Response to Stige LC, Vollset KW, Diserud O, et al. ‘Comment on Van Nes, Imsland and Jones “Salmon Lice Biology, Environmental Factors, and Smolt Behaviour With Implications for the Norwegian Salmon Farming Management System: A Critical Review”’

The Review [1] demonstrates inaccuracies (uncertainties) in the assumptions and practices in the current aquaculture management system in Norway (the Traffic Light System [TLS]). It includes knowledge relating to lice biology and smolt behaviour which is either not included in the TLS or implemented only to a limited extent. The Review [1] shows how the inclusion of this knowledge would reduce uncertainty and improve the accuracy of the system thereby increasing the value of the TLS to wild salmon conservation and aquaculture management.

As is evident from the referenced literature [1], the suggested inclusion of knowledge and needed improvements to the system are mainly based on substantial and independent research. Key points of uncertainty are the same as described in the annual assessment reports [2, 3] by the expert group (EG), consisting of several of the same authors as listed for the Comment [4]. Key points of critique and suggested improvements also agree with the evaluation by experts appointed by the Norwegian research council [5]. Also, the authors would like to emphasise the Review [1] was subject to extensive scientific peer review prior to publication. Hence, our main findings are scientifically justified and the claim in the Comment [4] that the Review [1] appears ‘biassed’ in evidence and is based on apparent ‘misunderstanding of the literature and the methodology’ is neither scientifically justified nor does it contribute to a constructive scientific debate.

In relation to smolt migration, the Comment [4] describes the use of extensive data and claims the data referred to in the review [1] are ‘also used by the EG’ to estimate migration times. However, the modelled migration time applied by the EG is derived from selected data limited to observations of 25% smolt emigration [6] and from duration of the migration period being set/fixed to 40 days for all rivers [2, 3]. Whereas the data presented in the Review [1] are based on acoustic tracking of migrating smolt including both progression of emigration (fjord entrance) and progression throughout fjord and over several years and fjords and demonstrate that the majority of the smolt migrate earlier and migration period is significantly shorter in these watercourses. Contrary to the claim in the Comment [4], these extensive data are not included in estimates of migration time in the TLS. Therefore, the Review [1] argues that the period adopted by the TLS both for retrieving observational- (calibration-) data and for modelling effects from lice mortality extends too long into season when lice levels are higher than the levels to which migrating salmon are realistically exposed. The significance of the timing and duration of migration used by the TLS to estimate smolt mortality and as a key source of uncertainty is clearly stated in the annual EG reports [2, 3] and is also emphasised in the article that forms the basis for the applied/modelled migration time [6]. These considerations support our claim that migration time as currently applied for assessment of lice induced mortality is a key source of unreliability in the TLS.

The Review [1] demonstrates that input data to all models, namely estimates of lice larvae from fish farms are likely overestimates, since these estimates exclude all reducing biological and ecosystem effects on lice on the adult female-, egg- and larval stages described in literature. This in turn can affect the outcome of the models. The Comment [4] states that the Review [1] ‘fails to mention that such possible overestimation is unlikely to bias mortality estimation, as models scale [calibrate] the calculated infestation pressure to observed lice counts on salmon smolts’. Firstly, this claim is incorrect since quite the contrary, the Review [1] elaborates on this and highlights calibration (observational-) data as a point of uncertainty. Secondly, if both input data to the models (estimated lice larvae) and the model calibration data (lice counts on fish) have several sources of unreliability, any counteracting effect as claimed in the Comment [4], would be negated. Uncertainty in the TLS model products in relation to calibration data is also highlighted by the external evaluation committee, which state that some of the calibration data ‘demonstrate high levels of variability and uncertainty’ and that ‘this reliance on calibration data potentially impairs the quality of model products’ [5].

The Review [1] presents a substantial number of experimental studies that demonstrate a significant reduction in post attachment lice loads prior to development to the more harmful pre-adult stages. Since observed and modelled lice loads form the basis of mortality estimates in all seven assessment end points of the TLS, the review argues this is crucial knowledge of lice biology that must be considered for accurate assessment of lice-induced mortality. The Comment [4] claims that the Review [1] ‘under communicates the uncertainty and over-assesses the most likely magnitude of such loss’. This claim contradicts the EG own practice/scientific opinion, since the EG itself implemented a 40% correction/reduction in the NVI model only recently as an improvement of the model (thus acknowledging the need for correction of the model to increase accuracy) and also state that this change/correction will result in a 40% reduction in mortality estimates for this model. Since reduction in lice load is corrected for in only two of the models, whereas in the remaining five assessment end points of lice-induced mortality assumes that 100% of the infesting lice will stay attached/survive, the claim made in the Review [1] stands correct; the current practice will contribute to overestimates of lice induced smolt mortality.

The Comment [4] hints at the presence of uninformed conflicts of interest in the Review [1]. This is a serious and unwarranted allegation consistent with the statement that the authors of the Review [1] have ‘misunderstood’ the literature or methodology. All relevant potential conflicts of interest have been honestly and openly stated in the process of publication. There are no competing interests that influence the scientific content of the published review.

Despite the inclusion of an extensive and independent knowledge base in the Review [1] as well as the alignment of its conclusions with the EG and the external evaluation committee, the Comment [4] argues against all proposed constructive critiques and relevant knowledge that could improve accuracy in key areas of the TLS. It is important to recall that the Review [1] reflects the recommendations of five reviewers.

Solveig van Nes: conceptualization; investigation; writing – original draft; methodology; writing – review and editing; formal analysis; validation. Albert Kjartan Dagbjartarson Imsland: conceptualization; validation; writing – review and editing, review final draft. Simon R. M. Jones: conceptualization; validation; writing – review and editing, review final draft.