Skip to contents

Slick Developer's Guide

DevelopersGuide.Rmd

Introduction

Slick is an R package designed for visualizing and exploring results of a Management Strategy Evaluation (MSE).

Slick allows MSE developers and analysts to prepare their MSE results in a standardized data structure, easily generate publication quality figures and tables, and present and explore the MSE results in an interactive application.

This document is a guide for analysts who wish to use Slick to summarize and present their MSE results.

Installing Slick

Slick will be released on the CRAN in the future.

In the meantime, Slick can be installed from Github:

# install.packages('pak')
pak::pkg_install('blue-matter/Slick')
library(Slick)

Reporting Issues and Bugs

Please use the Issues feature in the GitHub repository to report issues or bugs. Be sure to include a reproducible example so the issue can be recreated for debugging. Pull Requests are also welcome.

Acknowledgements

Slick was developed by Blue Matter Science and designed and commissioned by The Ocean Foundation’s International Fisheries Conservation Project and harveststrategies.org, with support from The Pew Charitable Trusts and the Common Oceans Tuna Fisheries Project which is funded by GEF and implemented by the FAO.

The prototype figure designs were developed by 5W Infographics.

Slick is under going further development. All feedback is welcome. Please contact Shana Miller with any comments or suggestions for further development.

The Slick Object

To use Slick, MSE results must be compiled into a Slick Object. The object is designed to be compatible with results from any MSE framework.

The Slick Object contains all of the information shown in the plots and tables, as well as metadata information such as title, introduction, and author details.

The contents of the Slick Object can be grouped into the following categories:

  1. Metadata: information including the title, introduction, and author details
  2. MPs: details of the Management Procedures
  3. OMs: factors, levels, and other information relating to the Operating Models
  4. Plots: details and values of the Performance Indicators for the Slick Plots

The Slick Plots

A Slick Object can include the following plots:

  1. Time Series: Plots the time series of a user-specified variable (e.g., yield, biomass) during the historical period and the projection period for each management procedure.
  2. Boxplot: Displays the minimum, first quartile, median, third quartile, and maximum from a set of results. Includes an option to display the data as a violin plot.
  3. Kobe: Compares the performance of MPs with respect to biomass (on the x-axis) and fishing mortality (on the y-axis). Includes an option to display the data as a Kobe Time plot, which shows the percentage of runs that fall in each of the Kobe quadrants in each projection year.
  4. Quilt: A table of performance indicators for each management procedure, with colored shading to indicate better/worse performance.
  5. Spider: Shows results for three or more performance indicators, each represented on an axis starting from the same center point. Sometimes also called a Radar plot.
  6. Tradeoff: Scatter plot comparing two performance indicators.

A valid Slick Object must include at least one Slick Plot.

The Slick App

Slick can be used to rapidly create standardized plots and tables of MSE results directly from the R console. See the sections below for examples.

The plots and tables can also be viewed in the Slick App, an interactive application developed with Shiny. The Slick App allows users to upload their data and customize the plots by filtering by management procedures, operating models, and performance indicators.

The App can be launched directly from the R console with:

App()

The App is also hosted online.

The online App is publicly accessible and can be used for presenting and examining MSE results to stakeholders and decision-makers.

Analysts can either share their Slick Object with their audience directly for them to upload into the App, or analysts can upload their Slick Object to the Slick Object Library, where it will be added to the list of Slick Objects that can be loaded directly within the App.

Example MSEs

In order to populate the example Slick Object, we need to define some operating models (OMs), and some management procedures (MPs), and generate some quantitative values for the Performance Indicators relating to each Slick Plot.

To do this, we will be using the openMSE platform to create example MSE objects for a set of MPs and several different OMs.

To reproduce this example, you will need to install the openMSE package and run the code in the following sub-sections.

Alternatively, you can follow the code in the plotting sections below and replace the example OMs, MPs, and Performance Indicator values shown here with those from your own MSE.

Install openMSE 

pak::pkg_install('blue-matter/openMSE')

Generate Operating Models

In this example, we are creating three operating models:

  1. Base: Base Case model for an (entirely fictional) Albacore fishery
  2. Low M: A robustness model where natural mortality (M) is 25% lower than the Base Case model
  3. High M: A robustness model were M is 33% higher than the Base Case model
nsim <- 12
OM_Base <- new('OM', Albacore, Generic_IncE, Generic_Obs, nsim=nsim)
OM_Base@D <- c(0.3,0.4)

OM_LowM <- OM_HighM <- OM_Base
OM_LowM@M <- OM_Base@M * 0.75
OM_HighM@M <- OM_Base@M * 1.3333

Each OM has 12 individual simulations (nsim=12). The OMs include uncertainty in the life-history parameters and the depletion at the end of the historical period (D=c(0.3,0.4)).

Specify Management Procedures

We are using four example management procedures from the DLMtool package:

  1. AvC: the TAC is fixed for all years to the average catch from the historical period
  2. Itarget1: the TAC is iteratively adjusted to reach a target CPUE
  3. DD: A simple delay-difference assessment model is used to estimate current biomass (B)(B) and UMSYU_{MSY}, and TAC=UMSYB\text{TAC}= U_{MSY}B
  4. SPmod: TAC is incrementally adjusted based on the apparent trend in surplus production.
MPs <- c('AvC', 'Itarget1', 'DD', 'SPmod')

These MPs are only being used here as examples, so the details don’t really matter. But if you’re interested you can see the DLMtool help documentation for more details on each MP.

Run Simulations

The runMSE function is used to spool up the historical fishery dynamics and run the closed-loop simulation testing of the four MPs for each of the three OMS:

MSE_Base <- runMSE(OM_Base, MPs=MPs, silent = TRUE)
MSE_LowM <- runMSE(OM_LowM, MPs=MPs, silent = TRUE)
MSE_HighM <- runMSE(OM_HighM, MPs=MPs, silent = TRUE)

In the sections below, these MSE objects will be used to calculate the performance indicators for each Slick Plot.

Creating a Slick Object

To create a Slick Object, simply use the Slick() function:

slick <- Slick()

slick is an S4 object of class Slick:

class(slick)
#> [1] "Slick"
#> attr(,"package")
#> [1] "Slick"

The slots in class Slick objects correspond to information such as the title, subtitle, and author of the Slick object, the operating models (OMs), and management procedures (MPs) included in the Slick object, and the details corresponding to each of the Slick Plots:

slotNames('Slick')
#>  [1] "Title"        "Subtitle"     "Date"         "Author"       "Email"       
#>  [6] "Institution"  "Introduction" "MPs"          "OMs"          "Boxplot"     
#> [11] "Kobe"         "Quilt"        "Spider"       "Timeseries"   "Tradeoff"

Each slot can be accessed and assigned using a function corresponding to the slot name; e.g.:

Date(slick)
#> [1] "2024-11-05"
Date(slick) <- Sys.Date()

The sections below provide examples for populating the Slick object.

Metadata

The Title, Subtitle, Date, Author, Email, and Institution functions are used to populate the corresponding information in the Slick Object.

These metadata are optional, but are useful when presenting the Slick results in the App:

Title(slick) <- 'An Example Slick Object'
Subtitle(slick) <- "For Testing and Demonstration Purposes"
Date(slick) <- Sys.Date()
Author(slick) <- 'Adrian Hordyk'
Email(slick) <-  "[adrian@bluematterscience.com](mailto:adrian@bluematterscience.com)"
Institution(slick) <- "[Blue Matter Science](bluematterscience.com)"

The same functions are used to return the information stored in the Slick Object:

Title(slick)
#> [1] "An Example Slick Object"
Subtitle(slick)
#> [1] "For Testing and Demonstration Purposes"
Date(slick)
#> [1] "2024-11-05"
Author(slick)
#> [1] "Adrian Hordyk"
Email(slick)
#> [1] "[adrian@bluematterscience.com](mailto:adrian@bluematterscience.com)"
Institution(slick)
#> [1] "[Blue Matter Science](bluematterscience.com)"

Markdown

With the exception of Date, Markdown can be used to include links, text styling, etc, for any of the text in the Slick Object. For example:

Introduction(slick) <- "

This *example* Slick Object has been designed to demonstrate the main features of **Slick**.

The code used to create this object is available in the [Slick Developer's Guide](https://slick.bluematterscience.com/articles/DevelopersGuide.html).

The information shown here is for demonstration purposes only.

"

The Markdown will be automatically converted to HTML in the App. Use the markdown=TRUE argument to print the HTML code in the console (in this document the HTML text is rendered directly between the horizontal lines):

Introduction(slick, markdown = TRUE)

This example Slick Object has been designed to demonstrate the main features of Slick.

The code used to create this object is available in the Slick Developer's Guide.

The information shown here is for demonstration purposes only.

Multiple Authors

Author should be a character vector with an element for each author:

Author(slick) <- c('A. Author', 'B. Author')

Email, and Institution must be character vectors of the same length as Author:

Email(slick) <- c('[a.author@email.com](mailto:a.author@email.com)',
                  '[b.author@email.com](mailto:b.author@email.com)')


Institution(slick) <- c('A Institution', 'B Institution')

Multi-Language Support

Title, Subtitle, and Introduction support multiple languages. Three languages are supported:

  • en: English
  • es: Spanish
  • fr: French

Multiple languages can included by providing a named list:

Title(slick) <- list(en='An Example Slick Object',
                     es='Un ejemplo de un objeto Slick',
                     fr="Un exemple d'objet Slick")
Title(slick)
#> [1] "An Example Slick Object"

Title(slick, 'es')
#> [1] "Un ejemplo de un objeto Slick"

Title(slick, 'fr')
#> [1] "Un exemple d'objet Slick"

Most text throughout the Slick Object supports both Markdown and multiple languages.

The App has a dropdown menu that allows users to select their preferred language.

MPs Object

The MPs Object contains the codes, labels, and descriptions of the management procedures (MPs) included in the Slick Object, as well as other options such as the colors to use in the plots, and preset buttons that will appear in the App.

All Slick Objects must have a completed MPs Object.

The easiest way to build a Slick Object is to create and populate the sub-components and then add them to the Slick Object.

Code, Label, and Description

Code, and Label are both required, as they are used in the plots and tables, while Description is optional.

In this example, there are 4 MPs:

mps <- MPs()
Code(mps) <- c('AvC', 'Itarget1', 'DD', 'SPmod')
Label(mps) <- c('Average Catch',
                'Index Target',
                'Delay Difference',
                'Surplus Production')

Description(mps) <- c('TAC is fixed for all years to the average catch from the historical period',
                      'TAC is iteratively adjusted to reach a target CPUE',
                      'Delay-difference assessment model used to estimate current biomass $(B)$ and $U_{MSY}$, and $\text{TAC}= U_{MSY}B$',
                      'TAC is incrementally adjusted based on the apparent trend in surplus production')

This could have also been done as:

mps <- MPs(Code=c('AvC', 'Itarget1', 'DD', 'SPmod'),
           Label=c('Average Catch',
                   'Index Target',
                   'Delay Difference',
                   'Surplus Production'),
           Description = c('TAC is fixed for all years to the average catch from the historical period',
                           'TAC is iteratively adjusted to reach a target CPUE',
                           'Delay-difference assessment model used to estimate current biomass $(B)$ and $U_{MSY}$, and $\text{TAC}= U_{MSY}B$',
                           'TAC is incrementally adjusted based on the apparent trend in surplus production'))

Or using the Metadata function:

Metadata(mps) <- data.frame(
  Code=c('AvC', 'Itarget1', 'DD', 'SPmod'),
  Label=c('Average Catch',
          'Index Target',
          'Delay Difference',
          'Surplus Production'),
  Description=c('TAC is fixed for all years to the average catch from the historical period',
                'TAC is iteratively adjusted to reach a target CPUE',
                'Delay-difference assessment model used to estimate current biomass $(B)$ and $U_{MSY}$, and $\text{TAC}= U_{MSY}B$',
                'TAC is incrementally adjusted based on the apparent trend in surplus production')
)

Metadata(mps)
#>       Code              Label
#> 1      AvC      Average Catch
#> 2 Itarget1       Index Target
#> 3       DD   Delay Difference
#> 4    SPmod Surplus Production
#>                                                                                                          Description
#> 1                                         TAC is fixed for all years to the average catch from the historical period
#> 2                                                                 TAC is iteratively adjusted to reach a target CPUE
#> 3 Delay-difference assessment model used to estimate current biomass $(B)$ and $U_{MSY}$, and $\text{TAC}= U_{MSY}B$
#> 4                                    TAC is incrementally adjusted based on the apparent trend in surplus production
#>     Color
#> 1 #C87A8A
#> 2 #909646
#> 3 #00A396
#> 4 #9189C7

Color

Color is used to set the colors of the MPs shown in the plots. If not specified, default colors will be used:

Color(mps)
#> [1] "#C87A8A" "#909646" "#00A396" "#9189C7"

Color(mps) <- colors()[sample(1:100,4)]
Color(mps)
#> [1] "darkorange1" "cyan1"       "coral"       "darkcyan"

# reset to default colors
Color(mps) <- default_mp_colors(4)

Preset

Preset is optional. It is used to add buttons to the App where users can rapidly select different sets of MPs.

For an MPs Object, Preset is an optional named list, with numeric values indicating the MPs to be selected by the Preset buttons in the App:

Preset(mps) <- list('All'=1:4,
                    'Model-Free'=1:2,
                    'Model-Based'=3:4)

Adding MPs to Slick Object

Once the MPs Object is complete, is can be added to the Slick Object:

MPs(slick) <- mps

OMs Object

The OMs Object describes the structure of the operating models contained in the Slick Object.

All Slick objects must have a completed OMs Object.

Create an OMs Object

A new OMs Object can be created with the OMs function:

oms <- OMs()

The OMs Object contains two mandatory slots (Factors and Design) and one optional slot (Preset).

Factors

Factors is a data.frame with column headings Factor, Level, and Description. It is used to describe the various factors, and levels within each factor, of the operating models included in the Slick Object.

Here is a simple example of a Slick Object that includes results from 6 OMs, spanning three levels on uncertainty in natural mortality (M) and two levels of steepness (h):

Factors(oms) <- data.frame(Factor='M',
                           Level=c('Base', 'Low M', 'High M'),
                           Description=c('Base Case',
                                         'Lower Natural Mortality',
                                         'Higher Natural Mortality')
)

Factors(oms)
#>   Factor  Level              Description
#> 1      M   Base                Base Case
#> 2      M  Low M  Lower Natural Mortality
#> 3      M High M Higher Natural Mortality

Design

The Design matrix is nOM rows and nFactor columns. The values in each column should either be numeric values indicating the levels for the corresponding factor, or the actual level values (i.e., Factors$Level) that correspond to each OM. The column names of Design must match the names of the factors (i.e, Factors$Level).

Here is the Design matrix corresponding with the Factors above:

Design(oms) <- data.frame(M=c('Base', 'Low M', 'High M'))

Design(oms)
#>        M
#> 1   Base
#> 2  Low M
#> 3 High M

It can sometimes be helpful to add names or labels to the operating models listed in Design. This can be done using rownames:

rownames(Design(oms)) <- c('Base Case', 'Less Productive', 'More Productive')

Preset

Preset is optional. It is used to add buttons to the App where users can rapidly select different sets of OMs.

For OMs objects, Preset is a multi-level list. Each element in first level lists must be named. The names are used to display the ‘Preset’ button in the App.

The second level of each named list element must be a list length nFactors, which each element containing numeric values representing the levels within each factor that should be included under that preset.

For example:

Preset(oms) <- list('Base Case'=list(1),
                    'Low M' = list(2),
                    'High M' = list(3),
                    'All'= list(1:3)
)

Adding OMs to Slick Object

The OMs object is now complete:

oms
#> 
#> ── An object of class `OMs` ────────────────────────────────────────────────────
#> 
#> ── `Factors` ──
#> 
#>   Factor  Level              Description
#> 1      M   Base                Base Case
#> 2      M  Low M  Lower Natural Mortality
#> 3      M High M Higher Natural Mortality
#> ── `Design` ──
#>                      M
#> Base Case         Base
#> Less Productive  Low M
#> More Productive High M
#> ── `Preset` ──
#> ── Base Case
#> 1
#> 
#> 
#> ── Low M 
#> 
#> 2
#> 
#> 
#> ── High M 
#> 
#> 3
#> 
#> 
#> ── All 
#> 
#> 1
#> 2
#> 3

The Check function can be used to make sure there are no errors:


Check(oms)
#> 
#> ── Checking: "OMs" ──
#> 
#> ✔ Complete

Once the OMs object is complete, it can be added to the Slick object:

OMs(slick) <- oms

Time Series Plot

Information for the Time Series plot is stored in a Timeseries Object.

A Timeseries Object can be created using the Timeseries function:

timeseries <- Timeseries()

Code, Label, and Description

Either the Code, Label, and Description functions or the Metadata function can be used to add the metadata information to a Timeseries Object.

In this example there are three performance indicators for the Timeseries Object:

Metadata(timeseries) <- data.frame(
  Code=c('SB/SBMSY', 'F/FMSY', 'Removals'),
  Label=c('SB/SBMSY', 'F/FMSY', 'Removals'),
  Description=c('Spawning biomass relative to equilibrium spawning biomass corresponding with maximum sustainable yield (MSY)',
                'Fishing mortality relative to F_MSY',
                'Removals (Landings + Discards)')
)

Code(timeseries)
#> [1] "SB/SBMSY" "F/FMSY"   "Removals"

All plotting functions support Multi-Language Support for the Metadata information.

Time, TimeNow, and TimeLab

The Time slot in the Timeseries Object is a numeric vector with the values for the historical and projection time-steps. In this example, we have 50 historical years and 50 projection years:

Time(timeseries) <- c(rev(seq(2024, by=-1, length.out=50)),
                      seq(2025, by=1, length.out=50))

TimeNow is a numeric value matching the last historical time step (i.e., current time) in Time:

TimeNow(timeseries) <- 2024

TimeLab is a character string providing the name of the time step in Time. A named list can be used for Multi-Language Support:

TimeLab(timeseries) <- 'Year'

Value

Value is a 5-dimensional array containing the values of the performance indicators for each simulation, OM, MP, performance indicator, and time-step (historical and projection):

nOM <- nrow(Design(slick))
nMP <- length(Code(mps))
nPI <- length(Code(timeseries))
nTS <- length(Time(timeseries))

Value(timeseries) <- array(NA, dim=c(nsim, nOM, nMP, nPI, nTS))

This chunk of code calculates the three performance indicators for each OM from the openMSE results generated above. You can ignore it if you have your own MSE data.

# SB/SBMSY
SB_SBMSY_Base <- abind::abind(replicate(nMP, MSE_Base@SSB_hist) |> aperm(c(1,3,2)),
                   MSE_Base@SSB,
                   along=3)/MSE_Base@RefPoint$SSBMSY

SB_SBMSY_LowM <- abind::abind(replicate(nMP, MSE_LowM@SSB_hist) |> aperm(c(1,3,2)),
                               MSE_LowM@SSB,
                              along=3)/MSE_LowM@RefPoint$SSBMSY

SB_SBMSY_HighM <- abind::abind(replicate(nMP, MSE_HighM@SSB_hist) |> aperm(c(1,3,2)),
                               MSE_HighM@SSB,
                               along=3)/MSE_HighM@RefPoint$SSBMSY

# F/FMSY
F_FMSY_Base <- abind::abind(replicate(nMP, MSE_Base@FM_hist) |> aperm(c(1,3,2)),
                            MSE_Base@FM)/MSE_Base@RefPoint$FMSY

F_FMSY_LowM  <- abind::abind(replicate(nMP, MSE_LowM@FM_hist) |> aperm(c(1,3,2)),
                             MSE_LowM@FM)/MSE_LowM@RefPoint$FMSY

F_FMSY_HighM  <- abind::abind(replicate(nMP, MSE_HighM@FM_hist) |> aperm(c(1,3,2)),
                              MSE_HighM@FM)/MSE_HighM@RefPoint$FMSY

# Removals
Removals_Base <- abind::abind(replicate(nMP, apply(MSE_Base@Hist@TSdata$Removals, 1:2, sum))
                              |> aperm(c(1,3,2)),
                              MSE_Base@Removals)

Removals_LowM  <- abind::abind(replicate(nMP, apply(MSE_LowM@Hist@TSdata$Removals, 1:2, sum))
                               |> aperm(c(1,3,2)),
                               MSE_LowM@Removals)

Removals_HighM  <- abind::abind(replicate(nMP, apply(MSE_HighM@Hist@TSdata$Removals, 1:2, sum))
                                |> aperm(c(1,3,2)),
                                MSE_HighM@Removals)

Once the performance indicators have been calculated for the Time Series plot, they can be added to the Value slot:

# SB/SBMSY
Value(timeseries)[,1,,1,] <- SB_SBMSY_Base
Value(timeseries)[,2,,1,] <- SB_SBMSY_LowM
Value(timeseries)[,3,,1,] <- SB_SBMSY_HighM

# F/FMSY
Value(timeseries)[,1,,2,] <- F_FMSY_Base
Value(timeseries)[,2,,2,] <- F_FMSY_LowM
Value(timeseries)[,3,,2,] <- F_FMSY_HighM

# Removals
Value(timeseries)[,1,,3,] <- Removals_Base
Value(timeseries)[,2,,3,] <- Removals_LowM
Value(timeseries)[,3,,3,] <- Removals_HighM

Preset

Each plotting object has a Preset slot. However, this feature is not currently supported for Timeseries plots.

Target and Limit

Target and Limit are optional. They are used to provide horizontal lines on the time series plot indicating the target and limit (if applicable) for a performance indicator. The color and labels can be

Target(timeseries) <- c(1, 0.8, NA)
Limit(timeseries) <- c(0.5, 1, NA)

Adding Timeseries Object to Slick Object

Once the Timeseries Object is complete it can be checked:

Check(timeseries)
#> 
#> ── Checking: "Timeseries" ──
#> 
#> ✔ Complete

and then added to our Slick Object:

Timeseries(slick) <- timeseries

Plotting the Timeseries Object

The plotTimeseries function is used to generate the Time Series plot:

plotTimeseries(slick) # PI 1


plotTimeseries(slick, 2, # PI 2, no quantiles
               includeQuants = FALSE)


plotTimeseries(slick,3, # PI 3, no quantiles, no MP labels
               includeQuants = FALSE,
               includeLabels = FALSE)

The plot can be faceted by management procedure:

plotTimeseries(slick, byMP=TRUE)

or by operating model:

plotTimeseries(slick, byOM=TRUE)

or both:

plotTimeseries(slick, byOM=TRUE, byMP=TRUE)

See ?plotTimeseries for information on more ways to modify the Time Series plot.

By default, the time series values are calculated as the mean over the operating models. The FilterSlick function can be used to filter the Slick Object to specific OMs, MPs, or PIs.

Here we are filtering to only include the Base Case OM (OM 1):

slick_OM1 <- FilterSlick(slick, OMs=1, plot='Timeseries')

plotTimeseries(slick_OM1, byOM=TRUE)

plotTimeseries(slick_OM1) # same plot

Checking and Testing

The Slick Object is now complete with the minimum requirements:

  1. A complete MPs Object
  2. A complete OMs Object
  3. and at least one complete Plot Object (Timeseries in this case)

The Slick Object can be checked for completeness with the Check function:

Check(slick)
#> 
#> ── Checking: "Slick" ──
#> 
#> ── Checking: "MPs"
#> ✔ Complete
#> 
#> ── Checking: "OMs"
#> ✔ Complete
#> 
#> ── Checking: "Boxplot"
#> ℹ Object is empty
#> 
#> ── Checking: "Quilt"
#> ℹ Object is empty
#> 
#> ── Checking: "Kobe"
#> ℹ Object is empty
#> 
#> ── Checking: "Spider"
#> ℹ Object is empty
#> 
#> ── Checking: "Timeseries"
#> ✔ Complete
#> 
#> ── Checking: "Tradeoff"
#> ℹ Object is empty

and loaded directly into the App as a named argument:

App(slick=slick)

Boxplot Plot

The documentation for the remaining plots is briefer and assumes you’ve read the Time Series Plot section.

Information for the Boxplot is stored in a Boxplot Object.

A Boxplot Object can be created using the Boxplot function. Here we are populating the metadata directly when we create the Boxplot object:

boxplot <- Boxplot(Code=c('SB/SBSMY 25', 'SB/SBSMY 50', 'TAC 25'),
                   Label=c('SB/SBMSY Year 25', 'SB/SBSMY Year 50', 'TAC Year 25'),
                   Description=c('SB/SBMSY in the 25th projection year',
                                 'SB/SBMSY in the last year',
                                 'TAC in the 25th projection year'))

Value is a numeric array containing the values of the performance indicators included in the Boxplot object. Value must have dimensions c(nsim, nOM, nMP, nPI):

nOM <- nrow(Design(slick))
nMP <- length(Code(mps))
nPI <- length(Code(boxplot))

Value(boxplot) <- array(NA, dim=c(nsim, nOM, nMP, nPI))

Value(boxplot)[,1,,1] <- MSE_Base@SB_SBMSY[,,25]
Value(boxplot)[,2,,1] <- MSE_LowM@SB_SBMSY[,,25]
Value(boxplot)[,3,,1] <- MSE_HighM@SB_SBMSY[,,25]

Value(boxplot)[,1,,2] <- MSE_Base@SB_SBMSY[,,50]
Value(boxplot)[,2,,2] <- MSE_LowM@SB_SBMSY[,,50]
Value(boxplot)[,3,,2] <- MSE_HighM@SB_SBMSY[,,50]

Value(boxplot)[,1,,3] <- MSE_Base@TAC[,,25]
Value(boxplot)[,2,,3] <- MSE_LowM@TAC[,,25]
Value(boxplot)[,3,,3] <- MSE_HighM@TAC[,,25]

Check the Boxplot object and add to the Slick Object:

Check(boxplot)
#> 
#> ── Checking: "Boxplot" ──
#> 
#> ✔ Complete
Boxplot(slick) <- boxplot

The plotBoxplot function is used to create the Boxplot from a Slick Object:

There is an option for violin plots:

plotBoxplot(slick, type='violin')

or both types:

plotBoxplot(slick, type='both')

The byOM=TRUE argument creates a boxplot for a given performance indicator for each OM in the Slick Object:

plotBoxplot(slick, byOM=TRUE)

See ?plotBoxplot for more options for the plotBoxplot function.

The FilterSlick can be used to filter the OMs and/or MPs to include in a given plot:

plotBoxplot(FilterSlick(slick, MPs=1:3, OMs=1:2, plot='Boxplot'),
            byOM=TRUE)

Kobe Plot

Information for the Kobe Plot is stored in a Kobe Object.

A Kobe Object can be created using the Kobe function. Here we are populating the metadata directly when we create the Kobe object:

kobe <- Kobe(Code=c('SB/SBMSY', 'F/FMSY'),
             Label=c('SB/SBMSY', 'F/FMSY'),
             Description = c('Spawning biomass relative to SB_MSY',
                             'Fishing mortality relative to F_MSY')
)

The Time slot of a Kobe Object contains numeric values for the projection time-steps:

Time(kobe) <- seq(2025, by=1, length.out=OM_Base@proyears)

Value is a numeric array containing the values of the performance indicators included in the Kobe Object. Value must have dimensions c(nsim, nOM, nMP, nPI, nTS):

nOM <- nrow(Design(slick))
nMP <- length(Code(mps))
nPI <- length(Code(kobe))
nTS <- length(Time(kobe))

Value(kobe) <- array(NA, dim=c(nsim, nOM, nMP, nPI, nTS))

Value(kobe)[,1,,1,] <- MSE_Base@SB_SBMSY
Value(kobe)[,2,,1,] <- MSE_LowM@SB_SBMSY
Value(kobe)[,3,,1,] <- MSE_HighM@SB_SBMSY

Value(kobe)[,1,,2,] <- MSE_Base@F_FMSY
Value(kobe)[,2,,2,] <- MSE_LowM@F_FMSY
Value(kobe)[,3,,2,] <- MSE_HighM@F_FMSY

Check and add to the Slick Object:

Check(kobe)
#> 
#> ── Checking: "Kobe" ──
#> 
#> ✔ Complete
Kobe(slick) <- kobe

Plot the Kobe plot:

plotKobe(slick)

Plot the Kobe Time plot:

plotKobe(slick, Time=TRUE)

See ?plotKobe for more options for the plotKobe function.

Quilt Plot

Information for the Quilt Plot is stored in a Quilt Object.

A Quilt Object can be created using the Quilt function. Here we are populating the metadata directly when we create the Quilt object:

quilt <- Quilt(Code=c('PGK',
                      'P100',
                      'PNOF',
                      'Mean TAC'),
               Label=c('Prob. Green Kobe',
                       'Prob SB>SBMSY',
                       'Prob. Not Overfishing',
                       'Mean Total Allowable Catch'),
               Description = c('Probability of being in the green region of Kobe plot over the projection period',
                               'Probability spawning biomass is greater than SB_MSY over the projection period',
                               'Probability of not overfishing over the projection period',
                               'Mean Total Allowable Catch over the projection period'))

Value is a numeric array containing the values of the performance indicators included in the Quilt Object. Value must have dimensions c(nOM, nMP, nPI):

nPI <- length(Code(quilt))

Value(quilt) <- array(NA, dim=c(nOM, nMP, nPI))

Value(quilt)[1,,1] <- apply(MSE_Base@SB_SBMSY > 1 & MSE_Base@F_FMSY < 1, 2, mean)
Value(quilt)[2,,1] <- apply(MSE_LowM@SB_SBMSY > 1 & MSE_LowM@F_FMSY < 1, 2, mean)
Value(quilt)[3,,1] <- apply(MSE_HighM@SB_SBMSY > 1 & MSE_HighM@F_FMSY < 1, 2, mean)

Value(quilt)[1,,2] <- apply(MSE_Base@SB_SBMSY > 1, 2, mean)
Value(quilt)[2,,2] <- apply(MSE_LowM@SB_SBMSY > 1, 2, mean)
Value(quilt)[3,,2] <- apply(MSE_HighM@SB_SBMSY > 1, 2, mean)

Value(quilt)[1,,3] <- apply(MSE_Base@F_FMSY < 1, 2, mean)
Value(quilt)[2,,3] <- apply(MSE_LowM@F_FMSY < 1, 2, mean)
Value(quilt)[3,,3] <- apply(MSE_HighM@F_FMSY < 1, 2, mean)

Value(quilt)[1,,4] <- apply(MSE_Base@TAC, 2, mean)
Value(quilt)[2,,4] <- apply(MSE_LowM@TAC, 2, mean)
Value(quilt)[3,,4] <- apply(MSE_HighM@TAC, 2, mean)

The MinValue and MaxValue functions can be used to set minimum and maximum values for calculating the colored shading for each PI:

MinValue(quilt) <- c(0,0,0,NA)
MaxValue(quilt) <- c(1,1,1,NA)

Check and add to the Slick Object:

Check(quilt)
#> 
#> ── Checking: "Quilt" ──
#> 
#> ✔ Complete

Quilt(slick) <- quilt

The default Quilt plot:

plotQuilt(slick)

Shading scaled between minimum and maximum for each PI:

plotQuilt(slick, minmax=TRUE)

A static table with no shading:

plotQuilt(slick, shading=FALSE, kable=TRUE)

MP

PGK

P100

PNOF

Mean TAC

AvC

0.89

0.90

0.93

930

Itarget1

0.91

0.95

0.95

940

DD

0.48

0.53

0.60

1,000

SPmod

0.16

0.22

0.20

870

See ?plotQuilt for more options for the plotQuilt function.

Spider Plot

Information for the Spider Plot is stored in a Spider Object.

A Spider Object can be created using the Spider function. Here we are populating the metadata directly when we create the Spider object:

spider <- Spider(Code=c('P100',
                      'P50',
                      'PNOF',
                      'Yield'),
               Label=c('Prob SB>SBMSY',
                       'Prob SB>0.5SBMSY',
                       'Prob. Not Overfishing',
                       'Prob. Mean Yield > 50%'),
               Description = c('Probability spawning biomass is greater than SB_MSY over the projection period',
                               'Probability spawning biomass is greater than 0.5 SB_MSY over the projection period',
                               'Probability of not overfishing over the projection period',
                               'Probability mean yield is greater than 50% of the reference yield'))

Value is a numeric array containing the values of the performance indicators included in the Spider Object. Value must have dimensions c(nOM, nMP, nPI):

nPI <- length(Code(spider))

Value(spider) <- array(NA, dim=c(nOM, nMP, nPI))

Value(spider)[1,,1] <- MSEtool::P100(MSE_Base)@Mean
Value(spider)[2,,1] <- MSEtool::P100(MSE_LowM)@Mean
Value(spider)[3,,1] <- MSEtool::P100(MSE_HighM)@Mean

Value(spider)[1,,2] <- MSEtool::P50(MSE_Base)@Mean
Value(spider)[2,,2] <- MSEtool::P50(MSE_LowM)@Mean
Value(spider)[3,,2] <- MSEtool::P50(MSE_HighM)@Mean

Value(spider)[1,,3] <- MSEtool::PNOF(MSE_Base)@Mean
Value(spider)[2,,3] <- MSEtool::PNOF(MSE_LowM)@Mean
Value(spider)[3,,3] <- MSEtool::PNOF(MSE_HighM)@Mean

Value(spider)[1,,4] <- MSEtool::LTY(MSE_Base)@Mean
Value(spider)[2,,4] <- MSEtool::LTY(MSE_LowM)@Mean
Value(spider)[3,,4] <- MSEtool::LTY(MSE_HighM)@Mean

There are a few things to consider for the Value of the Spider plot:

  1. The values for each Performance Indicator must be on the same scale, with high values indicating better performance
  2. Values cannot be negative
  3. All values should be between 0 and 1, or between 0 and 100, with 1 or 100 representing the highest possible value for a Performance Indicator.

Check and add to the Slick Object:

Check(spider)
#> 
#> ── Checking: "Spider" ──
#> 
#> ✔ Complete

Spider(slick) <- spider

The default Spider Plot:

plotSpider(slick)

Option to include fill color:

plotSpider(slick, fill=TRUE)

Plot by MP:

plotSpider(slick, byMP=TRUE)

Or by OM and MP:

plotSpider(slick, byOM=TRUE, incMean = FALSE)

See ?plotSpider for more options for the plotSpider function.

Tradeoff Plot

Information for the Tradeoff Plot is stored in a Tradeoff Object.

A Tradeoff Object can be created using the Tradeoff function. Here we are populating the metadata directly when we create the Tradeoff object:

tradeoff <- Tradeoff(Code=c('P100',
                        'P50',
                        'PNOF',
                        'Yield'),
                 Label=c('Prob SB>SBMSY',
                         'Prob SB>0.5SBMSY',
                         'Prob. Not Overfishing',
                         'Prob. Mean Yield > 50%'),
                 Description = c('Probability spawning biomass is greater than SB_MSY over the projection period',
                                 'Probability spawning biomass is greater than 0.5 SB_MSY over the projection period',
                                 'Probability of not overfishing over the projection period',
                                 'Probability mean yield is greater than 50% of the reference yield'))

Here we are using the same Performance Indicators as those in the Spider Plot.

Value is a numeric array containing the values of the performance indicators included in the Tradeoff Object. Value must have dimensions c(nOM, nMP, nPI):

nPI <- length(Code(tradeoff))

Value(tradeoff) <- array(NA, dim=c(nOM, nMP, nPI))

Value(tradeoff)[1,,1] <- MSEtool::P100(MSE_Base)@Mean
Value(tradeoff)[2,,1] <- MSEtool::P100(MSE_LowM)@Mean
Value(tradeoff)[3,,1] <- MSEtool::P100(MSE_HighM)@Mean

Value(tradeoff)[1,,2] <- MSEtool::P50(MSE_Base)@Mean
Value(tradeoff)[2,,2] <- MSEtool::P50(MSE_LowM)@Mean
Value(tradeoff)[3,,2] <- MSEtool::P50(MSE_HighM)@Mean

Value(tradeoff)[1,,3] <- MSEtool::PNOF(MSE_Base)@Mean
Value(tradeoff)[2,,3] <- MSEtool::PNOF(MSE_LowM)@Mean
Value(tradeoff)[3,,3] <- MSEtool::PNOF(MSE_HighM)@Mean

Value(tradeoff)[1,,4] <- MSEtool::LTY(MSE_Base)@Mean
Value(tradeoff)[2,,4] <- MSEtool::LTY(MSE_LowM)@Mean
Value(tradeoff)[3,,4] <- MSEtool::LTY(MSE_HighM)@Mean

Check and add to the Slick Object:

Check(tradeoff)
#> 
#> ── Checking: "Tradeoff" ──
#> 
#> ✔ Complete

Tradeoff(slick) <- tradeoff

The default Tradeoff Plot:

Tradeoff Plot with multiple Performance Indicators:

plotTradeoff(slick, 1:2, 3:4)

See ?plotTradeoff for more options for the plotTradeoff function.

Slick Object Library

You can add your Slick object to the list of case studies that can be directly loaded from the App by adding your Slick Object to the SlickLibrary repository.

This can be helpful if you wish to share your MSE results with others (e.g., a working group or stakeholders), as otherwise you will need to share the Slick Object file directly with them via email or some other method, and they will have to upload the file into the App.

This method also allows you to update the Slick Object easily.

The way to do this is to clone the SlickLibrary repo, than save your Slick Object to the Slick_Objects directory, and submit a pull request so that your change can be incorporated into the SlickLibrary repo.

A couple of things to note:

  1. The Slick Object must be saved with saveRDS with extension .slick
  2. The file name should be the name with wish to be displayed in the Case Study selection menu, with an underscore “_” to separate words.
saveRDS(slick, '../SlickLibrary/Slick_Objects/Example.slick')

Your Slick object will be available immediately once the SlickLibrary repo is updated.