Not Science’s Finest Hour

In closing months of last year a news release from the University of St Andrews was first to break the story titled ‘Large numbers of salmon are killed by parasites, finds new study’. The release referred to a paper published by an international team of authors in the Proceedings of the Royal Society B on the 7 November; this was based a meta-analysis of 24 experiments from 5 previous research papers concerning sea lice, 4 experiments conducted in Norway and 20 in Ireland. The ‘headline message’ in the press release was that the study had found ‘sea lice to be responsible for 39 per cent of mortalities amongst salmon in the Northeast Atlantic Ocean’. This figure was challenged and in a follow up statement the University repeated ‘The central, unequivocal finding of this research paper, as presented in our press release, is that parasites such as sea lice are responsible for an average of 39 per cent of all deaths of salmon at sea’. Needless to say the UK media gave the story high profile; even the BBC got involved and ran the story not only on its website but in its broadcasts. By the 15 November Inland Fisheries Ireland, who like St Andrews could lay claim to one of the meta-analysis authors, was keen to get involved. Their press release was even clearer. Titled ‘Devastating impact on Wild Salmon populations from Sea lice’, it said ‘In a newly published report, results reveal that on average 39% of salmon mortalities were attributable to sea lice, which impacts wild salmon numbers and therefore wild salmon fisheries’. This release carried particular authority since it offered a quote from Fergus O’Dowd, the Irish Minister of State with responsibility for Natural Resources. As was the case in the St Andrews press release, he made an inferred linkage between the ‘devastating impact’ and salmon farming. Over the intervening few weeks the angling columnists have had a ball with this, everyone and their dog has trotted out their favourite claims against salmon farming. The ‘scientific claim’ has even been used to support protests against the development of new government-sponsored fish farms on the west coast of Ireland. That 39% of wild salmon mortality at sea can be attributed to sea lice has been referred to repeatedly. So, what’s to be surprised about! Well, in this very curious case of the salmon louse, the oddity is that the meta-analysis on which all the brouhaha is based does not show anything like what has been claimed. Without descending into technical detail, it is necessary to explain that the meta-analysis essentially consisted of a statistical analysis of a series of experiments in which salmon smolts, were released into river sites in paired-groups either untreated (control) or treated with a prophylactic dose of sea lice therapeutant providing sea lice protection for about 9 weeks. The experiments then recorded how many salmon from each group were recovered as they returned from sea the following year and the percentage survival of salmon from each group. The difference between treatments in survival could then be attributed to the effects of sea lice during the early salmon migration. The design of the meta-analysis itself had some slightly odd technical features, but these are matters that will no doubt be debated by scientists elsewhere; so let’s concentrate on what was found. Essentially, in agreement with the studies on which the meta-analysis was based, prophylactic treatment of smolts, on average, resulted in a positive effect on salmon survival. Using the kind of simple arithmetic that everyone will understand, across all the 24 experiments, the survival percentage for the control fish was about 3.18 per cent of fish released and the survival figure for treated fish was 3.75 per cent, a difference of 0.57 per cent. So where did the figure of 39% come from? Well, the authors did their analysis on a logarithmic transformation of the data and then calculated from a mixed model the related percentage difference between the two survival figures: this difference was stated as the ‘percentage salmon recruitment lost to parasites’. To roughly illustrate what this means, if the survival (or if you prefer recruitment) figure was 5 per cent for the treated fish the 39% reduction would give a control survival figure of 3 per cent, so that mortality would be 95 per cent in the one case and 97 per cent in the other. Thus the meta-analysis showed that, as in previous studies, the average difference in mortality due to sea lice (i.e. 2%) is an extremely small component of overall marine mortality, which in this example is 93-95 per cent. Oh, and in case I forgot to mention, the experiments were all conducted with hatchery bred ranched salmon not wild salmon, and they provided no data related to any claimed impacts of salmon farms. Sea lice are ubiquitous in coastal waters around Norway and Ireland and none of the studies analysed made any comparisons between smolts released in areas where salmon farms exist and where they don’t. More scientific papers on this topic will be forthcoming and, as always in science, the true picture will ultimately be established. However, at the moment, the media claim ‘sea lice are responsible for 39 percent of mortalities amongst salmon in the Northeast Atlantic Ocean’ does not look like one of science’s finest hours.