The Environmental Niche of the Tuna Purse Seine Fleet in the Western and Central Pacific Ocean Based on Different Fisheries Data

Round 1

Reviewer 1 Report

Tunas are important economic species in marine fishery resources. This study analyzes the Environmental Niche of the Tuna Purse Seine Fleet, using data analysis, which is of great importance to the management and utilization of fishery resources, and also has a useful academic contribution. There are several issues with the manuscript:

1. Format of the manuscript: Heading 3.4 has repetitions (Line 288).

2. Line 113 Figure S1 does show that there is a significant difference in effort before and after 2015, but the study takes 2015 to 2020 as the research scope, and it is suggested that there can be a clearer discussion.

3. Line 230 “The thermocline, Chl and T200 were deleted due to their VIF values”. Could explain more clearly how to choose thermocline, SST, PP, S200, Chl and T200 in the research process, and finally delete thermocline, Chl and T200.

4. Whether the Line 310 study has any assumptions or expectations for using the data in 2020 to make predictions, and whether there are differences, "The purse seine tuna fleet were mainly located in the high value area of the habitat suitability index each month at the0. 25° spatial scales but that was not always the case in the model based on commercial fishery data.”, whether there is a better discussion.

5. Line 326 “However, the results from the models with these zones masked showed that there was little difference between models with masked data and those without masked data (Table S7-9), which suggests that the EEZs and MPAs have little effect on the model performance.” I’m not sure if such an inference is appropriate, but I suggest that it can be clarified.

6. The manuscript mentions that the distribution of the purse seine tuna fleet has related important variables, such as Dox, PP (Line351), SST, T100, O100 (Line409), water temperature (Line 462), etc. This part suggests It can be organized and explained clearly, such as using a table to explain, or there can be a conclusion chapter (this manuscript has no conclusion), providing a final and comprehensive discussion of the research results.

Author Response

Response to Reviewer 1 Comments

Review for manuscript: The environmental niche of the tuna purse seine fleet in the Western and Central Pacific Ocean based on different fisheries data.
Tunas are important economic species in marine fishery resources. This study analyzes the Environmental Niche of the Tuna Purse Seine Fleet, using data analysis, which is of great importance to the management and utilization of fishery resources, and also has a useful academic contribution. There are several issues with the manuscript:
 
1.Format of the manuscript: Heading 3.4 has repetitions (Line 288).
Response: Thank the reviewer very much! 
We have revised the title as ‘3.5’.
2.Line 113 Figure S1 does show that there is a significant difference in effort before and after 2015, but the study takes 2015 to 2020 as the research scope, and it is suggested that there can be a clearer discussion.
Response: Thank the reviewer very much! 
Figure S1 showed that the accumulate FE and vessel number sharply increased from 2012 and 2015 for more and more large vessels(>24m) Equipped with AIS devices,
and then oscillation during 2015 to 2020.The accumulate FE higher than 3.0×105,and the vessel number bigger than 200 during 2015 to 2020, which significant higher than that before 2015. 
We have replaced the text“before 2015 and after 2015” by “the accumulate FE and vessel number sharply increased from 2012 and 2015 for more and more large vessels(>24m) Equipped with AIS devices,and then oscillation during 2015 to 2020” in paper.

3.Line 230 “The thermocline, Chl and T200 were deleted due to their VIF values”. Could explain more clearly how to choose thermocline, SST, PP, S200, Chl and T200 in the research process, and finally delete thermocline, Chl and T200.
Response: Thank the reviewer very much! 
22 factors were selected based on previous literature. The correlations coefficients were calculated between all variables and the results showed than the correlations coefficients values between the thermocline and SST, Chl and PP, and T200 and S200 were greater than 0.9, respectively.Meanwhile, the variance inflation factor (VIF) of each variable and kappa coefficient were calculated by constructing linear model between ef and 22 factors. The results showed than the VIF values of the thermocline, SST, Chl, PP, T200, S200 were 39.91,35.51,26.62,21.17,17.07 and 15.28,respectively, greater than 10. The other VIF values lower than 10.The thermocline were deleted first due to it has the highest VIF value. 
We continue constructed linear model between ef and remaining factors. The VIF values of the Chl, PP, T200, S200 were 26.59,21.11,15.58 and 13.69,respectively.The other VIF values lower than 10.The Chl were deleted first due to it has the highest VIF value. We repeated the process until the highest VIF and kappa coefficient of model lower than 10 and 100,respectively.At last, the thermocline, Chl and T200 were deleted.
The detail of process was too long and this is a general operation. we didn’t write in the paper. We hope you can understand and accept.
4.Whether the Line 310 study has any assumptions or expectations for using the data in 2020 to make predictions, and whether there are differences, "The purse seine tuna fleet were mainly located in the high value area of the habitat suitability index each month at the0. 25° spatial scales but that was not always the case in the model based on commercial fishery data.”, whether there is a better discussion.
Response: Thank the reviewer very much! 
The purpose of this paper is to understand potential future interactions of fishing activities and the environment, mechanistic or correlative statistical models that explore the distribution of species in relation to environmental predictors. Furthermore, models for the distribution of fishing effort can be run under different climate scenarios to help understand how fishing effort may shift in the future and affect fishing communities. The prediction performance was used to test the generalization ability of model.
The advantage of FE data is spatial and temporal high-resolution than commercial data. We have taken your suggestion and added a sentence (This may due to the potential relationship information was captured by model with high spatial FE data, while was smoothed with low spatial commercial data.) at the last of 4.2 sections.

5.Line 326 “However, the results from the models with these zones masked showed that there was little difference between models with masked data and those without masked data (Table S7-9), which suggests that the EEZs and MPAs have little effect on the model performance.” I’m not sure if such an inference is appropriate, but I suggest that it can be clarified.
Response: Thank the reviewer very much! 
We compared the sample data selection location of model between with and without EEZs and MPAs, found that the sample data selection was similarly. The model didn’t select sample from the EEZs and MPAs location when their was included in study area.For example, The follow maps are the sample data locations in January.   We have taken your suggestion and revised the last sentence of the first paragraph in section 4.1 as “The MaxEnt model didn’t select sample from the EEZs and MPAs location when their was included in study area for their was no fishing activiy. The similarly sample data location suggests that the EEZs and MPAs doesn’t bring model biased.”

Figure 2. The spatial distribution of sample data in the GFW_0.25° model based on FE data in January.White dots show the presence locations used for training, while violet dots show test locations.(left: without EEZs and MPAs; right: with EEZs and MPAs)

6. The manuscript mentions that the distribution of the purse seine tuna fleet has related important variables, such as Dox, PP (Line351), SST, T100, O100 (Line409), water temperature (Line 462), etc. This part suggests It can be organized and explained clearly, such as using a table to explain, or there can be a conclusion chapter (this manuscript has no conclusion), providing a final and comprehensive discussion of the research results.
Response: Thank the reviewer very much! 
All the factors were selected in this paper were listed in Table S1. 
The relative important of environmental factors in section 3.3.
The contribution of each factor in each month of each model was listed in Table S3-S6. 
The average contribution of each factor for all month in each model was listed Table 2.
In section 4.3 we discuss why the SST, T100, O100 and PP have the highest contribution in model results.
At last, We have taken your suggestion and added a conclusion at the end of discussion section.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript entitled The Environmental Niche of the Tuna Purse Seine Fleet in the Western and Central Pacific Ocean Based on Different Fisheries data. Has the purpose to find a predictive model for fishing ground distributions based on future remote sensing environmental data using the MaxEnt method. The idea is well described and explained. The methods are adequate for the study. The failure of the manuscript is the lack of a very strong conclusion based on the results. In abstract section the authors tried to summarize in three lines a kind of conclusion. I suggest that author make a very practical conclusion at the end of discussion section.

MaxEnt method has been considered not to be adequate in marine environments, so I suggest to the authors considering describe why this method is useful at any type of environments, marine or terrestrial.

I highlighted some typing errors in the manuscript. In line 48 I suggest rewording the sentence to ovoid the repeated word DATA. In the second, they can use information instead data.

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 2 Comments

The manuscript entitled The Environmental Niche of the Tuna Purse Seine Fleet in the Western and Central Pacific Ocean Based on Different Fisheries data. Has the purpose to find a predictive model for fishing ground distributions based on future remote sensing environmental data using the MaxEnt method. The idea is well described and explained. The methods are adequate for the study. The failure of the manuscript is the lack of a very strong conclusion based on the results. In abstract section the authors tried to summarize in three lines a kind of conclusion. I suggest that author make a very practical conclusion at the end of discussion section.
Response: Thank the reviewer very much! 
We have taken your suggestion and added a conclusion at the end of discussion section.
MaxEnt method has been considered not to be adequate in marine environments, so I suggest to the authors considering describe why this method is useful at any type of environments, marine or terrestrial.
Response: Thank the reviewer very much! 
Maybe the reviewer is right. The MaxEnt method has been considered not to be adequate in marine environments recently. But I have not read relate literature. When I used this method, there is a review paper despict that the MaxEnt algorithm has been extensively used in marine ecosystems.(Melomerino S M , Fath B D . Ecological niche models and species distribution models in marine environments: A literature review and spatial analysis of evidence[J]. Ecological Modelling,2020, 415:108837.). Melomerino and Fath (2020) provided a systematic review of 328 articles on marine ENMs and SDMs published between 1990 and 2016.They found that the Maxent is one of the most popular modelling techniques(follow Fig.1), Since the appearance of Maxent in 2006.perhaps partly because it requires relatively little information, being designed to work with presence-only data (Phillips et al., 2006). 
In this paper, we used it because it has shown robust predictive accuracy (Elith et al., 2006; Graham et al., 2008).We have taken your suggestion and added the sentence (“it has shown robust predictive accuracy”) in introduction section.

Figure.1 Tendency in the use of environmental descriptors for ENM and SDM

highlighted some typing errors in the manuscript. In line 48 I suggest rewording the sentence to ovoid the repeated word DATA. In the second, they can use information instead data.
Response: Thank the reviewer very much! 
We have taken your suggestion and changed the sentence as “commercial fishing data”. 
We have deleted the repetitive word “the” in line 158.
But we don’t know what’s the problem in abstract.

Author Response File: Author Response.pdf

Reviewer 3 Report

General comments:

The topic of this paper is quite interesting and intriguing. The environmental niche characterization of the tuna purse seine fishing fleet in western-central Pacific Ocean is quite original and useful and deserves publication. A limit of the paper is that it is only based on information concerning the Fishery and environmental parameters. Abundance and spatial distribution of tuna species object of this study and possibly of their direct preys in the study area would probably have added more robustness and completeness to the results. Anyway this modeling effort is meaningful and may be applied also to other case studies.

All specific comments are given in the attached files. An extensive revision of English language is required.

Comments for author File: Comments.zip

Author Response

Thanks to the reviewer for his work and suggestions. We have made all the necessary changes to clarify issues identified by the reviewers on the manuscript (Please see our point-to-point responses to reviewers' comments below). 
 
1 A limit of the paper is that it is only based on information concerning the Fishery and environmental parameters. Abundance and spatial distribution of tuna species object of this study and possibly of their direct preys in the study area would probably have added more robustness and completeness to the results. 
Response: Thank the reviewer very much! 
Fig S3 is the spatial distribution of the catches of tuna purse seine in the WCPO during 2015-2020. The spatial resolution is 1 degree.
We have taken your suggestion and added a paragraph in section 4.4 to illustrate the spatial distribution of tuna species and relationship with FE.
The spatial distribution of tuna’s prey is difficult to map for tuna can easily adapt their survival strategy to the local prey composition. Different food consumptions of tuna between FAD-associated and -unassociated schools, and feeding habit variations in different current systems.
(Tanabe, T. Feeding habits of skipjack tuna Katsuwonus pelamis and other tuna Thunnus spp. juveniles in the tropical western Pacific. Fish. Sci. 2001, 67, 563–570. )
2 no relevant reports??
Response: Thank the reviewer very much! 
Maybe this sentence is inappropriate. We have deleted.
3 Rogerson (2001) suggests that VIF should not exceed 5 to avoid multicollinearity...
Response: Thank the reviewer very much! 
A VIF of 10 or even one as low as 5 have been used as rules of thumb to indicate excessive or serious multi-collinearity. Menard (1995: 66) states“A tolerance of less than 0.20 is cause for concern; a tolerance of less than 0.10 almost certainly indicates a serious collinearity problem.” Since VIF is the inverse of tolerance a tolerance of 0.20 corresponds to the rule of 5 and a tolerance of 0.10 to the rule of 10. Neter et al. (1989: 409) state “A maximum VIF value in excess of 10 is often taken as an indication that multi-collinearity may be unduly influencing the least square estimates.”Hair et al. (1995) suggest that a VIF of less than 10 are indicative of inconsequential collinearity. Marquardt (1970) uses a VIF greater than 10 as a guideline for serious multi-collinearity. Mason et al. (1989) cite a VIF of greater than 10 as reason for concern. The STATA manual (StataCorp 1997: 390) notes:“However, most analysts rely on informal rules of thumb applied to VIF (see Chaterjeeand Price 1991). According to these rules, there is evidence of multi-collinearity if 1.The largest VIF is greater than 10 (some chose the more conservative threshold valueof 30). 2. The mean of all of the VIF’s is considerably larger than 1.”Kennedy (1992:183) states that “for standardized data VIFi >10 indicates harmful collinearity.”
In latest paper,experience shows that when VIF is greater than 10, multicollinearity exists(R. Salmerón Gómez, A. Rodríguez Sánchez, C.G. García, J. García Pérez
The VIF and MSE in Raise Regression Mathematics, 8 (4) (2020), p. 605, 10.3390/math8040605)
We take 10 as the threshold in this paper.
4 I am confused. I understood that 2015-19 data were used for training models while 2020 was used for testing, but I got a different impression reading here.
Response: Thank the reviewer very much! 
The 2015-19 data were used for training models,but these data were divided into two datasets. 72% was used to train model and the rest was used to validate the model. After all the models were constructed, all the models were applied to predict the HSI value of each month in 2020.
5 Chl at which quota??
Response: Thank the reviewer very much! 
Chl represent the sea surface total Chlorophyll, that was described in Table S1.
6 “The spatial distribution of the suitability index predicted by models based on FE  data was different from that of the model based on commercial fishery data, and both showed an obvious variation in 2020
“you may try to explain why... Covid19 pandemic?
Response: Thank the reviewer very much! 
The sentence means the predicted HSI maps show seasonal variation and there is difference between predicted HSI based on FE and commercial fishery data.
The difference between predicted HSI based on FE and commercial fishery data has been discussed in section 4.2.
The seasonal variation mainly is caused by ocean environmental change of each month.
We have taken your suggestion and added a sentence in section 4.3.
“The predicted HSI maps show seasonal variation based on FE and commercial fishery data showed seasonal variation, which manily is caused by ocean environmental change of each month, especially the most important environmental variables.”.
7 “The number of BPs set in this paper were sufficient. ”they are judged sufficient on the base of which reference??
Response: Thank the reviewer very much! 
We judged sufficient based on the follow reference(6. BACKGROUND POINTS).In this paper at least more than 40% of cells were used in the analysis can represent all of the variety of combinations of natural factors in the study area;
A. Lissovsky1CA1;S. V. Dudov2CA2.Species-Distribution Modeling: Advantages and Limitations of Its Application. 2. MaxEnt[J].Biology Bulletin Reviews.2021,Vol.11(No.3)：265-275.
We have taken your suggestion and added a sentence “At least more than 40% of cells were used in the analysis can represent all of the variety of combinations of natural factors in the study area.’ in section 4.1.
8 This sentence seems contradictory with the previous one, but probably I didn't understand correctly what you meant here. Please rephrase in a clearer way.
Response: Thank the reviewer very much! 
We have replaced this sentence as “then sampling bias is negligible for this choice of background.”
9 “ Primary productivity provides nutrients for phytoplankton that attract the tuna’s prey, such as herrings, anchovies, and sardines.”
Please consider that a direct link between phytoplankton and small pelagic fish (herrings etc.) do exist, but often there is the intermediate level of zooplankton grazing on phytoplankton and constituting a food source for small pelagics
Response: Thank the reviewer very much! 
We have taken your suggestion and have rephrased. Please see the revised manuscript.
10 This collage of all tables together is not a good way to present the results. Tables should be separated.
Response: Thank the reviewer very much! 
We don’t understand this suggestion.Maybe there is a reference for example. There are several literature(reference 6 and 26 of this paper) use this form.
11 All specific comments are given in the attached files. An extensive revision of English language is required.
Response: Thank the reviewer very much! 
We have revised the language problem according the reviewer suggestion.Please see the revised manuscript.
12  In Supplementary file.
Response: Thank the reviewer very much! 
We have deleted figure 2 and changed corresponding figure number serial in manuscript.
We have change the figure 3,but there only small island could be seen in map.

Author Response File: Author Response.pdf