Numerical models for monitoring and forecasting ocean ecosystems: a short description of the present status

Libralato, Simone

doi:https://doi.org/10.5194/sp-5-opsr-13-2025

Articles | Volume 5-opsr

https://doi.org/10.5194/sp-5-opsr-13-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Report:

Ocean prediction: present status and state of the art (OPSR)

https://doi.org/10.5194/sp-5-opsr-13-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 5-opsr

02 Jun 2025

| OPSR | Chapter 5.4

| 02 Jun 2025 | OPSR | Chapter 5.4

Numerical models for monitoring and forecasting ocean ecosystems: a short description of the present status

Simone Libralato

Download

Final revised paper (published on 02 Jun 2025)
Preprint (discussion started on 24 Oct 2024)

Interactive discussion

Status: closed

RC1:
'Comment on sp-2024-42', Anonymous Referee #1, 15 Nov 2024
The paper provides an overview of six classes of marine ecosystem models and evaluates their current status in operational applications. I acknowledge that it’s a daunting task to review these many models in such limited space, and I appreciate the author’s efforts. However, I found some aspects of the paper challenging.
I wonder who the targeted audience for this review is. I’m afraid that the current version is not quite accessible to the wider audience, partiallydue to the use of jargon and acronyms—some of which are not explained—and by lengthy sentences from time to time. I have a background in ocean ecosystem modeling, but I have to admit that it’s not always easy to follow the description of all models. If the review is intended for a wider audience, I think it could be a bit more elementary.

The Introduction indicates that there are “available syntheses and reviews”. How does this short description complement those previous works? A brief summary highlighting the specific focus of the previous reviews in relation to this one would help readers identify which resources best align with their interests.

As a follow-up comment, given the limited space available for detailed information on each class of model, it would be beneficial if the author could recommend key references for further reading.

The criteria used to assess the readiness or maturity of each class of model for operational applications could be more explicitly defined. Is it based on the model complexity, computation cost, or technical challenges in implementation?

Given the limited space, the description of each model class is brief. The only exception is the “3 Population and stock assessment models.” It presents two complex model equations, which are a bit awkward and somewhat “too detailed” in this brief review.

Some technical corrections:
Page 2, line 31. Should be “six classes”
Page 5, line 151. What does “population dynamics under technical interactions” mean?
Page 7, line 195. “These WEM models…”
Page 15. Table. What is the gww?
Page 16. Table. Missing the “of” after the “Number”
Page 17. Table. “Ensemble” instead of “Ensamble”?
Citation: https://doi.org/10.5194/sp-2024-42-RC1
- AC1: 'Reply on RC1', Simone Libralato, 25 Feb 2025
  
  The aim of the paper is to provide a structured synthesis of models for the higher trophic levels of the oceans (essentially from zooplankton to top predators and fisheries) that can be linked to models for the lower trophic level (physics and biogeochemistry). The lower trophic level models currently provide operational products in the Copernicus Marine Service, and their potential to be linked to upper trophic level models opens up a wide range of new CMS products. The aim is to cover most approaches from single individuals through populations to multispecies, but the work can inevitably not be exhaustive. Conversely, the work provides a classification and examples of the most commonly used high trophic level tools with some indication of their potential use for operational linkage with lower trophic level models. Given the wide range of approaches developed in relation to specific scientific objectives, a sector-specific gergon may also have been used for the sake of brevity.
  1- In a revised version, I would ensure that all acronyms, parameters and gergons are explained and, if possible, simplified into a more readable text. I expect that a thorough revision for a native English language will also help to make the document more accessible to a wider audience by harmonizing the language in all 6 classes of models.
  2- This work summarises some information from various reviews and is not intended to replace them. The reviews each refer to one or two classes of models, while the work aims to build a synthetic bridge across several classes and approaches. The work does not claim to be exhaustive, but examples of tools for each class may shed light on their use for operational coupling with lower trophic level models. In a revised version, I can make these goals clear in a few sentences.
  3- In view of the above point, I can also refer in a revised version - for each class - to one or two review papers where readers can find more details.
  4- The readiness and maturity of each model was subjectively determined based on its current application rather than model complexity or computational cost. For example, tools that are routinely used in fisheries management were considered more mature because the code is publicly available and documented, and test cases for the inputs and outputs are developed and accessible. The most mature models also include routines for evaluating model performance and diagnostics and are used by a community of developers who can provide support, updates, and further development. In light of the suggested commentary, I believe that a revised version of the paper should distinguish between maturity and readiness. Maturity explains the above points (availability of code, documentation, test cases, diagnosis of model performance and reference community) and thus explains a subjective but very sound evaluation. Suitability for operational purposes is explicitly stated and defined based on existing knowledge of the model's potential connection to physical and biogeochemical spatio-temporal models. The existence of such applications, although rare, could shed light on the difficulties in linking (one- or two-way) with lower trophic level models. The applicability could be considered rather preliminary and less precise, as it is also more difficult to define objectively for the possibly very sparse application.
  5- I understand the need to strike a balance between in-depth discussion and presentation of each class of models. I am happy to reduce equations and explanations, keep the description equally synthetic for each model class, and perhaps leave more room for discussing readiness for operational applications as well as challenges.
  
  All changes proposed as “technical corrections”:
  Page 2, line 31. Should be “six classes”
  Answer: Yes, needs to be corrected.
  Page 5, line 151. What does “population dynamics under technical interactions” mean?
  Answer: This is an example of gergon. Marine populations exploited by two different fisheries is called technical interaction in fisheries science.
  Page 7, line 195. “These WEM models…”
  Answer: Yes, needs to be corrected.
  Page 15. Table. What is the gww?
  Answer: gww= gram wet weight clearly another acronym to correct.
  Page 16. Table. Missing the “of” after the “Number”
  Answer: Yes, needs to be corrected.
  Page 17. Table. “Ensemble” instead of “Ensamble”?
  Answer: Yes, needs to be corrected.
  
  Citation: https://doi.org/10.5194/sp-2024-42-AC1
RC2:
'Comment on sp-2024-42', Anonymous Referee #2, 28 Nov 2024
This paper provides a classification of numerical models and toolboxes developed in recent years for marine ecosystem modelling with a view to ecosystem management and prediction. Given the limited space, the description is succinct and the overview cannot meet the objective of a full review. The readers will inevitably have to delve deeper, consulting additional reviews, synthesis or original articles to access the information they're looking for. The diversity of models available today has become very large, and the classification proposed here is useful in guiding the user not directly involved in specific model developments. Therefore, the paper should be published but some aspects need to be improved to make the contribution more impactful:
The purpose of the paper and target audience should be clarified and made more explicit. Is it a “review of reviews” (as suggested page 2, line 31), or an attempt at classification, or a user's guide, or something else ?

Is the classification mutually exclusive, or can there be models that fall into several classes, or combine concepts from several classes? A short discussion on possible intersections between classes should be added.

One of the aims of the paper seems to be the assessment of the “operational maturity” of the models listed. The notion of “operational model” needs to be clarified in the text. Is it a question of high TRL ? of ability to produce information routinely ? in real time ? The title mentions “Numerical Models for Monitoring and Forecasting”, but the text suggests that the focus is more on management issues. It would be useful to identify the reasons why some models are still far from operational status, once the notion is clarified.

The content of the sections on the 6 classes might be harmonized. Equations or GitHub references are used in some sections, but not in others. Sections 6 and 7 do not give any conclusions on the operational maturity (or TRL) of the models, unlike the others. Maybe provide one example of emblematic model for each class ? I suggest to remove the equations in section 3.

From a user’s perspective, it will be very useful/important to identify in Table 1 (i) one reference paper for each model ; (ii) which models are open access, and how to access the code in practice (e.g. on GitHub).

Minor comments:
- page 2, line 31: five -> six classes
- page 2, line 40: “modelled dynamically or derived from energy requirement”: I don’t understand the alternative in the sentence.
- page 2, line 43: “clear empirical interpretation”: interpretation capability ?
- page 3, line 67: how B differs from Bt ? Ct not in (eq 1)
- page 3, line 79: meaning of SS3, a4a, XSA ? please remove unnecessary acronyms.
- page 4, line 119: would it make sense to cite ichthyop (Lett 2008) explicitly ? https://ichthyop.org/ which has been used in a variety of projects ? (at least to be added to Table 1)
- page 5, line 147: please provide the GitHub reference
- page 6, line 174: is DBEM the same as the one cited in Section 5 (species distribution models) ? Is this an example of hybrid class model ?
- page 7, line 204: “general improvement of realism but also lower accuracies”: this is confusing. What is meant by “realism” ?
- Table 1: the legend should be checked and further developed to explain to meaning of each column (e.g. time units ? is it time scale ? “Name” should be “Model” and Model should better be “Acronym” ?)
- Table 1: a new column indicating open access, web site and (one) reference paper would be very helpful
- Table 1: the last row (Apecosm) should be filled in
Citation: https://doi.org/10.5194/sp-2024-42-RC2
- AC2: 'Reply on RC2', Simone Libralato, 25 Feb 2025
  
  The paper aims to provide a structured synthesis of numerical approaches to link lower trophic level models (physics, biogeochemistry and plankton) with higher trophic levels. The paper aims to cover in an organized way approaches from single individuals to populations to multispecies, and more emphasis could be put on their use as tools that can be linked to the products of the Copernicus Marine Service.
  1- This work does not claim to be exhaustive of the approaches for modelling marine ecosystems and therefore cannot be regarded as a user manual. Rather, it offers a classification of a spectrum of approaches that is broader than any single review considered and cited. Each class is described using examples of the most commonly used tools at a high trophic level to help the reader navigate the vast amount of models. What is probably underweighted is the fact that this review includes references to the potential use of each model for operational linkage with models at lower trophic levels and can therefore be considered a comprehensive map of ecosystem models for operational oceanography. Stressing this in the aims and discussing more the Table (to be improved) might help explaining shaping this work better.
  2- The classification of numerical approaches shows certain overlaps. For example, the distinction between minimal realist models and whole ecosystem models can be subjective, and furthermore, the complexity of single species to whole ecosystem models is constantly increasing. I therefore agree with the reviewer that a brief discussion of overlap and continuity between model classes could be beneficial.
  3-I concede that the concept of operational maturity has been little explained and is therefore obviously subjective and has been mixed up with the operational use of numerical tools in management. I suggest revising the paper to add (in Table 1) a) maturity of the numerical approach, b) use in management practise, and c) readiness for monitoring and forecasting marine ecosystems.
  Maturity is defined by the free availability of the code, documentation of the model and applications, test cases, diagnosis of model performance, and the reference community that updates and contributes to the model: This is a fairly objective assessment that can easily be added to the table and discussed to facilitate dissemination of the approach.
  The utility for management practise is defined based on the current use of the numerical tool in formal analyses validated by a scientific community and used for the regular issuance of management recommendations. This will provide information on the robustness of the approach.
  Suitability for monitoring and prediction will be based on the ability of the tool to provide large-scale analyses, possibly integrated with lower trophic level models that can assimilate real data and predict the dynamics of marine resources in space and time.
  The explanation and evaluation of all tools with respect to these three features will clarify the operational status of the models and allow a discussion of the six model classes.
  4- I agree with this suggestion (which is similar to that of RC1) on the need to harmonize the description of the 6 classes in terms of length, examples, equations, details. I propose to revise the classes description by keeping it at a length that permits to explain the most relevant features avoiding the gergon, to give one or two examples per class and to discuss more the maturity, use and readiness, in order to make the content more useful in terms of marine ecosystem monitoring and forecasting.
  5- I agree and I’m happy to revise the Table by adding reference, code availability and eventual repository. Considering the modifications suggested on the Table (see also previous points) the table will give more ground for discussion in the last section of the paper.
  Answers to Minor comments:
  - page 2, line 31: five -> six classes
  Answer: Yes, needs to be corrected.
  
  - page 2, line 40: “modelled dynamically or derived from energy requirement”: I don’t understand the alternative in the sentence.
  Answer: The sentence must be corrected as follows: “The energy input can be a model output or input, depending on whether it is modeled dynamically or not.”
  - page 2, line 43: “clear empirical interpretation”: interpretation capability ?
  Answer: Yes, needs to be corrected
  - page 3, line 67: how B differs from Bt ? Ct not in (eq 1)
  Answer: Yes, must be corrected, since B should be Bt in the equation. Ct is not in the equation, but corresponds to Ft*Bt.
  - page 3, line 79: meaning of SS3, a4a, XSA ? please remove unnecessary acronyms.
  Answer: Yes, needs to be corrected
  - page 4, line 119: would it make sense to cite ichthyop (Lett 2008) explicitly ? https://ichthyop.org/ which has been used in a variety of projects ? (at least to be added to Table 1)
  Answer: I agree. Could be the case, at least to add in the table. In case two examples per class are considered, ichthyop could also be described in the main text.
  - page 5, line 147: please provide the GitHub reference
  Answer: Yes, will be added
  - page 6, line 174: is DBEM the same as the one cited in Section 5 (species distribution models) ? Is this an example of hybrid class model ?
  Answer: DBEM is similar to SDM, but I would use it more as a hybrid class of models.
  - page 7, line 204: “general improvement of realism but also lower accuracies”: this is confusing. What is meant by “realism” ?
  Answer: Realism here means realism in the variety of processes and compartments represented. Accuracy is the ability of the model to represent the monitoring data. Thus, a model can be very realistic but have low accuracy. Definitions need to be added.
  - Table 1: the legend should be checked and further developed to explain to meaning of each column (e.g. time units ? is it time scale ? “Name” should be “Model” and Model should better be “Acronym” ?)
  Answer: Yes, needs to be corrected
  - Table 1: a new column indicating open access, web site and (one) reference paper would be very helpful
  Answer: Agreed. The table will be updated with these columns and those relating to maturity, utility and readiness.
  - Table 1: the last row (Apecosm) should be filled in
  Answer: Yes, needs to be corrected
  
  Citation: https://doi.org/10.5194/sp-2024-42-AC2
RC3:
'Comment on sp-2024-42', Anonymous Referee #3, 02 Dec 2024

The author reviews current numerical marine modeling approaches, from bioenergetic models for individual organisms to ecosystem-scale models, with a stated focus on understanding and managing marine systems under stress from human activities and climate change. The manuscript attempts the difficult task of briefly assessing the maturity of various modeling approaches while discussing their applications in marine resource management.
The manuscript's assessment of model maturity in the earlier sections (2-4) appears subjective and would benefit from more rigorous supporting evidence and citations. Often, it is not even clear what factor contributed to the assessment. An example is the apparent contradiction in the utility of spatial explicitness and oceanographic forcing for stock assessment models and connectivity models that need to be resolved or better explained: "Although in most cases, stock assessment models are not spatially explicit and do not consider explicitly oceanographic forcings they are routinely used in formal assessments for management and might be considered as ready for operational oceanographic applications." (l. 101) Yet, the text describing connectivity models seems to suggest that adding the missing features makes the models less mature: "Connected with oceanographic variables and spatially explicit these models however, appear less mature for operational applications." Perhaps rather than attempting maturity assessments, I would suggest using the approach taken in Section 5, which focuses on describing actual use cases and operational status.
In general, I would suggest revising each section to follow a consistent format: beginning with a brief description of the model class, followed by current use cases -- if available -- specifically related to human activities and climate change. Each section could conclude with an assessment of which current environmental challenges are already being addressed or need attention. This would be more valuable than the current approach, which primarily describes existing models and their structure. The list of models in Table 1 already serves this purpose, allowing the main text to focus more on the present status and outstanding issues in the field.
In Section 2, the description switches from a general description of bioenergetic models to the DEB subclass. If all current bioenergetic models are DEB models, this should be pointed out in the manuscript. If not, what are the alternatives and when are they used?
Starting from the title to about the last sentence of the abstract, a reader might think that the topic of the manuscript are coupled physical-biogeochemical models, as these are often referred to as ecosystem models. Here it would be useful to differentiate the type of model earlier, preferentially in the title.

In this context, it would be helpful to more clearly state what kind of organisms are being modeled using the approaches. Neither the abstract nor the introduction explicitly mention what is being modeled: The abstract speaks of "individuals, populations, communities and ecosystems", the introduction mentions "marine biological resources" without providing any examples. As stated above, I would suggest making use cases of each type of model more explicit.
# specific comments
L 22: "numerical models can be divided into six broad classes [...] These five classes of models are reviewed in the following sections ...": It should be six classes in both cases.
L 51: the "DEB" abbreviation was introduced in the preceding sentence.
L 53: "abstract concepts that are more challenging to measure empirically": What would be an example of that?
L 54: "The presence of the storage": Here it would be easier for the reader to refer to "storage tissue" again, using the exact term introduced earlier.
Eq 1: What is n here?
L 67: "fishing mortality": So this type of model is only used to model populations of species that are being fished? Why not mention this early on explicitly?
L 79: "such as SS3, a4a, XSA, etc.": Listing these names/abbreviations here is not useful. Readers unfamiliar with these approaches are only given 3 letter abbreviations without any context or citations, and readers who know these approaches probably don't need the see the abbreviations again. Here, it would be much more useful to describe these sophisticated approaches in words.
Eq. 2: I think it is not useful to show the equation here, it can be more easily explained in words. If it is kept, the initial N should have a subscript t, and the "a=" is missing from some of the subscripts.
L 153: "considering the specific problem": What is the specific problem?
L 157: "MICE can simultaneously represent focal populations in age-structured classes, while others take a surplus production approach": This could be described better, what is the surplus production approach, and how does it compare to an age-structured approach?

Citation: https://doi.org/10.5194/sp-2024-42-RC3
- AC3:
  'Reply on RC3', Simone Libralato, 25 Feb 2025
  The reviewer's comment on the maturity assessment is completely in line with the previous reviewers' comments. Basically, not only did I fail to clearly define and measure maturity, but a possible confusion between different aspects such as maturity and use in management became apparent. I suggest defining and adding in the table: a) maturity of the numerical approach, b) application in management practise, and c) readiness for monitoring and prediction of marine ecosystems.
  Maturity is defined by the free availability of the code, documentation of the model and applications, test cases, diagnosis of model performance, and the reference community that updates and contributes to the model: This is a fairly objective assessment that can easily be included in the table and discussed to facilitate dissemination of the approach. b) The utility for management practise will be defined based on the current use of the numerical tool in formal analyses validated by a scientific community and used for the regular issuance of management recommendations. This will provide information on the robustness of the approach. c) readiness for monitoring and forecasting will be based on the ability of the tool to produce large-scale analyses, potentially integrated with lower trophic level models that can assimilate real data and predict marine resource dynamics in space and time.
  
  This should contribute to greater coherence, pragmatism and objectivity, although there will still be a degree of subjectivity in the assessment of readiness to monitor and predict.
  In agreement with other reviewers, there is also a need for RC3 to have a consistent format for the different model classes. An explanation of the current approach, minimal descriptions and applications will be brief, but I think they absolutely need to be included. However, as the reviewer suggests, I am happy to improve the description of the environmental aspects addressed for each class and include them at the end of each section. In some cases (whole ecosystem models) this is already explored to some extent, but I agree with harmonizing this structure for all classes.
  I agree that was posed an exaggerated attention to DEB modelling in section 2. This is a widely used approach but not the only one. I would revise in order to downplay the role of DEB.
  The title was given and cannot be changed as it is part of a collection of pre-defined titles. Nevertheless, I understand the question about the coupled physical-biogeochemical models: The solution I propose is to start the paper with a better description of the objectives. Namely, the aim of the paper is to provide a structured synthesis of models for the higher trophic levels of the oceans (essentially from zooplankton to top predators and fisheries) that can be linked to models for the lower trophic level (physics and biogeochemistry). The lower trophic level models currently provide operational products in the Copernicus Marine Service, and their potential to be linked to upper trophic level models opens up a wide range of new CMS products. The aim is to cover most high trophic level approaches, from single individuals to populations and multispecies, but the work may not be exhaustive. Conversely, the paper provides a classification and examples of the most commonly used high trophic level tools with some indication of their potential use for operational linkage to lower trophic level models.
  In the same vein, further explanation of the organisms being modelled is added in the introduction and in each section, with explicit mention of the most common applications, typical species and challenges.
  Answers to specific comments
  L 22: "numerical models can be divided into six broad classes [...] These five classes of models are reviewed in the following sections ...": It should be six classes in both cases.
  Answer: Yes, needs to be corrected
  L 51: the "DEB" abbreviation was introduced in the preceding sentence.
  Answer: Yes, needs to be corrected
  L 53: "abstract concepts that are more challenging to measure empirically": What would be an example of that?
  Answer: The intention here was to point out the “storage” of the DEB modelling approach, which is quite difficult to measure empirically. However, I concede that it is not an abstract concept. The text needs to be revised here.
  L 54: "The presence of the storage": Here it would be easier for the reader to refer to "storage tissue" again, using the exact term introduced earlier.
  Answer: Yes, needs to be corrected
  Eq 1: What is n here?
  Answer: It is a typo, but this equation will be removed for reasons of homogeneity between the sections.
  
  L 67: "fishing mortality": So this type of model is only used to model populations of species that are being fished? Why not mention this early on explicitly?
  Answer: Yes, I agree. It needs to be added
  
  L 79: "such as SS3, a4a, XSA, etc.": Listing these names/abbreviations here is not useful. Readers unfamiliar with these approaches are only given 3 letter abbreviations without any context or citations, and readers who know these approaches probably don't need the see the abbreviations again. Here, it would be much more useful to describe these sophisticated approaches in words.
  Answer: I understand that these acronyms confuse the general public and are useless (or incomplete) for fishing experts. I will use general descriptions in the maintext and refer to the table to describe these instruments in more detail.
  
  Eq. 2: I think it is not useful to show the equation here, it can be more easily explained in words. If it is kept, the initial N should have a subscript t, and the "a=" is missing from some of the subscripts.
  Answer: similar comment also from other reviewers. I will correct and remove the equation.
  
  L 153: "considering the specific problem": What is the specific problem?
  Answer: The intent here was to explain that MICE is a problem-specific model. Basically it is adapted to describe the ecosystem on the basis of the question to answer or issue to assess. This text will be revised for increase clarity.
  
  L 157: "MICE can simultaneously represent focal populations in age-structured classes, while others take a surplus production approach": This could be described better, what is the surplus production approach, and how does it compare to an age-structured approach?
  Answer: agree, more details and explanations are needed in order to allow non expert of this kind of models to appreciate their characteristics.
  
  Citation: https://doi.org/10.5194/sp-2024-42-AC3

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

ED: Publish subject to revisions (further review by editor and referees) (06 Mar 2025) by Kirsten Wilmer-Becker

AR by Simone Libralato on behalf of the Authors (14 May 2025) Author's response Author's tracked changes Manuscript

ED: Publish as is (14 May 2025) by Kirsten Wilmer-Becker

AR by Simone Libralato on behalf of the Authors (15 May 2025) Author's response Manuscript

Short summary

This work examines the current classification of numerical models of increasing complexity – from individuals and population and stock assessment models to models representing the whole ecosystem by covering all trophic levels – and presents examples and their operational maturity, finally demonstrating their use for supporting marine resource management, conservation, planning, and mitigation actions.