SEPA's Sea Lice Regulatory Framework

Annex B: Discussion regarding issues raised

16. This Annex provides detailed responses to discussions with Salmon Scotland. To aid navigation the topics are first summarised in Table B1. Links are provided from the tabulated topics to the detailed paragraphs which follow the table. Where the consideration corresponds to a recommendation in the report from the Chief Scientific Adviser (CSA) Marine, this is also indicated.

Table B1. of the issues raised by the sector in discussions with CSA Marine

Topic	Specific Issues / Requests	Cross reference to text
Baseline Data & Evidence Gaps	Lack of baseline data on natural sea lice levels in Scottish waters	#Plankton_sea_lice §32
	No population-level impact	#Population_impact §41 - 73
	Absence of quantified scale of any impact	#Population_impact §41 - 73
	Need for east/west coast comparison studies	#Plankton_sea_lice §32 #East-West Comparison §71
	Failure to analyse existing Environmental Management Plan (EMP) data	#EMP §74 - 76
	Post-smolt monitoring data not considered in SLRF development	#EMP §74 - 76
Scientific Evidence Base	Interpretation of RCT papers cited by SEPA regarding population impact	#Population_impact §41 - 73
	Need for independent analysis of wild salmonid catch data	#EMP §74 - 76
	Examination of global research reframing aquaculture sea lice contributions	#Population_impact §41 - 73
	Assessment of Canadian research on sea lice trends after farm cessation	#Canadian_removal_study §38 - 40
	Review of 10-year MD project outputs (or lack thereof)	Partially addressed #MD_Pilot_study §73
Risk Assessment Methodology	Compound conservatism in risk assessment parameters	Recommendation 2: Full sensitivity analyses to be undertaken.
	Use of unvalidated conceptual model of sea lice attachment	Recommendation 2: Full sensitivity analyses to be undertaken. Murray & Moriarty (2021)[6] base their theoretical model parameter choice on experimental observation (Tucker 2001[7])
	Reliance on Scottish Shelf Model with aggregated wind data from 2013	SSM not individually addressed, however broader Recommendation 2: Full sensitivity analyses to be undertaken.
	Black box nature of decision-making process	R1: Improve documentation and communication .
	Arbitrary 100m influence area for marine environment	This is in relation to medicinal use not directly related to SLRF
	Lack of transparency in how data feeds into framework	R1: Improve documentation and communication.
Regulatory Framework Issues	Standstill conditions based on flawed methodology	#Methods §125 - 178
	Penalisation of operators with good track records	#Track_record §172 - 175
	No mechanism for adaptive management or getting off standstill	R3: Full framework to be implemented.
	Disproportionate response to available evidence	#Expediency §77 – 102
	Failure to align with SIWG recommendations as a package	#Pressures §19 - 31
Fish Health & Welfare Concerns	Conflict between environmental controls and fish health requirements	#Fish_health §103 - 124
	Forced treatment regardless of fish size or health status	#Fish_health §103 - 124
	Veterinary ethics compromised by arbitrary thresholds	#Fish_health §103 - 124
	Medicine restrictions based on questionable environmental standards	#Fish_health §103 - 124
	Impact on fish welfare through over-treatment	#Fish_health §103 - 124
Monitoring & Validation	No framework for monitoring success of SLRF	R4: Shape of success to be defined.
	Expensive and problematic sentinel pen validation methods	#Cage_validation §34, §37 §82-83, §139, §180 - 182
	Lack of real-time data integration	R3: Full framework to be implemented.
	No assessment of framework effectiveness	R4: Shape of success to be defined.
	Missing adaptive framework for incorporating new evidence	R3: Full framework to be implemented.
Comparative Analysis Needs	Assessment of sea lice relative to other 40+ pressures on wild salmon	#Pressures §19 - 31
	Review of River Basin Management Plans showing "good environmental status"	It is not clear what the relevance is between a river’s good environmental status (which is not salmonid focused) to the aim of avoiding further marine environment deterioration with regards to sea lice.
	Comparison with Norwegian traffic light system limitations	#Expediency §77 – 102
	Analysis of proportionality compared to other environmental impacts	#Pressures §19 - 31
Geographic & Regional Issues	Northern Isles regulation with zero baseline data	Recognition of application to #Northern_Isles_trout §200
	West coast vs east coast impact differences	#Pressures §19 - 31
	Application to sea trout without scientific justification	#Trout §66
	Site-specific risk assessment inadequacies	#Site_specific_assessment §155 - 160

17. Within Annex B three ‘voices’ are represented:

• Voice 1(italics) reflects the issues raised by the industry.
• Voice 2 (plain font) represents CSA Marine review of relevant evidence.
• Voice 3 (bold) is the opinion/conclusion of CSA Marine.

18. The voices are intertwined and differentiated as above.

Annex B1. Context: Wild Salmon Pressures in Scotland

Issues Addressed in This Section:

• Assessment of sea lice relative to other 40+ pressures on wild salmon
• West coast vs east coast impact differences
• Failure to align with SIWG recommendations as a package

19. The Regional and national assessment of the pressures acting on Atlantic salmon in Scotland, 2021^[8] documented the pressures salmon face in Scotland. This indicates that the major threats identified nationally were predation by birds and seals and upstream barriers as the most severe pressures acting on wild salmon across Scotland. These pressures are considered stable but ongoing, representing the greatest current threats to salmon populations.

20. Of the emerging national threats identified, high river temperatures emerge as the most significant developing pressure, expected to increase due to climate change. Other emerging pressures include extreme high flow events, marine developments, and invasive North American signal crayfish, though confidence in assessing these impacts remains lower.

21. The declining pressures identified were exploitation through coastal and in-river netting; now considered historical, reflecting the success of the 2016 legislative ban on salmon retention in coastal waters.

Regional Variation in Sea Lice Pressure

22. Sea lice, however, are seen as a concern in regional contexts, primarily on the west coast where sea lice derived from salmonid aquaculture represents the greatest contemporary pressure, where salmon farming is concentrated. This regional dominance reflects the geographic distribution of marine finfish farms, with a presumption against further salmon farm developments on the north and east coasts of Scotland to protect migratory fish species.

23. Whilst sea lice do not feature prominently in the national summary, which is dominated by east coast data where salmonid aquaculture facilities are absent and impacts are likely minimal, it remains important where salmonid aquaculture occurs. The assessment reveals sea lice affects approximately 27% of Scottish river areas where severity could not be determined, indicating both widespread concern and significant uncertainty about impacts.

24. On the west coast, sea lice rank as the primary threat, followed by bird and seal predation. This contrasts sharply with other regions where physical barriers, predation, and climate-related pressures dominate, with netting eliminated. The localised nature of sea lice impacts demonstrates how pressure profiles vary across Scotland's coastal environments.

25. On the east coast the dominant pressure is bird predation and migration barriers whilst sea lice are perceived as negligible. The emerging concern on the east coast is climate change (temperature and flow extremes).

26. The north coast has generally fewer pressures affecting salmon although the primary concern is again predation from seals and birds, with sea lice providing minimal pressure.

27. In the southwest the major concerns are migration barriers and bird predation, and some netting persisted at the time of the review (although the latest fishery catch statistics^[9] indicate nine caught and retained salmon in southwest Scotland)

28. The assessment reveals significant uncertainties, with marine developments affecting over 50% of river areas classified as "unknown severity". Sea lice, farmed escapees, and predation also feature prominently among pressures where impact severity remains uncertain, highlighting the need for improved evidence and monitoring.

29. This assessment provides guidance for targeted management action. The regional variation in pressure profiles suggests that effective salmon conservation requires locally tailored approaches rather than uniform national strategies. For sea lice specifically, the concentrated impact on west coast populations requires attention in areas where salmon farming and wild salmon migration routes intersect.

30. The report emphasises that whilst sea lice represent a regionally critical pressure, Scotland's salmon face a diverse array of threats requiring comprehensive, adaptive management strategies that address the full spectrum of contemporary and emerging pressures across all regions.

31. There is clear evidence that there are multiple pressures faced by wild salmon populations and that these differ regionally around Scotland, however sea lice represent a particular pressure which can be managed, whereas survival at sea or impacts of climate change are not directly controllable.

Annex B2. Evidence Base for Sea Lice Impacts

Issues Addressed in This Section:

• Lack of baseline data on natural sea lice levels in Scottish waters
• No population-level impact quantification
• Absence of quantified scale of any impact
• Interpretation of RCT papers cited by SEPA
• Examination of global research reframing aquaculture sea lice contributions
• Assessment of Canadian research on sea lice trends after farm cessation
• Review of 10-year MD project outputs (partially addressed)
• Need for east/west coast comparison studies
• Need for independent analysis of wild salmonid catch data
• Application to sea trout without scientific justification
• Failure to analyse existing EMP data
• Post-smolt monitoring data not considered in SLRF development

Annex B2.1 Environmental Sea Lice Observations

32. When considering salmon aquaculture contribution to environmental levels of sea lice, Harte et al (2017)^[10] described the presence of sea lice in plankton samples collected from east and west coast sampling sites over a period of a decade. There were elevated plankton sea lice on the west coast compared to the east coast with the maximum recorded density value in Shieldaig at 14.2 sea lice m⁻³, while the Stonehaven site had density values of <1 louse m⁻³. Moreover, the species present on the west coast was the salmon specialist Lepeophtheirus salmonis, whilst the east coast was the generalist Caligus elongatus.

33. Although this is not an exhaustive exploration of environmental sea lice levels, these long-term series of east and west coast demonstrate a substantial difference in the abundance and species composition between two indicator sites.

34. Sentinel cage studies on the west coast (Pert et al 2014) ^[11] identified that Infestation pressure on the sentinel fish was correlated with gravid L. salmonis counts from nearby salmon farms, the paper demonstrates sentinel cage proximity to a salmon farm did not determine the prevalence of sea lice on sentinel salmon. A similar relationship was identified using plankton sampling by Penston & Davies (2009) ^[12], who found that numbers of gravid females on fish farms correlated with the numbers of sea lice copepodids in surrounding waters.

35. Similar observations on the proximity of salmon farms and infestation levels on wild salmonids have been demonstrated from data collected from west coast of Scotland which showed that the proportion of individual sea trout (Salmo trutta) with stress-inducing sea louse burdens increased with the mean weight of salmon on the nearest salmon farm and decreased with distance from that salmon farm (Middlemas et al. 2013^[13] ). Levels of sea lice on wild sea trout also relate to fluctuations on salmon farms associated with stage of production cycle (Middlemas et al. 2010^[14]).

36. Although these studies provide observations only in specific sites, this evidence demonstrates the scale of the difference in baseline sea lice level in an area in the presence (west coast), and absence of salmon aquaculture facilities (east coast), and that presence to salmon farms influences the environmental incidence of sea lice.

37. Rabe et al (2024)^[15] provide a review of the suite of environmental monitoring methods for assessing sea lice concentrations, including sentinel caged fish both static and towed, plankton sampling and wild fish surveillance and eDNA techniques. The paper includes examples of where sentinel cage data has been used to assess sea lice dispersal modelling outputs for both Scotland (e.g. Salama et al 2018^[16]) and Norway (Sandvik et al 2020)^[17], where there is a correspondence between modelled relative sea lice concentrations and observed sea lice concentrations.

38. The recent analysis by Jones et al. (2025)^[18] considered sea lice infestations on juvenile wild Pacific salmon (which have differing susceptibility to L. salmonis ^[19] , thus comparisons to the salmonids in Scotland may exhibit different influences on the environmental prevalence of L. salmonis) in British Columbia Discovery Islands over an eight-year period spanning both active and inactive phases of Atlantic salmon aquaculture. The study reports a 96% decline in sea lice abundance between 2020 and 2022, coinciding with the phased removal of salmon farms, which suggests that salmon aquaculture operations were a contributor to infestation pressure. However, sea lice persisted and even increased in 2024 despite the continued absence of salmon farming, indicating that environmental factors, such as temperature, salinity, and natural reservoirs, also play important roles in sea lice dynamics. While the influence of salmon farms is substantial, the findings demonstrate the complexity of infestation patterns and the need for environmental monitoring to better understand and manage risks to wild salmon populations.

39. This Canadian work demonstrates the complexity of relating presence of salmon farms and the interaction with environmental conditions to sea lice infestation on wild salmon. It is noted that this report is of a short time series; the elevated observation in 2024, after two low year recordings of sea lice on wild salmon after the removal of salmon farms in the area. It is not scientifically justifiable to make a definitive conclusion either way of the long-term natural variability of infestation experienced by wild salmon in the presence or absence of salmon farms. One extreme year in a three-time series can skew results, for long-term changes in climate studies time series generally need to exceed two decades in length and have 5-year running means applied (or similar statistical treatment).

40. Although this evidence demonstrates that salmon farms can act as a source of environmental sea lice, it does not demonstrate that this environmental exposure results in wild salmon risk. For this, assessment of salmon detriment is required.

Annex B2.2 Impacts on Wild Salmonid Condition and Survival

41. Assessment of somatic condition in wild adult salmon sampled from sites not located near salmon farms, revealed significant negative associations between sea lice burden and weight-at-length relationships that persist throughout the marine growth phase (Susdorf et al., 2018)^[20]. Analysis of over 6,000 return-migrant salmon captured at three locations across Scotland, England, and Ireland demonstrated consistent reductions in body condition associated with mobile sea lice density, with effect magnitudes varying seasonally and geographically in response to environmental conditions and salmon aquaculture intensity. The presence of sea lice in returning samples indicated a median reduction in weight by 3.7% representing a median of 85g reduced mass.

42. The relationship between sea lice burden and growth performance exhibits interactions with host characteristics, environmental conditions, and seasonal timing that modulate the severity of observed effects (Shephard et al., 2016)^[21]. Sea trout monitoring across 94 river systems in Ireland and Scotland revealed that proximity to salmon farms significantly reduced weight-at-length relationships, with the strongest impacts occurring during dry years when environmental stress was elevated. These context-dependent effects highlight the vulnerability of wild fish populations to cumulative stressor impacts that exceed individual tolerance thresholds.

43. Long-term consequences of condition reduction extend beyond immediate survival impacts to encompass reproductive success, competitive ability, and population-level demographic changes (Thorstad et al., 2015)^[22]. Reduced somatic condition typically reflects depleted lipid reserves that are critical for successful upstream migration and spawning performance, creating potential intergenerational effects that amplify population-level impacts beyond direct mortality estimates.

44. Laboratory-based dose-response relationships have enabled quantification of specific sea lice burden thresholds associated with different severities of impact on juvenile salmonids (Ives et al., 2023)^[23]. The identification of two critical thresholds at points where 50% probability of mortality occurs (equivalent to LD₅₀) provides empirical foundations for regulatory decision-making and monitoring programme design. The lower threshold (T1) of approximately 0.08 sea lice g⁻¹ host weight represents the onset of systemic sub-lethal effects likely to compromise long-term survival prospects, whilst the upper threshold (T2) of 0.24 sea lice g⁻¹ indicates 50% probability of direct mortality under controlled conditions.

45. Recent work by Vollsett et al (2025)^[24] reports that studies from PIT tagging studies of brown trout in Norway demonstrated a statistically significant reduction in survival from increase from 0 to 1 louse per gram reduced the survival probability by approximately 73% in 2020 and 58% in 2021. These are higher limits than observed in laboratory settings for Atlantic salmon and is hypothesised to be a result of brown trout having behavioural responses enabling the opportunity to recuperate from the osmotic damages caused by sea lice exposure.

46. Long-term salmon population modelling simulations indicate that sea lice induced reduction of the body condition of multi-sea-winter fish has long term declines in modelled population trajectories (Susdorf et al 2018)^[25].

47. Although sea lice parasitism has the potential for deterioration in fish condition, studies of actual direct mortality are required to be considered.

Annex B2.3 Experimental Evidence: The RCT Debate

48. Meta-analytical assessment of manipulative field experiments conducted across Ireland and Norway has provided quantitative estimates of sea lice-induced mortality in wild Atlantic salmon populations (Krkošek et al., 2013)^[26]. The analysis encompassed 24 experimental trials involving 283,347 tagged smolts released as paired control and prophylactically treated groups between 1996 and 2008, representing an empirical assessment of sea lice impacts on wild fish populations.

49. Results demonstrated a significant positive effect of parasite protection on survival to recruitment, with an overall odds ratio (odds ratio is a way to compare the chances of something happening in one group compared to the chances of it happening in another group) of 1.29 corresponding to an estimated 39% reduction in adult salmon recruitment attributable to sea lice infestation (Krkošek et al., 2013) as a result of treated versus untreated fish.

50. Regional variation in mortality estimates reveals heterogeneity that reflects differences in sea lice pressure, environmental conditions, and baseline survival rates between study locations (Vollset et al., 2016)^[27]. This Norwegian study reported of hatchery-reared Atlantic salmon smolts untreated with sea lice medicines experienced mortality from 0.6% to 31.9% between different river systems, with the most severe impacts occurring in areas characterised by intensive salmon aquaculture development and elevated environmental sea lice concentrations. Importantly, the magnitude of sea lice effects demonstrates strong negative correlation with baseline survival rates of anti-parasite medicine treated hatchery reared fish, suggesting that salmon populations already experiencing poor marine survival face additional mortality from parasite exposure.

51. Recent meta-analyses (Gargan et al 2025)^[28] of 43 paired releases spanning the period 2001 to 2019 show a significant treatment effect against sea lice with a risk ratio of 1.22, equating to 18% [CI: 8%–27%] fewer returns of untreated adult hatchery salmon. Meta-regression further demonstrated that the risk increased with sea lice infestation pressure from salmon farms. Infestation pressure was also significantly associated with declining return rates in both the treated and untreated groups, corroborating earlier findings that the medicinal treatment against sea lice may not completely shield the post-smolt salmon against the virulent effect of sea lice.

The Jackson-Krkošek Debate

52. There is discourse concerning interpretation of randomised control trials of sea lice treatment release and recapture studies. This is exemplified by the debate as a result of Jackson et al’s (2013) work in Ireland.

53. The quantification of sea lice impacts on wild Atlantic salmon populations has generated significant academic debate, centered primarily on methodological approaches to analysing randomised controlled trial (RCT) data. This discourse has evolved from disagreements about statistical methodology and interpretation – there is no clear “right” or “wrong”.

54. Jackson et al. (2013)^[29] presented a large-scale analysis of 28 paired releases involving 352,142 salmon across Ireland's west coast surveyed between 2001-2009. The study treated fish with prophylactic emamectin benzoate as the experimental intervention, they applied generalised logistic regression models incorporating random effects for location and temporal variation. Their primary findings indicated an odds ratio of 1.14 (95% CI: 1.07-1.21), corresponding to approximately 1% absolute difference in return rates between treated and control groups.

55. The authors' interpretation emphasised the magnitude of this effect relative to overall marine mortality (>94%), concluding that sea lice-induced mortality represented "a minor and irregular component of marine mortality" unlikely to significantly influence salmon conservation status. This conclusion was predicated on their statistical approach treating the data as individual fish-level observations within a mixed-effects framework, adjusting for inter-annual and spatial variation.

56. Krkošek et al. (2014)^[30] challenged Jackson's analysis arguing that Jackson failed to apply the paired structure of the experimental design in their meta-analytical approach, instead treating releases as independent observations without proper within-pair comparisons. They also contended that survival data required log-scale analysis rather than arithmetic comparison of proportions, citing established survival analysis methodology. They distinguished between absolute percentage point differences and proportional mortality effects, arguing that Jackson's 1% figure misrepresented the actual biological impact. Applying random-effects meta-analysis with appropriate weighting, Krkošek et al. calculated an odds ratio of 1.41 (95% CI: 1.25-1.60), translating to an estimated 34% loss of adult recruitment attributable to sea lice infestation. This represented a 30-fold difference in estimated impact magnitude.

57. Jackson et al.'s (2014)^[31] response defended their methodological choices maintaining that their logistic regression approach was appropriate for binary response variables from designed experiments, citing established statistical literature. They highlighted that Krkošek's meta-analysis exhibited extreme heterogeneity (I² = 93%), arguing that combined estimates under such conditions were "highly questionable" and violated fundamental meta-analytical assumptions. They clarified that their chi-squared tests did incorporate within-river comparisons, addressing the "paired structure" criticism as a mischaracterisation of their approach. They also noted that concurrent Norwegian studies using identical methodology (Skilbrei et al., 2013)^[32] produced comparable results, suggesting methodological robustness.

58. The recent Gargan et al. (2025) study addresses aspects of the original debate. Using the updated Irish dataset to date (43 paired releases, 545,893 fish, 2001-2019), they employed both meta-analytical and generalised linear mixed model approaches. Unlike previous studies, Gargan et al. (2025) incorporated quantitative estimates of sea lice infestation pressure, derived from Marine Institute monitoring data and salmon farm stock density estimates. They explicitly modelled the relationship between effect size and infestation pressure through meta-regression, addressing the heterogeneity concerns raised in the earlier Jackson-Krkošek debate. They investigated incomplete protection in treated groups, accounting for drug resistance and incomplete medication uptake.

59. Analysis by Larsen et al (2024)^[33] demonstrated a negative association of sea lice from salmon farms on recreational fishing catches of Atlantic salmon in Norway. Their modelling estimated below-average catches when the total sea-louse load exceeds the limit of 0.1 average adult female sea louse per farmed salmon and that the risk of below-average catches increases by approximately 47% when salmon farms exceed 0.1 louse per farmed salmon (estimated risk ratio of 1.47, 95% CI [1.10, 1.96]). The meta-analysis revealed a risk ratio of 1.22 (95% CI: 1.09-1.37), representing 18% fewer untreated fish returning. The meta-regression demonstrated that this effect varied significantly with infestation pressure, ranging from minimal impact during low-pressure periods to risk ratios >1.9 during high-pressure events.

60. The GLMM analysis revealed a significant treatment × infestation pressure interaction, indicating that both treated and untreated groups experienced reduced survival under high sea lice pressure, but with differential magnitude. This finding suggests that RCT approaches may underestimate total sea lice impacts due to incomplete treatment protection.

Interpretation and Implications

61. The debate about interpreting sea lice impacts on return rates of release treated fish highlighted that the heterogeneity in Krkošek's analysis emphasises the importance of exploring sources of between-study variation rather than simply computing pooled estimates.

62. There is a distinction between absolute and relative effect measures remaining an important part of the debate relating to sea lice influence on wild salmonids. Both paired and unpaired analytical approaches have merit, but the choice should align with research questions and data structure. Questions relating to absolute impact on salmonid survival, and relative effects on salmon survival from parasitic sea lice.

63. The Irish study demonstrates that there is a high level of marine mortality in wild salmon and that sea lice adds a small component to overall marine mortality, however when considering the influence of sea lice on return rates there is significantly greater return rate of fish treated against sea lice parasitism.

64. As a descriptive example (figure Annex B1). Consider a population of 100 salmon smolt emerging from a river. Jackson (2013) indicates that marine mortality is approximately 94%, with less than 1% attributed to sea lice, so in this example 95% die at sea, one of which is due to sea lice infestation. This results in five salmon undertaking return migration. As a result, sea lice would have caused 1/100 fish to suffer mortality from the total outcomes of the exemplar population. However, had that smolt not been exposed to sea lice in the coastal environment, six fish would have returned. In this example, one of the potential six fish which could have returned, suffers mortality as a result of being susceptible to sea lice parasitism. This is the equivalent of sea lice reducing the returning number of salmon from six fish to five fish which is a reduction of some 17% (1 fish/ 6 fish), which is a realistic representation in this example of the scale of sea lice induced mortality given the findings of the RTC studies (e.g. Gargan et al. 2025).

65. These global studies demonstrate the impact of sea lice on individuals and populations of salmon and trout, and although there is variability in survivability observed in RTC studies, the general conclusion is that sea lice are able to decrease the survivability of individuals within populations, and that sea lice influences can lead to long term population reductions.

Annex B2.4 Sea Trout Impacts

66. Although there is limited direct evidence of sea lice impact on trout populations, recent work by Vollsett et al (2025)^[34] in Norway of PIT tagged Salmo trutta found a significant negative correlation between sea lice per gram of fish weight and the survival probability. Increasing sea lice load from 0 to 1 louse per gram fish reduced the survival probability by approximately 73% in 2020 and 58% in 2021.

Confounding Factors in RCT Studies

67. It is prudent to recognise that these RCT described above use generalist parasitic medicinal treatments such as emamectin benzoate. These medicines have anthelmetic properties^{[35] [36]} and efficacy against monogenean parasites^[37]. As an example, Kent et al (2020)^[38] reported 100% prevalence of Anisakis simplex in wild Atlantic salmon from Scotland, with the same 100% prevalence in Norway (Mo et al 2021)^[39]. Whilst farmed Atlantic salmon are absent of Anasakis^{[40] [41]} due to the use of pellet feed. This indicates that the source of infection in wild salmon is highly unlikely to be of salmon farm derived. Atlantic salmon are reported to have an inflammatory response to infection by A. simplex and effects the organ functions of salmon^[42]. This led the Atlantic Salmon Trust report by Whelen & Mo (2022)^[43] to hypothesise that “……salmon with numerous Anisakis larvae in the musculature has a reduced swimming capacity and thus is an easy prey to a predator, e.g. a whale.”

68. Assessments of parasitism of cod has demonstrated that the body condition of infected individuals was lower than that of those free of parasites and declined with the intensity of infection; the condition of most infected fish was up to 20% lower than that of uninfected individuals ^[44].

69. These salmon RCT assume that the reduction in return rate is attributed to sea lice infestation being impeded through prophylactic treatments, however it has not been ascertained what component of the increased survivability because of anti-parasitic medicinal treatment providing protection against sea lice or other highly prevalent parasites which are known to effect survivability.

70. However, the protective period of emamectin benzoate is of the order of up to 60 days (Aldrin et al 2023)^[45], creating a situation whereby the acquisition of endoparasites is most likely to occur after such time. Although it cannot be discounted, the protective attributes of anti-parasite medicines occur a relatively short time after use, whilst salmon are in the coastal zone exposed to elevated levels of sea lice. As such it is likely that parasites excluded in this early period are predominantly sea lice and that this enhanced survivability from treatment is through protection against sea lice as opposed to other parasites.

Need for Comparative East-West Studies

71. RCT are typically conducted in areas with salmon aquaculture. There are no simultaneous RCT of east and west coast Scottish wild Atlantic salmon populations. If an RCT was conducted in an area completely free of salmon aquaculture (e.g. east coast Scotland), it might be the case that there would be no significant difference in the survival between treated and untreated groups. RCT determine the baseline survival in the absence of infestation of sea lice through medicinal treatments. RCTs in areas with salmon aquaculture facilities measure the combined effect of salmon farm-derived sea lice population (all sea lice are of wild origin, as salmon farms are free of sea lice when stocked), and direct-wild origin sea lice. The untreated group can be infested by both sources. An RCT simultaneously undertaken in a salmon aquaculture and salmon aquaculture-absent area enables an assessment of baseline for all other causes of mortality including parasitism from direct wild-origin hosts. It could also enable quantification of any potential survival benefit obtained as a result of being treated with veterinary medicines. This would enable quantification of the impact of salmon farm-derived sea lice exposure.

72. This may lead to greater understanding of the attributable contribution of salmon aquaculture to marine mortality of wild salmon.

73. A pilot study (Morris et al 2018)^[46] conducted trials of tagged anti-parasite treated and untreated fish from the Lochy river system on the west coast and Conan river on the east coast. The study resulted in low recapture of returning fish. In 2016, no returning tagged fish were recorded on the west coast Lochy sites, whereas 23 returning tagged salmon grilse were recorded on the East coast Conon site. In 2017, 2 tagged fish were detected on the Lochy sites and 29 on the Conon site (4 Multisea winter fish, 25 grilse). On the east coast, there were no significant differences between the treated and control groups, regarding return rates or the condition of the adult fish.

Annex B2.5 Environmental Management Plan Data

74. As a result of the concerns over the potential environmental impacts of the industry, particularly regarding interactions between farmed and wild salmon, Environmental Management Plans (EMPs) for salmon farming in Scotland emerged. EMPs were interim regulatory tools introduced by the Scottish Government in 2019 as conditions for new marine aquaculture planning applications where there is potential for wild/farmed salmon interaction in order to monitor salmon farming impacts on wild fish populations.^[47] Originally designed as temporary measures, EMPs are being phased out in favour of SEPA's Sea Lice Risk Assessment Framework. EMPs were designed to provide the feedback mechanism between salmon farmers and local District Salmon Fishery Boards to generate management responses when elevated sea lice numbers were detected.

75. The SEPA SLRF consultation response^[48] provides a description of the transition from EMPs to the SLRF. SEPA state:

“The environmental monitoring element of EMPs will be delivered through the new regulatory framework’s WSPZ monitoring programmes and the new nationally overseen sea trout monitoring programme. As with EMP monitoring, these programmes will be collaborative programmes funded by operators and designed to support adaptive management. However, they will be led by us, and their targeting and design will be driven by the needs of the regulatory framework.

During 2024, SEPA will commence work with local authorities to review EMPs for existing salmon farms. The review will consider:

• The information already generated by the EMP.
• The monitoring approaches used.
• Further monitoring information required to support implementation of the new regulatory framework.

Once a review has been completed and its outcome agreed with the local authority, we will work with the local authority and the operator concerned to incorporate any continued, or revised, monitoring required into the framework’s monitoring programmes.”

76. A documented description of the process of moving from EMPs to SLRF would aid understanding of how, if any, EMP monitoring and wild catch data supported the development of SLRF – and if it did not, an explanation of why.

Annex B3. Comparative Analysis: Norwegian Traffic Light System

Issues Addressed in This Section:

• Disproportionate response to available evidence
• Comparison with Norwegian traffic light system limitations
• Expensive and problematic sentinel pen validation methods
• Analysis of proportionality compared to other environmental impacts.

77. The Norwegian Traffic Light System (TLS) provides empirical evidence for evaluating the necessity and expediency of salmon farm regulatory frameworks for the protection of wild salmonids, including Scotland's sea lice regulatory framework. Examining the Norwegian Traffic Light System's operational history since 2017 provides some insight into the effectiveness of area-based regulatory approaches to sea lice risk from salmon aquaculture production and the impact on wild Atlantic salmon. It is noted that although there are shared characteristics of regulating aquaculture activities in the aim of supporting wild salmonid conservation, the underlying process of the TLS differ from the SLRF, so direct assessments between mechanisms and outcomes are not directly comparable.

78. The TLS divides Norway's coast into 13 production areas with capacity adjustments based on estimated wild salmon mortality from salmon farm-derived sea lice (Ministry of Trade, Industry and Fisheries, 2015)^[49]. The Norwegian TLS quantifies and regulate salmon lice impacts through modelling and observational data. However, recent evaluations raise questions about both the scientific validity and economic efficiency of such approaches.

Scientific Foundation Concerns

79. An Evaluation committee (EvalComm) undertook an independent assessment of the TLS in 2021 (Eliasen et al., 2021)^[50] and reported weaknesses in the TLS's scientific foundations. Central to these concerns are mortality thresholds that lack empirical validation: "The EvalComm is of the view that a solid empirical basis for the thresholds has not been provided to date and that such is required to underwrite key assessments arising from the TLS."

80. Van Nes et al. (2025)^[51] demonstrated systematic overestimation across multiple system components. Mortality thresholds, derived primarily from laboratory studies using inappropriate fish sizes and species, may overestimate effects by some 50%. The TLS also assumed smolt migration timing and duration inconsistent with acoustic telemetry data, with actual migration occurring approximately two weeks earlier (median date 9 May versus 23 May assumed) and lasting 6.9-14 days rather than the assumed 28 days.

81. The SLRF does not replicate these TLS shortcomings: the SLRF thresholds resulting from laboratory assessments are over double that of the original TLS, and the migratory period is informed by tracking work.

82. Within the TLS, sentinel cage monitoring is applied, however the review by Van Nes et al. (2025) report that Norwegian Sentinel cages used fish 4-5 times heavier than wild smolts without size correction, whilst data collection occurs in areas with optimal sea lice survival conditions rather than representative spatial sampling. Sea trout data also served as proxies for salmon despite differing life-history. This results in compounding model inaccuracy as they depend on calibration against these potentially flawed observational datasets.

83. The SLRF must account for these biological realities in any assessment and future adaptive implementation of using sentinel caged fish.

84. Further criticism of the traffic light system exists in that the TLS overestimates sea lice larvae production by assuming 100% reproductive activity among adult female sea lice whereas actual reproductive proportion of the population is 10-33% and the TLS fails to incorporate salinity-dependent mortality of lice larvae (Van Nes et al., 2025).

85. Not all female sea lice contribute to sea lice production. It currently appears that the sea lice budget per salmon farm is calculated as the product of Number of Fish, Sea Lice per Fish and New Sea Lice Per Day and is uncorrected to reflect that not all lice on fish will be reproductively contributing to the production of new sea lice per day. Reasons should be published describing why it is appropriate in SLRF to assume that all adult female sea lice counted are, or will become, ovigerous.

Structural and Economic Issues

86. Jensen et al. (2024)^[52] provides a theoretical context by analysing the TLS as a nonpoint pollution problem. In that individual salmon farm contributions to environmental impact cannot be directly observed or are expensive to monitor.

87. The salmon lice problem exhibits nonpoint characteristics: wild smolts migrate past multiple salmon farm sites making individual contributions unidentifiable. Sea lice population dynamics depend on numerous environmental factors beyond salmon farm operations; and empirical studies demonstrate "very weak causal relation between salmon lice at the salmon farm site level and the infection of wild salmon stocks" (Jensen et al., 2024). The authors propose that the sea lice problem exhibits classic nonpoint characteristics in that wild salmon smolts migrate past multiple salmon farm sites with each making individual contributions unidentifiable to the overall impact. Furthermore, sea lice population dynamics depend on environmental factors (e.g. salinity, temperature) beyond salmon farm operation control, and they assert that empirical studies demonstrate "very weak causal relation between salmon lice at the salmon farm site level and the infection of wild salmon stocks" (Jensen et al., 2024).

88. Jensen et al (2025) argue that the TLS represents an application of mechanisms for regulating aggregated pollution but suffers from deficiencies in that salmon farm production areas contain between 2-133 salmon farm sites, whilst experimental evidence demonstrates such mechanisms only function effectively with fewer than eight participants. Consequently, individual operators perceive minimal influence over aggregated outcomes, reducing the system to economically inefficient lump-sum transfers rather than effective incentive mechanisms.

89. This is an example of an alternative regulatory mechanism for protecting wild salmonids from sea lice emanating from salmon farms. To provide reassurance, the SLRF documented technical guide (Recommendation 1) could include a section on alternative regulatory mechanisms, concluding with the decision that the SLRF regulatory tool was the preferred mechanism for Scotland.

90. The TLS exhibits "collective punishment" characteristics whereby all producers face consequences based on aggregated performance regardless of individual actions, a principle inconsistent with established legal frameworks and perceived as inherently unfair (Jensen et al., 2024).

91. The SLRF appears not to suffer from this paradox in that individual salmon farm budget and non-deterioration conditions are applied on individual sites.

92. The TLS relies on what Jensen et al. (2024) characterise as a "highly uncertain measure of the aggregated salmon lice-induced mortality," with assessments varying substantially depending on methodology applied. This measurement of uncertainty, combined with weak causal relationships between regulated activities and environmental outcomes, undermines the theoretical foundation necessary for effective nonpoint pollution regulation.

Absence of Demonstrable Environmental Benefits

93. The TLS provides an absence of measurable environmental benefits. Larsen and Vormedal (2021)^[53] found "no or weak statistical correlations between the share of severely infested wild salmon smolts and measures for the prevalence of salmon lice at the farm site level." Environmental factors (e.g. temperature, salinity) and fish size consistently outperform farm-derived infection pressure as predictors of wild salmon lice levels.

94. The Norwegian experience reveals a contradictory position: stricter sea lice regulation has successfully reduced salmon farm lice levels without corresponding improvements in wild salmon populations, whilst causing "extensive" welfare impacts on farmed salmon (Jensen et al., 2024). This pattern suggests that current regulatory approaches in Norway may target sea lice epidemiology rather than genuine causal factors affecting wild salmon survival.

95. The SLRF documentation appears not to currently detail proposals for the evaluation of mortality of wild salmon, although monitoring is referred to. If one purpose of the framework is to support the survival of wild Atlantic salmon and sea trout, then an evaluation process should be described.

Alternative Regulatory Approaches

96. Jensen et al (2024) suggests market based regulatory alternatives for Norway, which may also be alternatives for Scotland. Converting the nonpoint pollution problem to point pollution through enhanced monitoring technologies to enable instruments such as Pigouvian taxes or tradeable permits. Input regulation targeting specific risk factors (delousing measures, preventive technologies) could achieve environmental objectives more efficiently than complex outcome-based systems.

Implications for Scotland

97. The Norwegian evidence raises questions about the efficacy of complex sea lice risk assessment frameworks. If the causal relationships between salmon aquaculture operations and wild fish impacts remain empirically undemonstrated since 2017 despite intensive study, the scientific foundation for their TLS becomes questionable.

98. The overestimation of risks identified in the TLS suggests that similar frameworks may generate disproportionate regulatory responses to uncertain environmental problems. SLRF does take heed of some of the shortcomings of TLS, nevertheless when implementing similar frameworks for Scotland, policy evaluators require robust empirical demonstration of causal relationships between proposed regulatory targets and environmental impacts.

99. The evaluation by Jensen et al (2024) identifies TLS's economic inefficiencies and may provide cautionary lessons about regulatory design for Scotland. Complex, model-dependent systems generate high transaction costs for the sector whilst failing to apply market incentives effectively. The Norwegian experience suggests that perhaps simpler, more targeted approaches focusing on demonstrable risk factors (e.g. sea lice levels per salmon farm) may achieve environmental outcomes at lower economic cost. The SLRF in essence does implement this approach but still determines which individual salmon farms are subject to sea lice limits using a complex risk assessment process with potentially high transactional costs associated with farm development or modification.

100. The absence of demonstrable environmental benefits in Norway alongside substantial economic impacts raises proportionality concerns relevant to any similar regulatory framework for Scotland. Regulatory authorities must carefully balance precautionary protection against the economic and social costs of potentially ineffective interventions towards a target aim of supporting the survival of wild salmonids

101. These findings suggest that area-based regulatory systems relying on complex modelling and uncertain causal relationships may be neither necessary nor expedient responses to salmon lice concerns. Alternative approaches emphasising direct monitoring, market-based incentives, and adaptive management frameworks based on demonstrated outcomes may achieve lower sea lice release with an economic balance. The SLRF has built on the Norwegian experience and focused on direct site monitoring and aims to have a site level adaptive process, this is welcomed.

102. In summary, if one purpose of the framework is to support the survival of wild Atlantic salmon and sea trout, then an evaluation process designed to demonstrate the effect of implementing the SLRF on wild Atlantic salmon and sea trout should be described and implemented.

Annex B4. Fish Health and Welfare Implications

Issues Addressed in This Section:

• Conflict between environmental controls and fish health requirements
• Forced treatment regardless of fish size or health status
• Veterinary ethics compromised by arbitrary thresholds
• Medicine restrictions based on questionable environmental standards
• Impact on fish welfare through over-treatment

103. The management of sea lice infestations in Atlantic salmon aquaculture represents a welfare challenge facing industry. Sea lice are naturally occurring ectoparasitic copepods that have co-evolved with their salmonid hosts (Torrissen et al., 2013)^[54]. However, the intensive nature of salmon farming creates high host density which in turn creates increased transmission rates, necessitating active management interventions (Pike & Wadsworth, 1999)^[55].

Regulatory Framework and Counting Requirements

104. Regulatory frameworks employ count-based threshold systems which have been a long-standing practice as detailed in the industry’s Code of Good Practice ^[56] which stipulates:

• An average of 0.5 adult female L. salmonis per fish during the period 1^st February to 30th June inclusive.
• An average of 1.0 adult female L. salmonis per fish during the period 1st July to 31st January inclusive.

105. The Scottish Government’s Fish Health Inspectorate have reporting requirements including different response levels of 2 and 6 sea lice per fish^[57].

106. These management regimes aim to balance industry productivity with protection of wild salmonid populations, recognising that salmon lice derived from salmon farms have the potential to interact with wild stocks of salmonids. Norway's traffic light system and the SLRF approach, where salmon farm sea lice capacity is regulated based on estimated sea lice-induced mortality of wild salmonids, also requires counting and treatment responses.

Welfare Costs of Count Management Systems

107. The SLRF is not prescriptive in how sea lice are managed on salmon farms, and there are wide range management options available to salmon farmers. However, the sector raised implication concerns for farmed salmon under such rigid count management systems pose a risk to salmon health and welfare. Sea lice have been shown to reduce fish growth and appetite and cause substantial costs to salmon farmers worldwide, with economic losses of £65 million in Scotland as a result of sea lice management, representing approximately 9% of total production value (Boerlage et al., 2024)^[58] Although there are non-treatment preventative measures available to salmon farmers such as fallowing, on-site growing periods, and barriers, the following focuses on the concern raised with relation to treatments and animal health and welfare. The treatments currently employed to maintain counts below regulatory thresholds may themselvesimpose significant welfare costs. These welfare costs are weighed against the benefit of treatment by trained professionals and veterinarians.

108. The counting of sea lice involves manual handling and inspection of farmed salmon. The process of counting to ensure statistical robustness was explored by Heuch et al (2011) which led to the Code of Good Practice and the Fish Health inspectorate to require a minimum of five fish per cage and twenty-five per salmon farm. However, for increased lower numbers of permitted sea lice, additional sampling of stocked fish may be required, Mes et al (2024) , which in turn requires more handling. The RSPCA identifies handling as a welfare issue for farmed salmon and handling must only occur when necessary. Handling of salmon has been attributed to between 13.6% - 16.6% of mortality on salmon farms between 2018-2020, greater than the on-farm mortality attributed to sea lice over the same period ranged from 7.5% - 13.4% . Not only does handling cause direct mortality, the starvation of fish in advance of in feed treatment has a subsequent diminishing effect of growth rates of salmon (Wolde et al 2022) . As an economic pest control measure, increased infeed treatment may not be a cost-effective measure given the starvation effects on farmed salmon (Boerlage et al 2024).

109. All sea lice treatments require fish handling, crowding, and transfer procedures that represent significant acute stressors, all of which the RSPCA identify as welfare concerns. The netting and pumping processes expose fish to elevated densities, leading to scale loss, fin damage, and elevated post-treatment mortality. Frequent repetitive handling is likely to chronically stress and compromise the animal's physiological and immune status towards a higher risk of secondary infections.

Treatment Efficacy and Resistance

110. Aldrin et al (2023)^[59] report that infeed treatments with emamectin benzoate are estimated to kill around 35% of the sea lice, whilst bath treatments with hydrogen peroxide kill around 74% and pyrethroids 50% of the sea lice, however when comparing against similar earlier analysis of sea lice treatments, there has been declining efficacy for hydrogen peroxide, pyrethroids and azamethiphos and necessitates more frequent treatments to maintain count thresholds.

111. Under a targeted treatment threshold regime there may be an increase in treatment frequency and potentially augmenting the loss of efficacy. As loss of efficacy develops, salmon farms may resort to using methods in their integrated pest management arsenal such as: higher concentrations of existing treatments with potential for increased toxicity risks, more frequent treatment cycles that compound stress effects, consider novel treatment combinations with unknown welfare impacts and multiple sequential treatments that prolong recovery periods.

Specific Treatment Welfare Impacts

Emamectin Benzoate

112. Emamectin benzoate (EMB), administered in-feed over seven consecutive days, accumulates in salmon tissues and can affect fish physiology beyond its antiparasitic effects. Different doses of EMB or repeated treatment did not affect feeding behaviour in salmon, however neurotoxicity was observed following multiple treatments (Whyte et al., 2019)^[60]. Feed withdrawal protocols, typically lasting 24-48 hours pre-treatment, ensure drug efficacy but create additional welfare concerns. Prolonged fasting can trigger aggressive behaviours, compromise immune responses, and delay recovery from treatment stress, particularly problematic during growth periods or in smaller fish.

Hydrogen Peroxide

113. Hydrogen peroxide treatments are an alternate method which is not without challenges. Other intervention strategies, such as thermal treatments, also present potential risk factors, which are evaluated in relation to the health and welfare benefits afforded by treatments delivered by trained professionals and veterinarians. Hydrogen peroxide toxicity increases with dose, exposure duration, and temperature, and can cause an acute stress response, oxidative stress, gill damage, respiratory acidosis and altered gas and ion exchange (Wood et al., 2021)^[61]. Salmon exposed to hydrogen peroxide treatments show time-dependent effects on physiological stress (glucose, lactate, and cortisol) and antioxidant enzyme expression in liver and gills (Vera & Miguard, 2016)^[62]. Use of hydrogen peroxide treatments has been reported as having increased mortality by 21% on salmon farms (Overton et al 2019)^[63].

Non-Medicinal Treatments

114. The shift towards non-medicinal treatments, driven partly by loss of efficacy concerns and environmental interactions, has introduced new welfare challenges. Thermal treatments had the greatest level of increased monthly mortality compared to the month before treatment, with 31% of treatments increasing registered mortality rates, followed by mechanical (25%) treatments (Overton et al., 2019). These known increased welfare risks to farm salmon would be incurred to meet sea lice count budgets to potentially benefit wild salmonid survival.

115. Other intervention strategies, such as thermal treatments, also present potential risk factors exposing fish to water temperatures of 28 – 34°C for up to 30 seconds have shown a significant increase in mortality as a function of temperature after treatment (14°C: 6.5%, 27°C: 5.3%, 30°C: 12.4% and 33°C: 18.9% mortality) (Bui et al., 2022)^64]. Critically, the group of fish that were not subjected to any treatments had no mortality throughout the experimental period. Eye damage was more prevalent in warm water treated groups, thermal treatments have led to injuries including gill haemorrhage, scale and skin loss, haemorrhage and vacuolation of thymic tissue, degeneration of nasal epithelium and brain haemorrhage. Research indicates that temperatures used during thermal de-lousing (28 – 34°C) are most likely painful to the fish, with pain thresholds identified between 26°C and 28°C, below the temperatures used in thermal delousing treatments (Nilsson et al., 2019 ^66]).

116. Treatment stress results in production impacts that indicate compromised welfare. Treatment groups lost up to 4.6% of their body weight over approximately 50 days including two treatment applications compared to slight growth of control fish that did not undergo any procedures (Bui et al., 2022)^[67]. This growth suppression reflects the metabolic costs of repeated stress responses and reduced appetite. Alternatively, the industry response has included earlier harvesting to avoid repeated treatments, with early harvests being a response by salmon farmers to increased risks of sea lice treatments, choosing to slaughter the fish rather than performing another delousing (Barrett et al., 2022)^[68].

117. The deployment of cleaner fish, particularly wrasse species, has emerged as a promising biological control strategy that exploits natural predator-prey relationships to reduce sea lice burden on farmed salmon (Torrissen et al., 2013). Wrasse demonstrate selective feeding behaviour that targets sea lice, providing continuous control pressure that complements periodic medicinal treatments whilst reducing overall therapeutic requirements.

118. Selective breeding programmes offer an alternative long-term potential for developing salmon strains with enhanced resistance to sea lice infestation, with heritability estimates of 0.3 confirming the genetic basis for individual variation in susceptibility (Gharbi et al., 2015)^[69]. Population genetic modelling suggests that 10 generations of selective breeding could substantially reduce medicinal treatment requirements and potentially eliminate the need for therapeutic intervention under optimal management conditions.

Broader Environmental Implications

119. The use of cleaner fish (wrasse and lumpsuckers) has wider ecological implications as currently some 21 tonnes of cleaner fish are cultured annually (Munro, 2023)^[70] whilst almost 80 tonnes of wrasse are harvested per year (Scottish Government, 2024)^[71] for use in salmon farming. Productivity and susceptibility assessments has demonstrated that ballan wrasse (Labrus bergylta) and cuckoo wrasse (Labrus mixtus) have the potential to be vulnerable to overexploitation by the wrasse fishery (Pritchard et al 2025)^[72] as well as two bycatch species viviparous eelpout (Zoarces viviparus) and spurdog (Squalus acanthias). However, work by Phillis et al (2022)^[73] determined that when considering the impacts of farmed lumpfish, farmed wrasse, and fished wrasse to calculate the environmental footprint of these Norwegian biological sea lice treatments, per ton of salmon produced, they found that wrasse fishing generates considerably lower impacts than farmed lumpfish and farmed wrasse.

120. Whilst not directly affecting farmed salmon welfare, repeated medicinal treatments contribute to pharmaceutical accumulation in marine sediments that may alter the environment around salmon farms. Negative effects of the sea lice therapeutant emamectin benzoate at low concentrations on benthic communities around salmon farms have been documented (Bloodworth et al., 2019)^[74]. These risks are managed through SEPAs regulation for environmental compliance in this matter.

121. A further potential unintended consequence is that medicinal sea lice treatments are seen as unfavourable by the public (Zhou et al. 2024) ^[75]. Targeting site sea lice limits through additional medicinal management could inadvertently lead to diminished perception of the industry from the public.

The Fish Welfare Paradox

122. Sea lice count limits have the potential to support wild fish populations. However, they may inadvertently compromise farmed salmon welfare through treatment-induced stress, accelerated resistance development, and suboptimal intervention timing. Furthermore, there are the possibilities for increased environmental degradation as well as increased dependency on cleaner fish.

123. Management regimes will require an acceptable compromise between industry functioning and impacts on ecosystems, but this compromise must also encompass the welfare of farmed salmon . The evidence reviewed suggests that current count-driven systems, whilst serving legitimate regulatory functions, require welfare-sensitive modifications, such as reducing biomass, to avoid undermining fish health.

124. Management approaches should integrate count monitoring with direct welfare assessment, employ adaptive thresholds that account for fish physiological and health status, and prioritise prevention strategies that reduce treatment dependence. There must be an acknowledgement that regulatory compliance and welfare optimisation may not always align, necessitating management flexibility that serves both ecological protection and animal welfare objectives.

Annex B5. SLRF Methodology Assessment

Issues Addressed in This Section:

• Standstill conditions based on flawed methodology
• Compound conservatism in risk assessment parameters
• Use of unvalidated conceptual model of lice attachment
• Reliance on Scottish Shelf Model with aggregated wind data
• Black box nature of decision-making process
• Lack of transparency in how data feeds into framework
• Site-specific risk assessment inadequacies
• Penalisation of operators with good track records

125. The claim that "standstill conditions are based on flawed and irrelevant methodology" represents a challenge to SEPA's regulatory framework for managing sea lice interactions between salmon farms and wild salmonids. This discussion examines the evidence base supporting and potentially undermining this methodology through critical analysis of the available documentation.

Annex B5.1 Methodological Foundations

Evidence Supporting the Methodological Approach

126. The standstill methodology is based on scientifically justified principles:

127. The exposure threshold of 0.7 sea lice per m² days exposure threshold is demonstrated through convergent validation. Norwegian research observed infestation pressure provided empirical validation of a similar order (Bui et al 2024), while Scottish modelling (Murray & Moriarty^[76]) and experimentation (Ives et al 2024) using infection rates, sea lice development, and post-smolt growth parameters suggested similar values. This convergence across different methodological approaches strengthens confidence in the threshold's validity. Nevertheless, sensitivity of SLRF outcomes to variation in exposure threshold has not been explored.

128. The underpinning modelling framework incorporates hydrodynamic models, particle tracking, and virtual salmon post-smolt migration which has been considered extensively by Scottish and international expert groups. The translation from validated models provides a foundation for applying the thresholds to the framework Salmon tracking data from the West Coast Tracking Project adds empirical grounding to the virtual post-smolt swim-speed component.

129. The methodology avoids regulatory overreach by applying standstill conditions only to salmon farms categorised as medium to high risk (61 of 164 salmon farms, 37% in SEPAs December 2023 response documentation), exempting lower-risk operations entirely. This proportionate approach aligns with established principles of environmental regulation.

130. The no-deterioration principle has established precedent in environmental regulation. SEPA's approach mirrors practices where new regulatory coverage is introduced to existing operations - initially requiring maintenance of current performance before considering active improvement measures. This staged approach provides regulatory certainty whilst building the evidence base for adaptative future action.

131. SEPA is publicly committed to methodological transparency through sharing of model files, having undertaken extensive stakeholder consultation, and commitment to adaptive management as part of the framework. The acknowledgment of uncertainties and plans for iterative improvement suggest that there is scope for amendments to be brought forward to the process. However, as it stands, the documenting of the SLRF is not comprehensive nor collated in an easily available manner.

Methodological Concerns

132. Several aspects of the modelling approach present concerns:

133. By their nature, models require simplification and approximation. The screening models employ simplifications that may introduce uncertainties. Each of the parameters described in Figure 2 will have a statistical range associated with the point parameter estimate. These include, for example, constant post-smolt swimming speeds (12.5 cm per second), average meteorological conditions, and uniform sea lice mortality rates (17% per day). Real-world variability in these parameters could substantially affect model predictions.

134. For example, the application of a single threshold (0.7 sea lice per m² days) across variable environmental conditions and potentially different salmon populations may be overly simplistic. The SEPA publications provide limited discussion of threshold uncertainty bounds or environmental context dependencies.

135. There is the potential for methodological translation errors in the conversion from FVCOM to MIKE 3 modelling systems, whilst maintaining similar results, this represents a departure from the original calibration and validation. SEPA acknowledges this limitation but provides insufficient detail on validation of the translated models against observational data.

136. According to consultation responses, the historical data reveal substantial gaps in sea lice counting, with "a large proportion of farms [...]missed providing counts in at least 1 week" during critical periods. Basing regulatory limits on incomplete datasets may compromise both the accuracy of baseline assessments and the validity of derived limits. There was also the concern raised relating to whether a minimum of three years data are representative. In reality SLRF uses a minimum of four years (not three as expressed), but this still discounts the possibility of on-site long-term variability in observations.

137. The use of estimated rather than actual fish numbers (calculated from maximum permitted biomass rather than reported stocking) introduces additional uncertainty into both risk assessment and limit-setting processes. Although it would be impractical to provide continual real-time actual numbers of onsite fish. There is limited validation of fish numbers from biomass across all the Wild Salmon Protection Zones and therefore spatial transfer of model parameters and thresholds developed in specific locations will incur additional uncertainty.

138. The methodologies focus on a representation of the exposure being used in the assessment without considering the status of wild salmon populations, resilience, or cumulative stress factors. For example, a resilient population with no major threats from predation may be more tolerant to exposure. This approach may lead to either over- or under-protection depending on population-specific vulnerabilities. This represents another uncertainty.

139. There is a reliance on sentinel cage studies for model validation and although it is suitable for indicative corroboration, it provides limited temporal and spatial coverage in comparison to the vast number of migratory routes migratory salmon could take. The proposed validation approach may not capture the full range of environmental conditions affecting salmon farm derived sea lice - wild salmon interactions.

140. However, the broad SLRF principles (notwithstanding the discussed methodological limitations, monitoring approaches, documentation/communication, the undefined commitments to adaptive management processes for assessment and monitoring refinement) are suitable for use in the relative risk assessment based on the adopted modelling processes.

Assessment Summary

141. The methodological concerns above list some of the limitations of complex environmental modelling: reliance on simplified point assumptions, incomplete validation across variable conditions, and data quality constraints. These limitations are neither unique to this framework nor necessarily limiting to its application, but they do point to the need for a more formal uncertainty analysis and continued framework refinement.

142. The characterisation of standstill conditions as "flawed and irrelevant" appears overstated based on the available evidence. Whilst the methodology contains uncertainties and limitations typical of complex environmental modelling practices, it has strong scientific foundations. The approach is more appropriately characterised as "precautionary but uncertain" rather than fundamentally flawed.

143. However, scientific concerns exist regarding model description, lack of a single reference guide/manual containing the documentation of assumptions and limitations of its simplifications, influence of data quality, and validation and scope of refined models (as they are developed). These limitations suggest the need for continued monitoring, validation, and refinement rather than wholesale rejection of the approach.

144. The most constructive critique would focus on specific technical improvements - enhanced validation studies, uncertainty quantification, population-specific thresholds, and expanded monitoring programmes - rather than fundamental methodological rejection. Such targeted improvements could further strengthen the scientific basis whilst maintaining the regulatory framework's core structure and objectives.

145. Ultimately, the standstill methodology represents a reasonable application of available science to a complex regulatory challenge, though it would benefit from continued development and validation as more data become available.

146. Onsite variability needs to be accounted for, if the principle of no deterioration from historical observations is the underlying factor in determining the permitted sea lice budget on salmon farms.

Annex B5.2 Site-Specific Risk Assessment

147. Building on the methodological foundations discussed above, this consideration focuses specifically on whether standstill conditions adequately capture site-specific environmental risks and wild salmon impacts, examining aspects not previously covered.

Geographic Heterogeneity Issues

148. There is the potential for geographic heterogeneity issues, such as bathymetric and oceanographic variation. The SEPA documentation presents differences in how the framework handles varying coastal environments. For instance, SEPA acknowledges that "sea lice from farms in some locations disperse out of the WSPZ before reaching the infective stage". It transpires that the standstill methodology does differentiate limits based on these dispersion characteristics, but available documentation is unclear on precisely how.

149. For some WSPZs on the west coast, the resolution of the hydrodynamic models is limited. This technical limitation makes it difficult to determine site-specific risk characterisation, as coarse model resolution cannot capture fine-scale oceanographic features that may substantially influence sea lice transport and retention.

150. Unlike the general threshold discussions covered previously, the site-specific criticism highlights that standstill conditions apply uniform performance-based limits regardless of the specific local wild salmon populations potentially being exposed. SEPA’s documents describe situations where "more than one river catchment's population of wild salmon emigrate through some WSPZs". Whilst there is no specified mechanism for adjusting limits based on different populations migrating through a single WSPZ, the regulator has powers to review and vary permit conditions and farm operators can apply for permit variation. The evaluation process of such application is unclear. If SLRF is to be adaptive then locally if historic limits are shown beyond reasonable doubt to be causal in an ongoing wild population decline (noting that the time scale of decline may not match the time scale of limit evaluation) then SLRF should have a way to mitigate that.

Migration Timing and Cumulative Effects

151. There may also be migration mismatches. The applied week 12 -22 control period (approximately 16th March to 30th May) may not align with actual migration timing for salmon from different river systems. Observations of salmon migration timing in Scotland (Malcolm et al 2015)^[77] indicate timings of mid-April to mid-May (not accounting for coastal residency). Earlier or later migration windows may be inadequately protected by standstill conditions calibrated to average timing patterns.

Environmental Context and Historical Performance

152. SLRF overall show no evidence of incorporating local habitat quality factors that might influence salmon vulnerability to sea lice impacts. For example, areas with degraded freshwater habitat might support salmon populations with reduced resilience, warranting more stringent sea lice management, but the performance-based approach cannot capture this variation. Whilst the control period is standardised, local environmental conditions (e.g. temperature, salinity gradients, predator/prey abundance) that influence both sea lice development and salmon vulnerability vary substantially across sites and seasons. It is noted that a future phase of SLRF implementation may rectify this.

153. Standstill limits are derived from an analysis of a salmon farm’s performances in managing sea lice in previous years. A salmon farm with historically relatively poor sea lice management in a high-capacity environment might receive lower limits than a salmon farm with good historical performance in a vulnerable area. Even though this is a simple consequence of SLRF logic, it is still perceived by some to be unfair.

154. The four-year approach for setting standstill limits may perpetuate site-inappropriate management levels as the limits are set based on historical performance. As environmental conditions (or wild salmon population status) changes, the historical performance may not reflect current actual site-specific risks, and in that sense the historic management of lice to historic levels could be labelled as site-inappropriate.

Trade-offs and Conclusions

155. Site-specific risk reflection might require more responsive limit-setting mechanisms that could adjust, for example, real-time environmental monitoring (e.g. temperature, salinity, currents), wild salmon population monitoring indicators, neighbouring salmon farm operational status and sea lice loads.

156. However, attempting to have site specific adaptability creates administrative complexity in that site-specific limit setting would require more complex administrative systems and potentially site-specific expertise that may exceed current regulatory capacity. It might also create unreasonable reporting burdens on producers; comprehensive site-specific risk reflection would demand extensive local environmental and biological monitoring that may be cost-prohibitive across all salmon farm sites.

157. There are also practical concerns in having more granular approaches. The current understanding of site-specific sea lice-salmon interactions may be insufficient to support highly differentiated regulatory approaches without substantial additional research investment. There is also the potential issue that wild salmon population responses to management changes may occur over timescales that exceed regulatory decision and response cycles, complicating site-specific adaptive management.

158. The criticism that standstill conditions do not reflect actual site-specific risks contains significant validity beyond the general methodological concerns discussed above. A performance-based approach to limit-setting, combined with standardised implementation across diverse environments, creates a disconnect between regulatory requirements and site-specific environmental realities.

159. However, the criticism also highlights the trade-off between scientific precision and administrative feasibility. Perfect site-specific risk reflection may be neither technically achievable nor regulatorily practical given current scientific understanding and regulator capacity. The gradual approach SEPA is implementing of increasing refined modelling incorporating site-specific elements which prioritises high-risk locations for detailed site-specific assessment whilst maintaining standardised approaches for lower-risk sites, accounts for this trade-off issue.

160. The criticism ultimately points to a limitation of the current approach rather than a fatal flaw, suggesting that the adaptive nature of the framework rather than replacement of the regulatory framework, is appropriate.

Annex B5.3 Data Quality and Statistical Issues

161. The historical sea lice count data forming the basis for standstill limits suffers from data quality issues. The SEPA consultation documents indicate that "a large proportion of farms, excluding those that were fallow, missed providing counts in at least 1 week with many not providing counts for 2 or more weeks" during critical management periods. This results in incomplete datasets upon which regulatory limits are then based. The quality problems extend beyond missing data points. Salmon farms report the top two reasons for non-reporting as "withdrawal period prior to harvesting" and "veterinary advice" - suggesting that data gaps may be biased toward periods of either high sea lice loads (requiring treatment) or compromised fish health. This creates a dataset that may have a systematic low bias.

162. The SEPA response to consultation indicates that sea lice limits will be set for existing salmon farms from statistical analysis of their performance over the recent past. However, there is no published detail of what this statistical approach entails.

163. SEPA have advised that sea lice limits will be set for existing salmon farms from statistical analysis of their performance over the recent past. Three scenarios are applied depending on data availability.

164. SEPA's standstill limits are derived using the 95th percentile of historical weekly total adult female sea lice numbers during the control period (weeks 12-22). For salmon farms with sufficient data (≥6 weeks in each of 4 Spring periods between 2018-2023), limits are calculated by identifying the 95th percentile of total sea lice, determining what sea lice-per-fish rate would produce this total at maximum stocking, adding 0.2 to this rate (capped at 2.0 sea lice per fish), then multiplying by maximum fish numbers to establish the 4-week rolling mean limit. A single-week limit is set at 4 times this value.

165. Salmon farms with insufficient data but ≥4 years operation use benchmark comparisons from similar salmon farms in the same area. Salmon farms operating <4 years use area-wide averages of calculated limits, adjusted for their peak fish numbers. Operators have been consulted on benchmark selection in both scenarios.

166. This percentile-based approach addresses any potential statistical concerns about frequent non-compliance. Salmon farms would historically have exceeded the 95th percentile in approximately 5% of weeks, meaning the limits permit operation within normal historical performance ranges rather than requiring performance substantially more stringent than historical norms.

167. It is unknown whether the reliance on the period between 2018-2023 captures longer-term environmental variability affecting both on-farm sea lice populations and wild salmon vulnerability. If this period exhibited atypical conditions (e.g. temperature patterns, treatment efficacy, reporting practices), derived limits may not reflect appropriate long-term thresholds. However, practically this is not possible to determine as mandatory weekly reporting is a relatively recent introduction^[78].

168. The requirement for only 6 weeks of data per year (approximately 55% coverage of the 11-week control period) means limits may be based on incomplete seasonal representation. Combined with the historical data quality issues noted above where "a large proportion of farms missed providing counts in at least 1 week with many not providing counts for 2 or more weeks", this creates uncertainty about whether the 95th percentile accurately represents typical high-sea lice conditions or is artificially influenced by systematic data gaps.

169. The 0.2 upwards adjustment may compensate for potential under-representation of low-count weeks but provides no mechanism for addressing other sources of baseline uncertainty, such as year-to-year environmental variation, changes in treatment protocols, or shifts in wild salmon migration timing that might alter the relevance of historical performance to current conditions.

170. Missing data can introduce biases, as discussed in previous sections. Systematic gaps in historical data collection may skew historical performance although the additional concession of 0.2 sea lice may allow for this. Missing counts during high sea lice periods would artificially lower calculated averages, creating even more stringent and unrealistic limits. However, this reflects operational reporting , delaying implementation of salmon farm site limits until a complete perfect time series is established is impractical as demonstrated through the historical reporting and would result in a limit never being determined.

Environmental Influences on Historical Data

171. There is also the need to consider environmental influences on on-farm historic sea lice populations. Modelling studies have demonstrated that sea lice populations are more sensitive to environmental parameters (such as Temperature and Salinity) than biological parameters (Rittenhouse et al 2016^[79]) . This is demonstrated by recent work in Canada which showed that in the absence of salmon farms, sea lice prevalence on wild salmon is at elevated level compared to when salmon farms were located within the system (Jones et al 2025). It may be the case that historic sea lice counts were reported during environmental periods conducive or restrictive to on-farm sea lice population growth and thus applying limits on these anomalous periods may be inappropriate. However, the continued reporting of sea lice counts will enable the development of a continued series of counts, but this will inherently be biased by the new regime (although the historic data is similarly biased by the previous sea lice counting regime). The adaptive nature of the framework should account for emerging count status as the time series develops.

Annex B5.4 Track Record Penalisation

172. The framework states that standstill limits "will be derived from an analysis of their farms” performances in managing sea lice in previous years" and "will be based on available data for at least 4 years."

The Performance Paradox

173. This approach creates an environment in which the limits placed on different salmon farms with differing track records potentially confer benefit to weaker historically performing salmon farms over stronger performing ones. As an example, consider a pair of hypothetical salmon farms, with different historical performance, both consented to stock 1000 tonnes. During the historical reporting period they both stocked 100,000 fish (corresponding to ~400 tonnes). Salmon farm A, which consistently maintained 0.1-0.2 adult female sea lice per fish over four years could face a standstill limit around 0.15, whilst salmon farm B with variable performance ranging from 0.3-0.9 sea lice per fish, for example, 0.6 sea lice per fish. In this hypothetical example, Farm A would have a farm limit of 15,000 sea lice and Farm B 60,000 sea lice. The better performer faces four times tighter regulatory constraints despite demonstrating historically better environmental stewardship. The historically poorer performer (Farm B) may now be able to increase their total number of fish more easily than the better performer (within their biomass limit), by adopting practices already pioneered by the better performer. Should Farm B adopt some of the practices of Farm A and be able to reduce their sea lice per fish count to 0.4, they would be able to raise their onsite number of fish to 150,000 and remain well within both their consented biomass and their farm sea lice limit yet still have greater sea lice per fish, and greater input of sea lice from the farm. Wild salmonids are no better protected as the number of sea lice entering the environment remains the same, yet the poorer historically performing farm is able to increase its production by 50%. Early adopters, such as Farm A, feel unrewarded and now disadvantaged. This may not be the regulators’ concern, but it is a potentially perverse outcome

174. It may be the case that some “better performing” salmon farms are located in more optimal environmental surroundings. Equally, it could be the case that some operators invested in improved sea lice management systems, staff training, or operational protocols during the historical reference period, and will now face regulatory disadvantages compared to operators who made minimal investments in environmental performance.

175. The SEPA documents demonstrate substantial variation in sea lice management performance across the sector. The observation that "close to 60% of reported weekly averages were less than or equal to 0.2 adult female sea lice per fish" in 2021-2022 indicates significant performance differentiation that would translate into widely varying standstill limits. The Loch Fyne and Loch Linnhe examples show how the same salmon farms can contribute differently to environmental risk depending on their sea lice management performance, yet the framework would fix these historical differences rather than rewarding continued improvement.

Contradictions and Perverse Incentives

176. The standstill element of SLRF adheres to no deterioration principles. Each salmon farm must maintain its current performance level regardless of how that performance compares to others. From this perspective, the framework prevents environmental degradation rather than promoting improvement. Whilst this does satisfy the aim of “no deterioration” the Scottish Government’s Biodiversity Strategy which states: “ By 2045, Scotland will have restored and regenerated biodiversity across our land, freshwater and seas” ^[80]. The relative risk assessment matrix suggests that farm-specific limits reflect local environmental capacity rather than pure salmon population performance outcomes. Salmon farms in high-risk locations appropriately face tighter controls regardless of their historical performance quality. The full implementation of SLRF (not just standstill limits) needs to be implemented as soon as possible.

177. The Performance-based limits provide operators with clear expectations based on their demonstrated capabilities rather than subjective assessments of what constitutes "good" performance. This approach ensures that historical good performers cannot deteriorate whilst poor performers cannot worsen, potentially representing a regulatory floor rather than ceiling approach.

178. The standstill framework acknowledges that "a very small number of farms contribute a large proportion of exposure" in high-risk areas. Environmental protection logic suggests tighter controls on salmon farms that historically contributed most to environmental risk, not on salmon farms that demonstrated proficient sea lice management. Until the full SLRF approach is implemented there may remain some instances to a reversing of this logic.

Annex B6. Monitoring and Adaptive Management Gaps

Issues Addressed in This Section:

• No framework for monitoring success of SLRF
• Expensive and problematic sentinel pen validation methods
• Lack of real-time data integration
• No assessment of framework effectiveness
• Missing adaptive framework for incorporating new evidence

179. A weakness of the SLRF is the absence of clearly defined mechanisms for evaluating its success and adapting to new evidence. While the framework establishes regulatory limits and monitoring requirements, it provides insufficient detail on how effectiveness will be measured and how the framework will evolve based on observed outcomes.

Annex B6.1 Sentinel Cage Validation Issues

180. Sentinel cage studies represent a key validation tool for the SLRF modelling framework, but their implementation raises several concerns. The Norwegian experience demonstrates that sentinel cages used fish 4-5 times heavier than wild smolts without size correction, whilst data collection occurred in areas with optimal sea lice survival conditions rather than representative spatial sampling (Van Nes et al., 2025). These methodological limitations must be considered if sentinel cage data are used to validate refined modelling.

181. The SEPA documents indicate that refined model validation through sentinel cage studies will be prioritised for only 8 WSPZs initially, meaning standstill conditions for the majority of sites will remain based on screening models (who’s sensitivities have not been fully explored). This limited initial spatial coverage undermines confidence in the site-specific appropriateness of regulatory limits across the rest of the framework.

182. Sentinel cage deployment is expensive and logistically challenging, limiting the temporal and spatial resolution of validation data. The vast number of potential migratory routes salmon could take means that even comprehensive sentinel cage networks cannot capture the full range of environmental conditions affecting farm-derived sea lice - wild salmon interactions. However, it is currently the only suitable method to establish motile stage sea lice resulting from environmental exposure, and the limitations of such approaches require acknowledging in any validation programme.

Annex B6.2 Success Measurement Framework

183. The framework's stated objective is to support the survival of wild Atlantic salmon and sea trout, yet it lacks defined metrics for evaluating whether this objective is being achieved. The Norwegian TLS experience is instructive: eight years of operation reduced salmon farm sea lice levels but without corresponding improvements in wild salmon populations, revealing that reducing salmon farm sea lice counts does not necessarily translate to wild salmon benefits over that time frame.

184. The SLRF appears not to currently incorporate detailed proposals for the evaluation of survival of wild salmon, although monitoring is referred to. The SLRF framework requires a process for measuring its success or otherwise. The purpose of the framework is to support the survival of wild Atlantic salmon and sea trout; therefore, an evaluation process is required.

185. SEPA has indicated that they have determined the types of monitoring and modelling programmes to assess adverse impacts and framework effectiveness, set a 5–6 year timescale for using results, and adopted a generalised random tessellation stratified survey (GRTS) design for juvenile salmon and trout monitoring. The recommendation is to publish these details, including metrics, statistical approaches, and governance arrangements.

186. Without published success metrics, the framework cannot determine whether:

• Wild salmonid survival is not deteriorating, or indeed improving in WSPZs
• Site-specific limits are having the intended protective effect

Annex B6.3 Adaptive Management Mechanisms

187. The framework includes provisions for annual risk reassessment but provides no mechanism for mid-season adjustment based on emerging evidence or changing environmental conditions. Real-time data on wild salmon stress indicators, unusual environmental circumstances (jellyfish blooms, algal blooms, temperature anomalies), or neighbouring salmon farm operational changes cannot be incorporated into regulatory decisions under the current standstill approach.

188. Some in the industry expressed the concern that there is no clear pathway for salmon farms to demonstrate improved performance and "get off standstill" conditions. This is an issue of regulatory approach that should be carefully explored by dialogue between industry and regulator.

189. A documented description of the process of moving from EMPs to SLRF would aid understanding of how, if any, EMP monitoring and wild catch data supported the development of SLRF. The transition from Environmental Management Plans to the SLRF represents a potential opportunity to evaluate what monitoring approaches proved most informative and which metrics best correlated with wild salmon outcomes.