Meiofauna in a Potential Deep-Sea Mining Area—Influence of Temporal and Spatial Variability on Small-Scale Abundance Models

Round 1

Reviewer 1 Report

This study looks at meiofaunal community in deep sea mining area, in relationship with environmental parameters, applying a random forest algorithm to model both environment and fauna. The study is an interesting research area, and poses indeed interesting and necessary questions that  should be understood in order to understand impacts of mining.

 

My issues with this study are various:

 

Firstly I find it lacks a clear aim , for example in the introduction The aims could be expanded upon and made clearer. Is this study aiming to develop a tool that can be used for predictions ? I don’t think one can aim to predict , in my view the whole points of predictions is in the future so cannot test if you have predicted well.Unless one is testing a prediction model - in which case the aim is to develop a prediction tool. This should then be reflected in the discussion, which for now is a bit vague in this respect , for example mentioning ‘Here, we show that random forest regression can also help’. But the rest of the discussion mostly focuses on discussing the results of the monitoring and not such the power of the model per se. Aside from paragraph 313 to 326 which I don’t understand. It discusses things that could be included but weren’t? If so then why not and why so early in the discussion ? At this stage I would like to know what worked and why and what didn’t and why if I had to reuse such a model for something else So either the aim of these study are refocused or the aim of prediction needs to be discussed further.

Furthermore, i would keep in mind that  the special issue is on the environmental drivers of meiofaunal composition, thus would aim to show the reader more of this aspect.

 

Secondly, the sampling campaign is poorly defined in the method. The number of cores should be presented divided per year and station, only seeing the graphs it can be seen that 2018 has an extremely low number of replica. It is not enough to say that abundance is decreasing. Furthermore, with this low replicates I am really unsure of the permanova design. Were the data tested for dispersion ? Seems that the dark blue points are much more dispersed.( Also the two blues are hardly distinguishable. ). And I am wondering with all of these pairs and such low replica whether pairwise comparisons can be considered meaningful.

 

Minor comments

The abstract should avoid saying that there was a change but mentioning what the change is

Generally, the writing should be improved for better flow

Figures would also benefit for clearer captions, e.g. what it is on the axis ? Assume is a measure of space, but are we talking meters, kilometres..

 

 

 

Author Response

Firstly I find it lacks a clear aim , for example in the introduction The aims could be expanded upon and made clearer. Is this study aiming to develop a tool that can be used for predictions ? I don’t think one can aim to predict , in my view the whole points of predictions is in the future so cannot test if you have predicted well.Unless one is testing a prediction model - in which case the aim is to develop a prediction tool. This should then be reflected in the discussion, which for now is a bit vague in this respect , for example mentioning ‘Here, we show that random forest regression can also help’. But the rest of the discussion mostly focuses on discussing the results of the monitoring and not such the power of the model per se. Aside from paragraph 313 to 326 which I don’t understand. It discusses things that could be included but weren’t? If so then why not and why so early in the discussion ? At this stage I would like to know what worked and why and what didn’t and why if I had to reuse such a model for something else So either the aim of these study are refocused or the aim of prediction needs to be discussed further.

Answer:

Your first issue is a lack of aim in our study. To strengthen it, I added a more precise description of our aim at the end of the Introduction (ll. 97-98). It is neither to develope a tool for prediction neither for specific predictions, but to investigate if distribution models can be used as an additional tool to investigate the variations of meiofauna in space and time. Furthermore, I hope that in this context, the paragraph ll. 331-344 is now understandable.

Furthermore, i would keep in mind that  the special issue is on the environmental drivers of meiofaunal composition, thus would aim to show the reader more of this aspect.

Answer:

The manuscript is now assigned to the Marine Diversity Section instead of the special issue.

Secondly, the sampling campaign is poorly defined in the method. The number of cores should be presented divided per year and station, only seeing the graphs it can be seen that 2018 has an extremely low number of replica. It is not enough to say that abundance is decreasing. Furthermore, with this low replicates I am really unsure of the permanova design. Were the data tested for dispersion ? Seems that the dark blue points are much more dispersed.( Also the two blues are hardly distinguishable. ). And I am wondering with all of these pairs and such low replica whether pairwise comparisons can be considered meaningful.

Answer:

The sampling design is now described in greater detail and, additionally, a table containing all meiofauna samples used in this study is added as Appendix 2. Furthermore, dispersion was investigated with the R-functions betadisper and permutest and results are added (see ll. 171-174 and Table 2).

The abstract should avoid saying that there was a change but mentioning what the change is

Answer:

The change observed  was added to the abstract (ll. 27-29).

Generally, the writing should be improved for better flow

Answer:

Revising the manuscript, the writing style was also revised and adapted.

Figures would also benefit for clearer captions, e.g. what it is on the axis ? Assume is a measure of space, but are we talking meters, kilometres..

Answer:

I assume, Your comment on figure captions was referring to the distribution maps. As a measure of space, I added a scale bar on each map (see Figure 3 & 5). 

Reviewer 2 Report

The paper seems to me interesting and well written and therefore worthy of publication. I appreciate the use of meiofauna for this type of problem. On one hand a weakness can be represented by the fact that many statements are not conclusive and therefore difficult to apply, but on the other hand interesting hypotheses are reported that could be verified with further research.

Author Response

The paper seems to me interesting and well written and therefore worthy of publication. I appreciate the use of meiofauna for this type of problem. On one hand a weakness can be represented by the fact that many statements are not conclusive and therefore difficult to apply, but on the other hand interesting hypotheses are reported that could be verified with further research.

Answer:

Although we have an extensive dataset for a deep-sea area, it is still impossible to present greatly conclusive statements. Still, we hope to have made the best of it.

Reviewer 3 Report

The manuscript Meiofauna standing stocks in a potential deep-sea mining area – Influence of spatial environmental variability and sampling in different years on small scale abundance models by Uhlenkott et al. is interesting and provides important results that are also well introduced in the context of the topic. The manuscript is kind of continuation of the topic introduced by Authors in the previous publication published this year in Journal of Applied Ecology. Here, Authors take another step and use their distribution model as a tool in predicting inter-annual variability and small-scale patterns of meiofauna. The manuscript is very well written, correctly organized and methods are adequate. Results are in general well exposed and discussion brings the current context of the topic. References are formatted with great care and according to journal’s requirements. The manuscript is within the Diversity scope and can do well in this journal.

There are, however, some considerations that should been taken into account prior acceptance.

Title: I don’t think that ‘standing stocks’ is a correct expression in the context of presented results. Authors focus only on abundance data, while standing stocks refers rather to abundance and biomass data.

Introduction:

• Please add information about CCZ location. It is already in the abstract, but the first sentence of introduction should contain such essential information either. Not all readers might be familiar with CCZ geography.
• L51: Although biodiversity[…..]- I believe that Authors were meant ‘abundance’ not ‘biodiversity’
• L52-54: please be more clear and explain where the abundance and diversity is higher, in nodule-free areas or nodule-covered ones.
• L54-56: It is not clear what biodiversity Authors refer here- Nematode diversity, meiofauna diversity? Does it refer to number of species/taxa or to other diversity indices? Please also explain at the beginning of the sentence what is Peru Basin- another example of nodule field?
• L60-62: this sentence does not have any connection with the previous one. Please rewrite. I also think that it would be more clear for the readers first describe and explain environmental factors that affect meiofauna and then make a comment about temporal variability.
• L61: GSR- this abbreviation is introduced here for the first time and is not explained
• L83-86: I can’t agree that environmental data are generally obtained from different positions than meiofauna abundance. Contrary, for most meiofauna investigations environmental data come from the same positions, the same multicorer/box corer. Please rewrite or make a comment about this specific study.

Materials and methods:

            This part needs a little more explanation, especially in terms of sampling strategy.

• Figure 1: this figure is somehow unclear and confusing. Not sure what is a white rectangle. The magenta square is hardly seen. There are some yellow points that they are invisible on the yellowish background. Why showing the whole German license area, when study is focused only on the area marked by the magenta square? Map of the BGR area could be much smaller and the main attention should be given to sampling area. Please mark clearly stations where meiofauna samples were collected and stations with environmental variables. As it is now, there is no way to figure this out. Schemes of the sampling gears are not necessary. I guess that most of the readers are familiar with them. Besides, in the text Fig. 1b is cited before Fig 1a. It should be like this.
• Add information about water depth range in the sampling area
• L103: what does it mean ‘highest repetition’? one sample for 2018 doesn’t seem as high number. Please gather all information about sampling in a table (even in a supplements).
• There is lack of information about cores diameter and depth of the sediment taken for meiofauna analysis. Later, in the results section Authors mention something about the layers- does it mean that the cores were sliced? If so, this information (together with the number of layers) have to be included in the M&M.
• L111:information about ‘smaller diameter in 2013’ is not clear, because the is no context and no information about diameter for other years. How does it influence the expression of abundance values? I cannot see the connection.
• L116: please add information about size of a boxcorer. How much of the sediment was taken for sediment characteristics? What is wet and dry bulk density?
• L138:what does ‘available layers’ mean?

Results

• L150: Authors mention Appendix 1 that should include all sediment and nodule characteristics that were available in this study. But Appendix 1 contains information about the chla concentration for several months.

Please make sure to provide the list with all characters taken into account in the study. It is very importnat.

• Table1 caption:L157 there is a mistake- first sediment characteristics, then nodule parameters are presented
• MDS and PERMANOVA results: on the one hand, Authors present MDS with no clear differences in terms of meiofauna structure between years (as they say in L160-161). On the other, they present results of pair-wise comparisons for factor year, but without results for main test, that would test differences for factor year. Besides it would be more logical first to perform PERMANOVA test and then visualize its result using MDS. Please also explain how the MDS was done- based on all available samples (replicates/ layers) or some mean values?
• L163: it looks that 2013 was also different from most of other years
• Table 2: in my opinion this table is redundant here. It can be moved to the appendix.
• Table 3: please add a, b, c abbreviations for different sets of environmental predictors, to be consistent with the abbreviations presented in Table 4
• Table 4: use term ‘Nematoda abundance’ rather than ‘Nematoda’. It would be more consistent with ‘overall abundance’ term
• L235: Between all sampling years, […]. – this sounds a little awkward. Please rephrase.
• Table 5 is placed in the text before Figure 4, although Fig. 4 is cited in the text first. Please move Table 5 further down.
• Figure 4: Please make both figures a & b more comparable. I don’t understand why asterisks have been used and why each point is shifted to the right. Purple (2013) points represent samples from one station, so they should be aligned. I would also suggest to use the same scale on both figures. There is also no clear what exactly boxplots on fig a show- median values with 25-75% and min-max values? Please be more clear.
• L259, 260, 261, 263, 266: I think that Authors wanted to cite Table 5 rather than Table 6
• L272-273: I cannot see the pattern described in this sentence.

Discussion

This part is well written. However, I think it would be interesting to make a comment about using presented method for other type of biological data, e.g. Nematode community structure. In many ecological studies, identifications of meiofaunal organisms are done to lower taxonomical levels, especially for such dominant group as Nematodes. Does random forest modelling would be a useful tool in predicting distribution pattern of dominant genera or some specific indicator species?  

It would be also important to comment on the sediment characteristics in the light of presented results. Why e.g. total inorganic carbon in 4 cm layer might be better predictor than carbon content in surface sediment layer, etc. What might be the reason of the fact that any food-related (except total organic carbon) variables were not successful in explaining variance in meiofauna abundance?

L341: […decrease in meiofauna abundance..]- decrease means some change, it is a process. Better would be to write that high nodule abundance can lead to lower abundance of meiofauna

L355-360: this sentence is not well connected with the previous one. It describes completely different scale of variation. Please rephrase or explain more clearly.

L362: I wouldn’t pay too much attention to particular years itself. Observed inter-annual shift between 2013 and 2014 might resulted from the highest number if samples collected for these years. What is important is a question if presented method indeed is useful in predicting temporal changes of seafloor dwelling meiofauna.

 

 

 

Author Response

Title: I don’t think that ‘standing stocks’ is a correct expression in the context of presented results. Authors focus only on abundance data, while standing stocks refers rather to abundance and biomass data.

Answer:

The title was adapted according to Your suggestion, removing the term "standing stock".

Introduction:

• Please add information about CCZ location. It is already in the abstract, but the first sentence of introduction should contain such essential information either. Not all readers might be familiar with CCZ geography.
• Answer: In the introduction, the information given on the location of the CCZ was added.
• L51: Although biodiversity[…..]- I believe that Authors were meant ‘abundance’ not ‘biodiversity’
• Answer: Although the sentence (now l. 52) did refer to diversity, it makes more sense if it indeed refers to abundance.
• L52-54: please be more clear and explain where the abundance and diversity is higher, in nodule-free areas or nodule-covered ones.
• Answer: The following sentence was specified as suggested (ll. 53-55).
• L54-56: It is not clear what biodiversity Authors refer here- Nematode diversity, meiofauna diversity? Does it refer to number of species/taxa or to other diversity indices? Please also explain at the beginning of the sentence what is Peru Basin- another example of nodule field?
• Answer: This also applies for the next sentence referring now to the taxonomic composition of the benthic community and giving more information on the Peru Basin (ll. 56-58).
• L60-62: this sentence does not have any connection with the previous one. Please rewrite. I also think that it would be more clear for the readers first describe and explain environmental factors that affect meiofauna and then make a comment about temporal variability.
• Answer: To accomplish a better connection within the text, the following paragraph was restructured (ll. 62-68).
• L61: GSR- this abbreviation is introduced here for the first time and is not explained
• Answer: The abbreviation GSR was removed and the sentence now refers to the Belgian license area instead (l. 67).
• L83-86: I can’t agree that environmental data are generally obtained from different positions than meiofauna abundance. Contrary, for most meiofauna investigations environmental data come from the same positions, the same multicorer/box corer. Please rewrite or make a comment about this specific study.
• Answer: The sentence on the availability of environmental data was rephrased to account for the fact, that environmental variables are usually obtained from the same site as meiofauna samples (ll. 86-89).

Materials and methods:

            This part needs a little more explanation, especially in terms of sampling strategy.

• Figure 1: this figure is somehow unclear and confusing. Not sure what is a white rectangle. The magenta square is hardly seen. There are some yellow points that they are invisible on the yellowish background. Why showing the whole German license area, when study is focused only on the area marked by the magenta square? Map of the BGR area could be much smaller and the main attention should be given to sampling area. Please mark clearly stations where meiofauna samples were collected and stations with environmental variables. As it is now, there is no way to figure this out. Schemes of the sampling gears are not necessary. I guess that most of the readers are familiar with them. Besides, in the text Fig. 1b is cited before Fig 1a. It should be like this.
• Answer: Figure 1 was revised and is not divided into a and b anymore.
• Add information about water depth range in the sampling area
• Answer: Information on the water range of sampling positions was added (l. 105).
• L103: what does it mean ‘highest repetition’? one sample for 2018 doesn’t seem as high number. Please gather all information about sampling in a table (even in a supplements).
• Answer: All sampling information was added to supplement table (Appendix 2) and the sentence was addapted to avoid confusion with the term "highest repetition" (ll. 106-108).
• There is lack of information about cores diameter and depth of the sediment taken for meiofauna analysis. Later, in the results section Authors mention something about the layers- does it mean that the cores were sliced? If so, this information (together with the number of layers) have to be included in the M&M.
• Answer: Core treatment and diameter were added (ll. 111-114 and Appendix 2).
• L111:information about ‘smaller diameter in 2013’ is not clear, because the is no context and no information about diameter for other years. How does it influence the expression of abundance values? I cannot see the connection.
• Answer: Also to better explain the need for a conversion of meiofauna abundances due to the smaller core diameter in 2013.
• L116: please add information about size of a boxcorer. How much of the sediment was taken for sediment characteristics? What is wet and dry bulk density?
• Answer: Information on the size of the boxcore was added (l. 123). Bulk density refers to how packed the particles are regarding a defined volume. As there is also no explanation for grain size or shear strength, this explanation was not added to the manuscript.
• L138:what does ‘available layers’ mean?
• Answer: The sentence was rewritten to explain, that "available layers" meant backscatter and bathymetric layers in the first sentence and to all (including the newly computed) environmnetal layers in the second sentence (ll. 154-157).

Results

• L150: Authors mention Appendix 1 that should include all sediment and nodule characteristics that were available in this study. But Appendix 1 contains information about the chla concentration for several months.
• Answer: Appendix 1 is now the correct table and includes all environmental parameters that were modelled but did not provide suitable models and, hence, were not used to model meiofauna abundance.

Please make sure to provide the list with all characters taken into account in the study. It is very importnat.

• Table1 caption:L157 there is a mistake- first sediment characteristics, then nodule parameters are presented
• Answer: The references to nodule and sediment characteristics were swaped in the caption of table 1.
• MDS and PERMANOVA results: on the one hand, Authors present MDS with no clear differences in terms of meiofauna structure between years (as they say in L160-161). On the other, they present results of pair-wise comparisons for factor year, but without results for main test, that would test differences for factor year. Besides it would be more logical first to perform PERMANOVA test and then visualize its result using MDS. Please also explain how the MDS was done- based on all available samples (replicates/ layers) or some mean values?
• Answer: The order of the presentation of PERMANOVA and NMDS was changed and adapted as suggested (ll. 169-193). (Additionally, investigations of dispersion effects were added based on suggestions by reviewer 1).
• L163: it looks that 2013 was also different from most of other years
• Answer: As far as I can see regarding table 2, 2013 is only significantly different compared to 2016.
• Table 2: in my opinion this table is redundant here. It can be moved to the appendix.
• Answer: As now the investigations of dispersion are also added to this table, I would rather keep it within the text.
• Table 3: please add a, b, c abbreviations for different sets of environmental predictors, to be consistent with the abbreviations presented in Table 4
• Answer: Table 3 was adapted as suggested to match with table 4.
• Table 4: use term ‘Nematoda abundance’ rather than ‘Nematoda’. It would be more consistent with ‘overall abundance’ term
• Answer: As suggested, "Nematoda" was changed to "Nematoda abundance" in table 4.
• L235: Between all sampling years, […]. – this sounds a little awkward. Please rephrase.
• Answer: "Between sampling years" was changed to "Regarding all years".
• Table 5 is placed in the text before Figure 4, although Fig. 4 is cited in the text first. Please move Table 5 further down.
• Answer: The order of table 5 and figure 4 was changed, so that they are now cited in the correct order in the text.
• Figure 4: Please make both figures a & b more comparable. I don’t understand why asterisks have been used and why each point is shifted to the right. Purple (2013) points represent samples from one station, so they should be aligned. I would also suggest to use the same scale on both figures. There is also no clear what exactly boxplots on fig a show- median values with 25-75% and min-max values? Please be more clear.
• Answer: Figure 4b was revised and is now containing the same kind of points as used in figure 2 and the points are aligned in a manner mirroring the boxplot in figure 4a. A more detailed explanation of the boxplot was added to the caption.
• L259, 260, 261, 263, 266: I think that Authors wanted to cite Table 5 rather than Table 6
• Answer: The mentioned citations of table 6 indeed refer to table 5 and were adjusted accordingly.
• L272-273: I cannot see the pattern described in this sentence.
• Answer: I again looked at figure 5 and still see the pattern described in ll. 290-291, although it large differences in general abundance mask it slightly.

Discussion

This part is well written. However, I think it would be interesting to make a comment about using presented method for other type of biological data, e.g. Nematode community structure. In many ecological studies, identifications of meiofaunal organisms are done to lower taxonomical levels, especially for such dominant group as Nematodes. Does random forest modelling would be a useful tool in predicting distribution pattern of dominant genera or some specific indicator species?

Answer:

Some additional thoughts on the application of distribution models on other groups was added (ll. 395-401).

It would be also important to comment on the sediment characteristics in the light of presented results. Why e.g. total inorganic carbon in 4 cm layer might be better predictor than carbon content in surface sediment layer, etc. What might be the reason of the fact that any food-related (except total organic carbon) variables were not successful in explaining variance in meiofauna abundance?

Answer:

The main reason, why predictor variables do not influence distribution models and community analysis in an expected manner, is possibly based on differences on the scale of sampling. Meiofauna is possibly influenced in smaller scale than is obtained with boxcore samples and mirrored distribution models (ll. 355-358).

L341: […decrease in meiofauna abundance..]- decrease means some change, it is a process. Better would be to write that high nodule abundance can lead to lower abundance of meiofauna

Answer:

"[...] decrease in meiofauna abundance" was changed to "lower meiofauna abundance" (l. 362) as suggested.

L355-360: this sentence is not well connected with the previous one. It describes completely different scale of variation. Please rephrase or explain more clearly.

Answer:

The not-good-connected paragraph on intra-annual differences in meiofauna abundance was deleted, to focus more strongly on our study.

L362: I wouldn’t pay too much attention to particular years itself. Observed inter-annual shift between 2013 and 2014 might resulted from the highest number if samples collected for these years. What is important is a question if presented method indeed is useful in predicting temporal changes of seafloor dwelling meiofauna.

Answer:

To further address the question of usefulness, a summarizing sentence was added at the end of that paragraph (ll. 385-387).