doi:10.1111/jfd.12157
Journal of Fish Diseases 2013
Short Communication
Comment on Jackson et al. ‘Impact of Lepeophtheirus
salmonis infestations on migrating Atlantic salmon,
Salmo salar L., smolts at eight locations in Ireland with
an analysis of lice-induced marine mortality’
ek1, C W Revie2, B Finstad3 and C D Todd4
M Krkos
1
2
3
4
Department of Ecology and Evolutionary Biology, University of Toronto, Toronto, ON, Canada
Atlantic Veterinary College, University of Prince Edward Island, Charlottetown, PE, Canada
Norwegian Institute for Nature Research, Trondheim, Norway
Scottish Oceans Institute, University of St Andrews, St Andrews, UK
Keywords: Atlantic salmon, Lepeophtheirus salmonis,
mortality,
parasite,
population
regulation,
recruitment.
Marine fishes are host to a diverse array of metazoan parasites, including nematodes, monogeneans,
digeneans, cestodes and crustaceans (Rohde 2005).
However, understanding the effects of infection on
host population dynamics is limited by ecosystemscale experimental data. Jackson et al. (2013)
reported data from 28 experimental trials (in multiple years and rivers; total 352 142 fish) designed
to test and quantify the effects of parasitic crustaceans – specifically, salmon lice (Lepeophtheirus salmonis) – on marine survival of Atlantic salmon
(Salmo salar). For each trial, paired groups of
tagged, hatchery-reared juvenile salmon smolts
were released into rivers in Ireland – one group
received an in-feed parasiticide and the other was a
control. Differential survival between control and
treatment groups was calculated using recovery
programmes of returning adults in coastal marine
and freshwater fisheries and, in some cases, in-river
fish traps. These exceptionally valuable data comprise an ecosystem-scale manipulative experiment
Correspondence M Krkosek, Department of Ecology and Evolutionary Biology, University of Toronto, 25 Harbord St, Toronto,
ON M5S 3G5, Canada (e-mail: [email protected])
Ó 2013
John Wiley & Sons Ltd
1
that has the power to reveal the effects of parasitic
crustaceans on the recruitment of an ecologically
important and iconic marine fish. Such insight is
of considerable importance to the basic understanding of host–parasite ecology as well as to
managers, regulators and legislative authorities with
responsibility for fisheries and aquaculture.
However, Jackson et al. (2013) incorrectly lead
the reader to a conclusion that sea lice play a
minor, perhaps even negligible, role in salmon survival. Such a conclusion can be supported only if
one is prepared to accept at least three fundamental
methodological errors. The first is that the paired
control–treatment structure of the data was not
utilized in their meta-analysis (only as separate pertrial chi-squared tests), thereby allowing interannual fluctuations in overall survival among trials
to obscure (or bias downward) the overall effects of
parasiticide treatment relative to controls. The second is the use of arithmetic averages for comparing
survival proportions between control and treatment
groups; these data are log-normally distributed,
and appropriate survival analysis involves the calculation of differences in groups on the log scale.
Finally, their measurement of the difference in survival between control and treatment groups in
absolute percentage points (their overall final result
is approximately 1% point) does not equate to the
percentage of salmon that are lost to mortality
M Krkos ek et al. Comment on Jackson et al. (2013)
Journal of Fish Diseases 2013
Study reference
caused by parasites. Critically, as we show below,
this last estimate is actually 30 times higher.
It is recognized, and we agree, that salmon
experience high marine mortality, which is a natural part of their life history; this is a trait shared
by many marine fish species collectively known
for their high production of eggs but low survival
from egg to adult recruitment. Furthermore, these
life histories are characterized by high inter-annual
variability in marine survival for multiple reasons,
many of which are poorly understood. It therefore
is of no surprise that survival from release through
to recovery in the experimental trials reported in
the study by Jackson et al. (2013) was, overall,
quite low and variable among trials. It is important to note that time-series changes in overall survivorship are of no relevance to the assessment of
the effects of treatment in pairwise releases such as
the present: the focus has to be on the relative
effect for each independent trial. The critical feature of the data – which allow powerful insight
into the effect of parasiticide treatment on salmon
survival – is that releases were structured as paired
control–treatment groups. Thus, each release
shared the same external environmental effects,
allowing an analysis that controls for this external
variability by means of a paired-sample analysis.
The statistical methods for analysing such data
come from the fields of survival analysis and epidemiological meta-analysis, which we have
described and applied to many of the data in
the study by Jackson et al. (2013) in a previous
publication (Krkosek et al. 2013).
However, Jackson et al. (2013) did not report
an appropriate pairwise analysis. The key feature
of these analyses is that survival data are analysed
Delphi 01*
Burr 01
Burr 02
Gowla 03
Invermore 03
Burr 03
Gowla 04
Invermore 04
Erriff 04
Burr 04
Erriff 05
Gowla 05
Invermore 05
Delphi 05
Burr 05
Lee 06
Burr 06*
Screebe 06
Delphi 06
Erne 06
Burr 07
Delphi 07
Delphi 08 DEL
Burr 08*
Burr 09
Summary
0.40
1.00
2.51
6.31
15.85
39.81
Odds ratio
Figure 1 Forest plot displaying a random-effects weighted meta-analysis of the effect of treatment on the likelihood of a one sea
winter (1SW) adult salmon returning the year after release as a smolt. Weightings are inversely proportional to the standard errors
of the odds ratios. Horizontal lines represent the 95% confidence intervals of the effect size in each trial, and the relative sizes of
solid squares reflect the percentage weighting (based on standard errors of effect sizes). The vertical dotted line represents the null
hypothesis – an odds ratio of 1. The diamond shows the overall meta-analytic effect across all studies, with its width corresponding
to the 95% confidence interval. Results are given by trial, annotated according to the same identifiers in the study by Jackson et al.
(2013). The data analysed are taken unaltered from Table 1 in the study by Jackson et al. (2013), with the three trials indicated by
asterisk being those where two releases into the same river in the same year were aggregated. Also note that Table 1 in the study
by Jackson et al. (2013) includes a data transposition for numbers of treated and control fish released in 2001 (‘Delphi 01 DEL’)
compared with the study by Jackson et al. (2011b).
Ó 2013
John Wiley & Sons Ltd
2
Journal of Fish Diseases 2013
on a log scale with Gaussian error to estimate differences between treatment and control groups via
a paired structure such as paired-sample t-test,
general linear mixed model or odds ratio (Krkosek
et al. 2013). In the present re-analysis of the data
reported by Jackson et al. (2013), there were three
occurrences of two releases in the same river and
year, which we aggregated into a single release per
year to maintain independence among trials. Our
re-analysis, using a paired-sample t-test of log survival estimates, reveals an overall significant effect
of parasiticide treatment on survival (t = 2.83,
df = 24, P-value = 0.0092). The estimated effect
size is φ = 0.42 (95% CI = 0.73 to 0.11),
which equates to an estimated average annual loss
of salmon recruitment due to parasites equal to
1 exp(φ) = 0.34 (95% CI 0.11–0.52), or onethird. Furthermore, a weighted meta-analysis
reveals an overall odds ratio between treatment
and control groups not of 1.14 (95% CI 1.07–
1.21) as Jackson et al. (2013) claim, but actually
1.41 (95% CI 1.25–1.60; Fig. 1).
Our foregoing re-analyses depart substantially
from those reported and interpreted by Jackson
et al. (2013), and in their previous publications
that utilized the same data (Jackson et al. 2011a,
b). Whereas they assert that sea lice cause 1% of
mortality in Atlantic salmon, the correct estimate
is actually a one-third loss of overall adult recruitment. To further illustrate, if, in the absence of
parasites, final adult recruitment is 6% of smolt
production, then the effect of parasite mortality
reduces that recruitment to 4%. According to
interpretations used by Jackson et al. (2013), that
is a change of 2%, which we agree is a small number. However, the realized effect is that it reduces
the abundance of adult spawners returning to a
river from, say, 6000 down to 4000; this one-third
loss of salmon returns could have significant conservation or fishery implications. Our purpose here
is not to downplay factors other than parasites that
Ó 2013
John Wiley & Sons Ltd
3
M Krkos ek et al. Comment on Jackson et al. (2013)
may also have a large influence on marine survival
of Atlantic salmon: we acknowledge that few
smolts survive to return in any wild salmon population and that recent declines in the survival of
Irish Atlantic salmon cannot be solely explained by
sea lice. Rather, our purpose is to highlight that
parasites can and, in this case, do have a large
effect on fisheries recruitment, irrespective of
apparent changes in overall marine mortality over
time, and with important implications for the
management and conservation of wild salmon
stocks.
References
Jackson D., Cotter D., OMaoil
eidigh N., O’Donohoe P.,
White J., Kane F., Kelly S., McDermott T., McEvoy S.,
Drumm A., Cullen A. & Rogan G. (2011a) An evaluation
of the impact of early infestation with the salmon louse
Lepeophtheirus salmonis on the subsequent survival of
outwardly migrating Atlantic salmon, Salmo salar L., smolts.
Aquaculture 320, 159–163.
Jackson D., Cotter D., OMaoil
eidigh N., O’Donohoe P.,
White J., Kane F., Kelly S., McDermott T., McEvoy S.,
Drumm A. & Cullen A. (2011b) Impact of early infestation
with the salmon louse Lepeophtheirus salmonis on the
subsequent survival of outwardly migrating Atlantic salmon
smolts from a number of rivers on Ireland’s south and west
coasts. Aquaculture 319, 37–40.
Jackson D., Cotter D., Newell J., McEvoy S., O’Donohoe P.,
Kane1 F., McDermott T., Kelly S. & Drumm A. (2013)
Impact of Lepeophtheirus salmonis infestations on migrating
Atlantic salmon, Salmo salar L., smolts at eight locations in
Ireland with an analysis of lice-induced marine mortality.
Journal of Fish Diseases 36, 273–281.
Krkosek M., Revie C., Gargan P., Skilbrei O. T., Finstad B.
& Todd C.D. (2013) Impact of parasites on salmon
recruitment in the Northeast Atlantic Ocean. Proceedings of
the Royal Society of London. Series B 280, 20122359.
Rohde K. (2005) Marine Parasitology. CABI Publishing,
Wallingford, pp. 565.
Received: 26 February 2013
Accepted: 28 May 2013