1 Introduction

1.1 MERA is under development!

Please note that MERA is currently under development. MERA is beta software and subject to ongoing testing.

If you find a bug or have a suggestion please contact

1.2 Overview

Fisheries managers are in need of tools to inform decision making often in the face of imperfect information and high uncertainty. By linking a simple user interface and questionnaire (Figure 1.1) to an operating model and sophisticated simulation software, MERA provides managers with an accessible and powerful tool for identifying species at risk, selecting management procedures that can achieve their performance objectives, and calculating stock status.

MERA is intended to better account for uncertainty in the fishery system, prioritizing robust management options and identifying value in alternative data collection and research programs. By focusing on operational modelling, MERA can provide quantitative outputs that are central to fishery legal frameworks and eco-certification standards, for example probabilistic estimates of stock status relative to reference levels. MERA lessens the reliance on subjective, qualitative scoring systems, increasing transparency and accountability in decision making.

Furthermore since the App is compatible with the R statistical software operating models, management procedures and diagnostics are all customizable allowing for bespoke state-of-the art closed-loop simulation including MSE.


Figure 1.1. The MERA user interface.

1.3 How to use MERA

MERA has two inputs: a mandatory quantitative questionnaire and optionally, a standardized format for fishery data (Figure 1.2). The questionnaire contains a total of 30 questions, 19 regarding the fishery dynamics, 7 questions about the management system and a further 4 about the types and quality of data that are available (see Section 2 for details on all questions).

Once completing the questionnaire users must select from four use cases:

  1. Risk Assessment – quantifying future biological risk of status quo management

  2. Management Planning – determining a suitable management mode

  3. Management Performance – evaluating current management mode

  4. Status Determination – calculating stock status.

The first three use the quantitative questionnaire to construct an operating model while the status determination mode requires additional user data to estimate stock status. Any operating model can be further conditioned (‘fitted’) using fishery data. Where operating model features are required that are not included in the quantitative questionnaire it may be exported into the R statistical environment, modified and imported back into the MERA software. The R packages DLMtool (Carruthers and Hordyk 2018) and MSEtool (Huynh et al. 2019) are compatible with MERA operating models and include a wide range of tools for modifying these to include custom dynamics such as ontogenetic habitat shifts, fine-scale movement dynamics, time varying movement, temporal changes in growth and natural mortality rate, shifting fleet size selectivity and economic constraints.


Figure 1.2. MERA components and workflow.

1.4 A note on skins

MERA includes considerable flexibility on how results of the four use-cases are presented. This is to allow various user groups to obtain results that are pertinent to their application and setting.

For the purpose of this documentation results are presented for a prelminary beta version of the the Marine Stewardship Council skin.


1.5 Background to MERA

1.5.1 The need for informed management

Fisheries management systems typically involve an ongoing cycle of harvest, data collection, data processing, resource assessment, management recommendations and enforcement of management measures (Walters and Martell 2004). Fisheries managers are embedded in these systems and must make critical decisions, many of which are often poorly informed.

For example, managers must apportion budgets among various species and competing programs such as data-collection, scientific research, resource assessment and enforcement. Managers must select an appropriate time interval for renewing management advice and the correct level of model complexity for assessing a stock. They must also select among many ways to interpret stock assessment outputs in terms of management advice (i.e. a harvest control rule). Additionally managers may be expected to guide their science program in the direction of the most critical uncertainties in the system that are most in need of research.

Given that it generally involves the use of public money in the stewardship of public resources it is problematic that even in developed fishery management systems, decision making is often ad-hoc, lacking both transparency in the basis for decisions such as those listed above, and a coherent overall strategy (but see IWC, other exceptions in Australian / South African fisheries REFS).

1.5.2 Options for informing strategic management

There are two options for evaluating competing modes of management in order to inform management strategy: experimentation and theoretical modelling. Exploited ecological systems are a notoriously challenging subject for theoretical systems modelling. Due to biological, behavioral and ecological complexities (Rouyer et al. 2008, Sugihara et al. 2011), shifting productivity (Vert Pre et al. 2013), changing environmental conditions, unobserved exploitation (Agnew et al. 2009), variable quality of data and many other factors besides, models may be an inadequate representation of the system, failing to capture critical system dynamics and providing unreliable predictions.

Empirical testing recognizes these difficulties and evaluates alternative management modes applied in practice to the real system. However, for practical reasons the experimental approach is not feasible for most fisheries. Experimental replication is not possible when the subject is a single fish stock existing in a unique ecological setting precluding rigorous and statistically valid empirical evaluation. Even if it were feasible, the statistical power to detect system changes over relevant time horizons may be expected to be low and experimentation expensive, accounting for the costs of additional data collection and lost fishery yields. For such reasons, proposed experimental approaches to fishery management such as adaptive management (Holling 1978, Walters 2002) have not seen widespread adoption despite their empirical advantages (Walters 2007, Westgate et al. 2013).

The alternative, theoretical testing, relies on the development of representative systems dynamics models (operating models) to evaluate the expected performance of candidate management modes. While operating models can be used to inform a wide range of management questions, previous applications in fisheries have generally focused on Management Strategy Evaluation (MSE) in which various ways of setting management advice using data (‘management procedures’) are comparatively evaluated (Butterworth and Punt 1999, Punt et al. 2016).

While there have been criticisms of operational modelling (Rochet and Rice 2009, in reference to MSE), the potential advantages of the approach have made it on ongoing priority for developed fishery management systems at various scales for example, US state fisheries (CDFW 2018), federal fisheries in the US (NOAA 2019) and Canada (Kronlund et al. 2013) and for high seas tuna stocks (Anon 2018). Prevailing obstructions to more widespread development and adoption of operating models include relatively high costs (compared with a one-off assessment of the population) and the availability of suitably qualified analysts and data to inform various scenarios for system dynamics.

1.5.3 Contemporary approaches for informing fisheries management

In the absence of rigorous operational modelling, fisheries resource management has often relied on subjective, qualitative frameworks for guidance. A large number of such frameworks exist that aim to evaluate biological risks (Productivity Susceptibility Analysis, PSA: Hobday 2007), stock status (Rapfish; Pitcher 1999), select management options (FISHE: EDF 2019) and prioritize management issues (Fletcher 2005). Such systems have been used widely. For example, PSA has been used in the eco-certification of global fish stocks (Seafoodwatch 2016) and satisfied legal requirements for ecosystem-based fishery management (Hobday et al. 2011).

Since they simply formalize expert judgement into a scoring system that accepts subjective inputs, it is hard to objectively evaluate both the validity of their assumptions and the quality of advice that they provide. When such systems have been codified and subject to theoretical testing their performance was found to be poor (e.g. biological risk assessment using PSA, Hordyk and Carruthers 2018).

1.5.4 MERA objectives

The primary objective of the MERA software (Carruthers et al. 2019) was to provide an accessible tool for the construction of operating models enabling fishery managers to make informed strategic decisions at various levels including data collection, species prioritization, MP selection and enforcement.

A secondary objective was to support an accessible user interface with sophisticated models and statistical libraries to allow for state-of-the-art operational modelling and MSE when required. Thirdly, MERA was designed to be applicable to the widest range of fisheries possible, various in their scientific understanding, data availability, biological, ecological and exploitation characteristics.

Lastly, MERA had to be customizable and inform diverse user groups including regional fishery management organizations, international development agencies such as the UN Food and Agricultural Organization and seafood certification bodies such as the Marine Stewardship Council


2 Questionnaire

2.1 Fishery Questions

The Fishery panel is a set of questions about the characteristics of the fish population and its fishery. The operating model parameter mappings of each answer are provided in Appendix Table A.1

2.1.1 Fishery description

Describe the fishery you are modelling and identify yourself and the relevant management agency.

The ‘number of years’ is important and will be used to simulate your historical fishery. It is the total number of years that substantive fishing (that could impact the stock) has been operating for. So for example, if it is 2020 this year, and the fishery began in 1981, this would be 40.

The text input box provides a place for you to document important qualitative aspects of the fishery that provide necessary context for the questionnaire including

  1. The history and current status of the fishery, including fleets, sectors, vessel types and practices/gear by vessel type, landing ports, economics/markets, whether targeted/bycatch, other stocks caught in the fishery.

  2. The stock’s ecosystem functions, dependencies, and habitat types.

  3. Any relevant reference materials and supporting documents, such as stocks assessments, research, and other analyses.

Figure 2.1. The fields of the fishery description question.

2.1.2 Longevity

How long lived is the fish species? This is a critical input determining stock productivity. The parameter M is the instantaneous natural mortality rate. For a review of data-limited methods of estimating M see Kenchington (2014).

Figure 2.2. An example of MERA specified longevity expressed as ‘maximum age’ and the instantaneous natural mortality rate M.

2.1.3 Stock depletion

Depletion D, refers to current spawning stock biomass relative to unfished.

Since depletion is a data-rich quantity it may not be readily quantified and it may be necessary to specify a wide range of uncertainty for this input to identify MPs that are suitably robust.

In a data-limited situation, coarse information regarding depletion may be obtained from examining length compositions, historical versus current catch rates, or by use of so-called Robin-Hood approaches.

For further information see Carruthers et al. (2014) and Punt et al (2011)

Figure 2.3. An example of MERA specified stock depletion - current spawnign stock biomass relative to ‘unfished’ levels.

2.1.4 Resilience

How resilient to exploitation is the stock? This question controls recruitment compensation - the extent to which recruitment is reduced from unfished levels (R0) as the spawning stock becomes increasingly depleted below unfishe levels (SSB0). Here resilence is expressed in terms of steepness (h): the fraction of unfished recruitment at 1/5 unfished spawning biomass.

For a useful review of compensatory density dependence in fish populations see Rose et al. (2001).

Figure 2.4. An example of MERA specified recruitment compensation (resilience) expressed as steepness: the fraction of unfished recruitment expected to occur at 1/5 unfishned spawning stock levels.

2.1.5 Historical effort pattern

What temporal pattern best describes the trend in historical annual fishing effort (e.g. boat-days per year, number of trips per year)?

If more than one answer is given, historical fishing will be simulated subject to all trends in equal frequency

If a very specific pattern of effort is required, you can use the sliders to warp the effort patterns.

This question specifies the possible range of mean trends, you will have an opportunity to adjust the extent of inter-annual variability and changes in fishing efficiency (catchability) in the following questions.

Here is an introduction to fishing effort courtesy of the UN FAO.

Figure 2.5. An example of MERA specified historical trend in fishing effort.

2.1.6 Inter-annual variability in historical effort

The extent of interannual variability in historical exploitation rates around the mean trend(s) specified in Fishery question #5. Again, here is the introduction to effort and exploitation rate by the UN FAO..

Figure 2.6. An example of MERA specified historical trend in fishing effort rate subject to additional inter-annual variability.

2.1.7 Historical fishing efficiency changes

The annual percentage increase or decrease in historical fishing efficiency. In targeted fisheries gear efficiency may improve over time given techological improvements in the gear, changes in fishing behavior, fish distribution and information sharing among fishers, among other things. Conversely, non-target or bycatch species may be subject to declining fishing efficiency due to regulations or avoidance behaviors. The catchability (q) is the fraction of available fish caught per unit of effort. For example, a 2% per annum increase in fishing efficiency means that after 35 years twice as many fish will be caught for the same effort as today.

The introduction to fishing efficiency by the FAO provides a basic summary and Arrenguin-Sanchez provide a more comprehensive review of catchability.

Figure 2.7. An example of MERA specified historical fishing efficiency changes.

2.1.8 Future fishing efficiency changes

This is similar to the previous question but determines the future relationship between fishing mortality rate and effort. This is a principal driver determining the performance differential of management procedures that set catch (TAC) versus effort advice (TAE).

Figure 2.8. An example of MERA specified future fishing efficiency changes.

2.1.9 Length at maturity

Size a maturity relative to asymptotic length (LM).

Note 1: ‘maturity’ as used by this model (and most fish population dynamics models) is not really whether a fish has fully developed gonads, but rather the fraction of maximum spawning potential per weight. For example, some fishes mature early, but at small sizes they spawn infrequently and their recruits have poor survival (low spawning fraction).

Note 2: asymptotic length is not the maximum length observed but rather the mean expected size of fish at their maximum age under unfished conditions

An ICES workshop report provides an overview of maturity estimation

Figure 2.9. An example of MERA specified maturity at length, relative to asymptotic length.

2.1.10 Selectivity of small fish

Fishing gear selectivity relative to asymptotic length (S) (ascending limb selectivity). For example, if 50% of 40cm fish are caught and maximum length is 100cm, S = 0.4.

The UN FAO provides an introduction to gear selectivity and how it may be quantified. For a more involved discussion on selectivity see the IATTC CAPAM workshop report

Figure 2.10. An example of MERA specified selectivity at length (relative to asymptotic length) for the ascending (smaller fish) limb of the curve.

2.1.11 Selectivity of large fish

Fishing gear selectivity of the largest individuals (SL). For example, if only 20% of the longest fish are caught by the gear SL = 0.2. Again here is the FAO introductory document and the IATTC CAPAM workshop report.

Figure 2.11. An example of MERA specified selectivity at length (relative to asymptotic length) for the descending (larger fish) limb of the curve.

2.1.12 Discard rate

Discard rate: what fraction of fish that are caught are discarded (includes fish that are dead and alive)?

The US National Marine Fisheries Service have a general guide to Understanding Fish Bycatch Discard and Escapee Mortality and one of the authors of that guide, Michael Davis also has a useful article: Key principles for understanding fish bycatch discard mortality.

Figure 2.12. An example of MERA specified discarding rate (fraction of fish caught that are released).

2.1.13 Post-release mortality rate

Post-release mortality rate (PRM). What fraction of discarded fish die after release?

Again here is NOAA’s general guide to Understanding Fish Bycatch Discard and Escapee Mortality and one of the authors and the article fo Michael Davis: Key principles for understanding fish bycatch discard mortality.

Figure 2.13. An example of MERA specified post-release mortality rate (fraction of fish that are discarded that subsequently die due to capture).

2.1.14 Recruitment variability

Interannual variability in recruitment (the coefficient of variation in log-normal recruitment deviations, sigma R). Recruitment is expected to change among years in response to changing spawning biomass levels. On top of this is additional variability that may be driven by many factors including varying ocean conditions, amount of spawning habitat, food availability and predation. Sigma R controls the extent of this additional variability in annual recruitments. For example, a sigma R of 10% means that 95% of recruitments will fall in approximately 80-120% of the mean recruitment predicted from spawning biomass.

Edward Houde authored a Comprehensive Review of Recruitment and Sources of Variability and if that isn’t sufficient there is Chambers and Trippel (1997).

Figure 2.14. An example of MERA specified annual recruitment variability expressed as a log-normal standard deviation.

2.1.15 Size of existing spatial closures

The size of a existing spatial closure (e.g. Marine Protected Area, MPA). The size A, is the % of habitat that is protected (the same fraction closed is applied to the habitats of all life stages, for example spawning and rearing grounds.

The FAO provides a comprehensive review of Marine Protected Areas.

Figure 2.15. An example of MERA specified historical spatial closure.

2.1.16 Spatial mixing in/out of existing spatial closures

The degree of stock mixing in/out of existing spatial closure. The degree of the spatial mixing of the fish stock is represented as the probability (P) of a fish leaving the spatial closure (i.e. the marine protected area, MPA) between years

Figure 2.16. An example of MERA specified mixing among existing open/closed areas.

2.1.17 Size of existing spatial closures

The size of a hypothetical future spatial closure (e.g. Marine Protected Area, MPA). The size A, is the % of habitat that is protected (the same fraction closed is applied to the habitats of all life stages, for example spawning and rearing grounds. This question addresses a possible future MPA allowing for the testing of MPs that use this closed area.

Figure 2.17. An example of MERA specified hypothetical future spatial closure.

2.1.18 Spatial mixing in/out of existing spatial closures

The degree of stock mixing in/out of the future hypothetical spatial closure. The degree of the spatial mixing of the fish stock is represented as the probability (P) of a fish leaving the spatial closure (i.e. the marine protected area, MPA) between years

Figure 2.18. An example of MERA specified mixing among future hypothetical open/closed areas.

2.1.19 Initial stock depletion

Initial depletion of the stock relative to asymptotic unfished levels (D1: spawning stock biomass in year 1 relative to equilibrium unfished conditions).

Many fisheries undertake large fluctuations in productivity. In some of these cases, a fishery may have began at a time when the stock was naturally low. This question provides an opportunity to specify this initial depletion. The default however is that the stock was at asymptotic unfished levels in the first year of the fishery

Figure 2.19. An example of MERA specified initial stock depletion (spawning stock biomass relative to unfished levels at the start of the historical simulated time period)

2.2 Management Questions

The Management panel is a set of questions about what fishery management options are available and how well management advice is followed.These questions:

  • identify what management procedures are feasible given the types of management measures.
  • determine the relative success of various management procedures that provide different types of advice.

The operating model parameter mappings of each answer are provided in Appendix Table A.2.

2.2.1 Types of fishery management that are possible

Here you indicate which MPs are feasible given the management options that are available. Management procedures can provide management advice in terms of:

  • Total Allowable Catch limits (TACs, e.g. 20,000 metric tonnes).
  • Total Allowable Effort (TAE, e.g. 800 trap days per year).
  • Size limits (e.g. minimum size of 45cm).
  • Time-area closures (e.g. closing an area to fishing, an MPA or a Winter closure.

For more information see the UN FAO guide to fishery management types.

Or alternatively, Steffanson and Rosenberg describe and discuss fishery managment types in their 2005 paper.

2.2.2 TAC offset

The possible extent to which fishing operations may exceed (overages) or fall short (underages) of the specified Total Allowable Catch (TAC)? For example, given a TAC of 1000 tonnes a 10% offset (overage) would on average lead to 1100 tonnes of fish taken.,

The FAO provides a cursory introduction to uncertainties in fisheries management including implementation error here.

Fulton et al. provide a discussion of implementation error in their 2011 paper.

Figure 2.20. An example of MERA specified TAC offset.

2.2.3 TAC implementation variability

In the previous question you specified the range of the possible TAC offset (mean overage or underage).In this question you add the variability (V) in the implementation of TACs among years.

For example, if on average thereis no TAC offset, a V of 10% leads to annual overages/underages within 20% of the annual TAC recommendation (the black line in the figure opposite) for 95% of cases.

The colored lines show the minimum and maximum variability superimposed on the lowest (dashed line) and highest (solid line) levels of overages/underages specified in the previous question.

Figure 2.21. An example of MERA specified TAC implementation variability.

2.2.4 TAE offset

What is the possible extent to which fishing operations may exceed (overages) or fall short (underages) of the specified Total Allowable Effort (TAE)?

For example, given a TAE of 2000 boat-days of fishing a 10% overage would on average lead to 2200 boat days of effort.

Note: you have the option of selecting MATCH TAC IMPLEMENTATION and mimicking the TAC offset (e.g. assuming that if more than the TAC is taken due to lack of enforcement there would be a similar discrepancy between recommended and implemented TAE)

Figure 2.22. An example of MERA specified TAE offset.

2.2.5 TAE implementation variability

In the previous question you specified the range of possible TAE offset (mean overages/underages). In this question you add the variability (V) in the implementation of TAEs among years.

For example, if on average there is no TAE offset, a V of 20% leads to annual TAE overages/underages within 40% of the annual TAE recommendation (the black line in the figure opposite) for 95% of cases.

The colored lines show the minimum and maximum variability superimposed on the lowest (dashed line) and highest (solid line) levels of overages/underages specified in the previous question.

As with the offset, you have the option of matching the TAC implementation variability.

Figure 2.23. An example of MERA specified TAE implementation variability

2.2.6 Size limit offset

What is the possible extent to which fishing operations may exceed (catch larger) or fall short (catch smaller) fish than the specified minimum size limit? For example, given a size limit of 20cm (e.g. escape hole size of a trap), a value of 20% would lead to a mean minimum size in the catch of 24cm.

Note that if you match TAC implementation variability this will be inverted for size limits (going under the minimum size limit is assumed equivalent to going over the TAC).

Figure 2.24. An example of MERA specified size-limit offset

2.2.7 Size limit implementation variability

In the previous question you specified the range of possible mean violations of a minimum size limit. In this question you add variability (V) in size limit implementation among years.

For example, a size limit of 90cm is exceeded by an average of 10cm, a value of 5% leads to minimum catch sizes of between 90cm and 110cm (the black line in the figure opposite) for 95% of cases. The colored lines show the minimum and maximum variability superimposed on the lowest (dashed line) and highest (solid line) offset in size limit specified in the previous question.

Figure 2.25. An example of MERA specified size-limit implementation variability

2.3 Data Questions

The Data panel is a set of questions about what types of data are available and quality of the data that are available. These questions: * identify what management procedures are feasible given the types of data available. * determine the relative success of the various management types that rely on differing types of data.

2.3.1 Types of data that are available

Management procedures require various data. Where data types are unavailable some MPs may not be feasible.

Annual catches are yearly reporting landings (e.g. 135 tonnes in 1998, 159 tonnes in 1999, etc).

Relative abundance indices may be fishery-dependent such as catch-per-unit-effort data or fishery-independent such as an annual abundance survey.

In the context of annual catches and relative abundance indices, ‘historical’ refers to data going back to ‘unfished conditions’ (pre industrial fishing) such that catches may be used to reconstruct stock trajectory and indices may infer current stock depletion. In contrast, ‘recent’ refers to data available for least 5 years from today.

Effort data are annual observations of fishing effort such as boat days in 2002.

Growth data refers to parameter estimates for growth parameters such as von Bertalanffy growth parameter K and mean asymptotic length, L-infinity.

Size and age composition data are samples of size and ages in the catch going back at least 2 years from today.

2.3.2 Catch reporting bias

In some data-limited fisheries, incomplete monitoring of fishing operations may lead to under-reporting (and to a lesser extent over reporting) of annual catches.

For further discussion of catch under reporting see Agnew et al. (2009).

Figure 2.26. An example of MERA specified catch reporting bias

2.3.3 Hyperstability in indices

Is the primary index of relative abundance proportional to real biomass? Indices of relative abundance derived from fishery catch-per-unit effort may decline faster than real abundance (hyperdepletion) in cases where, for example, the species is being avoided or there has been attrition of high-density sub-population structure during early commericial fishing.

Conversely catch per unit effort data may respond slower than real biomass changes if the species is being targetted, there is range contraction of fishing toward high density areas as the stock declines or the population naturally forms aggregations. For example purse-seine fisheries are often strongly hyperstable since the fish per aggregation may remain high even at low stock sizes.

It may be generally assumed that a well designed fishery-independent survey is proportational to abundance but there are notable exceptions.

See Erisman et al. (1992) or Maunder et al. (2006).

Figure 2.27. An example of MERA specified non-linearity among indices and stock biomass

2.3.4 Overall data quality

What is the overall quality of data that are available?

  • Perfect Information: An unrealistic and idealized observation model for testing the theoretical performance of MPs.
  • Good quality: annual catches and abundance indices are observed with low error (<20% CV) and length/age composition data are numerous (~100 independent observations per year).
  • Data moderate: annual catches and abundance indices are observed with greater error (<30% CV) and length/age composition data are fewer (~40 independent samples per year).
  • Data poor: annual catches and abundance indices are imprecisely observed (<50% CV) and length/age composition data are sparse (~15 independent samples per year).

A description of the observation error model is included in Carruthers et al (2013) and a similar model was used by Carruthers et al. (2015).

2.4 Extra

In addition to the mandatory questions of the Fishery, Management and Data panels, users have the option of further specifying their operating models. The ‘Extra’ tab includes features to load fishery data, condition operating models, add economic features and extend the model for short-lived species.

2.4.1 Fishery data uploading and OM conditioning

By uploading their fishery data, the user unlocks a number of MERA features:

  1. Conditioning of operating models (management planning and management performance)
  2. Realistic MP feasibility analysis to identify MPs that will run with the exact configuration of data that you have (management planning)
  3. Quantification of auxilliary indicators (management performance)
  4. Status determination

A comprehensive guide to data formatting for MERA can be found here

2.4.2 Closed loop simulation controls

Management interval controls how frequently new management advice is calculated. For example, given a management interval of 4 years a new Total Allowable Catch may be set in 2020, 2024, 2028 (and so on) that is kept constant in the interval between these updates.

You can control the number of simulated realizations of your fishery using ‘No. simulations’. Each simulation takes a draw of model parameters from the ranges specified by the MERA questionnaire. This also controls the number of simulations generated if you condition your operating model on data. Generally you can obtain meaningful early results with just 48 simulations, stable MP performance ranking with 96 simulations and stable absolute MP performance with 192. In general 192 or greater simulations are required to quantify value of information and cost of current uncertainties.

If greater than 48 simulations are specified, the user has the option to distribute calculations over a cluster using parallel computation. Note however that you will lose the progress bar.

2.4.3 Detailed operating model controls

It is possible to save MERA operating models, modify these and then loadt them back into the software. This allows for fully customizable operating models including, for example, correlated parameters, complex spatial dynamics including ontogyny and age-base mortality.

The ability to load OMs also allows for the importing of externally derived operating models that are DLMtool and MSEtool compatible e.g. those in the DLMtool fishery library

The detailed operating model controls also allow for a comprehensive documentation of all OM features (whether derived by MERA or otherwise) using the ‘Detailed OM Report’ function.

2.4.4 Condition operating models

If compatible data are loaded in ‘Extra’ panel 1, it is possible to condition operating models using the same population dynamics models as those used in Status Determination Mode.

2.4.5 Bio-economic dynamics

< Under construction >

Simple economic models have been added to the DLMtool and MSEtool operating models and are currently subject to beta testing.

The current options are for demonstration purposes. This feature is currently under development.

2.4.6 Short-lived extensions

< Under construction >

DLMtool now includes functionality to shorten the time-step of the default operating models (annual) to allow for the better approximation of the dynamics of short-lived speies.

The current options are for demonstration and beta testing purposes, This feature is currently under development.


3 Risk Assessment Use Case

3.1 Background to risk assessment

The risk assessment use-case is the simplest of the four options and directly takes the operating model generated from the questionnaire to project four management scenarios: status quo catches (e.g. tonnes of fish), status quo fishing effort (e.g. days of fishing), ‘optimal’ fishing mortality rate commensurate with maximum sustainable yield (FMSY) and zero catches. The output from this analysis is a summary of biological risk (Figure 3) over projected years.

The purpose of the risk assessment use case is to provide an alternative to contemporary subjective scoring systems such as productivity susceptibility analysis (PSA, Hobday et al. 2011) that may not necessarily accurately quantify biological risk (Carruthers and Hordyk 2018).

3.2 Running a risk assessment

Running the risk assessment is straightforward. An operating model is constructed directly from the questionnaire and future projections are carried out for

  • 196 simulations
  • for a standard 50 year projection
  • for 4 alternative management scenarios: status quo fishing effort, status quo annual catches, zero catch and fishing exactly at FMSY.

Since it involves a relatively large number of simulations, the risk assessment can take a couple of minutes to run.

3.3 Interpreting results of risk assessment

Since it is intended to quantify biological risk, the Risk Assessment use case only provides outputs with respect to projected stock biomass.

Here risk is determined relative to biomass commensurate with maximum sustainable yield (BMSY) and half of BMSY - a reference biomass level that is commonly used in fisheries management to determine rebuilding thresholds (e.g. US federal fisheries) and certfication standards to evaluate stock status (e.g. the Marine Stewardship Council).

3.3.1 Projection plots

The risk assessment use case provides a projection plot (Figure 3.1) showing the 90% (light blue) and 50% (dark blue) probability intervals, the median estimate (white line) and two example simulations (dark blue lines). The two horizontal grey lines represent BMSY and half of BMSY.

In this example current effort and catches have similar biomass projections with the current catches somewhat less aggressive, resulting in higher median biomass levels and a fractionally higher probability of being above the limit reference level of 50% BMSY.

Figure 3.2. Risk assessment projection plots for current effort (CurE), current catches (CurC), FMSY fishing (FMSYref) and zero fishing (NFref).

3.3.2 Risk assessment table

The results of the projection are also tabulated (Table 3.1), and phrased in terms of the probability (the fraction of simulations) that biomass is below half of BMSY in future years.

The zero fishing scenario frames the maximum rebuilding potential, FMSY fishing represents ‘perfect management’ highlighting that even then, there is some probability of biomass dropping below the limit reference level (half BMSY) due to natural fluctuations in stock productivity.

Table 3.1. The probability that biomass is greater than half BMSY in future years.


4 Management Planning Use Case

4.1 Background to management planning

Given what is currently known (and not known) about a stock, its fishery and research program:

  • what management approach is likely to best achieve management objectives (MP selection)?
  • what data collection processes are most important in determining management performance (Value of information)?
  • which current uncertainties have the greatest influence of management performance (Cost of current uncertainties)?

Management Planning is the most complex use-case of MERA and provides closed-loop simulation testing of numerous management procedures and diagnostics to help managers identify research priorities.

4.2 Running a closed-loop simulation

Closed-loop simulation iteratively projects the stock and fishery forwards in time, simulating data, generating management advice from MPs and then calculating the impact of this advice on the stock.

It follows that the user must select both a management interval (the duration before new management advice is calculated) and the MPs that should be tested.

The default settings for the management planning mode (Figure 4.1) are deliberately intended to be computationally less demanding for demonstration purposes. An Management interval of 8 years is selected which is relatively long for most fisheries and the ‘Demo’ MP set is a small subset of 5 MPs that run quickly (curE75 is 75% of current fishing effort, DCAC is depletion-corrected average catch (MacCall 2009), IT10 is an index target management procedure that allows for increases/decreases in TAC of 10%, matlenlim is a minimum size limit at the size at 50% maturity, MRreal is a marine reserved in area 1 with reallocation of fishing effort to the open areas).

Figure 4.1. The controls for the management planning use-case.

Users can choose appropriate management interval and a larger range of MPs. The ‘All’ MP set includes the 80+ data-poor and data-moderate MPs included in the DLMtool package including a few simple data-rich assessments. The ‘Top 20’ MP set is a subset of these that includes 20 MPs that are generally among the best performing across a varied set of operating models. The user can select ‘No ref. MPs’ to exclude reference MPs (e.g. FMSY fishing and zero catches) that are included to frame MP performance. Alternatively the user can choose ‘Data-rich MPs’ to include in the analysis, 8 data rich state-space stock assessment based MPs from the MSEtool package. For operating models with more than 48 simulations, the user can select ‘Parallel comp.’ to distribute calculation over numerous processors (note that although it will generally run faster, this breaks the progress bar).

The management-planning mode runs two closed-loop simulations. The first (base) is for the depletion specified by the questionnaire (and optionally informed by operating model conditioning). The second set of simulations (rebuilding) quantifies hypothetical rebuilding performance and starts the forward projections at a user-specified depletion range. This is controlled by a slider ‘Start % BMSY from which to evaluate rebuilding’ and the default starting level is 50% BMSY levels.

4.3 Interpreting results of Management Planning

4.3.1 Biomass projection table for base operating model relative to limit reference point

The biomass trajectories are tabulated (Table 4.1), and phrased in terms of the probability (the fraction of simulations) that biomass is above half of BMSY in future years (half of BMSY is the limit reference point).

Table 4.1 shows an example projection for 20 MPs. In general projections of all MPs achieved relatively high probabilities of being over the limit reference level. The exceptions were delay-difference MPs providing effort advice (DDe and DDe75) that could drop to a 30-35% probability of being over the limit reference point.

Table 4.1. The probability that biomass is greater than half BMSY in future years for the base operating model. Probabilities of 50% or less are color-coded red, those above 90% are color coded green. MP type denotes the class of MP according to the type of advice it provides be it Total Allowable Catch (TAC), Total Allowable Effort (TAE), a Size Limit (SzLim) or spatial closure (MPA). The feasibility column indicates whether an MP can be applied in practice. When an MP is not available due to data-deficiencies the feasibilty column includes a ‘D’. When an MP cannot be applied due to restrictions on the type of advice management can provide (e.g. a size limit isn’t an option) then the feasibility column includes an ‘M’.

4.3.2 Biomass projection table for base operating model relative to target reference point

In addition to biomass relative to the limit reference point, biomass trajectories are tabulated (Table 4.2), and phrased in terms of the probability that biomass is above BMSY in future years (BMSY is the target reference point).

Table 4.2. The probability that biomass is greater than BMSY in future years for the base operating model.

4.3.3 Projection plot for base operating model

Accompanying Tables 4.1 and 4.2 is Figure 4.2 which shows a graphical representation of projected biomass for each MP. These plots include 90% (light blue) and 50% (dark blue) probability intervals, the median estimate (white line) and two example simulations (dark blue lines). The two horizontal grey lines represent BMSY and half of BMSY (target and limit reference points, respectively).

Figure 4.2. Biomass projection plots for eight management procedures given the base operating model.

4.3.4 Projection table for rebuilding operating model

Similarly to the base operating model, tables and figures are provided for the rebuilding scenario (the second closed-loop simulation starting from user-specified status) (Table 4.3), and phrased in terms of the probability (the fraction of simulations) that biomass is below half of BMSY in future years.

Again the DDe and DDe75 MPs provided poor biomass performance. The remaining MPs can be expected to rebuild the stock to some extent for most of the simulations.

Table 4.3. The probability that biomass is greater than half BMSY in future years for the rebuilding operating model.

4.3.5 Projection plot for rebuilding operating model

Accompanying Table 4.3 are a pair of projection plots (Figures 4.3 and 4.4, respectively) showing the long- and short-term biomass outcomes for each MP under the rebuilding scenario.

Figure 4.3. Long-term biomass projection plots for eight management procedures given the rebuilding operating model.

Figure 4.4. Short-term Biomass projection plots for eight management procedures given the rebuilding operating model.

4.3.6 Cost of current uncertainties

For each type of operating model uncertainty (that is specified in the answers of the MERA questionnaire), a range of values are sampled. After the closed-loop simulation is run, it is possible to evaluate how yield performance varied across the range in these answers, identifying those uncertainties that carry the highest yield differential (are potentially the most costly).

Cost of current uncertainties is illustrated in Figure 4.5. On the x-axis are the top-7 most contributory sources of uncertainty. The left-most questions are the most influencial with larger bars indicating the degree of impact on yields. The uncertainties on the x-axis are labelled according to their question number in the MERA questionnaire. For example, for the DBSRA MP (second from left panel, top row) the most important uncertainty was fishery question 10 (F10) selectivity. Across the range of sampled values for this parameter there was a different of 37% in long-term yield.

Figure 4.5. Cost of current uncertainties analysis. Each panel represents the closed-loop simulation testing of a single MP using the base operating model. The variability in long-term yield (expressed as a %) is evaluated across the uncertainty specified for each MERA question. Questions numbered F, M and D correspond to those in the Fishery, Management and Data panels.

4.3.7 Value of Information

Similarly to the cost of current uncertainties analysis, it is possible to evaluate the impact of observation processes on MP performance (Figure 4.6). The degree of observation error and bias was specified in the Data questions of the MERA questionnaire.

Perhaps not surprisingly, for the DBSRA MP (that takes stock depletion as an input) depletion bias was the most important process determining long-term fishery yields.

Figure 4.6. Each panel represents the closed-loop simulation testing of a single MP using the base operating model. The variability in long-term yield (expressed as a %) is evaluated across the uncertainty specified for various observation processes. Biases are long-term persistent over / under estimation of quantities. Errors are generally log-normal annual errors in quantities.

4.3.8 Yield projection

So far the Management Planning results have focused solely on biomass performance. However in most fishery management settings there is a well-established trade-off between yields and biomass outcomes. The yield projections (Figure 4.7) show future trajectories in fishery yield for each MP.

Figure 4.7. Yield projections phrased as a fraction of current yield for the base operating model.

4.3.9 Fishing mortality rate projection

In addition to biomass and yield projections, fishing mortality rate projections relative to FMSY are included in Figure 4.8to illustrate chronic over or under exploitation.

Figure 4.8. Fishing mortality rate projections relative to FMSY fishing, for the base operating model. .

4.3.10 Yield - Biomass trade-offs

Higher yields may be expected to be inversely related to biological status: a prevailing performance trade-off that fishery managers must navigate. The yield-biomass trade-off plots (Figure 4.9) provide a summary of the probability of achieving target and limit biomass levels whilst obtaining reasonably high yields over the long-term.

Figure 4.9 plots these probabilities for each MP, color coded according to the type of advice they provide. The top-right represents better performance, the bottom-left, worse performance.

In this example, output control (TAC) MPs are generally outperforming input controls. Among the best performing MPs are ‘DCAC’, ‘DD4010’, ‘DD’, and ‘HDAAC’. There is a relatively steep reduction in expected yield to obtain the better biomass outcomes obtained by the MP ‘MCD4010’.

Figure 4.9. Long-term biomass-yield performance trade-offs. The probability of exceeding the limit reference point (half BMSY, lefthand panel) and target reference points (BMSY, righthand panel) plotted against the probability of obtaining more than half reference yield (MSY). The legend refers to the class of MP according to the type of advice they provide. ‘Input’ are input controls such as size limits, effort controls or spatial closures. ‘Output’ MPs provide TAC advice. ‘Reference’ MPs are those that show theoretical performance of reference MPs as a yardstick for what is possible or desirable under idealised management situations.


5 Management Performance Use Case


5.1 Background to Management Performance and Auxilliary Indicators

In situations where an MP has been adopted and used for management, it is possible to develop indicators that can detect whether the real fishery conditions have departed substantially from those that were simulated (and used to select the MP).

Often referred to as ‘Exceptional Circumstances’ protocols, these compare the posterior predicted data from the operating model with those that have been observed since MP adoption.

In most MSE settings, quite simple exceptional circumstances protocols are established that compare the posterior predicted probability intervals of operating model indices of abundance with those indices as they are gathered.

MERA uses a somewhat more advanced multivariate approach in which multiple types of data can be observed and compared with operating model predictions (Carruthers and Hordyk 2018) under a more general umbrella term ‘Auxilliary Indicators’.

5.2 Running a management performance evaluation

In the options panel of the MERA questionnaire there is the option to load fishery data. These data can include time series of relative abundance indices, mean length and catch data.

This data file follows the standard Data format of DLMtool (Carruthers and Hordyk 2018) and includes a slot ‘LHYear’. LHyear is the last historical year of data prior to MP adoption. After this reference year, all observed data can be compared to simulated data using the multivariate auxiliary indicators included in MERA to detect departures in system dynamics.

If a suitably formatted data file is uploaded to MERA, the option to calculate auxilliary indicators becomes available.

5.3 Interpreting results of Management Performance

5.3.1 Posterior versus observed data

When an MP was tested using closed-loop simulation, for every simulation, in every year of the projection simulated data were generated. These multiple simulated data sets provides a basis for comparison with a single observed data from the real fishery system whilst an MP was in use.

Since an MP is codified, the only explanations for the observation of data that differ from those of the simulation is that either the operating model is misspecfied (either historically and/or in the future) or the operating model is correct and by chance atypical data have been collected.

Figure 5.1. Posterior predicted (blue) data cross-correlation plots (48 simulations) for the base operating model over the first 5 years of a projection compared with observed (orange) data over the same time period during which the DCAC MP was in use. The data codes include a data type and quantity. ‘C’, ‘I’ and ‘ML’ refer to catch, relative abundance index and mean length, respectively. ‘M’, ‘S’ and ‘V’ refer to the mean, slope and variance respectively. Hence CV is the variance in catches, MLS is the slope in mean length of fish caught.

5.3.2 Multivariate distance

The blue simulated data points of Figure 5.1 have a mean in multivariate space. It is possible to quantify the distance over all dimensions (data types) and simplify the problem into a univariate test of similarity. This is the basis for common methods of multivariate outlier detection and relies on a calculation of Mahalanobis distance which is similar to standard Euclidean distance but accounts for correlations among variables (observed data types). It sounds more complicated than it is - for more information see Carruthers and Hordyk (2018).

Figure 5.2. Multivariate distance (Mahalanobis distance) of the simulated data (blue density) versus the observed data (orange line) collected over the first 5 years that the MP DCAC was in use. A type-I error rate of 5% (alpha) corresponds with a distance of 13.79. With a distance of 14.14, the observed data are inconsistent with those simulated, an outlier is detected and there may be considered sufficient evidence to initiate exceptional circumstances protocols.

6 Status Determination Use Case

6.1 Background to status determination

A large number of proposed methods of estimating stock status have been suggested in the literature that encompass formal stock assessments all the way to approaches that claim to rely on only catch data. A serious problem with the literature is that the comparative performance of these approaches has not be evaluated. It can be argued that fishery system dynamics are sufficiently unique that few general rules might exist regarding which methods of status evaluation are the most reliable - that these are eccentric to the particularities of a specific fishery.

To address this MERA includes a range of status determination approaches and tests their expected performance given the simulated conditions described by the questionnaire. This simulation testing provides a context for interpreting real status determinations that can be calculated from real fishery data, uploaded in the Options panel of the MERA questionnaire.

6.2 Running a Status Determination

In the options panel of the MERA questionnaire there is the option to load fishery data. These data can include time series of relative abundance indices, mean length and catch data.

Once loaded status determination can be run - it automatically detects what data types are available and identifies those status estimation methods approaches that are compatible.

Each modelling approach for estimating status relies on a varying combination of data types that are coded according to the data used:

  • C: catch data (annual)
  • I: index of relative abundance (annual)
  • M: mean length of fish in the catch (annual)
  • L: length composition data (year by length class)
  • A: age composition data (year by age)

Approaches that use only catch data or length compositions assume a pattern in annual fishing mortality rate defined by the annual fishing effort of Fishery Question 5 and the catchability changes of Fishery Question 7.

6.3 Interpreting results of Status Determination

6.3.1 Simulation testing

Under development

6.3.2 Status estimates

Under development


7 Support

7.1 MERA webpage

Under development

7.2 About DLMtool

For more information about DLMtool and MSEtool see Carruthers and Hordyk 2018

7.3 Other operating models

A range of operating models can be loaded into MERA from the DLMtool fishery library

7.4 More information on Management Procedures

You can view further documentation on any of the MPs featured in MERA by clicking on the following links:


8 Acknowledgements

8.1 Collaborators

MERA has benefitted greatly by the feedback and oversight of many people. Particular thanks to Katie Longo, Keith Sainsbury, Tony Smith, Sandy Morison, Kevin Stokes and Dave Newman for their careful feedback and guidance throughout software testing.

Thanks to Carlos Montero, Ricky Amoroso, Abdul ben Hasan, Roberto Licandeo and Brett van Poorten for their feedback during the testing phase.

8.2 MERA

MERA benefits from the ongoing support of the Packard Foundation, the Marine Stewardship Council, the Natural Resources Defense Council and the United Nations Food and Agricultural Organization.

8.3 DLMtool and MSEtool

Many thanks for the ongoing support of the Natural Resources Defense Council, and in particular DLMtool team members David Newman and Lisa Suatoni who have provided input at every stage. The development of DLMtool has been funded by the Gordon and Betty Moore Foundation, the Packard Foundation, Fisheries and Oceans Canada, the Walton Foundation, Resources Legacy Fund, the Natural Resources Defense Council, the United Nations Food and Agricultural Organization and the California Department of Fish and Wildlife.


9 References

Agnew D.J., Pearce J., Pramod G., Peatman T., Watson R., Beddington J.R., et al. 2009 Estimating the Worldwide Extent of Illegal Fishing. PLoS ONE 4(2): e4570. https://doi.org/10.1371/journal.pone.0004570

Anon. 2018. Report of the 2018 joint tuna RFMO management strategy evaluation working group meeting. Seattle. 24 pp. Available at http://www.tuna-org.org/Documents/tRFMO_MSE_2018_TEXT_final.pdf [accessed June 2019]

Butterworth, D. S., & Punt, A. E. 1999. Experiences in the evaluation and implementation of management procedures. ICES Journal of Marine Science, 56(6), 985–998. https://doi.org/10.1006/jmsc.1999.0532

EDF. 2019. FISHE: Framework for Integrated Stock and Habitat Evaluation. Environmental Defense Fund. Available at: http://fishe.edf.org/ [accessed June 2019]

Fletcher WJ. The application of qualitative risk assessment methodology to prioritize issues for fisheries management. ICES J Mar Sci. 2005;62(8):1576–87.

Hobday AJ, Smith A, Webb H, Daley R, Wayte S, Bulman C, et al. 2007. Ecological Risk Assessment for Effects of Fishing: Methodology. Australian Fisheries Management Authority R04/1072. Canberra.

Hobday AJ, Smith ADM, Stobutzki IC, Bulman C, Daley R, Dambacher JM, et al. 2011. Ecological risk assessment for the effects of fishing. Fish Res.108(2–3):372–84.

Holling, C.S. 1978. Adaptive Environmental Assessment and Management. John Wiley and Sons, Chichester, UK.

Kronlund, A.R., Holt, K.R., Cleary, J.S., Shelton, P.A. 2013. Current approaches for the provision of scientific advice on the precautionary approach for Canadian fish stocks: Sections 8 – Management Strategy Evaluation. Canadian Science Advisory Secretariat. Research Document - 2013/081. Available at: http://www.dfo-mpo.gc.ca/csas-sccs/publications/resdocs-docrech/2013/2013_081-eng.html [accessed July 2019]

CDFW. 2018. Master plan for fisheries: a guide for implementation of the Marine Life Management Act. California Department of Fish & Wildlife. Available from: https://nrm.dfg.ca.gov/FileHandler.ashx?DocumentID=159222&inline [accessed July 2019]

MERA. 2019. Version XXX. Available at: MERA.netlify.xxx [accessed July 2019]

NOAA. 2019. Integrated Ecosystem Assessment, U.S. National Oceanic and Atmospheric Administration. Available at: https://www.integratedecosystemassessment.noaa.gov/publications [accessed June 2019]

Pitcher, T.J. 1999. Rapfish, A Rapid Appraisal Technique For Fisheries, And Its Application To The Code Of Conduct For Responsible Fisheries. FAO Fisheries Circular No. FIRM/C: No. 947: 47pp

Punt, A.E., Butterworth, D.S., de Moor, C.L., De Oliveira, J.A.A. & Haddon, M. 2016. Management strategy evaluation: best practices. Fish and Fisheries, 17, 303–334, http://dx.doi.org/10.1111/faf.12104.

Rochet, M-J. & Rice, J.C. 2009. Simulation-based management strategy evaluation: ignorance disguised as mathematics? ICES Journal of Marine Science. 66:754-762.

Rouyer, T., Fromentin, J.-M., Ménard, F., Cazelles, B., Briand, K., Pianet, R., et al. 2008. Complex interplays among population dynamics, environmental forcing, and exploitation in fisheries. Proc. Natl. Acad. Sci. U.S.A. 105, 5420–5425. doi: 10.1073/pnas.0709034105

SeaFood Watch. 2016. Seafood Watch Standard for Fisheries [Internet]. Monterey Bay Aquarium Seafood Watch. Available from: https://www.seafoodwatch.org/seafood-recommendations/our-standards [accessed July 2019]

Sugihara, G., Beddington, J., Hsieh, C.-H., Deyle, E., Fogarty, M., Slaser, S. M., et al. 2011. Are exploited fish populations stable? Proc. Natl. Acad. Sci. U.S.A. 108, E1224–E1225. doi: 10.1073/pnas.1112033108

Thorson, J.T., Munch, S.B., Cope, J.M., Gao, J. 2017. Predicting life history parameters for all fishes worldwide. Ecological Applications. 27(8): 2262-2276. Doi: 10.1002/eap.1606.Walters, C. 2002. Adaptive management of renewable resources. Macmillan Publishing Co. New York. 347 pp. 

Vert-pre, K. A., Amoroso, R. O., Jensen, O. P., and Hilborn, R. 2013. The frequency and intensity of productivity regime shifts in marine fish stocks. Proc. Natl. Acad. Sci. U.S.A. 110, 1779–1784. doi: 10.1073/pnas.1214879110

Walters, C.J. 2007. Is adaptive management helping to solve fisheries problems? Ambio, 36, 304-307.

Walters, C.J., Martell, S.J.D. 2004. Fisheries Ecology and Management. Princeton University Press. 448 pp. 

Westgate, M.J., Likens, G.E. & Lindenmayer, D.B. 2013. Adaptive management of biological systems: A review. Biological Conservation, 158, 128-139.


10 Appendices

10.1 Appendix A

This Appendix details the parameter mappings of the various answers of the MERA questionnaire.

Table App.1. The operating model parameter mapping of the answers for the various fishery questions.


Table App.2. The operating model parameter mapping of the answers for the various management questions.


Table App.3. The operating model parameter mapping of the answers for the various data questions.