Parallel Universe: I have written previously about the book ‘Escape from Model Land’ by Dr Erica Thompson, but last week I found that I was actually part of what she calls Model Land’s ‘mirror world’. I had spent two and a half hours online watching SEPA’s latest Sea Lice Risk Framework workshop. Although this was effectively a tick box exercise for SEPA, I was increasingly horrified by what I heard. Unlike some members of the industry who have been in discussion about the framework, SEPA have been extremely reluctant to speak to me, so much of what I heard I found to be new and as the revelations of how the framework is intended to work unfolded, I felt increasingly detached from the real world, transferring instead to a parallel world where it seems common sense no longer exists. Instead, this new mirror-world is governed entirely by SEPA’s model and whatever the model says is the new truth to which everyone must conform.
In my opinion, SEPA have been seduced by their model, for as Erica Thompson writes ‘Models have an almost magical capacity to lure their users into mistaking the sharp, tidy and analytically accessible world of a model with actual reality.’ At the same time, Dr Thompson highlights that the greatest problem with models is something no modeller will see as a problem – using them is good fun. Model Land is a wonderful place because in Model Land all the assumptions are true.
For me, this is where SEPA and I are at odds because I would argue that the fundamental assumptions of the model are not true at all. The science they use is simply wrong and as Erica Thompson highlights, it is all too easy to approach models as sources of objective scientific fact, but she adds that such naïve Model Land Realism can have catastrophic effects, and this is exactly what I believe will happen to the industry.
I appreciate that the industry feels it needs to be seen to be working with SEPA on this framework but when will the lure of Model Land sufficiently fade for the sector to say SEPAs proposals are not only unworkable, but also unreasonable. I would argue that we are already at that point, but I don’t see why full scientific discussion on sea lice cannot occur whilst plans for the framework progress.
I am not going to go through the science again to explain why SEPA are wrong, but rather pick up how being trapped in Model Land has led SEPA to make some unbelievable decisions. SEPA’s Head of Ecology told the REC Committee in November 2020 that SEPA do not believe that sea lice from salmon farms are responsible for the decline in wild fish. Yet, the Sea Lice Risk Framework now includes an assessment of the juvenile population to be paid for by the industry at an annual cost of £205,000 to monitor a total of 280 sites each year. Yet, there is already a national programme of juvenile salmon assessment in place, which already covers the west coast. Details of the national programme are available in the Marine Scotland topic sheet 144 and an example of the results of this survey from 2017 onwards can be seen on the West Sutherland Fisheries Trust website. However, the biggest question is how SEPA will be able to relate the results of these assessments to salmon farming? Maybe the answer is buried deep into the model but it must be able to predict how many adult fish return to the catchment, how many are intercepted by seals and other predators,; how many are caught and killed by anglers, directly or indirectly; whether there is sufficient water for the fish to ascend the river, find a mate and lay eggs; whether the redd survives or is washed out in any spate event; whether the eggs are successful and hatch and whether there is sufficient food in the river to sustain the juveniles as long as they can avoid predation. The number of juvenile salmon in any catchment is dependent on many factors and attributing any decline in numbers due to salmon farming is simply a step too far.
SEPA also propose that the industry should contribute £300,000 towards a sea trout monitoring programme that has been running since 1997 and is currently paid for by the Crown Estate. The results of these surveys when properly analysed, show that salmon farming has little impact on wild fish, although one recently published paper (2024) from Government scientists which used selective data collected from 2013 to 2017 suggests otherwise. They claim that changes in the way the data was reported prevented them analysing any other time period. As the reporting was changed by Scottish Government scientists, this is hardly an excuse not to analyse the whole data set.
However, the fundamental issue with the SEPA model is that it has not been validated. The model predicts where lice will be and in what number, so it should be possible to sample specific sites and compare the numbers of lice detected with the number predicted by the model. It is unclear why SEPA have not undertaken such validation except that in SEPA’s Model Land, validation can be achieved by other methods such as sentinel cages. It is also unclear why they did not attempt to deploy sentinel cages this year even in one location to provide their perceived validation and thus give more confidence that their model works according to the theory. Instead, I believe that this is another part of the Framework that is to be paid for by the industry. SEPA clearly have great faith in sentinel cages because they are paying IMR in Norway £200,000 for their expertise in sentinel deployment and analysis. It is worth pointing out that the last time sentinel cages were deployed was in Loch Linnhe between 2011 and 2013 and none of the fish in around ten cages suffered any measurable infestation in any of the three years during the critical smolt emigration period. Surely, if sentinel cages had any real value they would have been used in Scotland over the last ten years and they have not. It seems that SEPA’s model does not take such evidence into account.
The final word goes to Erica Thompson who writes that scientists who become trapped in Model Land through blind trust in their models do so at their peril.
On Guard: Sentinel or Guard cages had first been proposed as a method of assessing sea lice infestation during the 1980s. They have been in sporadic use since with Norway now appearing as their main proponent. Sentinel cages have been used as part of the Traffic Light System assessment since the outset. The data collected from lice on fish stocked in sentinel cages in the Hardangerfjord from 2012 to 2017 was used to calibrate the model of sea lice density.
According to the 2023 Sea Lice Expert Group report, the sentinel cage is a closed 1m3 cage hanging from 0.5-1.5 m depth and anchored on fixed station. 30 salmon smolts are stocked in between 17 and 20 cages in each fjord system. The measurement of lice infestation was initially set at less than 2 lice per fish for low infestation, between 2 and 6 lice per fish as moderate and above 6 lice per fish as high. However, in 2019, this was changed to 1 and less lice per fish per 14 days as low with moderate being 1-10 lice and more than 10 lice as high. As far as I understand, it, these changes were made to make it easier to compare the sentinel cage data with the modelled data.
In their report, the Expert Group do list some of the uncertainties that surround the use of sentinel cages including the possibility that there could be freshwater flows through the cages, which lice will avoid and this they say will be reflected in the lice infestation. However, clearly if the cages are sited well away from river mouths, then the dilution of normal salinity water will be avoided. Surely, the sites are researched before the cages are deployed eliminating such uncertainly. However, the fact that the Expert Group mention this at all must be of concern. They say another area of uncertainty arising from the use of sentinel cages is that the fish contained in them are at rest whilst migrating smolts are swimming. It must be assumed therefore that infestation rates would be higher in sentinel cages than in the wild If infestation occurs as the Expert Group claim. Other areas of uncertainty relate to differences in local currents, uneven lice distribution, fouling of the cages and different behaviour of the fish in the cages.
However, the Expert Group fail to mention other areas of uncertainty that are of greater concern. The Expert Group assume that any infestation must come from dispersed larval sea lice but fail to consider that infestation can come naturally from passing wild fish. They also fail to take into account that salmon smolts held in sentinel cages for two weeks without food will become stressed and more susceptible to infestation than healthy wild fish. Why fish have to be held for two weeks is unclear when fish can have left the fjords within a few days. In Scotland, this might be as little as one or two days.
There are undoubtably other areas of uncertainty relating to local conditions that have not been considered, all of which add to whether sentinel cage data has any validity, especially in relation to validating a model.
SEPA are to pay IMR £200,000 for help and advice on sentinel cage data but how much expertise IMR really have is open to question, Certainly, the annual reports from the Expert Group suggest a lack of consistency and an increasing lack of faith in sentinel cage data. The Expert Group have produced reports from 2016 onwards and using information from these reports, I have plotted the deployment of sentinel cages in the Traffic Light System’s thirteen production areas.
I would have thought that if IMR really wanted to validate their model, they would deploy sentinel cages in all 13 production areas but at most they achieved seven in 2017 and 2018. Since then, the number deployed has varied ranging from one to four with just two in the last assessment.
The results from PO3 for 2023 as translated by Google Translate state:
Guard cages: The guard cages indicate a generally low infection pressure in the outermost and innermost parts and moderate in the middle in the first round. In the second round, the infection pressure was generally somewhat lower in the middle and somewhat higher in the outer parts of the fjord. The results support a categorization in moderate salmon lice-related mortality. Given a median emigration of salmon in week 20, the first round covers from the median date of emigration and the following two weeks. The second period covers late-migrating salmon, and salmon that spend a long time migrating through the fjord. The uncertainty is assessed as medium as the cages do not cover the first part of the smolt migration, and do not cover Bjørnafjorden.
So that is as clear as mud.
Finally, from 2016 to 2019 the Expert Group included a table in their reports which compared the results of catching fish by trawling with fish caught by trapping and fish in the sentinel cage with IMR’s infestation pressure model, their virtual smolt model, with the Veterinary Institute smolt model and SINTEF’s infection pressure and virtual smolt models and their final conclusion in terms of low, moderate or high ratings. It is possible to see that the eventual ratings seem to have little to do with the sentinel cages assessments. It’s all a bit random. The following tables are for the years 2016 to 2019.
Since 2019. The Expert Group have not published these tables presumably because they amended the number of lice per fish used to assess the various levels of infestation pressure, as discussed above, in order to make sentinel cage data fit in better with their models. A moderate infestation pressure is now assessed as between 1 and 10 lice per fish over the 14-day period which is a big range for a short period of exposure.
I repeat what I said earlier that if sentinel cages are such a good source of information relating to infestation pressure, why have they not been used in Scotland since 2013?
Merry-go-round: periodically, I write to the Cabinet Secretary for Rural Affairs to express my concerns about the SEPA Sea Lice Risk Framework and the fact that neither SEPA nor Scottish Government scientist appear willing to want to discuss these concerns. Last week I received a response to my latest concerns and received the usual dismissal.
I appreciate that the respondent probably has little knowledge of the subject, and the content comes from someone, probably from within the scientific section who has more of a view. But again, there is a repeat that the international consensus of people who refuse to discuss the subject is that sea lice pose a risk to wild salmon. They also say that it is not over-precautionary, which it is, is transparent, which it is not, reasonable, which it is not, and protects the environment, which is certainly will not, whilst supporting social and economic development, which it doesn’t.
However, most interesting is the fact that the Sea Lice Risk Framework is adaptive which means that it can respond to new scientific evidence as it becomes available. Of course, the framework is only adaptive if SEPA and Government scientists are willing to discuss new evidence, which they clearly are not.
I would hope that the Cabinet Secretary is aware of the claims that are written in her name, but I fear she is not.
I repeat again that if SEPA and Scottish Government scientists are so confident in their approach, then they should have no hesitation is being willing to stand up and defend their science.