Guest Post by Willis Eschenbach
There’s a new study out called Increase in mercury in Pacific yellowfin tuna by Paul E. Drevnick, Carl H. Lamborg, and Martin J. Horgan. It claims that:
By compiling and re-analyzing published reports on yellowfin tuna (Thunnus albacares) caught near Hawaii (USA) over the past half century, the authors found that the concentration of mercury in these fish currently is increasing at a rate of at least 3.8% per year.
That seemed a bit too neat for me, so I took a deeper look. To their credit, they posted the data as used along with the study. As usual, I started by taking a look at all of the data. There were three samples of tuna studied, which were caught in 1971, 1998, and 2008. Here is a boxplot of the raw data.
Figure 1. Boxplot of the tuna mercury data by year of collection. Width of the box is proportional to the number of data points. Boxes show where half of the data is located. Heavy black line is the median of the data. Notches are the error intervals on the median. Units are parts per million (ppm)
OK, so far, so good. Next, they removed both the big fish and the small fish. They also removed two outliers, which had mercury values of 1.32 and 0.015 ppm. Once those are removed, the result is shown in Figure 2.
Figure 2. As in Figure 1, but for the reduced dataset.
Note that because of the greatly reduced numbers in the 2008 data, the uncertainty notch has become much wider, and the width of the box is smaller.
Finally, they adjust the mercury content for the weight of the fish. This is important because as the fish gains weight, it bioaccumulates mercury. Figure 3 shows what happens to the reduced dataset once the mercury content has been adjusted (either upwards or downwards) depending on the weight of the individual fish.
Figure 3. As in Figure 2, but with the mercury levels adjusted for the weight of the individual fish.
This has made some obvious changes to the results. First, the outliers have been greatly reduced, as has the range of the data. This is because the outliers were heavy fish with lots of mercury, so when they were adjusted their mercury levels came down. And curiously, while there is not a lot of change in the median and spread of the 1971 and 1998 data, the 2008 data has risen significantly. Finally, while the adjustment process reduced the error of the median in 1971 and 1998 data, it actually increased the error of the median in the 2008 data.
Now, these are the results that they claim show that mercury in these tuna is “increasing at a rate of at least 3.8% per year.” I’m sorry, but I’m not seeing that. For starters, if anything the mercury levels fell during the period where we have good data, from 1971 to 1998. That means that the entirety of the purported increase occurred over 10 years, after being stable for nearly thirty years? I’m not buying that claim at all.
So why did the results in 2008 move up so much due to the adjustment by weight? The problem is in the weight distribution of the fish in the three groups. Figure 4 shows the same three groups, but this time it shows the weights of the fish instead of the mercury levels.
Figure 4. Boxplot of the weights of the fish involved in the tuna study.
As you can see, while the distribution of the weights of the fish caught in 1971 and 1998 are quite similar, the 2008 sample are predominantly small fish. In theory, then, the mercury levels in these fish should be increased to bring them in line with the larger fish.
However, there are a couple of problems with that. First, the mercury/weight relationship gets flat down at the lower end. As the authors say:
It was necessary to remove the fish of less than 22 kg from the analysis, because these fish did not adhere to the assumption of linearity. Mercury concentrations in young tuna tend to be low but highly variable [18]. A diet shift occurs in young tuna when a critical body mass is developed that enables endothermic capability to allow access to prey in deeper, colder water [19]. At a certain size (depending on species), likely because of this ontogenetic diet shift, the relationship of mercury concentration versus size conforms to expectations (i.e., a linear relationship).
But here’s the problem with that theory … ugly data. Figure 5 shows the scatterplot of mercury levels versus fish weight.
Figure 5. Scatterplot, fish weight versus raw (unadjusted) mercury levels. Colors indicate the years as in previous figures (red-1971, gold-1998, and blue-2008)
A couple of points stand out here. First, their 22 kg cutoff seems way too low. According to their own data, there is little difference between mercury levels in tuna up to about 40 kg. This means that there will be errors in the adjusted mercury for fish less than 40 kg or so. Second, most of the blue 2008 data is low-weight fish (blue dots) … and as a result, the adjusted mercury levels of the 2008 data will be overestimated. Finally, this preponderance of light weight fish in 2008 is also the reason that the mercury adjustment, rather than reducing the spread of the 2008 data, actually increased the spread of the data.
So to summarize. The 1971 mercury data is statistically indistinguishable from the 1998 data, and the fish have about the same weight distribution. Together, these two groups comprise 94% of the data. They show no change in mercury levels over that twenty-seven year period.
They’ve built their entire claim of an increase in mercury on a mere 14 fish, 6% of the data, which are significantly lighter in weight than the other 94% of the sample. And as Figure 5 shows, it is likely that their adjusted mercury content is overestimated. Fourteen small-fry fish are all they have to hold up their claims? Really? This is almost to the level of the One Yamal Tree farrago.
And in any case, the idea that there would be absolutely no increase in mercury levels for nearly thirty years and then the mercury would jump significantly over the next ten years doesn’t pass the laugh test.
Best to all,
w.
PS-if you disagree with someone, please QUOTE THEIR EXACT WORDS that you disagree with, so that we can all understand the exact nature of your objection.
ALSO-
Folks not familiar with them might be interested in my other posts on mercury, viz:
In the process of writing my piece about Lisa Jackson and the EPA, I got to reading about the EPA passing new mercury regulations. Their regulations are supposed to save the lives of some 11,000 people per year. So I figured I should learn something about mercury. It turned out…
I’ve been puzzling for a while about why the areas with the most power plants aren’t the areas with the worst levels of mercury pollution. Why aren’t the areas downwind from the power plants heavily polluted? I keep running across curious statements like “There was no obvious relationship between large-mouth bass or yellow perch fish…