Long-Term Trends in Freshwater and Marine Growth Patterns in Three Sub-Arctic Atlantic Salmon Populations

Round 1

Reviewer 1 Report

The data and publications relating to the River Teno in northern Finland have for some time now provided an extremely valuable resource for researchers and managers of wild Atlantic salmon throughout NW Europe.  This importance is manifest in the fish exploiting sub-Arctic and Arctic waters further north compared to most other European and Icelandic populations migrating to the NE Atlantic.

The power and scientific/management value of long-term time-series data cannot be over-emphasised.  The Teno data clearly are a very valuable resource.

I write my comments here from the point of view of a reader having no other or prior knowledge of the Teno data.  For such a reader I feel that it will be necessary for the authors to provide an extensive and comprehensive set of information, as Supplementary files, to provide the necessary background information upon which to judge the present data, results and interpretations.

Page 2, paragraph 2.  Timing of smolt emigration is mentioned, but there is no clarification of variation in smolt river age/emigration timing and any associated complexities.  For example, if older (S4, S5) smolts emigrate early in the smolt run season, and the S3 cohort leaves late, then there is marked potential here for environmental match/mismatch for the smolt river ages entering seawater in early summer time.  This needs to be considered.

Page 3, para 1.  There is a need here to explain (in Supplementary information) the details and complexities of the raw data.  What are the seasonal restrictions? – e.g. late-returning cohorts not being sampled because of season end or closure of fisheries? (Page 4, para 1).  Possible long-term variation in run-timing of return adults is crucial here.  If the data rely predominantly upon commercial catches (with a fixed season close) then an increasing proportion of a cohort will not be sampled if run-timing is delayed.

Page 4, para 4.  Here you are estimating emigrant smolt length from measures of the adult scale.  This will almost certainly include an error (see Hanson, N.N. et al 2018.  J Fish Biol. 10.1111/jfb.13863 for one river system).  This may or may not have important implications for the modelled assessments of smolt size; in Hanson’s  paper they showed that the simple linear interpolation markedly under-estimated small smolts and over-estimated large smolts.

Page 5, para 2.  I have some important concerns over the Air temperature and SST data utilised in these models, in that autocorrelation has not been considered.  Autocorrelation is an important problem (that needs addressing) for many time-series and especially so for sea temperatures.  Regression analysis requires the data to be independent – this is not the case when autocorrelation I present.

Page 6, para 1.  Following on from my concerns about autocorrelation, I am not clear that simple linear modelling is the correct approach here.  It is not possible for the reader to infer or interpret any subtleties in the overall declines shown in Fig. 3.  But I suspect that a linear approach cannot be approproiate, in that surviving smolt size (for example) has a physiological minimum.  Even if there is a general decline it will not continue to do so.  That is, non-linear models should be assessed?

Fig. 3.  It is not at all clear to me what is the numerical value of the y-axis in these graphs.  “Total freshwater growth (mm) …”?

Fig. 4.  You need to explicitly state that the May-September temperatures here are Air temperatures.

Fig. 5.  Setting aside my concerns that SST autocorrelation has not been considered, I am not at all sure that an annual mean Barents Sea temperature is ecologically meaningful in terms of the present data.  One ought to be assessing monthly, or seasonal, data to try and interrogate the biological data.

Fig. 6.  I am aware that even professional salmon biologists are not confident in their accuracy in allocating sex to salmon, based on external characteristics only.  This is especially difficult for 1SW adults returning early in the summer.

Fig. 6 shows some evidence of slight increased in mean length over these time series.  I wonder, however, if length tells the full story.  What about weight at length and condition factor?

I have one concern over cause-effect interpretations with data such as the present.  I fully appreciate the challenges, and am aware that one can generate endless possible scenarios.  But in relation to Page 10, paras 2 & 3, consider one simple hypothetical situation for salmon at sea.

1.        The salmon migrate to two distinct areas of the Barents Sea.  In Area A the fish generally grow slowly and mature/return as adults at a small size.

2.       In Area B they grow faster and larger.

3.       Collapse in prey availability in Area A results in poorer growth and excessive mortality of salmon in Area A

4.       Salmon in Area B continue to grow and survive as normal.

5.       End result, the average size of return adults increases over time.  But this is most definitely not a positive effect of marine environment change.  This arises because of a spatial deterioration and one is able to assess only the survivors returning to the river.  You emphasize yourselves that one is looking explicitly at survivors.

Hence, I would urge caution in that an increase in a factor might not necessarily be a positive response.  I suggest this illustration because you mention (Page 3, para 1) that abundances have fallen markedly over recent years in your time-series.

The data and publications relating to the River Teno in northern Finland have for some time now provided an extremely valuable resource for researchers and managers of wild Atlantic salmon throughout NW Europe.  This importance is manifest in the fish exploiting sub-Arctic and Arctic waters further north compared to most other European and Icelandic populations migrating to the NE Atlantic.

The power and scientific/management value of long-term time-series data cannot be over-emphasised.  The Teno data clearly are a very valuable resource.

I write my comments here from the point of view of a reader having no other or prior knowledge of the Teno data.  For such a reader I feel that it will be necessary for the authors to provide an extensive and comprehensive set of information, as Supplementary files, to provide the necessary background information upon which to judge the present data, results and interpretations.

Page 2, paragraph 2.  Timing of smolt emigration is mentioned, but there is no clarification of variation in smolt river age/emigration timing and any associated complexities.  For example, if older (S4, S5) smolts emigrate early in the smolt run season, and the S3 cohort leaves late, then there is marked potential here for environmental match/mismatch for the smolt river ages entering seawater in early summer time.  This needs to be considered.

Page 3, para 1.  There is a need here to explain (in Supplementary information) the details and complexities of the raw data.  What are the seasonal restrictions? – e.g. late-returning cohorts not being sampled because of season end or closure of fisheries? (Page 4, para 1).  Possible long-term variation in run-timing of return adults is crucial here.  If the data rely predominantly upon commercial catches (with a fixed season close) then an increasing proportion of a cohort will not be sampled if run-timing is delayed.

Page 4, para 4.  Here you are estimating emigrant smolt length from measures of the adult scale.  This will almost certainly include an error (see Hanson, N.N. et al 2018.  J Fish Biol. 10.1111/jfb.13863 for one river system).  This may or may not have important implications for the modelled assessments of smolt size; in Hanson’s  paper they showed that the simple linear interpolation markedly under-estimated small smolts and over-estimated large smolts.

Page 5, para 2.  I have some important concerns over the Air temperature and SST data utilised in these models, in that autocorrelation has not been considered.  Autocorrelation is an important problem (that needs addressing) for many time-series and especially so for sea temperatures.  Regression analysis requires the data to be independent – this is not the case when autocorrelation I present.

Page 6, para 1.  Following on from my concerns about autocorrelation, I am not clear that simple linear modelling is the correct approach here.  It is not possible for the reader to infer or interpret any subtleties in the overall declines shown in Fig. 3.  But I suspect that a linear approach cannot be approproiate, in that surviving smolt size (for example) has a physiological minimum.  Even if there is a general decline it will not continue to do so.  That is, non-linear models should be assessed?

Fig. 3.  It is not at all clear to me what is the numerical value of the y-axis in these graphs.  “Total freshwater growth (mm) …”?

Fig. 4.  You need to explicitly state that the May-September temperatures here are Air temperatures.

Fig. 5.  Setting aside my concerns that SST autocorrelation has not been considered, I am not at all sure that an annual mean Barents Sea temperature is ecologically meaningful in terms of the present data.  One ought to be assessing monthly, or seasonal, data to try and interrogate the biological data.

Fig. 6.  I am aware that even professional salmon biologists are not confident in their accuracy in allocating sex to salmon, based on external characteristics only.  This is especially difficult for 1SW adults returning early in the summer.

Fig. 6 shows some evidence of slight increased in mean length over these time series.  I wonder, however, if length tells the full story.  What about weight at length and condition factor?

I have one concern over cause-effect interpretations with data such as the present.  I fully appreciate the challenges, and am aware that one can generate endless possible scenarios.  But in relation to Page 10, paras 2 & 3, consider one simple hypothetical situation for salmon at sea.

1.        The salmon migrate to two distinct areas of the Barents Sea.  In Area A the fish generally grow slowly and mature/return as adults at a small size.

2.       In Area B they grow faster and larger.

3.       Collapse in prey availability in Area A results in poorer growth and excessive mortality of salmon in Area A

4.       Salmon in Area B continue to grow and survive as normal.

5.       End result, the average size of return adults increases over time.  But this is most definitely not a positive effect of marine environment change.  This arises because of a spatial deterioration and one is able to assess only the survivors returning to the river.  You emphasize yourselves that one is looking explicitly at survivors.

Hence, I would urge caution in that an increase in a factor might not necessarily be a positive response.  I suggest this illustration because you mention (Page 3, para 1) that abundances have fallen markedly over recent years in your time-series.

I fully appreciate the difficulties and challenges for Non-English speakers writing in English.  But the text would benefit markedly from some additional assistance

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper presents some interesting data, but I cannot get past the total lack of integration with existing data for Atlantic salmon in the subject at large. There is a rich database of information on these same growth parameters and an existing framework of hypothesis building and testing that must be addressed. It is scientific malpractice to totally ignore this body of science on salmon post-smolt growth ad climate related effects that have shaped salmon populations. I cannot support this paper as it is presently composed.

The writing is fair, but could use some editing.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

My comments, in relation to the changes/additions made by the authors in response to my initial review are shown below in bold face.

The data and publications relating to the River Teno in northern Finland have for some time now provided an extremely valuable resource for researchers and managers of wild Atlantic salmon throughout NW Europe.  This importance is manifest in the fish exploiting sub-Arctic and Arctic waters further north compared to most other European and Icelandic populations migrating to the NE Atlantic.

The power and scientific/management value of long-term time-series data cannot be over-emphasised.  The Teno data clearly are a very valuable resource.

I write my comments here from the point of view of a reader having no other or prior knowledge of the Teno data.  For such a reader I feel that it will be necessary for the authors to provide an extensive and comprehensive set of information, as Supplementary files, to provide the necessary background information upon which to judge the present data, results and interpretations.

Page 2, paragraph 2.  Timing of smolt emigration is mentioned, but there is no clarification of variation in smolt river age/emigration timing and any associated complexities.  For example, if older (S4, S5) smolts emigrate early in the smolt run season, and the S3 cohort leaves late, then there is marked potential here for environmental match/mismatch for the smolt river ages entering seawater in early summer time.  This needs to be considered.

Additional sentences added on this (lines 297-303)

Page 3, para 1.  There is a need here to explain (in Supplementary information) the details and complexities of the raw data.  What are the seasonal restrictions? – e.g. late-returning cohorts not being sampled because of season end or closure of fisheries? (Page 4, para 1).  Possible long-term variation in run-timing of return adults is crucial here.  If the data rely predominantly upon commercial catches (with a fixed season close) then an increasing proportion of a cohort will not be sampled if run-timing is delayed.

No response or addition

Page 4, para 4.  Here you are estimating emigrant smolt length from measures of the adult scale.  This will almost certainly include an error (see Hanson, N.N. et al 2018.  J Fish Biol. 10.1111/jfb.13863 for one river system).  This may or may not have important implications for the modelled assessments of smolt size; in Hanson’s  paper they showed that the simple linear interpolation markedly under-estimated small smolts and over-estimated large smolts.

Additional sentences (lines 153-157).  This problem now is explicitly acknowledged, but it then is essentially ignored.  I do not wish to sound pedantic, but it has to be said that this problem extends beyond simply estimating the body length of the fish at smolt (=river) emigration.  If you wish to accurately estimate changes in body length, after smolt emigration and during the marine migration (e.g. at the end of one or more sea winters), then you have to use a corrected smolt length in relation to the known final length to make those marine growth estimates.  Thus, this problem for the present data and analyses goes further than simply estimating by back-calculation the length of the emigrant smolt.

Page 5, para 2.  I have some important concerns over the Air temperature and SST data utilised in these models, in that autocorrelation has not been considered.  Autocorrelation is an important problem (that needs addressing) for many time-series and especially so for sea temperatures.  Regression analysis requires the data to be independent – this is not the case when autocorrelation I present.

No response or change/addition.  This  is a fundamental problem that cannot be ignored.

Page 6, para 1.  Following on from my concerns about autocorrelation, I am not clear that simple linear modelling is the correct approach here.  It is not possible for the reader to infer or interpret any subtleties in the overall declines shown in Fig. 3.  But I suspect that a linear approach cannot be appropriate, in that surviving smolt size (for example) has a physiological minimum.  Even if there is a general decline it will not continue to do so.  That is, non-linear models should be assessed?

No response or change/addition. 

Fig. 3.  It is not at all clear to me what is the numerical value of the y-axis in these graphs.  “Total freshwater growth (mm) …”?

I am not sure that these data are especially valuable, being the raw measures.

Fig. 4.  You need to explicitly state that the May-September temperatures here are Air temperatures.

Done

Fig. 5.  Setting aside my concerns that SST autocorrelation has not been considered, I am not at all sure that an annual mean Barents Sea temperature is ecologically meaningful in terms of the present data.  One ought to be assessing monthly, or seasonal, data to try and interrogate the biological data.

No response or change/addition. 

Fig. 6.  I am aware that even professional salmon biologists are not confident in their accuracy in allocating sex to salmon, based on external characteristics only.  This is especially difficult for 1SW adults returning early in the summer.

No response or change/addition. 

Fig. 6 shows some evidence of slight increased in mean length over these time series.  I wonder, however, if length tells the full story.  What about weight at length and condition factor?

No response or change/addition. 

I have one concern over cause-effect interpretations with data such as the present.  I fully appreciate the challenges, and am aware that one can generate endless possible scenarios.  But in relation to Page 10, paras 2 & 3, consider one simple hypothetical situation for salmon at sea.

1.       The salmon migrate to two distinct areas of the Barents Sea.  In Area A the fish generally grow slowly and mature/return as adults at a small size.

2.      In Area B they grow faster and larger.

3.      Collapse in prey availability in Area A results in poorer growth and excessive mortality of salmon in Area A

4.      Salmon in Area B continue to grow and survive as normal.

5.      End result, the average size of return adults increases over time.  But this is most definitely not a positive effect of marine environment change.  This arises because of a spatial deterioration and one is able to assess only the survivors returning to the river.  You emphasize yourselves that one is looking explicitly at survivors.

Hence, I would urge caution in that an increase in a factor might not necessarily be a positive response.  I suggest this illustration because you mention (Page 3, para 1) that abundances have fallen markedly over recent years in your time-series.

The additional paragraph in the Discussion (lines 366-379) do not provide a particularly informative context of change in the Teno populations in relation to other European and North American populations.

I fully appreciate the difficulties and challenges in writing scientific papers in English, when it is not one’s native language.  While it generally reads informatively, this script would require a considerable amount of editing of the syntax, grammar and occasional choice of specific words by an English speaker.

Author Response

Our replies to referee 1’s comments, including review round 1 and 2

The data and publications relating to the River Teno in northern Finland have for some time now provided an extremely valuable resource for researchers and managers of wild Atlantic salmon throughout NW Europe.  This importance is manifest in the fish exploiting sub-Arctic and Arctic waters further north compared to most other European and Icelandic populations migrating to the NE Atlantic.

The power and scientific/management value of long-term time-series data cannot be over-emphasised.  The Teno data clearly are a very valuable resource.

Our reply: We thank the reviewer for encouraging words to our manuscript and other valuable comments that helped us to improve the work. Please see our case-by-case comments below.

I write my comments here from the point of view of a reader having no other or prior knowledge of the Teno data.  For such a reader I feel that it will be necessary for the authors to provide an extensive and comprehensive set of information, as Supplementary files, to provide the necessary background information upon which to judge the present data, results and interpretations.

Our reply: We are doubtful what does the reviewer ask us to provide more. Teno River system is well-studied, and the salmon population has been under extensive research for decades. We have explained the study area (2.1) with the figure 1, and sampling procedure (2.2) with relevant citations of previous studies made in Teno salmon [23, 26-30, 32, 36-37]. We are unsure what would be the missing but still relevant piece of information to this study that is missing from the manuscript.

Page 2, paragraph 2.  Timing of smolt emigration is mentioned, but there is no clarification of variation in smolt river age/emigration timing and any associated complexities.  For example, if older (S4, S5) smolts emigrate early in the smolt run season, and the S3 cohort leaves late, then there is marked potential here for environmental match/mismatch for the smolt river ages entering seawater in early summer time.  This needs to be considered.

Our reply: We must admit that no tributary-specific, sub-population specific nor life-history-specific potential differences in smolt migration timing is not known in Teno river system. Anyhow, we have added the discussion of this possibility, please see lines 300-306.

Page 3, para 1.  There is a need here to explain (in Supplementary information) the details and complexities of the raw data.  What are the seasonal restrictions? – e.g. late-returning cohorts not being sampled because of season end or closure of fisheries? (Page 4, para 1).  Possible long-term variation in run-timing of return adults is crucial here.  If the data rely predominantly upon commercial catches (with a fixed season close) then an increasing proportion of a cohort will not be sampled if run-timing is delayed.

Our reply: In this study, we focus on growth patterns of 1 and 2SW salmon. These two are the main life-history strategies that Teno salmon display. Multi-Sea Winter fish exist, but our scale measurement data is not (yet) available for the further analysis as the measuring process is ongoing. The fishing season (i.e., sampling period) lasts from the beginning of the spawning migration window to the beginning of the spawning, please see further details from

Vähä, J.P., Erkinaro, J., Niemelä, E., Primmer, C.R., Saloniemi, I., Johansen, M., Svenning, M. & Brørs, S. 2011. Temporally stable population-specific differences in run timing of one-sea-winter Atlantic salmon returning to a large river system. – Evolutionary Applications 4: 39–53.

Niemelä, E., Orell, P., Erkinaro, J., Dempson, J.B., Brørs, S, Svenning, M. & Hassinen, E. 2006.  Previously spawned Atlantic salmon ascend a large subarctic river earlier than their maiden counterparts. – Journal of Fish Biology 69: 1151–1163.

Our additional reply to round 2 comments: We have now added these references in our manuscript and explained that the sampling season from June to August virtually covers the entire run timing of salmon in Teno system. Please see lines 131-132

Page 4, para 4.  Here you are estimating emigrant smolt length from measures of the adult scale.  This will almost certainly include an error (see Hanson, N.N. et al 2018.  J Fish Biol. 10.1111/jfb.13863 for one river system).  This may or may not have important implications for the modelled assessments of smolt size; in Hanson’s  paper they showed that the simple linear interpolation markedly under-estimated small smolts and over-estimated large smolts.

Referee’s round 2 comment:

Additional sentences (lines 153-157).  This problem now is explicitly acknowledged, but it then is essentially ignored.  I do not wish to sound pedantic, but it has to be said that this problem extends beyond simply estimating the body length of the fish at smolt (=river) emigration.  If you wish to accurately estimate changes in body length, after smolt emigration and during the marine migration (e.g. at the end of one or more sea winters), then you have to use a corrected smolt length in relation to the known final length to make those marine growth estimates.  Thus, this problem for the present data and analyses goes further than simply estimating by back-calculation the length of the emigrant smolt.

 Our reply: We have not estimated the smolt length, but the growth based on the scale (measured as increment in mm on the scale), so no back-calculation of smolt length has been done here.   As we are showing in the figure 2, the freshwater phase can be separated from the sea phase. We are aware that the scale measures are not 1:1 comparable to the total length of the fish. We have added this note in the manuscript and cited accordingly.

Our additional response to round 2 comments: We have added two references describing the issue with back-calculations, and why we have abandoned such approach in our study. Please see lines 148-158.

Page 5, para 2.  I have some important concerns over the Air temperature and SST data utilised in these models, in that autocorrelation has not been considered.  Autocorrelation is an important problem (that needs addressing) for many time-series and especially so for sea temperatures.  Regression analysis requires the data to be independent – this is not the case when autocorrelation I present.

Our reply: We are slightly unsure what does the reviewer refers by commenting that the autocorrelation has not been considered. The fact that annual mean temperatures have been rising within the time series has been taken into account by using linear mixed-effects models.  In the LMM this effect has been split into variance components of the sampling year so the temperatures as continuous variable can be interpreted independently. The air temperature and the SST have not been used in the same models, so their probable autocorrelation does not affect the results. This could have been potentially avoided by creating a new “sea area” variable and include in the model with an interaction with the temperature. Anyhow we preferred congruently the parsimonious model selection procedure.  The reason why we have used air temperatures for the freshwater growth models is that we do not have a water temperature series long and comprehensive enough to be used.

Our additional response to round 2 comments:  We have included a relevant reference describing the issue with using a year as a fixed effect factor in growth analysis of fish. We now explain also in the manuscript, why we have used mixed effects models in our study. Please see the lines 196-199.

Page 6, para 1.  Following on from my concerns about autocorrelation, I am not clear that simple linear modelling is the correct approach here.  It is not possible for the reader to infer or interpret any subtleties in the overall declines shown in Fig. 3.  But I suspect that a linear approach cannot be approproiate, in that surviving smolt size (for example) has a physiological minimum.  Even if there is a general decline it will not continue to do so.  That is, non-linear models should be assessed?

Our reply: As we have explained the modelling procedure in the materials and methods section in the paragraph 2.4, we have not used a simple linear model but linear mixed-effects model approach in order to take the autocorrelation into account by adding the year in the random factor. This allows to interpret the variance component that annual variation has. The LMM approach is rather robust and therefore we reckon it will be much more conservative and applicable modelling approach than non-linear models. LMM assumptions apply to model residuals, which are required to be independent, homoscedastic, and normally distributed.

Fig. 3.  It is not at all clear to me what is the numerical value of the y-axis in these graphs.  “Total freshwater growth (mm) …”?

Our reply: We have rephrased the figure caption.

Fig. 4.  You need to explicitly state that the May-September temperatures here are Air temperatures.

Our reply: We have added a sentence: “Mean T refers monthly mean air temperatures” in the figure caption.

Fig. 5.  Setting aside my concerns that SST autocorrelation has not been considered, I am not at all sure that an annual mean Barents Sea temperature is ecologically meaningful in terms of the present data.  One ought to be assessing monthly, or seasonal, data to try and interrogate the biological data.

 Our reply: To take the auto-correlation issues with the time and temperature, we have taken this into account by analysing the data with linear mixed-effects models, where the year has been used as random factor. By doing this we aimed to take annual variation into account and then the mean temperature can be interpreted “independently”. Barents Sea West and East region SST have been decided to include in the model based on the model selection procedure, despite their potential auto-correlation. The monthly mean temperatures we have included in the model that is reported in the supplementary materials (Suppl. Table 3).

Fig. 6.  I am aware that even professional salmon biologists are not confident in their accuracy in allocating sex to salmon, based on external characteristics only.  This is especially difficult for 1SW adults returning early in the summer.

Our reply: The samples used in this study are catches of local fishermen exploiting the salmon populations in these tributaries of the Teno River. All fish have been culled and the sex determined based on opened fish.

Fig. 6 shows some evidence of slight increased in mean length over these time series.  I wonder, however, if length tells the full story.  What about weight at length and condition factor?

 Our reply: While it is true that Fulton’s K or weight could give an interesting information about the condition of the migrant before spawning, we have considered that the length provide clearer realization of the growth in the individual life-history perspective. Also, the length has been linked with the scale measurements in previous studies.

Our additional reply to round 2 comments: We have now included extra panels in the figure 6 that show how the weight of fish display similar trends as length.

I have one concern over cause-effect interpretations with data such as the present.  I fully appreciate the challenges, and am aware that one can generate endless possible scenarios.  But in relation to Page 10, paras 2 & 3, consider one simple hypothetical situation for salmon at sea.

1. The salmon migrate to two distinct areas of the Barents Sea.  In Area A the fish generally grow slowly and mature/return as adults at a small size.
2. In Area B they grow faster and larger.
3. Collapse in prey availability in Area A results in poorer growth and excessive mortality of salmon in Area A
4. Salmon in Area B continue to grow and survive as normal.
5. End result, the average size of return adults increases over time.  But this is most definitely not a positive effect of marine environment change.  This arises because of a spatial deterioration and one is able to assess only the survivors returning to the river.  You emphasize yourselves that one is looking explicitly at survivors.

Hence, I would urge caution in that an increase in a factor might not necessarily be a positive response.  I suggest this illustration because you mention (Page 3, para 1) that abundances have fallen markedly over recent years in your time-series.

Our reply: This is an interesting and yet plausible hypothesis that is of worth to digest and study further. Unfortunately, at this point with the data we have, we cannot make direct assumptions of population-specific foraging areas of the salmon in the sea. We have avoided to report the increase of sea growth as a positive signal throughout the manuscript, as greater size generally comes with other potentially negative impacts on salmon populations such as greater energy demands, higher metabolic rates, inter-individual competition for resources etc. To test this suggested alternative hypothesis, more research is needed. Especially the clearer view of migration routes of salmon at the Barents Sea, and samples of immature salmon at the sea will be required to test such hypothesis.

The additional paragraph in the Discussion (lines 366-379) do not provide a particularly informative context of change in the Teno populations in relation to other European and North American populations.

Our reply: We have now revised the paragraph and linked the discussion with our results.

Reviewer 2 Report

The authors miss important papers on the subject (see below).

These papers and including those they have already cited show that southern tier European salmon appear to have post-smolt growth signatures that correlate with post-smolt survival over time. Summer thermal conditions also align with these patterns, suggesting the change in temperature is associated with a change feeding conditions during the first year at sea; the result is a change in smolt growth and condition. These changes in growth and temperature occur over a protracted period (the summer feeding season) for post-smotls. Since growth has been reduced for such a long period of time, the smolt class accumulates greater mortalities since they stay at predation vulnerable sizes for a longer period. The same citations acknowledge that the information on European origin salmon is limited to stock originating from the southern part of the range; hence, the patterns among northern tier stocks may be different.

Post-smolt growth is not correlated with post-smolt survival in the North American stocks studied. Instead, post-smolt survival is correlated with thermal conditions associated with the start of the ocean migration of post-smolts. The hypothesis is that with the change in spring thermal conditions, the nature of the predation fields affecting post-smolts has changed, resulting in greater mortality during that time. Since the recruitment patterning mortality has already occurred, the post-smolt growth pattern can be independent of the survival pattern. The results of this paper are not novel in showing an absence of a linkage between post-smolt growth and post-smolt survival. It does add to our understanding of the range of response of Atlantic salmon to varying ocean conditions.

Friedland KD, MacLean JC, Hansen LP, Peyronnet AJ, Karlsson L, Reddin DG, Maoileidigh NO, McCarthy JL. 2009. The recruitment of Atlantic salmon in Europe. Ices J Mar Sci 66(2): 289–304. doi: 10.1093/icesjms/fsn210

Friedland KD, Manning JP, Link JS, Gilbert JR, Gilbert AT, O’Connell AF. 2012. Variation in wind and piscivorous predator fields affecting the survival of Atlantic salmon, Salmo salar, in the Gulf of Maine. Fisheries Manag Ecol 19(1): 22–35. doi: 10.1111/j.1365-2400.2011.00814.x

Friedland KD, Shank BV, Todd CD, McGinnity P, Nye JA. 2014. Differential response of continental stock complexes of Atlantic salmon (Salmo salar) to the Atlantic Multidecadal Oscillation. J Marine Syst 133: 77–87. doi: 10.1016/j.jmarsys.2013.03.003

Hogan F, Friedland KD. 2010. Retrospective growth analysis of Atlantic salmon Salmo salar and implications for abundance trends. J Fish Biol 76(10): 2502–2520. doi: 10.1111/j.1095-8649.2010.02650.x

Mccarthy JL, Friedland KD, Hansen LP. 2008. Monthly indices of the post-smolt growth of Atlantic salmon from the Drammen River, Norway. J Fish Biol 72(7): 1572–1588. doi: 10.1111/j.1095-8649.2008.01820.x

Author Response

Our replies for the referee 2’s round 2 comments

The authors miss important papers on the subject (see below).

These papers and including those they have already cited show that southern tier European salmon appear to have post-smolt growth signatures that correlate with post-smolt survival over time. Summer thermal conditions also align with these patterns, suggesting the change in temperature is associated with a change feeding conditions during the first year at sea; the result is a change in smolt growth and condition. These changes in growth and temperature occur over a protracted period (the summer feeding season) for post-smotls. Since growth has been reduced for such a long period of time, the smolt class accumulates greater mortalities since they stay at predation vulnerable sizes for a longer period. The same citations acknowledge that the information on European origin salmon is limited to stock originating from the southern part of the range; hence, the patterns among northern tier stocks may be different.

Post-smolt growth is not correlated with post-smolt survival in the North American stocks studied. Instead, post-smolt survival is correlated with thermal conditions associated with the start of the ocean migration of post-smolts. The hypothesis is that with the change in spring thermal conditions, the nature of the predation fields affecting post-smolts has changed, resulting in greater mortality during that time. Since the recruitment patterning mortality has already occurred, the post-smolt growth pattern can be independent of the survival pattern. The results of this paper are not novel in showing an absence of a linkage between post-smolt growth and post-smolt survival. It does add to our understanding of the range of response of Atlantic salmon to varying ocean conditions.

Friedland KD, MacLean JC, Hansen LP, Peyronnet AJ, Karlsson L, Reddin DG, Maoileidigh NO, McCarthy JL. 2009. The recruitment of Atlantic salmon in Europe. Ices J Mar Sci 66(2): 289–304. doi: 10.1093/icesjms/fsn210

Friedland KD, Manning JP, Link JS, Gilbert JR, Gilbert AT, O’Connell AF. 2012. Variation in wind and piscivorous predator fields affecting the survival of Atlantic salmon, Salmo salar, in the Gulf of Maine. Fisheries Manag Ecol 19(1): 22–35. doi: 10.1111/j.1365-2400.2011.00814.x

Friedland KD, Shank BV, Todd CD, McGinnity P, Nye JA. 2014. Differential response of continental stock complexes of Atlantic salmon (Salmo salar) to the Atlantic Multidecadal Oscillation. J Marine Syst 133: 77–87. doi: 10.1016/j.jmarsys.2013.03.003

Hogan F, Friedland KD. 2010. Retrospective growth analysis of Atlantic salmon Salmo salar and implications for abundance trends. J Fish Biol 76(10): 2502–2520. doi: 10.1111/j.1095-8649.2010.02650.x

Mccarthy JL, Friedland KD, Hansen LP. 2008. Monthly indices of the post-smolt growth of Atlantic salmon from the Drammen River, Norway. J Fish Biol 72(7): 1572–1588. doi: 10.1111/j.1095-8649.2008.01820.x

Our reply: We have addressed these points in the added paragraph (Please see lines 370-389), included some new references, and linked this discussion with our results. However, our paper does not describe trends in recruitment or survival (main topic in most papers listed above), which we found highly speculative to be included in the discussion.

Our previous replies for the referee 2’s round 1 comments

Referee 2’s comments

This paper presents some interesting data, but I cannot get past the total lack of integration with existing data for Atlantic salmon in the subject at large. There is a rich database of information on these same growth parameters and an existing framework of hypothesis building and testing that must be addressed. It is scientific malpractice to totally ignore this body of science on salmon post-smolt growth ad climate related effects that have shaped salmon populations. I cannot support this paper as it is presently composed.

Our reply: We regret that this review is rather vague and does not help much of us to improve our manuscript. We have tried to cover a reasonable amount of relevant as well as recent literature of growth of salmon at sea, e.g. [9, 13-16, 46, 48, 50]. As the referee states, there is a lot of studies published, and we cannot cite all of them. Based on the comments of the reviewer, it is difficult to understand what are those exact studies that should have been cited, instead of some that were, or in addition to those used. However, we have now included a couple of additional relevant citations for comparison to our results in our revised manuscript. We have added a paragraph to discuss the potential lack of correlation between growth and survival, and mechanisms behind the phenomenon. Please see lines 366-379.

Round 3

Reviewer 2 Report

I am now supportive of the paper. I think its add to our knowledgebase on Atlantic salmon.

Author Response

We thank the reviewer for they contribution to our work.