Ecopath with Ecosin - Network to Restore our Atlantic Fishery 2076

What do you know about ecopath with ecosin?

Ecopath with Ecosim (EwE) is a free, open-source ecological/ecosystem modeling software suite widely used for analyzing food webs, especially in aquatic (mainly marine and fisheries) systems, though it can apply to terrestrial ones too. (LINK)

It originated from work by Jeffrey Polovina at NOAA in the early 1980s (the original Ecopath approach for mass-balance modeling of ecosystems). It was significantly expanded and developed primarily at the University of British Columbia’s Fisheries Centre by researchers including Villy Christensen, Carl Walters, and Daniel Pauly, with ongoing support from the Ecopath International Initiative (a non-profit) and a global network of users and contributors. In 2007, NOAA recognized it as one of its top ten scientific breakthroughs in 200 years for revolutionizing the understanding of complex marine ecosystems. (LINK)

Main Components

EwE has three core modules:

Ecopath: Creates a static, mass-balanced “snapshot” of an ecosystem. It represents the system as trophically linked biomass “pools” (species, functional groups/guilds, or life stages). It quantifies energy/biomass flows, production, consumption, and other parameters under a mass-balance assumption (typically over a year). It builds on network analysis ideas (e.g., from Ulanowicz) and is parameterized largely from available stock-assessment and ecological data. (LINK)

Ecosim: A time-dynamic simulation module that takes the balanced Ecopath model and runs it forward (or backward) in time using differential equations. It explores how biomasses and flows change under varying fishing pressure, environmental drivers (e.g., primary productivity), or other forcing. It incorporates concepts like foraging arena theory to handle top-down/bottom-up control and vulnerability of prey. (LINK)

Ecospace: Adds spatial and temporal dynamics on a grid. It is used especially for evaluating the placement and impacts of marine protected areas (MPAs), spatial fishing patterns, and environmental heterogeneity. (LINK)

Additional capabilities include Ecotracer (for tracking contaminants or tracers) and support for end-to-end modeling.

Common Uses

– Addressing ecological questions about food-web structure and functioning.

– Evaluating ecosystem effects of fishing and other human pressures.

– Exploring management policy options and “what-if” scenarios.

– Analyzing impacts and optimal placement of protected areas.

– Modeling effects of environmental change (e.g., climate-related productivity shifts).

– Supporting ecosystem-based fisheries management.

It is applied worldwide (thousands of users in 150+ countries, hundreds of published models and papers). There is an open repository called EcoBase with metadata and downloadable models. (LINK)

Practical Details

Current status: Latest stable release is version 6.6.8 (November 2022). Version 6.7 has been in development/testing (features like improved fitting, UI updates, etc.), with builds available and a full release anticipated around 2025. New textbooks/user guides explaining theory and software use were released around 2024. (LINK)

– Primarily a Windows desktop application (requires Microsoft Access database drivers); the computational core can run on other systems via Mono.

– Free and open-source; development is supported by projects, user contributions, training, professional licenses, and collaborations.

– Official site: [ecopath.org](https://ecopath.org/).

Does ecopath use forage biomass or does it take into account energy density of the forage?

Ecopath with Ecosim (EwE) primarily uses forage (prey) *biomass*, not energy density automatically.

Core Approach

Ecopath is a mass-balance model. The fundamental units are almost always biomass (typically tonnes per km²).

– Biomasses of functional groups (including forage species or groups) are entered directly.

– Diet compositions are entered as proportions of the predator’s diet (usually by weight or volume).

– Consumption rates are handled via the consumption/biomass ratio (*Q/B*, year⁻¹), which is also biomass-based.

– The two master equations balance production, consumption, predation, respiration, unassimilated food, fishing, etc., in biomass (or sometimes carbon/nutrient) units.

In a standard biomass-currency model, two prey items of equal biomass are treated as equivalent food, regardless of differences in energy density (caloric content). The model does not automatically convert biomass to energy or adjust for energy density of the forage.

When Energy Density Can*Be Accounted For

The software is flexible about the “currency” of the model:

– You can build the entire model in energy units (e.g., kJ or kcal per km²). In that case, biomasses, diets, P/B, Q/B, etc., must all be expressed in energy terms. Energy density of each group is then incorporated by converting the original biomass data into energy before entering parameters.

– Diet data can be entered on an energy-content basis instead of wet-weight or volume basis (the user guide notes that diet inputs can represent “weight, volume or energy content of the preys”).

– The “unassimilated consumption” parameter is especially relevant when the currency is energy-related.

In practice, most published EwE models use wet-weight (or dry-weight) biomass as the currency. Energy density is only explicitly handled if the modeler deliberately converts everything to energy units or manually adjusts diet proportions / consumption rates to reflect differences in caloric value of different forage species.

Summary

| Aspect | Standard Ecopath practice | Energy-aware approach |

|——————————-|————————————|—————————————-|

| Primary unit | Biomass (t km⁻²) | Energy (e.g., kJ km⁻²) |

| Forage representation | Biomass of prey groups | Energy content of prey groups |

| Automatic energy-density adjustment | No | Only if model is built in energy units |

| Diet matrix | Usually by weight/volume | Can be by energy content |

So, by default, Ecopath uses forage biomass. Accounting for energy density of the forage requires deliberate parameterization by the modeler (either converting the whole model to energy currency or adjusting inputs accordingly).

Ecopath was used by David Chagaris on work with ecological reference points and menhaden. There was a YouTube video on the status in 2022. Do you have any more recent information on that study?

Yes — there are substantial updates since the 2020/2022 period.

David Chagaris (University of Florida) led the development of ecological reference points (ERPs) for Atlantic menhaden using an intermediate-complexity Ecopath with Ecosim model (NWACS-MICE). The core 2020 paper (Chagaris et al., Frontiers in Marine Science) and the associated 2020 Ecopath seminar (the YouTube video you mentioned) established the approach: ERPs based on the tradeoff between menhaden fishing mortality (F) and striped bass biomass (the most sensitive predator), while striped bass is fished at its own target F. Those original ERPs (Ftarget ≈ 0.19, Fthreshold ≈ 0.57) were adopted by the Atlantic States Marine Fisheries Commission (ASMFC) Menhaden Management Board in 2020 — the first interdependent multispecies reference points used for U.S. fishery management. (LINK)

Most Recent Developments (2023–2025/2026)

A full 2025 ERP Benchmark Stock Assessme (SEDAR 102) refined and updated the NWACS-MICE model and the ERPs. Key points:

– The model was recalibrated with newer data, updated natural mortality, and refinements to predator–prey links (including attention to species such as spiny dogfish and Atlantic herring).

– New ERP values (using the same management definition based on striped bass biomass targets/thresholds) are **lower** than the 2020 values: roughly *F*target ≈ 0.15 and *F*threshold ≈ 0.46 (exact figures vary slightly by final interpolation/run; one presented set was *F*target 0.151 / *F*threshold 0.458).

– The 2025 single-species assessment update applied these revised ERPs. Based on 2023 terminal-year estimates, **Atlantic menhaden is not overfished and overfishing is not occurring** (*F*2023 is below the ERP threshold but above the target; fecundity is above the ERP threshold but below the target).

– Chagaris remains actively involved (presentations at the August 2025 SEDAR 102 Review Workshop in Charleston, SC, ongoing ERP Work Group meetings through 2025, and model runs documented in working papers).

– Work continues on incorporating seasonal/spatial dynamics and alternative productivity scenarios.

The full 2025 ERP Benchmark Assessment Report + Peer Review Report is available via ASMFC and the SEDAR 102 project page. Supporting materials (presentations, working papers on model runs, natural mortality updates, etc.) are also posted there.

Related Gulf of Mexico Work

Chagaris co-authored a parallel effort for Gulf menhade (Berenshtein, Sagarese, Lauretta, Schueller & Chagaris, 2023, *Frontiers in Marine Science*). It used a U.S. Gulf-wide EwE model to quantify trade-offs between menhaden harvest and predator biomass (king mackerel, Spanish mackerel, red drum, etc.) and to explore candidate ERPs. Current (post-2008) fishing levels generally fall within the suggested ERP ranges. (LINK)

Chagaris’s lab page continues to track both the Atlantic and Gulf menhaden ERP efforts and lists the key publications and meeting summaries.

In short, the original 2020 NwACS-MICE ERP framework is still the foundation, but it received a major peer-reviewed update and recalibration in 2025 that produced more precautionary (lower) reference points now used for management advice.

– Official site: [ecopath.org](https://ecopath.org/).

Does his work ever use forage energy density in the Ecopath model or is it just using biomass?

Chagaris’s Ecopath models for menhaden ecological reference points use biomass (not energy density).

Atlantic Menhaden (NWACS-MICE)

The core model used for the 2020 ERPs (and the 2025 SEDAR 102 update) is a reduced-complexity Ecopath with Ecosim model of the Northwest Atlantic Continental Shelf (NWACS-MICE).

– It is parameterized in standard **biomass units** (typically metric tons per km²).

– Biomass, production/biomass (*P/B*), consumption/biomass (*Q/B*), and diet compositions are all treated on a wet-weight (or equivalent mass) basis.

– Diets are entered as proportional contributions by mass.

– There is no indication in the published papers (Chagaris et al. 2020 and related ERP documents), technical reports, or SEDAR materials that the model currency was converted to energy units or that forage energy densities were used to rescale biomasses, diets, or consumption rates.

Gulf of Mexico Menhaden Work

The related Gulf-wide EwE model (Berenshtein, Sagarese, Lauretta, Schueller & Chagaris 2023 and supporting NOAA technical documentation) is also constructed and calibrated in biomass units (mt/km²). Outputs such as biomass time series, predation mortality, and trade-off surfaces are reported in biomass terms. No conversion to energy density appears in the model description or results.

Context on Energy Density

Separate studies have measured energy density (caloric content) of Gulf menhaden and other forage species in the region and note that these values *could* improve ecosystem models. However, those data have not been incorporated into Chagaris’s published ERP models as the model currency or as a systematic adjustment to forage quality. In short, the models follow the standard Ecopath practice of mass-balance in biomass units. Energy density of the forage is not accounted for.

What actual testing has been done to test the accuracy of the Ecopath and ecosim systems?

Testing of Ecopath with Ecosim (EwE) focuses on sensitivity analyses, hindcasting (fitting to historical time series), skill metrics, and limited forecast evaluations rather than large-scale independent experimental validation There is no single definitive “accuracy test” of the overall system because it is a flexible modeling framework applied to many ecosystems; instead, individual models are assessed through standard practices and published studies.

1. Sensitivity and Precision Testing (Ecopath Component)

– Early work (e.g., Essington 2007) added error to inputs of nine published Ecopath models and measured how well the mass-balance process recovered true biomass and ecotrophic efficiency (EE). Prediction errors were generally comparable in magnitude to the input data uncertainty. Biomass and production/biomass (*P/B*) were the most influential (high-leverage) parameters; the balancing process itself did not greatly reduce error, and tightly linked food-web cycles could amplify errors. (LINK)

– A 2021 study (Susini & Todd) systematically added imprecision to four basic inputs across eight published models and examined effects on six ecosystem indicators. Kempton’s Q and total system throughput were most responsive; input biomass was again identified as high-leverage. The authors concluded that a dedicated sensitivity tool within EwE would be valuable. (LINK)

2. Time-Series Fitting / Hindcasting (Ecosim Component)

The primary practical test of Ecosim is calibration to historical biomass, catch, and effort time series by estimating vulnerability parameters (predator–prey interaction strengths).

– Goodness-of-fit is quantified with sum-of-squares (SS), Akaike Information Criterion (AIC), correlation, modeling efficiency, bias, and reliability metrics.

– Best-practice guidance (Heymans et al. 2016 and later reviews) emphasizes formal statistical fitting rather than default vulnerability settings. (LINK)

– Examples of validated models include a Norwegian Sea/Barents Sea EwE model fitted from 1950–2000 against VPA and survey data (reasonably good fits, improved by primary-production forcing). (LINK)

– Studies of data-limited cases (e.g., Lake Victoria) show that models without historical fitting can produce divergent predictions, especially at higher fishing pressures; adjusting biomass-accumulation terms or using short-term trends can reduce differences. (LINK)

– Recent work (Ren et al. 2025) compared vulnerability settings for hindcast and forecast skill. Vulnerability-fitted models best reproduced historical dynamics; among unfitted alternatives, trophic-level-based settings performed better for hindcasts, while depletion-related settings were more robust for forecasts under changed fishing. (LINK)

3. Skill Assessment and Forecast Testing

– Formal skill metrics (bias, error, reliability, modeling efficiency, Spearman rank correlation, etc.) are increasingly applied to both hindcasts and short-term forecasts.

– Spatial skill assessment tools exist in Ecospace (e.g., hotspot analysis against telemetry data, regional biomass comparisons).

– ICES Working Group on Multispecies Assessment Methods (WGSAM) “key runs” evaluate whether outputs (e.g., *F*MSY) are ecologically sensible and not outside historically observed ranges.

– Predictive skill can degrade if models over-fit historical data or miss processes such as recruitment variability. Hindcast success does not guarantee forecast reliability.

4. Broader Critiques and Limitations of Testing

– Early reviews (e.g., Plagányi & Butterworth 2004) noted a paucity of systematic stepwise investigations of model behavior, potential over-reliance on defaults, and incomplete treatment of uncertainty. (LINK)

– Common issues include the constraining nature of the initial mass-balance assumption, sensitivity to vulnerability parameters, and the fact that many published models are never rigorously fitted or cross-validated.

– Independent experimental validation (e.g., controlled microcosm tests of similar food-web models) has sometimes shown poor predictive power even when fits to data look good, raising general caution about complex ecosystem models. – Uncertainty is commonly explored via Monte Carlo routines, but these have known limitations for balancing and advice products.

Summary of Actual Testing Practices

|——————————-|—————————————————|—————————————————-|————————————-|

| Management “key runs” | Check ecological realism of reference points | Sensible *F*MSY ranges | Used by ICES and similar groups

|In practice, confidence in any particular EwE application rests on transparent documentation of data quality (pedigree index), formal fitting diagnostics, sensitivity analyses, and comparison with independent observations or alternative models. Models that successfully hindcast historical dynamics under fishing and environmental forcing are considered more credible for strategic (not tactical) advice, but residual uncertainty remains high, especially for forecasts.

Has it ever been tested comparing the EwE model using forage biomass against an EwE model using a variable forage biomass that includes energy density?

Yes — at least one published study has directly compared a standard biomass-based (mass-balanced) EwE model against energy-balanced versions that incorporate energy density of functional groups, including forage fish.

Key Study: Surma et al. (2018)

Szymon Surma, Evgeny A. Pakhomov, and Tony J. Pitcher published “Energy-based ecosystem modelling illuminates the ecological role of Northeast Pacific herring” in *Marine Ecology Progress Series* (2018). (LINK)

They took an existing mass-balanced Ecopath model of northern British Columbia / southeastern Alaska waters and converted it into a set of **energy-balanced models** by applying published energy-content (energy-density) values (from bomb calorimetry data) to all functional groups.

What they compared:

– Ecosystem size-spectrum slopes

– Proportions of Pacific herring in predator diets

– Contributions of major ecological categories (e.g., forage fish, pelagics, benthos, producers) to total ecosystem biomass vs. total ecosystem energy pools

– Dynamic Ecosim simulations of the trophic effects of recovering depleted whale populations (and related herring-focused scenarios)

Main findings relevant to the comparison:

– Energy-balanced models increased the apparent importance of herring (a high-energy-density forage fish) as prey. Herring contributions to the diets of several marine mammals, seabirds, and commercial fish were ~20–110% higher than in the original mass-balanced model.

– Forage fish and other pelagic groups made up higher proportions of the total ecosystem pool when expressed in energy rather than biomass.

– Balancing the high-energy version required a noticeable (but non-significant) steepening of the size-spectrum slope relative to the mass-balanced model.

– Dynamic Ecosim runs showed that the trophic impacts of whale recovery (and herring depletion scenarios examined in related work) were stronger for some endothermic predators in the energy-based models than in the biomass-based version.

The authors concluded that energy-based models are viable and useful for re-evaluating the ecological role of high-quality forage species, and that ignoring energy density can understate the importance of energy-rich prey such as adult Pacific herring.

Broader Context

– EwE itself supports either biomass or energy as the model currency (the user guide notes that diets can be entered by weight, volume, or energy content, and that unassimilated consumption is especially relevant for energy-based models).

– However, nearly all operational applications (including the menhaden ERP models discussed previously) remain in biomass units.

– Explicit head-to-head tests of biomass-only vs. energy-density-adjusted versions remain rare. The Surma et al. (2018) work is the clearest published example that systematically converts an entire food-web model and compares structural metrics plus dynamic simulations.

In short, EwE is one of the most widely used and accessible tools for quantitative food-web and ecosystem modeling in marine science and fisheries, bridging static description with dynamic simulation and spatial analysis.
No large-scale, multi-system validation exercise appears to have systematically tested whether energy-density-adjusted EwE models systematically outperform (or underperform) standard biomass versions against independent data. The Surma study demonstrates that the difference can be ecologically meaningful for forage-fish questions, particularly for endothermic predators.