but the
species are not given in the report. Easton et al (2002), referred to in the
EWS report analysed 4 farmed salmon from British Columbia, Canada, 1
Atlantic and 3 Chinook, and 4 wild salmon, 1 Chinook and 1 Chum from Alaska
and 2 Sockeye from British Columbia. They list in the paper concentrations
for 112 congeners or groups of congeners, fewer than listed in the EWG
report. The mean total concentration, picog/g are:
Farmed 51 216
Wild 5 302
Hence the EWG claim that PCB concentrations in farmed salmon are 10 times
those in wild. But note that the overall mean concentration in the EWG
sample of salmon is about 5 times that in the Easton et al sample of wild,
and that for more congeners analysed in the EWG survey. The mean
concentration in Chilean salmon is only one and half times that in the
Easton wild sample. The bald claim in the EWG report that 'Farmed salmon
contain five to 10 times the PCBs of wild salmon' is not true as a general
statement of the situation.
Of more relevance though for risks to human health are the concentrations of
the dioxin-like congeners. I have referred above differences in toxic
potencies of the various congeners and the use of TEFs as weighting factors,
but there is another consideration in the way that TEQ are reported. The
values of concentrations above are in picog/g wet weight, but in the
literature, and in regulations or advisories, TEQs are often, and typically,
expressed as picog/g lipid. This might not be all that helpful to a
nutritionist or toxicologist wanting to know how much PCBs are ingested in a
serving, and derives from the practice of in environmental studies. PCBs,
and other HCHs, strongly partition into the lipid phase of organisms and
when studying bioaccumulation through food chains it is more useful to
environmentalists to compare concentrations in lipid phases. Concentration
as wet weight and in lipid can be interconverted when the lipid
concentration of the sample is known, or assumed.
The EWG report lists the TEQs of the dioxin-like PCBs on a wet weight basis,
and also the lipid contents so that the TEQs on a lipid basis can be
calculated. The summary of mean values on a lipid basis, picog TEQ/g lipid,
for separate sources are:
Canada 4.03
Chile 3.10
USA 4.24
Iceland 4.92
Scotland 9.20
Unknown 4.55
All 4.48
The corresponding values for the 2 sets of samples in the Easton survey are:
farmed 6.80
wild 0.63
(If anyone looks at the Easton et al (2002) summary results, Table 5, they
should note that though congeners 170 & 180 are listed and included in the
total values based on wet wt., they are not included in the Total mammalian
TEQ values based on lipid weight. The Table shows TEFs for these congeners,
but they are the 1994 WHO recommendations and these congeners are not listed
in the most recent WHO, 1998, recommendations. It would have been less
confusing if TEFs for these congeners had been left blank in the table).
On this basis the farmed fish in the EWG survey has about 7 times the
toxicity of the wild fish in the Easton et al survey, but two thirds the
toxicity of the farmed. However, 4 farmed salmon from one location in Canada
is a very small, and unrepresentative, sample on which to base a broad
statement that farmed salmon contain more PCBs, on a mass basis or as TEQ,
than wild salmon. A recent paper, (Krümmel et al, 2003), declares that adult
sockeye salmon returning to spawn in lakes in Alaska contained 2 500 nanog/g
lipid. Assuming a lipid content of 6.5%, the mean value of wild salmon in
Easton et al., this equates to 162 500 pg/g wet wt.. This is to be compared
with the mean for the wild salmon in Easton et al. of 5 302 pg/g wet wt..
and 51 216 pg/g for the farmed salmon.
The EWG report has a figure with 4 histograms comparing PCBs, picog TEQ/g
wet wt., in farmed and wild salmon for 4 countries - Ireland, Canada, USA &
Scotland. Sources of data are cited in the legend to the figure. One is Axys
(2003) and appears to be the data in the EWG report; the height of the bar
for the farmed is consistent with the EWG results. The bar for the wild
alongside that for the farmed must come from the Easton report, which is
cited, because the EWG survey did not include any wild salmon The Canadian
results I assume are from the Easton et al (2002) reference, though the
conversion from TEQ on a lipid basis to wet wt. basis is incorrect; it is
shown in the figure as 0.63 whereas I calculate it as 1.01, (6.80 on a lipid
basis and 14.8% lipid). The legend cites a Canadian Fish Inspection Agency
report, but this does not have any data on PCBs in salmon. The source of the
data for the Irish salmon is not cited, but is a report by the Food
Standards Agency of Ireland (2002). The values, picog TEQ/g wet wt., in that
report,e based on 15 fish in each category are: farmed 3.2, wild 0.72.
(These equate to 27 and 6 picog/g on a lipid basis using typical values for
lipid contents of farmed and wild salmon in Ireland). The legend to the EWG
figure cites 2 papers for the Scottish data, Jacobs, et al., 2002a, & 2002b.
The values from the histograms are 0.69 for farmed and 0.29 for the wild,
but the results in these papers require discussion.
The Jacobs 2002a paper is the one that was alluded to in a TV program last
year that led to the exchange of messages on this Listserve last year. Seven
dioxin-like PCBs were measured in 8 samples of farmed salmon, Salmo salar,
and 2 samples of wild salmon, and the mean TEQs on a lipid basis were:
farmed, 9.55, wild, 9.33, and on a wet basis 1.61 and 1.62 respectively. The
differences are not statistically significant. The data displayed in the EWG
figure are not from this paper. The data in Jacobs et al., 2002b, are more
complicated. The authors analysed 6 farmed salmon and 2 wild salmon sampled
in Scotland and 5 salmon products sampled in Belgium. Fewer PCB congeners
were analysed for in the case of the Belgian samples and none of the
non-ortho congeners. (The dioxin-like congeners are grouped in 2 classes,
the mono-ortho and the non-ortho isomers; the more toxic congeners are in
the latter group). The TEQs of the Belgian samples then do not include the
non-ortho congeners. The table of results in the paper lists the results for
the congeners, but do not include values for any of the non-ortho congeners.
However, the summary results quotes TEQs for both the mono-ortho non-ortho
congeners, and their sums. On a wet wt basis the mean values for total TEQs
are: farmed 2.47, wild 1.66. There is a quite high variance among samples
and the difference between farmed and wild means is not significantly
different. The values for Scottish salmon in the EWG figure are not these.
What they are are the TEQ values for the smaller set of mono-ortho
congeners. The EWG report is misleading because it is not comparing like
with like.
The UK Food Standards Agency issued, (FSA, 1999), the results of a survey of
PCBs and dioxins in UK-landed and imported fish. This included 12 salmon, 11
domestic, 1 imported. The mean TEQ for PCBs was 18.4 picog/g lipid. The TEQs
on a lipid basis for Jacobs et al, 2002a, were 16.3 picog/g and for the
2002b paper, 17.6, consistent with the results of the FSA survey. The FSA
survey also measured PCBs in 6 marine commercial species of fish caught in
the wild, and in farmed trout. The individual results showed a very wide
range of PCB TEQs, 2.2-110 picog/g lipid, and all of the salmon samples were
well within this range.
So, where does all this leave us on this question of PCBs in salmon, and any
difference between farmed and wild varieties. In the case of salmon sampled
in the USA the comparison between contents in farmed and wild rests on a
comparison with a small number, 4, of wild salmon sampled in Canada.
Similarly, any comparison in Canada is based on this small sample of wild
fish. What is striking about the Canadian data in Easton et al (2002) is the
very low concentrations in the wild samples. One must question whether these
4 salmon are a representative sample of wild salmon consumed in North
America. Results for salmon sampled in Scotland show no difference between
PCB TEQs between farmed and wild salmon, again on a small data base for the
wild variety. There is certainly a marked difference in PCB TEQs between
farmed and wild salmon sampled in Ireland, but the values for farmed salmon
are similar to those sampled in Scotland and within the range found in
commercial wild marine fish.
Something that should be noted is that concentrations of dioxins and PCBs in
the environment and in foods in decreasing with the advent on restrictions
or bans on their manufacture and use. The Total Diet surveys carried out in
the UK show an 85% decrease in dioxin + PCB TEQs between 1981 and 2001, and
a 50% decrease between 1997 and 2001. The FSA survey of PCBs in salmon and
marine fish referred above was based on fish sampled in 1995, and the Jacobs
et al papers are based on salmon sampled in 1999. The Easton et al results
are based on salmon sampled in 1999 and 2000. The EWG report does not state
when the samples were taken. The fish meal manufacturers and fish feed
compounders are well aware of the situation concerning PCBs and dioxins in
their products and it is likely, though I have not come across and data,
that concentrations of these contaminants in fish feeds have gone down over
the last few years. It seems that dioxins and PCBs in fish pose a reducing
risk to human health.
References
Easton, M.D.L., Luszniak, D. & Von der Geest, D., 2002. Preliminary
examination of contaminant loadings in farmed salmon, wild salmon and
commercial salmon feed.
Chemosphere, 46,1053_1074.
FSA, 1999. Dioxins and PCBs in UK and imported marine fish. Food
surveillance Information Sheets, No. 184.
http://archive.food.gov.uk/maff/archive/food/infsheet/1999/no184/184diox.htm
Food Standards Agency of Ireland, 2002.
http://www.fsai.ie/surveillance/food/surveillance_food_summarydioxins.asp
Jacobs, M., Ferrario, J. & Byrne, C., 2002a. Investigation of
polychlorinated dibenzo_p_dioxins, dibenzo_p_furans and selected coplanar
PCBs in Scottish farmed Atlantic salmon (Salmo salar). Chemosphere, 47,
183_191.
Jacobs, M.N., Covaci, A., Schepens, P., 2002. Investigation of selected
persistent organic pollutants in farmed Atlantic salmon (Salmo salar),
salmon aquaculture feed, and fish oil components of the feed. Environmental
Science & Technology, 36, 2797_2805.
Krümmel et al, 2003. Nature, 425, 255-266
This archive was generated by hypermail 2b29 : Sun Sep 21 2003 - 09:32:30 PDT