The Effect of Fishery Management on the Yield of the Critically Endangered European Eel Anguilla anguilla in Mesotrophic Rivers and Streams in Central Europe

Round 1

Reviewer 1 Report

Referee report:  The effect of fishery management on the harvest rates of the critically endangered European eel Anguilla anguilla in Mesotrophic rivers and streams in Central Europe.

The author seeks to utilize a rich set of data to evaluate a significant (and probably costly) effort to enhance the size of the European eel stock in Czech rivers.  The management effort consists of “reintroduction activities” and increased minimum “landing size” of caught eal.  Data utilized consist of anglers logbooks and administrative data over the period 2005 to 2018 for 176 fishing sites.  A generalized regression model is used to estimate the effect of the management effort.  The author concludes that increasing the minimum size of landed eel increases the average size of landed eel.  Furthermore, that angling effort is the main driver of catch.  Lastly, the author rejects the hypothesis that stocking of juveniles increases catch.  These are important and interesting conclusions.  It would be interesting to see if these conclusions hold if the dataset is analysed using different tools.

The author does a good job at describing the conditions surrounding the angling “industry”.  The definition of variables in the dataset could be more extensive as I will return to below.  Furthermore, the fisheries “model” is not described at all.  All we have is Table 4 which hints the shape of the estimation equation.  We are informed that stocking is assumed to effect catch some years later.  No attempt is made to explain the complex relationship between growth of the biomass in fresh water, return of mature eels to the Saragossa Sea, recruitment and later return to fresh water in the Czech rivers.  Does restocking have any effect on recruitment of eel from the Saragossa?  If there is no such effect the whole program of restocking must be re-evaluated.  If stocking only adds to the fishable stock in fresh water and does not add to the reproductive capacity of the whole stock it might be better to ban catch of eel all together and encourage fishers to use their skills to catch other species that are more easily kept at sustainable levels.

The paper is not fit for publication in its present form.  To be considered for publication the author needs to address several issues:

1. Some sort of a biological model needs to be presented.  The author needs to discuss why some relationships may be ignored and which are important for the present study.
2. It is stated that European countries took aim at allowing 40% of adult eels to migrate to the sea. We are not informed if there was some sort of follow up on that aim by measurements of migration or in any other form.
3. The author should argue which estimation equation to use and even attempt more than one method to estimate the fundamental equations.
4. The author needs to define the variables used. For instance, what is meant by “Harvest rate” (the left-hand variable in table 4 equations).  Usually, harvest rate is the percentage of stock that is harvested (https://www.ices.dk/community/Documents/Advice/Acronyms_and_terminology.pdf), but it seems that the author uses the term synonymously with catch volume or yield.
5. What does it mean that harvest and stocking rates were not normally distributed? Are we talking about number of eels stocked and harvested per fishing site or are we talking about harvest and stocking per areal unit of fishable waters in a given geographical location?  One would expect the distribution for absolute number of eels fished or stocked to be different from distribution of harvest rates per areal units of fishable water.
6. At the beginning of the results section it is stated that European eel is harvested at lower intensity than other species. This statement begs some form of explanation.
7. Table 3: what is the unit of effort used in the study?
8. The table 4 regressions: Assume that the left hand variable is harvest rate (i.e. number of eels caught per hectare of water in a given geographical position), why should the absolute size of surface area or bank length influence the harvest rate?  Would it not be more natural to use the ratio of bank length to surface area as right hand variable?
9. Some measurement on environmental impact seems missing in the regressions. One such measure could be the catch per unit of effort of other species than eels in the rivers systems as the author does have access to those data.  Supposedly a negative environmental impact like accidental chemical pollution would affect all the caught species in similar manner.  Environmental factors other than fishing effort and stocking could exacerbate or nullify the effect of fishing effort and stocking and should be accounted for in some way.

The author should discuss other methods to measure the impact of increased minimum landing size and stocking on harvest rate of eels.  Some form of panel based difference in difference methods might be useful as an experiment.

Bottom line:  The paper is not publishable in its present form.  The author has a very rich data set that can be utilized much better than already done.  The author is strongly advised to work further with the data analysis.

Comments for author File: Comments.pdf

Author Response

Response of the author:

General comments:

I re-defined the variables in the dataset that I analysed. I described the models in the results more thoroughly. I added environmental and biological factors to the model and tested their effect on the eel harvest rates. I added the recommendation regarding the program of restocking to the conclusion section.

Specific comments:

1. I added the environmental and biological factors to the models (Table 4). I also discussed the addition of other methods and factors (lines 467-469), any why I did not include them in the study (lines 547-551).
2. I added the information regarding the progress of the measurements undertaken (lines 46-48, 53-54).
3. I added other methods of measurement, including a comparison to the harvest rates of other fish species as well as the correlations with environmental and biological factors (lines 144-147, 229-274, 356-366, Table 4), and a comparison to a study group of fishing sites (lines 135-138, 344-349, 407-411, 417-420, 426-428). I also added the environmental and biological factors to the models (Table 4).
4. I defined the variables including the variable “harvest rate” (lines 295-304). I did read the acronyms and terminology document that was suggested. However, I have been using the terminology “harvest rate” in my other manuscripts, and I wanted to be consistent, so decided to keep it.
5. I added the explanation regarding the distributions of the variables (lines 277-284). I tested the distributions per hectare of a fishing site. I only tested the distributions between all the fishing sites because that is what I am analysing further in the models and other statistical tests. I did not test the distributions within each fishing site, as that is not what I am testing in the further analyses.
6. I added an explanation about what it means (lines 326).
7. I highlighted the units that were used in the models and statistical analyses in this study (lines 295-304, line 335, Table 4).
8. I added the units of measurement to the Table 4. I also swapped the surface area and the bank length for the suggested ration between bank length and surface area (Table 4).
9. I added the environmental and biological factors to the analysis. The factors include harvest rates of other fishes (lines 255-263, Table 4).

I discussed the usage of other methods of data analysis and measuring the impacts of the increased angling size and stocking on eel harvest rates (lines 547-551). I hope I have broadened the analysis enough to be publishable now.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper assesses the effectiveness of regulation on the eel population in the Czech Republic, and finds that reducing the minimum landing size (MLS) of eel catches by anglers has reduced the harvest rate of eels, but increased eel stocking measures had no effect. I have three questions for the author to answer before recommending publication.

First, the author deals only with angling and never mentions netting. Is netting not a more serious threat then angling to the eel population in the Czech Republic?

Second, how does the author know what are the effects of increased MLS and increased stocking measures? On increased MLS, the author says:

I tested the effect of the restriction – the increased minimum angling size from 45 cm to 50 cm TL – on eel harvest rates (Figure 2b). While the eel harvest rates fluctuated over time before the restriction was introduced, they stabilized after…The median harvest rate of the eel was lower after the restriction was introduced (lines 257-261)

But how does he know that the increased MLS was the cause of lower harvest rates?

On increased stocking levels, the author says:

the intensive stocking of elvers and yellow eel did not increase the eel harvest rates (lines 267-268)

eel stocking had only a weak effect on eel harvest rates (line 351)

But how does he know that eel stocking did not increase the eel harvest rates? And does he know why this is the case?

Third, the author states in the Abstract that the paper assesses the impact of two fisheries measures on the eel population – MLS and stocking:

This paper aimed to analyze the effect of two fishery management actions – an increased minimum legal angling size, and eel stocking – on eel harvest rates (lines 9-11)   

Yet later he adds an additional factor – fishing effort:

the angling effort was the only important factor explaining the eel harvest rates. The angling effort was strongly negatively correlated with the eel harvest rates, meaning that a higher number of anglers led to a lower CPUE (catch per unit effort). (lines 263-266)

The study confirmed our initial hypothesis that the implementation of the minimum legal angling size leads to lower harvest rates of eels and to larger harvested eels. I also confirmed the hypothesis that eel harvest rates are driven by angling effort (lines 316-318)

 The confusion is compounded when the author later refers to ‘other factors’:

I tested whether the eel harvest rates are driven by eel stocking, by angling limitations, or by other factors (lines 92-93)

One other factor appears to be changes in anglers’ behaviour:

the behavior and preferences of anglers could also be responsible for decreased eel harvests. The critically endangered conservation status of eel could make anglers release caught eels back to the water instead of killing them (335-337)

In the following passage, the author refers to several additional factors causing decreased harvesting rates in fisheries in Europe:

 decreased harvest rate was not exclusive for eels – other fish species also experienced decreased landings in central Europe and outside of Europe [6, 40]. In general, fish harvest rates in Europe were decreasing mainly due to a lower fish abundance caused by lower input of nutrients (N, P, K) into water. Other reasons were the legislative protection of fish-eating otters Lutra lutra and cormorants Phalacrocorax carbo, introduction of parasites (e. g. bladder nematode Anguillicola crassus), increased angling pressure, water pollution (organic pollutants and drug residuals), and inappropriate modifications of rivers and streams (lines 327-334)

Is the author admitting that any or all of these other factors could account for the decreased eel harvest? If so, doesn’t this admission undermines his claim that increased MLS was the cause of reduced eel harvesting?

Author Response

Response of the author:

Question number one: netting is basically non-existent in the Czech Republic. I added this information to the manuscript to the methods section (lines 207-208).

Question number two: I tested the correlations between the harvest rates of eels, the important fisheries factors, and the important environmental and biological factors. The results were that only the fishing effort had any effect on it. Then I tested the models before and after the minimum legal angling size was increased. The results were that the models were different, signalling a significant change after the angling size was increased. It is true that I cannot be sure that it was only caused by this, but I added several more important environmental and biological factors to the model, and the harvest rate was not strongly correlated to neither of them (see Table 4). Therefore, I ruled out the effect od those factors, which ultimately led me to the conclusion that the fishing effort and the increased angling size drove the harvest rates. However, I added a paragraph to the conclusion stating that other factors that were not tested could be responsible as well (lines 552-560). I also described why I was not able to test the effect of those factors (lines 547-551).

Question number three: I unified the research aims, questions, and hypotheses throughout the whole manuscript. You can see that in the abstract (lines 10-11, 19-20), in the introduction (lines 94-103), and in the discussion (lines 467-469). And I added the previously mentioned paragraph that states that I was not able to test for every factor that could potentially affect the harvest rates (lines 547-551).

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Referee report:  The effect of fishery management on the harvest rates of the critically endangered European eel Anguilla anguilla in Mesotrophic rivers and streams in Central Europe; re-submitted.

Remarks:

• Study design:The author has added a statement on lines 136-139 of the resubmitted manuscript where it is stated that rivers are divided into (1) study group and (2) control group.  Fishing sites in both groups have similar environmental and biological characteristics.  But it is not stated what is different when the division is introduced.  We learn in lines 346 to 348 that the minimum angling size was not changed in sites of the study group.  The author needs to bring this fact up-front (i.e. in the statement in lines 136-139).  Furthermore, he needs to show a table reporting the means for critical variables for the study group and the control group and report results from statistical tests showing that hypothesis of same means and variances for variables studied over the study group and the control group.  e. is the difference between means (and variances) of stocking and harvesting etc in the control group and study group significantly different from 0 or not.

• ” Definition of variables:The author claims to have refined the definition of all variables.  It is correct that the sources of environmental variables are now explicitly given.  Furthermore the methods used to arrive at the stock estimates of eels is outlined:  “I estimated the biomass of eels in the rivers using the combination of the following   Firstly, I used the data collected by the Czech Fishing Union. However, since the eel is a rare species that is hard to detect by a random sampling, I also estimated the  number of eels based on the biomass of stocked elvers and yellow eels [kg per ha], the growth speed of eels, the ocean and upstream migration speed of young eels, the downstream migration speed of adult eels, the age of maturation of eels, and the estimatedage when the adult eels start to migrate downstream. I used the information regarding eel biology and ecology from other studies“.  The statement „I estimated....“ reveals that the authors is utilizing unique data produced by himself.  In name of reproducability and replicability he must be much more precise in explaining how his data is produced and he must also make his estimate of eel stock size available for other researachers.  I can agree that it would disturb the flow of the paper to include the derivation report of stock size in the main text of the paper.  But ist should definately be in an appendix, either published with the paper or obtainable from the editor upon request.  The report should include detailed information of formulas used to derive the stock size and all assumptions made in order to facilitate the derivation of the stock size of eels.  For other variables the author must clearly explain what data is taken „as is“ from the sources.  Furthermore, if some transformation has been conducted the author must clearly explain and reveal the mathematical expressions used.
• In line 278-280 it is stated that “…number of eels stocked and harvested per one hectare of a fishing site were tested.”I have hard time understanding the meaning of this statement.  Does this mean that the author is testing if stocking and harvesting is distributed in some specific manner?  Or is it tested if there is a relationship between stocking and harvesting (some form of regression?).
• Biological model:The mathematical biological model on which the estimating equation is based is not presented.  The estimating equation can be revealed by inspecting Table 4.  The estimation of the treatment effect is obtained by running two regression, one before and one after the change of angling size.   I find that line of attack disturbing.  Why not run a single regression with time and treatment dummies?  Then all the treatment group sites would get a treatment dummy valued as “1” for the period post introduction of the treatment.  The treatment dummy would have the value of “0” in all other cases.  Or alternatively use some form of difference-in-difference panel estimation instead?  The author might benefit from studying DiD in some detail, https://benjamindwilliams.weebly.com/uploads/6/8/5/7/68575765/lecture12.pdf.  Note the warning on slide 9 that a before-after estimator will be biased if there are time-trends as is the case in the present study.
• The author argues that it is passable for him to use “harvest rate” differently from how it is used by international organs and by most fisheries specialists, because how he has used the term in earlies papers.  I don’t find that line of argument satisfactory.

I can not recommend the publication of the paper until these concerns are addressed.

Author Response

Comments of the reviewer:

Reviewer comment 1:

1) Study design: The author has added a statement on lines 136-139 of the resubmitted manuscript where it is stated that rivers are divided into (1) study group and (2) control group.  Fishing sites in both groups have similar environmental and biological characteristics.  But it is not stated what is different when the division is introduced.  We learn in lines 346 to 348 that the minimum angling size was not changed in sites of the study group.  The author needs to bring this fact up-front (i.e. in the statement in lines 136-139).  Furthermore, he needs to show a table reporting the means for critical variables for the study group and the control group and report results from statistical tests showing that hypothesis of same means and variances for variables studied over the study group and the control group.  e. is the difference between means (and variances) of stocking and harvesting etc in the control group and study group significantly different from 0 or not.

Author response:

I added the information that the fishing sites in the control group had constant minimum legal angling size (lines 136-143). I added the table that describes the median values and variances of the parameters that were tested in the study, including the results of the tests (Table 1). The methodology that describes how the groups were compared is also described (lines 141-143). 

Reviewer comment 2:

2) ” Definition of variables: The author claims to have refined the definition of all variables.  It is correct that the sources of environmental variables are now explicitly given.  Furthermore the methods used to arrive at the stock estimates of eels is outlined:  “I estimated the biomass of eels in the rivers using the combination of the following   Firstly, I used the data collected by the Czech Fishing Union. However, since the eel is a rare species that is hard to detect by a random sampling, I also estimated the  number of eels based on the biomass of stocked elvers and yellow eels [kg per ha], the growth speed of eels, the ocean and upstream migration speed of young eels, the downstream migration speed of adult eels, the age of maturation of eels, and the estimated age when the adult eels start to migrate downstream. I used the information regarding eel biology and ecology from other studies“.  The statement „I estimated....“ reveals that the authors is utilizing unique data produced by himself.  In name of reproducability and replicability he must be much more precise in explaining how his data is produced and he must also make his estimate of eel stock size available for other researachers.  I can agree that it would disturb the flow of the paper to include the derivation report of stock size in the main text of the paper.  But ist should definately be in an appendix, either published with the paper or obtainable from the editor upon request.  The report should include detailed information of formulas used to derive the stock size and all assumptions made in order to facilitate the derivation of the stock size of eels.  For other variables the author must clearly explain what data is taken „as is“ from the sources.  Furthermore, if some transformation has been conducted the author must clearly explain and reveal the mathematical expressions used.

Author response:

In the appendix of the article, I described the whole process of estimating and calculating the eel stock (eel biomass per hectare) in the studied rivers. The appendix will be published with the paper as a supplementary material (lines 285-286). I added the formulas and their explanation to the appendix (see supplementary material). In case of the other variables, I described which source they came from, and if they were standardized or re-calculated (transformation etc.) (lines 160, 177-178, 191-192, 256, 265, 276-277, 280-281). 

Reviewer comment 3:

3) In line 278-280 it is stated that “…number of eels stocked and harvested per one hectare of a fishing site were tested.”I have hard time understanding the meaning of this statement.  Does this mean that the author is testing if stocking and harvesting is distributed in some specific manner?  Or is it tested if there is a relationship between stocking and harvesting (some form of regression?).

Author response:

I replaced the sentence with a sentence “I tested whether the eel yields are distributed normally across all studied fishing sites.“ (lines 293-294).

Reviewer comment 4:

4) Biological model: The mathematical biological model on which the estimating equation is based is not presented.  The estimating equation can be revealed by inspecting Table 4.  The estimation of the treatment effect is obtained by running two regression, one before and one after the change of angling size.   I find that line of attack disturbing.  Why not run a single regression with time and treatment dummies?  Then all the treatment group sites would get a treatment dummy valued as “1” for the period post introduction of the treatment.  The treatment dummy would have the value of “0” in all other cases.  Or alternatively use some form of difference-in-difference panel estimation instead?  The author might benefit from studying DiD in some detail, https://benjamindwilliams.weebly.com/uploads/6/8/5/7/68575765/lecture12.pdf.  Note the warning on slide 9 that a before-after estimator will be biased if there are time-trends as is the case in the present study.

Author response:

The biological model is described in the methods section (lines 336-340). In addition to the previously described statistical method, I also added the method suggested by the reviewer. It is described in the methods section (lines 327-328, 341-352) and in the results section (lines 407-410 and Table 6). However, I also decided to keep the first method after a consultation with a mathematician/statistician from the Faculty of Mathematics and Physics on the Charles University (https://www.mff.cuni.cz/en). He confirmed that the statistical analysis that I used was correctly used in this case. We concluded that there are certainly more ways to analyse the data, and neither of them is wrong.  

Reviewer comment 5:

5) The author argues that it is passable for him to use “harvest rate” differently from how it is used by international organs and by most fisheries specialists, because how he has used the term in earlies papers.  I don’t find that line of argument satisfactory.

Author response:

I changed the term “harvest rate” to “yield” in the whole manuscript (see the changes tracker in the MS Word version of the manuscript).

Author Response File: Author Response.pdf