A. General assessment.
Dr M Kent eloquently illustrated this problem at the Vancouver 2003 meeting on sea lice by pointing out the strong statistical association between the daily consumption of ice cream in the US and deaths of teenagers by drowning.
Models are most effective in indicating areas where additional scientific data to refine hypotheses are required, rather than being useful in producing conclusions such as is being proposed in this paper. In addition, the validity of conclusions made from statistical analysis is directly related to the quality and completeness of the data being analysed. However, the data presented in the report by Krkosek et al. is particularly weak and highly relevant publicly available data on sea lice in salmon farms and wild salmonids in BC have either been missed or ignored.
There are several important areas in the study where the science is sufficiently compromised to cast doubts on the conclusions being made:
Several assumptions that are made are critical to the model being employed, but either do not have supportive scientific evidence or there is actual contradiction in the scientific information available.
Sources of lice infection.
The statement on which the model is based Juvenile salmon --- first encounter infective copepodids that originate from two primary host populations : farm salmon and sympatric wild hosts assumes that there are no other major sources of infection. However, it is known and accepted by the authors, that Caligus clemensi has a wide range of fish hosts, some of which are known as permanent residents and occur in large numbers in inshore waters of BC. It can not be assumed that the majority of unidentified sea lice found in the study were not from such a source. Similarly, although the authors dismiss conspecifics as a possible source, they have not considered other possible salmonid hosts in the study area as potential sources of Lepeophtheirus salmonis.
The infection status of the farm(s) in the study area is (are) critical to the study. It is bizarre in the extreme to make conclusions on the transmission of sea lice from farm to wild salmon without including any information on the infection status of the farm involved at the time of study.
Similarly, the authors have failed to take account of available data on lice infections in farmed and wild fish in BC that are available to the general public in the BC MAFF and DFO websites during the period of their study, and possibly even in the same area as the study. The latter point is not possible to verify as the paper does not provide details on the location of the study.
The failure in the study to differentiate most sea lice found to the species level prevents even general conclusions being made on most of the data collected. These may or may not be on mixed sea lice species and if so, it is not possible to assign proportions of the different species present. The authors choose to group data on different lice species and different host species on the basis of similarities in host behaviour and parasite life cycle without full knowledge of these or the possible implications of even subtle differences between these on the patterns of infection.
It is justifiably stated that farms are a point source of potential sea lice infection but it is assumed, without any evidence (e.g. behaviour of sticklebacks) that wild hosts are a distributed source. Several European studies on Lepeophtheirus salmonis and Caligus have indicated discontinuous levels of infection in wild populations.
Dispersion of free living stages of sea lice and the use of advection-diffusion models to describe this. Many scientific studies have shown that the free living stages of sea lice do not behave as inert particles, but respond to a variety of stimuli. The consequence is a non-random pattern seen in the occurrence of larval stages and the process of infection, not described by such models.
The purpose of the Methods section in a scientific paper is to allow the study to be repeated and the results checked. Several critical pieces of information are missing from this paper.
The dismissal of the effects of a list of possible causes of variation in sea lice infection as trivial in favour of farm effects without proper evaluation of the completeness of this list, their potential interactive nature and the independent commonality of many to farms and to sea lice patterns of infection indicates an apparent lack of awareness of the authors of the complexity of infections in ecological systems. For example, the relationship could be investigated between the distance of the area of the farm in the study from the points of entry of smolts into sea water capable of sustaining infection in relation to the pattern of infection with Lepeophtheirus salmonis.
Only 12 parasites were identified as Lepeophtheirus salmonis from a total of 3645 parasite recorded in the study. As this species is the parasite specific to salmonids and causes disease on farms, either the scale of transfer is minimal or the development of the parasite to maturity and egg production (the identification feature used) does not occur on fish sampled in the study.
B. Specific comments on the contents of the paper
It is essential that the information used from quoted scientific papers is accurate. For example McVicar (1997, 2004) did not implicate farms in the infestations and collapse of wild sea trout and Atlantic salmon populations in Europe, but made reference to other publications where this was claimed. Both of these papers actually threw doubt on that claim.
Vertical transmission is a disease term normally used for transfer of infection from adults to offspring through eggs, not transfer between adults and juveniles through the environment. The authors should have used the term horizontal transmission.
The important statement that transmission of sea lice from conspecifics to juveniles did not occur until late July must be supported by published evidence. The occurrence and location of all potential sources of infection is central to this study including inshore feeding populations of any species of salmonid. The absence of information is not evidence for absence unless comprehensive studies have been undertaken. This has not been acceptable done throughout the year in BC and in particular for the period of the study by Krkosek et al.
Both DFO and BC MAFF published data on their web sites on sea lice infections from extensive studies in 2003, on wild and farmed salmonids in BC, i.e. the same Province being studied in this paper. The absence of any reference to this readily available and comparable data (in time and general area) either in the Introduction or in conclusions being reached in the Discussion is unacceptable. All relevant available information should be taken into account and the failure to use data not supporting a particular case is contrary to a scientific study seeking the truth. This and/or the failure of the study to collect their own data from farms leads to wide unsupported assumptions being made on the infection status of the farms at the time of their study that are critical to the conclusions being proposed.
An essential component of any scientific publication is that sufficient information is presented in the materials and methods to enable the study to be repeated and the results independently verified by members of the authors peer group. This is not possible in this paper with major gaps in details:
The exact location of the study area/migration routes sampled the position of sampling and the farms should be given. Ideally a map should be provided. Justification is required for industry anonymity being maintained as a reason for not providing study location details as no data are included from industry and they therefore have no justification to suppress other information. It should be clarified whether industry were asked and requested this anonymity.
Failure to identify the species of copepodid, chalimus and non-ovigerous stages to copepod species prevents comparison with DFO data from the same period in a farming area of BC. The data are meaningless unless it has been shown that the proportion of the different species remains constant in different parts of the sampling area and at different times of sampling. BC MAFF data from their website in the period 2003-05 indicates that adult Caligus are found only in small numbers on farmed salmon and can not be considered to be a major source of infection to wild fish in the area. Also, the DFO data have shown wild sticklebacks to be a major source of infection by this species.
Failure to measure each fish caught does not reveal the size structure sampled or an index obtained of time since entering sea water (and so exposure to infection)
As the two migration routes sampled share the same data landward of the salmon farm, only partially separate data sets were obtained i.e. there are only two sets for the 20 km before the farm. The implications of this should be discussed.
Statements are made on migration routes of juvenile salmonids in the study area e.g. ---migratory salmon indirectly passed within 7km --- of a second salmon farm without any supportive evidence in the scientific literature of the details of migratory routes of these salmonid species in the study region.
The failure to ---- sample for approximately 20-60 km of the migration routes ---- and to only take salinity data from around farm A raises questions why this decision was made and whether gaps in data, particularly salinity in relation to onset of infection, may have affected results.
In the Model the assumption being made that --- a farm is a point source of lice whereas wild hosts are a distributed source--- must be supported by evidence as many micro and macro infections in wild fish have highly geographically and temporally discontinuous distributions. This has even been demonstrated for Lepeophtheirus salmonis in Europe.
River estuaries have been shown to be a point source of infection to wild salmonids.
Several studies have shown that larval stages are not well mixed as being claimed by Krkosek et al. and even their limited powers of movement (in response to light, current and chemical stimuli) lead to non-random spatial and temporal aggregations. Advection-diffusion models are therefore not strictly applicable to dispersal of planktonic stages of lice. In the scientific literature, the typical pattern of infection found with sea lice is an overdispersion not poisson as being claimed by this study.
Similarly, some reports in Europe have indicated that sea lice infections in sentinel fish in the vicinity of farms can show an episodic (i.e. discontinuous) occurrence.
As the main conclusions being proposed in the paper depend totally on the validity of these assumptions which contradict current published knowledge on Lepeophtheirus salmonis, the authors need to provide appropriate scientific evidence to support their position.
The grouping of data on different lice and salmonid species for analysis on the basis of --- similarities in host behaviour and parasite life cycles --- must be supported by evidence to give confidence in this assumption. The life cycle of Lepeophtheirus salmonis is well known, but that of Caligus clemensi sufficiently less so to enable proper comparison to be made. Similarly, there is insufficient evidence on the behaviour of juveniles of the different species of salmonids in BC.
Only a total of 65 lice out of 5514 were identified to species level. However, the Model assumes a similar contribution of all the remaining 5449 from farmed and wild sources. This is contrary to the real data results posted by BC MAFF on their website which show that Caligus adults rarely occur on farmed salmon.
It is highly irregular in a results section that a conclusion is reached, namely that data can be subdivided into a farm and non-farm source. The lack of discussion suggests that the conclusion has been reached before consideration of the data obtained in the study.
It is not surprising that sea lice data are spatially structured as young salmonids enter the sea uninfected and infection accumulates with time, at least partially reflecting the time of residence in sea water. Without information on the size of fish sampled in relation to the position caught and the infection in individual fish meaningful interpretation of the data is not possible.
If distance from river/stream of origin is substituted for presence of the farm a similar conclusion of infection being related to sea water residence time could be made without involvement of a farm. The report does not discuss this possibility.
The study has omitted to assess the statistical significance of the differences found between samples from different areas, particularly in relation to the occurrence of copepodids in the area of the farm. The error bars shown in Figure 2 would suggest that there are no significant differences between most of the samples and that the differences being claimed are not real.
The presented list of possible unaccounted sources of noise variation in sea lice data can not be considered exhaustive as determinant factors causing variations in fish diseases, including other species of Lepeophtheirus, have been shown to be numerous, complex and also highly interactive. A superficial understanding of such relationships is shown by the authors when they consider that the cause of the signal they identified amongst in empirical data could only be due either to an aggregation of a wild population (then rejected)or to the presence of the farm. There are many possible influencing factors (e.g. size of wild post smolts/duration of exposure to infection, shoaling of juvenile wild fish) in addition to these listed.
The statement Given the paucity of confounding factors and abundant sources of noise ------ unaccounted sources of variation must be trivial relative to the effect of the salmon farm. has been made without any proper consideration of these and their dismissal is highly speculative and unsupported:
No information has been presented on the occurrence of ovigerous lice on the farm immediately prior to and during the study period that could cause a farm effect on lice infection levels in the local wild fish population. The source(s) of lice in the area therefore remains speculative.
In the study, only 12 sea lice were identified as Lepeophtheirus salmonis, the predominant species occurring on farmed stocks (BC MAFF website). As it is not possible to apportion the proportion of different sea lice species present in the samples but not identified, it is possible that infections were derived mainly or even fully from wild sources. Importantly, no information is presented on the local occurrence of sticklebacks and other sea lice hosts in the area (the DFO website found sticklebacks to be abundant in salmon farming areas of BC in 2003 and to carry high levels of Caligus clemensi infections).The best that can be said from the data presented in the study is that sea lice infections are increasing in magnitude in the same area as the farm is located.
There has been no consideration of the fact that the farm shares a potentially wide series of environmental variables with the population of wild fish being studied several of which could independently influence lice levels on wild fish resulting in correlations with the location of the farm without there being a cause-effect relationship between the two. For example if the data were analysed against distance from fresh water origins of the young fish (as an index of length of exposure to infection), an alternative conclusion independent of the farm could be reached.
The series of classic disease epizootic studies by Anderson and May (one of which is quoted) indicate the expectation of a series of peaks and troughs in the level of an infectious new disease occurrence in an animal population, these decreasing in magnitude with the length of time of disease occurrence. This pattern is similar to that shown in the data presented in this paper.
The comparison of sea lice levels in different areas is not possible without detailed knowledge of the factors causing disease variations. Many fish disease studies have demonstrated this including international studies by the International Council for the Exploration of the Seas (ICES)). Different levels of lice in two areas are determined by the full range of factors influencing infection processes and do not per se indicate a farm-related effect.
The rejection of the approach taken by Marshall (2003) and the findings in that report that non-farm factors were most important ignores that this study was conducted in a confined area and the factors mentioned by Krkosek et al. (seasonal variations in temperature and salinity) were common to both wild and farmed fish in that area.
The studies by Costelloe and co-workers in Ireland using plankton sampling and susceptible sentinel fish have not been able to detect infection with Lepeophtheirus salmonis above background levels more than approximately 1 km from the cages, not the transport over long distances as proposed in this report. For a significant infection pressure to occur on a mobile population of juvenile wild salmonids passing through such a small spatial window (if similar dispersion process occur in BC), and given the uncertainties of the infection process, an extremely high release of infection would have to be proposed for the farm. This is contrary to sea lice data from BC farms presented for the same year in the BC MAFF website.
The claim that con-specifics are not involved in transmission until months later in the life cycle ignores the findings reported in the DFO website of another major source of Caligus in sticklebacks (and other species of local fish) and the possibility that other species of salmonid present in the area during the study period could have provided a significant source of Lepeophtheirus salmonis. As most of the sea lice observed in the study were not identified to species level, it is not possible to exclude these possible sources.
The part of the discussion dealing with evolved mechanisms to contend with high sea lice infestation pressures is invalid for Caligus clemensi that are likely to be continuously present in large numbers in the area from a variety of resident hosts (DFO website). Similarly, it has not been shown that inshore populations of salmonids species capable of carrying Lepeophtheirus salmonis are not also continuously present in the area.