Package 'mizer'

Description

The mizer package implements multi-species size-based modelling in R. It has been designed for modelling marine ecosystems.

Details

Using mizer is relatively simple. There are three main stages:

1. Setting the model parameters. This is done by creating an object of class MizerParams. This includes model parameters such as the life history parameters of each species, and the range of the size spectrum. There are several setup functions that help to create a MizerParams objects for particular types of models:

   • newSingleSpeciesParams()

   • newCommunityParams()

   • newTraitParams()

   • newMultispeciesParams()

2. Running a simulation. This is done by calling the project() function with the model parameters. This produces an object of MizerSim that contains the results of the simulation.

3. Exploring results. After a simulation has been run, the results can be explored using a range of plotting_functions, summary_functions and indicator_functions.

See the mizer website for full details of the principles behind mizer and how the package can be used to perform size-based modelling.

Author(s)

Maintainer: Gustav Delius [email protected] (ORCID) [copyright holder]

Authors:

• Gustav Delius [email protected] (ORCID) [copyright holder]

• Finlay Scott [email protected] [copyright holder]

• Julia Blanchard [email protected] (ORCID) [copyright holder]

• Ken Andersen [email protected] (ORCID) [copyright holder]

Other contributors:

• Richard Southwell [email protected] [contributor, copyright holder]

See Also

Useful links:

• https://sizespectrum.org/mizer/

• https://github.com/sizespectrum/mizer

• Report bugs at https://github.com/sizespectrum/mizer/issues

Description

addPlot() adds another set of values to an existing ggplot, for example to compare the same rate before and after a model change. There are methods for all the mizer array classes. Each checks whether the existing plot uses a compatible x variable, and warns if the y variable or y-axis units appear to differ.

Usage

addPlot(
  plot,
  x,
  species = NULL,
  total = FALSE,
  background = TRUE,
  colour = NULL,
  linetype = "dashed",
  linewidth = 0.8,
  alpha = 1,
  ...
)

Arguments

Value

A ggplot2 object.

See Also

Other plotting functions: animate(), plot, plot2(), plotBiomass(), plotCDF(), plotCDF2(), plotDiet(), plotFMort(), plotFeedingLevel(), plotGrowthCurves(), plotMizerParams, plotMizerSim, plotPredMort(), plotRelative(), plotSpectra(), plotSpectra2(), plotSpectraRelative(), plotYield(), plotYieldGear(), plotYieldVsF(), plotting_functions

Examples

p <- plot(getEncounter(NS_params), species = "Cod")
addPlot(p, getEncounter(NS_params), species = "Cod")

pr <- plot(getResourceMort(NS_params))
addPlot(pr, getResourceMort(NS_params))

Description

Takes a MizerParams object and adds additional species with given parameters to the ecosystem. It sets the initial values for these new species to their steady-state solution in the given initial state of the existing ecosystem. This will be close to the true steady state if the abundances of the new species are sufficiently low. Hence the abundances of the new species are set so that they are at most 1/100th of the resource power law. Their reproductive efficiencies are set so as to keep them at that low level.

Usage

addSpecies(
  params,
  species_params,
  gear_params = data.frame(),
  initial_effort,
  interaction,
  steady = TRUE,
  info_level = default_info_level(),
  ...
)

Arguments

Details

The resulting MizerParams object will use the same size grid where possible, but if one of the new species needs a larger range of w (either because a new species has an egg size smaller than those of existing species or a maximum size larger than those of existing species) then the grid will be expanded and all arrays will be enlarged accordingly.

If any of the rate arrays of the existing species had been set by the user to values other than those calculated as default from the species parameters, then these will be preserved. Only the rates for the new species will be calculated from their species parameters.

After adding the new species, the background species are not retuned and the system is not run to steady state. This could be done with tuneSteadyState(). The new species will have a reproduction level of 1/4, this can then be changed with reproduction_level<-().

Value

An object of type MizerParams

See Also

removeSpecies(), renameSpecies()

Examples

params <- newTraitParams()
species_params <- data.frame(
    species = "Mullet",
    w_max = 173,
    w_mat = 15,
    beta = 283,
    sigma = 1.8,
    h = 30,
    a = 0.0085,
    b = 3.11
)
params <- addSpecies(params, species_params)
plotSpectra(params)

Description

This function adjusts the size grid in a MizerParams object to the desired minimum and maximum size. It can both expand and truncate the grid. If the grid is truncated, any data outside the new grid is discarded. A warning is issued if there is non-negligible biomass in the discarded size bins.

Usage

adjustSizeGrid(params, ...)

## S3 method for class 'MizerParams'
adjustSizeGrid(
  params,
  new_min_w = min(params@species_params$w_min),
  new_max_w = max(params@species_params$w_max),
  new_min_w_pp = min(params@w_full),
  preserve_species = params@species_params$species,
  tol = 1e-06,
  ...
)

Arguments

Value

A new MizerParams object with the updated size grid.

Description

Uses the size-dependent growth rate and the size at maturity to calculate the age at maturity.

Usage

age_mat(params, ...)

Arguments

Details

Using that by definition of the growth rate $g(w) = dw/dt$ we have that

$\mathrm{age_{mat}} = \int_0^{w_{mat}.}\frac{dw}{g(w)}$

In the implementation this integral is approximated on the model size grid by summing dw / g(w) over all size bins with w < w_mat.

Value

A named vector. The names are the species names and the values are the ages at maturity.

Examples

age_mat(NS_params)

Description

This is not a good way to determine the age at maturity because the von Bertalanffy growth curve is not reliable for larvae and juveniles. However this was used in previous versions of mizer and is supplied for backwards compatibility.

Usage

age_mat_vB(object, ...)

Arguments

Details

Uses the age at maturity that is implied by the von Bertalanffy growth curve specified by the w_inf, k_vb, t0, a and b parameters in the species_params data frame.

If any of k_vb is missing for a species, the function returns NA for that species. Default values of b = 3 and t0 = 0 are used if these are missing. If w_inf is missing, w_max is used instead.

Value

A named vector. The names are the species names and the values are the ages at maturity.

Description

Creates an interactive plotly animation in which a play button steps through time, drawing one line per species at each frame.

Usage

animate(
  x,
  species = NULL,
  log_x = TRUE,
  log_y = TRUE,
  log = NULL,
  wlim = c(NA, NA),
  llim = c(NA, NA),
  ylim = c(NA, NA),
  tlim = c(NA, NA),
  size_axis = c("w", "l"),
  per_log_size = NULL,
  total = FALSE,
  background = TRUE,
  frame_duration = 500,
  transition_duration = frame_duration,
  easing = "linear",
  ...
)

animateSpectra(sim, ...)

Arguments

Details

The function dispatches on the class of x:

• MizerSim — animates the community abundance spectra (number density or biomass density vs body size). Resource, background species, and a community total can be added via the resource, background, and total arguments. The total is the total of everything the model holds, whatever is drawn. The biomass and per_log_size arguments choose the plotted quantity in the same way as in plotSpectra(), and the power argument is available as the same alternative to them. Both axes are log10 by default and can each be switched to linear with log_x = FALSE or log_y = FALSE.

• ArrayTimeBySpeciesBySize — animates any per-species, size-resolved quantity returned by a MizerSim accessor, such as getFMort(), getFeedingLevel(), or getPredMort(). Both axes are log10 by default and can each be switched to linear with log_x = FALSE or log_y = FALSE. Background species and a total can be added via the background and total arguments. The total is the total over every species the array holds, whatever is drawn.

• ArrayTimeByResourceBySize — animates the size-resolved resource quantity returned by NResource() on a MizerSim object. There is only a single resource spectrum, so species, total and background do nothing and warn if set.

Species linecolours and linetypes follow params@linecolour and params@linetype.

animateSpectra() is retained as a backward-compatible alias.

Value

A plotly object with one animated line trace per plotted group. Use the play button or the slider to step through time.

See Also

Other plotting functions: addPlot(), plot, plot2(), plotBiomass(), plotCDF(), plotCDF2(), plotDiet(), plotFMort(), plotFeedingLevel(), plotGrowthCurves(), plotMizerParams, plotMizerSim, plotPredMort(), plotRelative(), plotSpectra(), plotSpectra2(), plotSpectraRelative(), plotYield(), plotYieldGear(), plotYieldVsF(), plotting_functions

Examples

# Animate biomass density spectra, showing only sizes above 0.1 g
animate(NS_sim, power = 2, wlim = c(0.1, NA), tlim = c(1997, 2007))

# Animate fishing mortality through time
animate(getFMort(NS_sim))

# Animate feeding level for two species only
animate(getFeedingLevel(NS_sim), species = c("Cod", "Herring"))

# Animate the resource spectrum
animate(NResource(NS_sim))

Description

Several functions in mizer return a vector over the full size grid holding a resource-related quantity such as the resource number density, the resource mortality, the intrinsic resource birth rate or carrying capacity. The ArrayResourceBySize class wraps these vectors to provide convenient print(), summary(), plot(), and as.data.frame() methods.

Usage

ArrayResourceBySize(
  x,
  value_name = NULL,
  units = NULL,
  type = NULL,
  params = NULL
)

is.ArrayResourceBySize(x)

Arguments

Details

An ArrayResourceBySize object behaves just like a regular numeric vector for arithmetic operations and subsetting. It carries three lightweight attributes:

• value_name – a human-readable name for the value (e.g. "Resource mortality").

• units – the units of the value (e.g. "1/year").

• params – the MizerParams object that the value was computed from.

Value

An ArrayResourceBySize object (inherits from numeric).

is.ArrayResourceBySize() returns TRUE if x is an ArrayResourceBySize object, FALSE otherwise.

See Also

print(), summary(), as.data.frame(), plot(), plot2(), plotRelative(), addPlot()

Examples

mort <- getResourceMort(NS_params)
is.ArrayResourceBySize(mort)
summary(mort)
plot(mort)

Description

Many functions in mizer return two-dimensional arrays (species x size) holding rates like encounter rate, feeding level, growth rate, mortality etc. The ArraySpeciesBySize class wraps these arrays to provide convenient print(), summary(), plot(), and as.data.frame() methods.

Usage

ArraySpeciesBySize(
  x,
  value_name = NULL,
  units = NULL,
  type = NULL,
  params = NULL,
  representation = c("point", "average")
)

is.ArraySpeciesBySize(x)

Arguments

Details

An ArraySpeciesBySize object behaves just like a regular matrix for arithmetic operations and subsetting. It carries a few lightweight attributes:

• value_name – a human-readable name for the value (e.g. "Encounter rate").

• units – the units of the rate (e.g. "g/year").

• type – the kind of quantity the values are.

Value

An ArraySpeciesBySize object (inherits from matrix and array).

is.ArraySpeciesBySize() returns TRUE if x is an ArraySpeciesBySize object, FALSE otherwise.

See Also

print(), summary(), as.data.frame(), plot()

Examples

enc <- getEncounter(NS_params)
is.ArraySpeciesBySize(enc)
summary(enc)

Description

The NResource() function returns a two-dimensional array (time x size) holding the resource number density through time. The ArrayTimeByResourceBySize class wraps this array to provide convenient print(), summary(), plot(), and as.data.frame() methods.

Usage

ArrayTimeByResourceBySize(
  x,
  value_name = NULL,
  units = NULL,
  type = NULL,
  params = NULL
)

is.ArrayTimeByResourceBySize(x)

Arguments

Details

An ArrayTimeByResourceBySize object behaves just like a regular matrix for arithmetic operations and subsetting. It carries these lightweight attributes:

• value_name – a human-readable name for the value (e.g. "Number density").

• units – the units of the value (e.g. "1/g").

• params – the MizerParams object that the value was computed from.

Value

An ArrayTimeByResourceBySize object (inherits from matrix and array).

is.ArrayTimeByResourceBySize() returns TRUE if x is an ArrayTimeByResourceBySize object, FALSE otherwise.

See Also

print(), summary(), as.data.frame(), plot(), plot2(), plotRelative(), addPlot(), animate()

Examples

nr <- NResource(NS_sim)
is.ArrayTimeByResourceBySize(nr)
summary(nr)
plot(nr)

Description

Some functions in mizer return two-dimensional arrays (time x species) holding quantities like biomass, abundance, or yield rate through time. The ArrayTimeBySpecies class wraps these arrays to provide convenient print(), summary(), plot(), and as.data.frame() methods.

Usage

ArrayTimeBySpecies(
  x,
  value_name = NULL,
  units = NULL,
  type = NULL,
  params = NULL
)

is.ArrayTimeBySpecies(x)

Arguments

Details

An ArrayTimeBySpecies object behaves just like a regular matrix for arithmetic operations and subsetting. It carries these lightweight attributes:

• value_name – a human-readable name for the value (e.g. "Biomass").

• units – the units of the value (e.g. "g", "g/year").

• params – the MizerParams object that created the values.

Value

An ArrayTimeBySpecies object (inherits from matrix and array).

is.ArrayTimeBySpecies() returns TRUE if x is an ArrayTimeBySpecies object, FALSE otherwise.

See Also

print(), summary(), as.data.frame(), plot()

Examples

bio <- getBiomass(NS_sim)
is.ArrayTimeBySpecies(bio)
summary(bio)

Description

Some functions in mizer return three-dimensional arrays (time x species x size) holding quantities like fishing mortality, feeding level, or predation mortality through time. The ArrayTimeBySpeciesBySize class wraps these arrays to provide convenient print(), summary(), plot(), animate(), and as.data.frame() methods.

Usage

ArrayTimeBySpeciesBySize(
  x,
  value_name = NULL,
  units = NULL,
  type = NULL,
  params = NULL,
  representation = c("point", "average")
)

is.ArrayTimeBySpeciesBySize(x)

Arguments

Details

An ArrayTimeBySpeciesBySize object behaves just like a regular array for arithmetic operations and subsetting. It carries these lightweight attributes:

• value_name – a human-readable name for the value (e.g. "Fishing mortality").

• units – the units of the value (e.g. "1/year").

• type – the kind of quantity the values are.

• params – the MizerParams object that the value was computed from.

Value

An ArrayTimeBySpeciesBySize object (inherits from array).

is.ArrayTimeBySpeciesBySize() returns TRUE if x is an ArrayTimeBySpeciesBySize object, FALSE otherwise.

See Also

print(), summary(), as.data.frame(), plot(), animateSpectra()

Examples

fmort <- getFMort(NS_sim)
is.ArrayTimeBySpeciesBySize(fmort)
summary(fmort)
plot(fmort, time = 2007)

Description

The as.data.frame() methods for mizer array classes turn matrix- and array-like results into tidy long-form data frames, with one row per observed combination of species, size and/or time. The numeric result is always stored in a column called value.

Usage

## S3 method for class 'ArraySpeciesBySize'
as.data.frame(x, row.names = NULL, optional = FALSE, ...)
## S3 method for class 'ArrayTimeBySpecies'
as.data.frame(x, row.names = NULL, optional = FALSE, ...)
## S3 method for class 'ArrayTimeBySpeciesBySize'
as.data.frame(x, row.names = NULL, optional = FALSE, ...)

Arguments

Details

The returned columns are:

• ArraySpeciesBySize: w, value, Species.

• ArrayTimeBySpecies: time, value, Species.

• ArrayTimeBySpeciesBySize: time, Species, w, value.

If the original object has non-numeric or missing dimension names, sequential indices are used for the time or w columns. Species names are taken from the row, column or dimension names of the original object.

Value

A data frame in long format.

See Also

print(), summary(), str(), plot(), ArraySpeciesBySize(), ArrayTimeBySpecies(), ArrayTimeBySpeciesBySize()

Examples

enc <- getEncounter(NS_params)
head(as.data.frame(enc))

biomass <- getBiomass(NS_sim)
head(as.data.frame(biomass))

Description

Takes the density-independent rates $R_{di}$ of egg production (as calculated by getRDI()) and returns reduced, density-dependent reproduction rates $R_{dd}$ given as

$R_{dd} = R_{di} \frac{R_{max}}{R_{di} + R_{max}}$

where $R_{max}$ are the maximum possible reproduction rates that must be specified in a column in the species parameter dataframe. (All quantities in the above equation are species-specific but we dropped the species index for simplicity.)

Usage

BevertonHoltRDD(rdi, species_params, ...)

Arguments

Details

This is only one example of a density-dependence. You can write your own function based on this example, returning different density-dependent reproduction rates. Three other examples provided are RickerRDD(), SheperdRDD(), noRDD() and constantRDD(). For more explanation see setReproduction().

Value

Vector of density-dependent reproduction rates.

See Also

Other functions calculating density-dependent reproduction rate: RickerRDD(), SheperdRDD(), constantEggRDI(), constantRDD(), noRDD()

Description

Prepares the weight $K(w)$ of an integral over the size spectrum so that the integral is evaluated with the quadrature scheme the model is actually using. Use this when writing your own indicator or diagnostic function; the built-in summary and indicator functions call it for you.

Usage

bin_average_weight(K, params)

Arguments

Details

An integral $\int N(w) K(w)\, dw$ is discretised on mizer's finite-volume grid as $\sum_j N_j \bar K_j \Delta w_j$, where $N_j$ is the cell average of the density over bin $[w_j, w_{j+1}]$. Only the weight is approximated: $N_j$ is already a cell average and $\Delta w_j$ is exact, so neither the abundance nor the bin widths should ever be passed through this function.

Whether the point weight $K(w_j)$ is replaced by the bin average

$\bar K_j = \frac{1}{\Delta w_j}\int_{w_j}^{w_{j+1}} K(w)\,dw \approx \tfrac12\big(K(w_j) + K(w_{j+1})\big)$

is controlled by the bin_average entry of the model's second_order_w() slot. When it is FALSE (the default) K is returned unchanged, so an indicator written with this function reproduces the left-edge Riemann sums of previous mizer versions byte-for-byte. When it is TRUE the trapezoidal bin average is returned, which is uniformly second order and exact whenever $K$ is linear in $w$ (e.g. the first moment $K = w$, for which it equals $(w_{j+1}^2 - w_j^2)/(2\Delta w_j)$).

Because the gating happens inside, always call this rather than averaging unconditionally: a hard-coded bin average silently changes the results of models that are on the default scheme.

If K is a product of several size-dependent factors, average the product and not the individual factors — the average of a product is not the product of the averages. Spawning stock biomass, for example, averages maturity * w as a single weight.

The top bin has no right-hand neighbour on the grid, so its weight is left unaveraged (one-sided); the density there is negligible, so this does not affect the second-order accuracy of the totals.

Value

The weight K, bin-averaged when params@second_order_w[["bin_average"]] is TRUE, otherwise returned unchanged.

See Also

second_order_w(), get_size_range_array(), encounter_kernel()

Examples

# Biomass of each species above 10g -- what getBiomass() does internally.
params <- NS_params
K <- get_size_range_array(params, min_w = 10)   # species x size, 0/1
K <- sweep(K, 2, params@w, "*")                 # weight by w to get biomass
K <- bin_average_weight(K, params)              # gated on second_order_w
rowSums(sweep(initialN(params) * K, 2, params@dw, "*"))

Description

A predation kernel where the predator/prey mass ratio is uniformly distributed on an interval.

Usage

box_pred_kernel(ppmr, ppmr_min, ppmr_max)

Arguments

Details

Writing the predator mass as $w$ and the prey mass as $w_p$, the feeding kernel is 1 if $w/w_p$ is between ppmr_min and ppmr_max inclusive and zero otherwise. ppmr_min must be strictly smaller than ppmr_max. The parameters need to be given in the species parameter dataframe in the columns ppmr_min and ppmr_max.

Value

A vector giving the value of the predation kernel at each of the predator/prey mass ratios in the ppmr argument.

See Also

setPredKernel()

Other predation kernel: gaussian_mixture_pred_kernel(), lognormal_pred_kernel(), power_law_pred_kernel(), truncated_lognormal_pred_kernel()

Examples

params <- NS_params
# Set all required paramters before changing kernel type
species_params(params)$ppmr_max <- 4000
species_params(params)$ppmr_min <- 200
species_params(params)$pred_kernel_type <- "box"
plot(w_full(params), pred_kernel(params)["Cod", 10, ], type="l", log="x")

Description

This function calculates the selectivity for each gear, species and size from the gear parameters. It is called by setFishing() when the selectivity is not set by the user. The returned array is initialised to zero, so gear-species combinations that are not listed in gear_params(params) remain zero. For each listed combination the function named in sel_func is called with w = params@w, the corresponding species parameters, and the selectivity parameters from the matching row in gear_params(params).

Usage

calc_selectivity(params)

Arguments

Value

An array (gear x species x size) with the selectivity values

Bin-averaged selectivity

By default the selectivity is point-sampled at the grid nodes params@w, i.e. at the left edge of each size bin. This is only first-order accurate in the bin size when the selectivity is used in the finite-volume update of the size spectrum. When the bin_average entry of the second_order_w() slot is TRUE, each selectivity function is instead integrated over its size bin, so that selectivity[g, i, j] holds the bin average

$\bar S_{g,i,j} = \frac{1}{\Delta w_j} \int_{w_j}^{w_{j+1}} S_{g,i}(w)\, dw.$

The integral is evaluated with a composite-midpoint rule on a log-spaced sub-grid of each bin, mirroring the bin-integrated predation kernel. This lifts the fishing mortality towards second order at no extra runtime cost (the integration happens once here, the rate functions are unchanged). A welcome side effect is that a knife-edge gear then gets the exact fraction of the straddling bin that lies above the knife edge, removing a grid artefact.

Examples

params <- NS_params
str(calc_selectivity(params))
calc_selectivity(params)["Pelagic", "Herring", ]

Description

Given a MizerParams object params for which biomass observations are available for at least some species via the biomass_observed column in the species_params data frame, this function returns an updated MizerParams object which is rescaled with scaleModel() so that the total biomass in the model agrees with the total observed biomass.

Usage

calibrateBiomass(params, ...)

Arguments

Details

Biomass observations usually only include individuals above a certain size. This size should be specified in a biomass_cutoff column of the species parameter data frame. If this is missing, it is assumed that all sizes are included in the observed biomass, i.e., it includes larval biomass.

After using this function the total biomass in the model will match the total biomass, summed over all species. However the biomasses of the individual species will not match observations yet, with some species having biomasses that are too high and others too low. So after this function you may want to use matchBiomasses(). This is described in the blog post at https://blog.mizer.sizespectrum.org/posts/2021-08-20-a-5-step-recipe-for-tuning-the-model-steady-state/.

Value

A MizerParams object. If no non-missing observed biomass values are provided, the original object is returned unchanged.

Examples

params <- NS_params
species_params(params)$biomass_observed <- 
    c(0.8, 61, 12, 35, 1.6, 20, 10, 7.6, 135, 60, 30, 78)
species_params(params)$biomass_cutoff <- 10
params2 <- calibrateBiomass(params)
plotBiomassObservedVsModel(params2)

Description

Given a MizerParams object params for which number observations are available for at least some species via the number_observed column in the species_params data frame, this function returns an updated MizerParams object which is rescaled with scaleModel() so that the total number in the model agrees with the total observed number.

Usage

calibrateNumber(params, ...)

Arguments

Details

Number observations usually only include individuals above a certain size. This size should be specified in a number_cutoff column of the species parameter data frame. If this is missing, it is assumed that all sizes are included in the observed number, i.e., it includes larval number.

After using this function the total number in the model will match the total number, summed over all species. However the numbers of the individual species will not match observations yet, with some species having numbers that are too high and others too low. So after this function you may want to use matchNumbers(). This is described in the blog post at https://blog.mizer.sizespectrum.org/posts/2021-08-20-a-5-step-recipe-for-tuning-the-model-steady-state/.

Value

A MizerParams object. If no non-missing observed number values are provided, the original object is returned unchanged.

Examples

params <- NS_params
species_params(params)$number_observed <-
    c(0.8, 61, 12, 35, 1.6, 20, 10, 7.6, 135, 60, 30, 78)
species_params(params)$number_cutoff <- 10
params2 <- calibrateNumber(params)

Description

Coerces a MizerParams or MizerSim object to the S3 class hierarchy corresponding to the object's own extension list. For MizerSim, the extension names are read from the embedded MizerParams object.

Usage

coerceToExtensionClass(object, extensions = objectExtensions(object))

Arguments

Details

This is infrastructure for extension package authors. Ordinary users should load saved models with readParams() or readSim(); validParams() and validSim() repair the class vector of an object that is already in memory.

Value

The same object coerced to the appropriate S3 class vector, or to the base class for an empty extension list.

See Also

"Creating a mizer extension package": Creating a mizer extension package

Other extension tools: NOther(), initialNOther<-(), other_mort(), recordExtension(), setComponent(), setRateFunction()

Description

Compare two MizerParams objects and print out differences

Usage

compareParams(params1, params2, ...)

Arguments

Value

Invisibly returns a character vector of difference messages, one element per difference. As a side effect, prints the differences in a human-readable format.

Examples

params1 <- NS_params
params2 <- params1
species_params(params2)$w_mat[1] <- 10
# Keep this example focused on the model parameter.
params2@time_modified <- params1@time_modified
compareParams(params1, params2)

Description

An alias provided for backward compatibility with mizer version <= 2.5.2

Usage

completeSpeciesParams(species_params, check_misspellings = TRUE)

Arguments

Details

validGivenSpeciesParams() checks the validity of the given species parameters. It throws an error if

• the species column does not exist or contains duplicates

• the asymptotic size w_inf is not specified for all species (but see the backwards-compatibility note below)

If a weight-based parameter is missing but the corresponding length-based parameter is given, as well as the a and b parameters for length-weight conversion, then the weight-based parameters are added. If both length and weight are given, then weight is used and an info_about_default condition is signalled if the two are inconsistent.

The required maximum-size parameter is w_inf, the von Bertalanffy asymptotic size of an average individual. For backwards compatibility, if no w_inf column is given, its values are taken from the w_repro_max column if that is present, or otherwise from the w_max column, and an informational message is issued. (w_repro_max is preferred over w_max because in earlier versions of mizer it was the size at which growth stopped and is therefore the closest analogue to the asymptotic size.)

Some inconsistencies in the size parameters are resolved as follows:

• Any w_mat that is not smaller than w_inf is set to w_inf / 4.

• Any w_mat25 that is not smaller than w_mat is marked as missing, as is the corresponding l_mat25, so that the default is used instead.

• Any w_min that is not smaller than w_mat is set to 0.001 or w_mat /10, whichever is smaller.

• Any w_repro_max that is not larger than w_mat is set to 4 * w_mat.

The row names of the returned data frame will be the species names. If species_params was provided as a tibble it is converted back to an ordinary data frame.

The function tests for some typical misspellings of parameter names, like wrong capitalisation or missing underscores and issues a warning if it detects such a name. Pass check_misspellings = FALSE to skip that test, as mizer does when it re-validates a table it has already checked, so that the report is made once, when the column is introduced.

Value

For validSpeciesParams(): A valid species parameter data frame with additional parameters with default values.

For validGivenSpeciesParams(): A valid species parameter data frame without additional parameters.

See Also

species_params(), validGearParams(), validParams(), validSim()

Description

Helper function to keep other components constant

Usage

constant_other(params, n_other, component, ...)

Arguments

Value

The current value of the component

Description

The new egg production is set to compensate for the loss of individuals from the smallest size class through growth and mortality. The result should not be modified by density dependence, so this should be used together with the noRDD() function, see example.

Usage

constantEggRDI(params, n, e_growth, mort, diffusion, ...)

Arguments

Value

Vector with the value for each species

See Also

Other functions calculating density-dependent reproduction rate: BevertonHoltRDD(), RickerRDD(), SheperdRDD(), constantRDD(), noRDD()

Examples

# choose an example params object
params <- NS_params
# We set the reproduction rate functions
params <- setRateFunction(params, "RDI", "constantEggRDI")
params <- setRateFunction(params, "RDD", "noRDD")
# Now the egg density should stay fixed no matter how we fish
sim <- project(params, effort = 10, progress_bar = FALSE)
# To check that indeed the egg densities have not changed, we first construct
# the indices for addressing the egg densities
no_sp <- nrow(params@species_params)
idx <- (params@w_min_idx - 1) * no_sp + (1:no_sp)
# Now we can check equality between egg densities at the start and the end
all.equal(finalN(sim)[idx], initialN(params)[idx])

Description

Simply returns the value from species_params$constant_reproduction.

Usage

constantRDD(rdi, species_params, ...)

Arguments

Value

Vector species_params$constant_reproduction

See Also

Other functions calculating density-dependent reproduction rate: BevertonHoltRDD(), RickerRDD(), SheperdRDD(), constantEggRDI(), noRDD()

Description

This function allows you to make arbitrary changes to how mizer works by allowing you to replace any mizer function with your own version. You should do this only as a last resort, when you find that you can not use the standard mizer extension mechanism to achieve your goal.

Usage

customFunction(name, fun)

Arguments

Details

If the function you need to overwrite is one of the mizer rate functions, then you should use setRateFunction() instead of this function. Similarly you should use ⁠resource_dynamics()<-⁠ to change the resource dynamics and setReproduction() to change the density-dependence in reproduction. You should also investigate whether you can achieve your goal by introducing additional ecosystem components with setComponent().

If you find that your goal really does require you to overwrite a mizer function, please also create an issue on the mizer issue tracker at https://github.com/sizespectrum/mizer/issues to describe your goal, because it will be interesting to the mizer community and may motivate future improvements to the mizer functionality.

Note that customFunction() only overwrites the function used by the mizer code. It does not overwrite the function that is exported by mizer. This will become clear when you run the code in the Examples section.

This function does not in any way check that your replacement function is compatible with mizer. Calling this function can totally break mizer. However you can always undo the effect by reloading mizer with

detach(package:mizer, unload = TRUE)
library(mizer)

Value

No return value, called for side effects

See Also

"Extending mizer": guide to extending mizer

Examples

## Not run: 
fake_project <- function(...) "Fake"
customFunction("project", fake_project)
mizer::project(NS_params) # This will print "Fake"
project(NS_params) # This will still use the old project() function
# To undo the effect:
customFunction("project", project)
mizer::project(NS_params) # This will again use the old project()

## End(Not run)

Description

Returns the mizer_info_level option if it is set and fallback otherwise. This is the default of the info_level argument of the functions that report information, so that options(mizer_info_level = 0) quietens mizer as a whole, including the functions that have no info_level argument of their own, such as species_params<-() and the rate setters.

Usage

default_info_level(fallback = 3)

Arguments

Details

Extension packages should use this as the default of their own info_level argument, so that a constructor or setter of theirs follows the option like mizer's own do:

newFooParams <- function(species_params, ...,
                         info_level = default_info_level()) {
    newMultispeciesParams(species_params, info_level = info_level, ...)
}

Take the argument explicitly like that rather than hard-coding a value in the call, which would make a user's own info_level collide with it.

Value

A single number, or NA to leave the reporting to a handler further out.

Examples

default_info_level()

# Setting the option changes what every reporting function defaults to.
old <- options(mizer_info_level = 1)
default_info_level()
options(old)

Description

If the predation kernel type has not been specified for a species, then it is set to "lognormal" and the default values are set for the parameters beta and sigma.

Usage

default_pred_kernel_params(object)

Arguments

Value

The object with updated columns in the species params data frame.

Description

Function to set and get which edition of default choices is being used.

Usage

defaults_edition(edition = NULL)

Arguments

Details

The mizer functions for creating new models make a lot of choices for default values for parameters that are not provided by the user. Sometimes we find better ways to choose the defaults and update mizer accordingly. When we do this, we will increase the edition number.

If you call defaults_edition() without an argument it returns the currently active edition. Otherwise it sets the active edition to the given value.

Users who want their existing code for creating models not to change behaviour when run with future versions of mizer should explicitly set the desired defaults edition at the top of their code.

The most recent edition is edition 2. It will become the default in the next release. The current default is edition 1. The following defaults are changed in edition 2:

• catchability = 0.3 instead of 1

• ⁠initial effort⁠ = 1 instead of 0

• gamma is set to ensure a feeding level of f0 for larvae with the current value of interaction_resource instead of interaction_resource = 1'.

• initial_n is set using get_steady_state_n() instead of the rather arbitrary old choice.

• In setReproduction(), psi is no longer forced to 1 above w_repro_max; its value is determined entirely by the maturity ogive and the reproductive proportion.

Value

If edition is NULL, the currently active edition number. If edition is supplied, the function sets the global mizer_defaults_edition option, emits a message, and returns the supplied value invisibly.

Description

Check whether two objects are numerically different, ignoring all attributes.

Usage

different(a, b)

Arguments

Details

We use this helper function in particular to see if a new value for a slot in MizerParams is different from the existing value in order to give the appropriate messages.

Value

TRUE or FALSE

Description

This function can be used in projectUntilSettled() to decide when sufficient convergence to steady state has been achieved.

Usage

distanceMaxRelRDI(params, current, previous)

Arguments

Value

The largest absolute relative change in rdi: max(abs((current_rdi - previous_rdi) / previous_rdi)). If any entry of previous_rdi is zero, the result can be infinite.

See Also

Other distance functions: distanceSSLogN()

Description

Calculates the sum squared difference between log(N) in current and previous state. This function can be used in projectUntilSettled() to decide when sufficient convergence to steady state has been achieved.

Only the size classes that hold fish are measured. A class whose density is zero in either state has no log to take, and one holding a negligible share of its species' biomass has a log that never stops moving: above a size where growth stops the density decays exponentially, so log(n) falls by the same amount between every pair of states, for ever. Left in the sum, a single such class holding $10^{-92}$ g of fish can hold the distance above any tolerance indefinitely and stop projectUntilSettled() from ever converging, while the biomass drift correctly reports a fixed point. Excluding it changes nothing for a model that does not have one, because the classes that hold the fish are exactly the classes that are kept.

Usage

distanceSSLogN(
  params,
  current,
  previous,
  biomass_share_cutoff = steady_share_cutoff(),
  ...
)

Arguments

Value

The sum of squares of the difference in the logs of the fish abundances n, over the size classes that hold fish in both states: sum((log(current$n) - log(previous$n))^2)

See Also

Other distance functions: distanceMaxRelRDI()

Description

A hump-shaped selectivity function with a sigmoidal rise and an independent sigmoidal drop-off. This drop-off is what distinguishes this from the function sigmoid_length() and it is intended to model the escape of large individuals from the fishing gear.

Usage

double_sigmoid_length(w, l25, l50, l50_right, l25_right, species_params, ...)

Arguments

Details

You would not usually call this function directly. Instead, set the sel_func column in gear_params() to "double_sigmoid_length" and provide the l25, l50, l50_right and l25_right values as additional columns. setFishing() will then call this function automatically when calculating the selectivity array.

The selectivity is obtained as the product of two sigmoidal curves, one rising and one dropping. The sigmoidal rise is based on the two parameters l25 and l50 which determine the length at which 25% and 50% of the stock is selected respectively. The sigmoidal drop-off is based on the two parameters l50_right and l25_right which determine the length at which the selectivity curve has dropped back to 50% and 25% respectively. The selectivity is given by the function

$S(l) = \frac{1}{1 + \exp\left(\log(3)\frac{l50 -l}{l50 - l25}\right)}\frac{1}{1 + \exp\left(\log(3)\frac{l50_{right} -l}{l50_{right} - l25_{right}}\right)}$

As the size-based model is weight based, and this selectivity function is length based, it uses the length-weight parameters a and b to convert between length and weight.

$l = \left(\frac{w}{a}\right)^{1/b}$

Value

Vector of selectivities at the given sizes. Requires ⁠l25 < l50 < l50_right < l25_right⁠.

See Also

gear_params() for setting the selectivity parameters.

Other selectivity functions: knife_edge(), knife_edge_length(), sigmoid_length(), sigmoid_weight()

Examples

# Hump-shaped selectivity: rises from l25=10 to l50=15,
# then drops back to 50% at l50_right=40 and 25% at l25_right=50
sp <- list(a = 0.01, b = 3)
w <- c(1, 10, 100, 500, 1000)
double_sigmoid_length(w, l25 = 10, l50 = 15,
                      l50_right = 40, l25_right = 50,
                      species_params = sp)

Description

An internal function. Sets up a valid MizerParams object with all the slots initialised and given dimension names, but with some slots left empty. This function is to be used by other functions to set up full parameter objects.

Usage

emptyParams(
  species_params,
  gear_params = data.frame(),
  no_w = 100,
  min_w = 0.001,
  max_w = NA,
  min_w_pp = 1e-12
)

Arguments

Value

An empty but valid MizerParams object

Size grid

A size grid is created so that the log-sizes are equally spaced. The spacing is chosen so that there will be no_w fish size bins, with the smallest starting at min_w and the largest starting at max_w. For the resource spectrum there is a larger set of bins containing additional bins below min_w, with the same log size. The number of extra bins is such that min_w_pp comes to lie within the smallest bin.

Changes to species params

The species_params slot of the returned MizerParams object may differ from the data frame supplied as argument to this function because default values are set for missing parameters.

See Also

See newMultispeciesParams() for a function that fills the slots left empty by this function.

Description

Returns the kernel array $\Phi_i(w_k, w_p)$ for which

$E_i(w_k) = \gamma_i(w_k) \sum_p \Phi_i(w_k, w_p) N^{eff}_i(w_p) w_p \Delta w_p$

reproduces exactly the available energy computed by mizerEncounter(), where $N^{eff}$ is the interaction-weighted prey density. It is the kernel that any summary function must use if its result is to be consistent with getEncounter().

Usage

encounter_kernel(params)

Arguments

Details

On the default first-order path this is just the point-sampled kernel returned by pred_kernel(). When second-order bin-averaging is switched on (see second_order_w()) the two differ: setPredKernel() then builds the Fourier-transformed kernel from the kernel integrated over the prey bin, divided by $\beta - 1$ so that the plain point weight $w_p \Delta w_p$ carried by the prey vector is cancelled. Those bin-integrated weights are recovered here from params@ft_pred_kernel_e by an inverse Fourier transform, which costs one FFT and keeps this helper automatically in step with whatever quadrature setPredKernel() used.

Pair it with the plain point prey weight params@w_full * params@dw_full. That weight is a normalisation which the kernel construction is built to cancel, not a first-order quadrature weight, so it must not be passed through bin_average_weight(): doing so applies the prey-bin integral twice. A summary function that instead pairs the point-sampled pred_kernel() with a bin-averaged prey weight double-counts that quadrature; that was the bug behind issue #474.

Value

An array (predator species x predator size x prey size).

See Also

pred_kernel() for the point-sampled kernel used for plotting and for supplying a custom kernel, second_order_w(), bin_average_weight()

Description

This function expands the size grid in a MizerParams object to the desired min and max size, preserving all existing species. The function is deprecated because you can achieve the same more flexibly with adjustSizeGrid().

Usage

expandSizeGrid(params, ...)

## S3 method for class 'MizerParams'
expandSizeGrid(
  params,
  new_min_w = min(params@w),
  new_max_w = max(params@w),
  preserve_species = params@species_params$species,
  ...
)

Arguments

Value

A new MizerParams object with the updated size grid.

Description

Size spectra at end of simulation

Usage

finalN(sim)

finalNResource(sim)

idxFinalT(sim)

Arguments

Value

For finalN(): An ArraySpeciesBySize object (species x size) holding the consumer number densities at the end of the simulation

For finalNResource(): A vector holding the resource number densities at the end of the simulation for all size classes

For idxFinalT(): An integer giving the index for extracting the results for the final time step

Examples

str(finalN(NS_sim))

# This could also be obtained using `N()` and `idxFinalT()`
identical(N(NS_sim)[idxFinalT(NS_sim), , ], finalN(NS_sim))
str(finalNResource(NS_sim))
idx <- idxFinalT(NS_sim)
idx
# This coincides with
length(getTimes(NS_sim))
# and corresponds to the final time
getTimes(NS_sim)[idx]
# We can use this index to extract the result at the final time
identical(N(NS_sim)[idx, , ], finalN(NS_sim))
identical(NResource(NS_sim)[idx, ], finalNResource(NS_sim))

Description

Puts the model onto a steady state of its own dynamics, changing no parameter: the reproduction rate, the resource and the consumer spectra all settle together at whatever the parameters you already have imply.

Usage

findSteadyState(
  params,
  solver = c("project", "newton"),
  effort = params@initial_effort,
  info_level = default_info_level(),
  ...
)

Arguments

Details

This is the counterpart of tuneSteadyState(), which instead holds the reproduction rate and the resource at the values you supplied and adjusts erepro/R_max and cc_pp to make those values steady. Use this function when the parameters are the thing you want to keep — when asking what state a given model settles into, for instance under a changed fishing effort — and tuneSteadyState() while calibrating.

Nothing being held fixed means the search has more ways to end badly. With solver = "project" the run can settle on a limit cycle or drive a species extinct rather than reach a fixed point, and with either solver reproduction can collapse. Check the "convergence" attribute rather than assuming a fixed point was reached; see the section below.

Value

A MizerParams object with the initial state replaced by the steady state found and no parameter changed. It carries a "convergence" attribute describing the solution found; see projectUntilSettled().

Choosing a solver

solver = "project" (the default) runs the dynamics until they settle, via projectUntilSettled(), and takes the final state. It is exactly that function with the trajectory thrown away; call projectUntilSettled() instead if you want to watch the approach.

solver = "newton" solves the steady-state equation directly with a Newton-type root finder from the nleqslv package, so it converges even when the steady state is dynamically unstable, where the time-stepping solver diverges away from it. This is the natural entry point for a stability analysis with getStability().

The Newton solver treats the resource densities as unknowns alongside the fish and appends the resource steady-state equation to the system, so the resource density and the feeding levels it implies are self-consistent even where consumers are satiated. That equation is written for the default semichemostat resource dynamics, so solver = "newton" stops with an error for any other resource_dynamics; use solver = "project" there.

The Newton iteration also needs the residual $F(N)$ to be continuous. A custom rate function registered with setRateFunction() that jumps as a function of the abundances makes $F$ discontinuous, and where the equilibrium lies on the switching threshold there is no root at all, because neither branch is in equilibrium there. The solver then stalls (nleqslv termination code 3) and returns an iterate pinned to the threshold. See Discontinuous rate functions.

It also respects the active transport scheme: if the experimental second-order scheme is enabled (see second_order_w()) it solves the steady-state equation of that scheme. With the van Leer reconstruction the residual is only Lipschitz, so the iteration converges to a fixed point of the dynamics but not to machine precision. The unlimited "centred" reconstruction admits an undamped odd-even mode at a steady state with no physical diffusion, giving an ill-conditioned steady-state Jacobian for which the solver is not expected to converge.

What you get back may not be a steady state

The stopping criterion is a proxy. It says that two states t_per years apart differ by less than distance_tol on whatever scale the criterion is measured on; it does not say that the state reached is a fixed point. There are four ways the returned object can fail to be one:

• the run converged on its own scale while the biomasses are still visibly drifting (termination = "distance_tolerance");

• the run reached t_max without converging (termination = "time_limit");

• the run settled on a limit cycle (termination = "cycle_detected"), in which case the state stored is one point on that cycle;

• the run stopped because a species was going extinct (termination = "extinction").

So treat the result as a claim to be checked rather than as a guarantee:

attr(params, "convergence")$attractor  # "fixed_point", "limit_cycle" or NA
attr(params, "convergence")$residual   # largest biomass drift, in 1/year
isSteady(params)                       # TRUE if within tolerance
summary(params)                        # includes the biomass-drift verdict
plot(getSteadyResidual(params))        # which species, and at which sizes

attractor is the field that answers the question: it is "fixed_point" only where the measured biomass drift is within residual_tol, so it cannot be satisfied by a distance function that has merely gone quiet. termination says how the run ended and converged whether the solver met its own criterion; neither is a claim about the state. The last line says where the model is not steady, which is the one to reach for when it is not: a model that is off steady state is usually off in one species or one part of the size range, and the plot names it. See getSteadyResidual() for why the verdict is phrased in terms of biomass drift rather than the largest per-capita rate.

The messages this function prints say the same thing — a converged run whose biomasses are still moving reports the drift and adds "Reduce the tolerance on the distance function to converge further." — but they are suppressed by info_level = 0, so in a script the "convergence" attribute is the reliable check.

Finally, a genuine fixed point need not be a stable one. Use getStability() to find out, and solver = "newton" to converge onto a fixed point that the dynamics themselves would run away from.

See Also

tuneSteadyState(), projectUntilSettled(), isSteady(), getSteadyResidual(), getStability()

Examples

params <- findSteadyState(NS_params, solver = "newton")
plotSpectra(params)

Description

A lookup table used by signal_frozen_changes() to decide whether a change the user made can take effect. Each entry is named after a slot of MizerParams that can be frozen and gives the quantity as the user knows it, the call that unfreezes it, the parameters that the setter and its default calculations read, and what kind of parameters those are.

Usage

frozen_rate_params()

Details

The list of parameters does not have to be exhaustive, and deliberately is not: it names the parameters that the setter reads directly together with the main inputs of the default calculations for those parameters. A parameter that is missing simply means that the user is not warned, and is left with the message that the setter itself gives, see signal_not_recalculated(). Listing a parameter that in fact has no influence is the worse mistake, because it warns about a change that did take effect.

Value

A named list of lists with entries quantity, reset_call, params and derived_from.

Description

A predation kernel for which the log predator/prey mass ratio follows a mixture of Gaussian distributions.

Usage

gaussian_mixture_pred_kernel(ppmr, kernel_p, kernel_mean, kernel_sd)

Arguments

Details

Writing the predator mass as $w$, the prey mass as $w_p$, and $x = \ln(w / w_p)$, the feeding kernel is

$\phi_i(w, w_p) = \sum_j a_{ij} \exp\left[-\frac{(x - \mu_{ij})^2}{2\sigma_{ij}^2}\right], \qquad a_{ij} = \frac{p_{ij}/\sigma_{ij}} {\sum_k p_{ik}/\sigma_{ik}}.$

for predator/prey mass ratios greater than or equal to one, and zero for smaller ratios.

This is proportional to the Gaussian-mixture probability density with mixing proportions $p_{ij}$, means $\mu_{ij}$, and standard deviations $\sigma_{ij}$. The scaling makes the sum of the component peak heights equal to one. Consequently the kernel is at most one, and a one-component mixture is identical to lognormal_pred_kernel() with beta = exp(kernel_mean) and sigma = kernel_sd.

The three component parameters are vectors of equal length. When this function is selected in a species parameter data frame, they should be held in the list-columns kernel_p, kernel_mean, and kernel_sd. The values in kernel_p must be non-negative with at least one positive value, but they do not need to sum to one because they are normalised by the function.

Value

A vector giving the value of the predation kernel at each of the predator/prey mass ratios in the ppmr argument.

See Also

setPredKernel()

Other predation kernel: box_pred_kernel(), lognormal_pred_kernel(), power_law_pred_kernel(), truncated_lognormal_pred_kernel()

Examples

ppmr <- exp(seq(0, 12, length.out = 200))
phi <- gaussian_mixture_pred_kernel(
    ppmr,
    kernel_p = c(0.3, 0.7),
    kernel_mean = c(4, 8),
    kernel_sd = c(0.8, 1.5)
)
plot(ppmr, phi, type = "l", log = "x")

Description

These functions allow you to get or set the gear parameters stored in a MizerParams object. These are used by setFishing() to set up the selectivity and catchability and thus together with the fishing effort determine the fishing mortality.

Usage

gear_params(object)

gear_params(object) <- value

is.gear_params(x)

Arguments

Details

The gear_params data has one row for each gear-species pair and one column for each parameter that determines how that gear interacts with that species. The columns are:

• species The name of the species

• gear The name of the gear

• catchability A number specifying how strongly this gear selects this species.

• sel_func The name of the function that calculates the selectivity curve.

• One column for each selectivity parameter needed by the selectivity functions.

For the details see setFishing().

There can optionally also be a column yield_observed that allows you to specify for each gear and species the total annual fisheries yield in grams per year. This is used by plotYieldObservedVsModel(), which adds the yields up over the gears to get the observed yield of each species, see get_yield_observed().

The fishing effort, which is also needed to determine the fishing mortality exerted by a gear is not set via the gear_params data frame but is set with initial_effort() or is specified when calling project().

If you change a gear parameter, this will be used to recalculate the selectivity and catchability arrays by calling setFishing(), unless you have previously set these by hand.

⁠gear_params<-⁠ automatically sets the row names to contain the species name and the gear name, separated by a comma and a space. The last example below illustrates how this facilitates changing an individual gear parameter.

Value

Data frame with gear parameters

is.gear_params() returns TRUE if x is a gear_params object, FALSE otherwise.

See Also

Other functions for setting parameters: setExtDiffusion(), setExtEncounter(), setExtMort(), setFishing(), setInteraction(), setMaxIntakeRate(), setMetabolicRate(), setParams(), setPredKernel(), setReproduction(), setSearchVolume(), species_params(), use_predation_diffusion()

Examples

params <- NS_params

# gears set up in example
gear_params(params)

# setting totally different gears
gear_params(params) <- data.frame(
    gear = c("gear1", "gear2", "gear1"),
    species = c("Cod", "Cod", "Haddock"),
    catchability = c(0.5, 2, 1),
    sel_func = c("sigmoid_weight", "knife_edge", "sigmoid_weight"),
    sigmoidal_weight = c(1000, NA, 800),
    sigmoidal_sigma = c(100, NA, 100),
    knife_edge_size = c(NA, 1000, NA)
    )
gear_params(params)

# changing an individual entry
gear_params(params)["Cod, gear1", "catchability"] <- 0.8

Description

Fills in any missing values for f0 so that if the prey abundance was described by the power law $\kappa w^{-\lambda}$ then the encounter rate coming from the given gamma parameter would lead to the feeding level $f_0$. This is thus doing the inverse of get_gamma_default(). Only for internal use.

Usage

get_f0_default(params)

Arguments

Details

For species for which no value for gamma is specified in the species parameter data frame, the f0 values is kept as provided in the species parameter data frame or it is set to 0.6 if it is not provided.

See the Target Feeding Level section of the "Calculation of Default Parameter Values" vignette for the mathematical derivation.

The available energy is measured with the predation part of mizer's own encounter rate, mizerEncounter(). It is a property of the species parameters and of the resource power law rather than of the model's dynamics, so external encounter and contributions registered with other_encounter() (including component encounter functions) are excluded. The measurement also deliberately does not go through the extension dispatch chain, nor through an encounter function registered with setRateFunction().

Value

A vector with the values of f0 for all species

See Also

Other functions calculating defaults: get_gamma_default(), get_h_default(), get_ks_default()

Description

Fills in any missing values for gamma so that fish feeding on a resource spectrum described by the power law $\kappa w^{-\lambda}$ achieve a feeding level $f_0$. Only for internal use.

Usage

get_gamma_default(params)

Arguments

Details

See the Search Volume Coefficient section of the "Calculation of Default Parameter Values" vignette for the mathematical derivation.

The available energy is measured with the predation part of mizer's own encounter rate, mizerEncounter(). It is a property of the species parameters and of the resource power law rather than of the model's dynamics, so external encounter and contributions registered with other_encounter() (including component encounter functions) are excluded. The measurement also deliberately does not go through the extension dispatch chain, nor through an encounter function registered with setRateFunction().

Value

A vector with the values of gamma for all species

See Also

Other functions calculating defaults: get_f0_default(), get_h_default(), get_ks_default()

Description

Sets h so that the species reaches maturity size w_mat at the maturity age age_mat if it feeds at feeding level f0.

Usage

get_h_default(params)

Arguments

Details

If age_mat is missing in the species parameter data frame, then it is calculated from the von Bertalanffy growth curve parameters k_vb and (optionally t0) taken from the species parameter data frame. This is not reliable and a warning is issued.

If no growth information is given at all for a species, the default is set to h = 30.

See the Maximum Intake Rate Coefficient section of the "Calculation of Default Parameter Values" vignette for the mathematical derivation.

Value

A vector with the values of h for all species

See Also

Other functions calculating defaults: get_f0_default(), get_gamma_default(), get_ks_default()

Description

This function uses the model parameters and other parameters to calculate initial values for the species number densities. These initial abundances are currently quite arbitrary and not close to the steady state. We intend to improve this in the future.

Usage

get_initial_n(params, n0_mult = NULL, a = 0.35)

Arguments

Value

An ArraySpeciesBySize object (species x size) of population abundances.

Examples

init_n <- get_initial_n(NS_params)

Description

Fills in any missing values for ks so that the critical feeding level needed to sustain the species is as specified in the fc column in the species parameter data frame. If that column is not provided the default critical feeding level $f_c = 0.2$ is used.

Usage

get_ks_default(params)

Arguments

Details

See the Standard Metabolic Rate Coefficient section of the "Calculation of Default Parameter Values" vignette for the mathematical derivation.

Value

A vector with the values of ks for all species

See Also

Other functions calculating defaults: get_f0_default(), get_gamma_default(), get_h_default()

Description

This involves finding the feeding kernel function for each species, using the pred_kernel_type parameter in the species_params data frame, checking that it is valid and all its arguments are contained in the species_params data frame, and then calling this function with the ppmr vector.

Usage

get_phi(species_params, ppmr)

Arguments

Value

An array (species x ppmr) with the values of the predation kernel function

Description

Helper function that returns an array (species x size) of logical values indicating whether that size bin is within the size limits specified by the arguments. Either the size limits can be the same for all species or they can be specified as vectors with one value for each species in the model.

Usage

get_size_range_array(
  params,
  min_w = min(params@w),
  max_w = max(params@w),
  min_l = NULL,
  max_l = NULL,
  ...
)

Arguments

Value

A logical array (species x size), with dimnames sp and w.

Length to weight conversion

If min_l is specified there is no need to specify min_w and so on. However, if a length is specified (minimum or maximum) then it is necessary for the species parameter data.frame to include the parameters a and b that determine the relation between length $l$ and weight $w$ by

$w = a l^b.$

It is possible to mix length and weight constraints, e.g. by supplying a minimum weight and a maximum length, but this must be done the same for all species. The default values are the minimum and maximum weights of the spectrum, i.e., the full range of the size spectrum is used.

Description

This function calculates the steady state abundance by solving the transport equation with given growth and mortality rates. It sets up a tri-diagonal system and solves it.

Usage

get_steady_state_n(
  params,
  g,
  mu,
  D,
  N0,
  max_iterations = 500,
  tol = 1e-10,
  relax = 0.3
)

Arguments

Details

The spatial discretisation of the advective flux is read from the flux entry of the second_order_w slot of params. With a second-order flux scheme active the steady state must match the one that project() converges to. Because the limiter depends on the solution, the steady state is then found by an under-relaxed Picard iteration: the limiter is frozen at the current iterate, the resulting tridiagonal system is solved, and the iterate is updated towards that solution, repeating until it converges. (At dt = 1 the limited operator is not diagonally dominant, so the plain fixed-point map only stalls; under-relaxation makes it converge.)

The returned abundance is held at zero above each species' w_max, the same upper boundary condition that project() imposes (via zero_above_support() in project_n()) and that the Newton solver solves on. Without this the bottom-up solve would carry density above w_max whenever growth is still positive there or diffusion pushes density past it.

Value

A matrix with the steady state abundance

Description

Internal helper to select the saved time points whose times lie between the smallest and largest values in time_range, inclusive.

Usage

get_time_elements(sim, time_range, slot_name = "n")

Arguments

Value

A named logical vector, with one entry for each saved time in sim, indicating whether that time lies in the requested range.

Description

The observed yield lives in the yield_observed column of the gear parameter data frame, see gear_params(), where it is given for each gear-species pair. This function adds the observations up over the gears to give the total observed yield of each species. With the gear argument you can restrict the sum to a subset of the gears.

Usage

get_yield_observed(params, gear = NULL)

Arguments

Details

Older models, and the examples in older versions of mizer, put yield_observed into the species parameter data frame instead. That is still accepted: a species that has no observation among the gear parameters takes its value from the species parameters. Where both tables give a value for a species, the gear parameters win. The species parameter observation is a total over all gears, so it is ignored when gear selects only some of the gears.

Value

A numeric vector with one entry for each species, named by species, holding the observed yield in grams per year, or NA for species without an observation. NULL if the observations are not available: if neither the gear parameters nor the species parameters have a yield_observed column, or, when gear is given, if the gear parameters have no such column.

Description

Calculates the total biomass through time within user defined size limits. The default option is to use the size range starting at the size specified by the biomass_cutoff species parameter, if it is set, or else the full size range of each species. You can specify minimum and maximum weight or length range for the species. Lengths take precedence over weights (i.e. if both min_l and min_w are supplied, only min_l will be used).

Usage

getBiomass(object, use_cutoff = FALSE, ...)

Arguments

Details

When no size range arguments are provided, the function checks if the biomass_cutoff column exists in the species parameters. If it does, those values are used as the minimum weight for each species. For species with NA values in biomass_cutoff, the default minimum weight (smallest weight in the model) is used.

Value

If called with a MizerParams object, a named vector with the biomass in grams for each species in the model. If called with a MizerSim object, an ArrayTimeBySpecies object (time x species) containing the biomass in grams at each time step for all species.

See Also

Other summary functions: getDiet(), getGrowthCurves(), getN(), getSSB(), getSteadyResidual(), getTrophicLevel(), getTrophicLevelBySpecies(), getYield(), getYieldGear()

Examples

biomass <- getBiomass(NS_sim)
biomass["1972", "Herring"]
biomass <- getBiomass(NS_sim, min_w = 10, max_w = 1000)
biomass["1972", "Herring"]

# If species_params contains a `biomass_cutoff`` column, it can be used
# as the minimum weight when use_cutoff = TRUE
species_params(NS_sim@params)$biomass_cutoff <- 10
biomass <- getBiomass(NS_sim, use_cutoff = TRUE)  # Uses biomass_cutoff as min_w
biomass["1972", "Herring"]

Description

Calculates the slope of the community abundance by performing a linear regression on the logged total numerical abundance at weight and logged weights (natural logs, not log to base 10, are used). You can specify minimum and maximum weight or length range for the species. Lengths take precedence over weights (i.e. if both min_l and min_w are supplied, only min_l will be used). You can also specify the species to be used in the calculation.

Usage

getCommunitySlope(object, species = NULL, biomass = TRUE, ...)

Arguments

Value

A data.frame with columns slope, intercept and the coefficient of determination R^2 (and a time step column when called with a MizerSim object).

See Also

Other functions for calculating indicators: getMeanMaxWeight(), getMeanWeight(), getProportionOfLargeFish()

Examples

# Slope based on biomass, using all species and sizes
slope_biomass <- getCommunitySlope(NS_sim)
slope_biomass[1, ] # in 1976
slope_biomass[idxFinalT(NS_sim), ] # in 2010

# Slope based on numbers, using all species and sizes
slope_numbers <- getCommunitySlope(NS_sim, biomass = FALSE)
slope_numbers[1, ] # in 1976

# Slope based on biomass, using all species and sizes between 10g and 1000g
slope_biomass <- getCommunitySlope(NS_sim, min_w = 10, max_w = 1000)
slope_biomass[1, ] # in 1976

# Slope based on biomass, using only demersal species and
# sizes between 10g and 1000g
dem_species <- c("Dab","Whiting", "Sole", "Gurnard", "Plaice",
                 "Haddock", "Cod", "Saithe")
slope_biomass <- getCommunitySlope(NS_sim, species = dem_species,
                                   min_w = 10, max_w = 1000)
slope_biomass[1, ] # in 1976

getCommunitySlope(NS_params)

Description

The critical feeding level is the feeding level at which the food intake is just high enough to cover the metabolic costs, with nothing left over for growth or reproduction.

Usage

getCriticalFeedingLevel(params)

Arguments

Value

An ArraySpeciesBySize object (species x size) with the critical feeding level

Examples

str(getFeedingLevel(NS_params))

Description

Calculates the rate at which a predator of a particular species and size consumes biomass of each prey species, resource, and other components of the ecosystem. Returns either the rates in grams/year or the proportion of the total consumption rate.

Usage

getDiet(object, proportion = TRUE, ...)

Arguments

Details

The rates $D_{ij}(w)$ at which a predator of species $i$ and size $w$ consumes biomass from prey species $j$ are calculated from the predation kernel $\phi_i(w, w_p)$, the search volume $\gamma_i(w)$, the feeding level $f_i(w)$, the species interaction matrix $\theta_{ij}$ and the prey abundance density $N_j(w_p)$:

$D_{ij}(w, w_p) = (1-f_i(w)) \gamma_i(w) \theta_{ij} \int N_j(w_p) \phi_i(w, w_p) w_p dw_p.$

The prey index $j$ runs over all species and the resource.

Extra columns are added for the external encounter rate and for any extra ecosystem components in your model for which you have defined an encounter rate function. These encounter rates are multiplied by $1-f_i(w)$ to give the rate of consumption of biomass from these extra components.

This function performs the same integration as getEncounter() but does not aggregate over prey species, and multiplies by $1-f_i(w)$ to get the consumed biomass rather than the available biomass. Outside the range of sizes for a predator species the returned rate is zero. Summing the result of getDiet(proportion = FALSE) over prey therefore reproduces getEncounter(params) * (1 - getFeedingLevel(params)), whichever quadrature scheme the model uses (see second_order_w()).

Value

• MizerParams: An array (predator species x predator size x (prey species + resource + other components)). Dimnames are "predator", "w", and "prey".

• MizerSim: A four-dimensional array (time x predator species x predator size x prey) with the diet at each selected saved time step. If drop = TRUE then dimensions of length 1 are removed.

See Also

plotDiet()

Other summary functions: getBiomass(), getGrowthCurves(), getN(), getSSB(), getSteadyResidual(), getTrophicLevel(), getTrophicLevelBySpecies(), getYield(), getYieldGear()

Examples

diet <- getDiet(NS_params)
str(diet)

# For a MizerSim the diet is returned at each saved time step
sim <- project(NS_params, t_max = 20, effort = 0.5)
# Diet at the saved time steps over years 15 - 20
diet <- getDiet(sim, time_range = c(15, 20))
str(diet)

Description

Calculates the diffusion rate $D_i(w)$ (grams^2/year) for each species. This diffusion rate has two components:

1. The diffusion due due to the variability in prey sizes. This is the diffusion term from the jump-growth equation.

2. Any externally specified diffusion, which is added via setExtDiffusion()

Usage

getDiffusion(object, ...)

Arguments

Details

The diffusion due due to the variability in prey sizes is determined by summing over all prey species and the resource spectrum and then integrating over all prey sizes $w_p$, weighted by predation kernel $\phi(w,w_p)$:

$d_i(w) = (1-f_i(w))(\alpha_i(1-\psi_i(w)))^2\gamma_i(w) \int \left( \theta_{ip} N_R(w_p) + \sum_{j} \theta_{ij} N_j(w_p) \right) \phi_i(w,w_p) w_p^2 \, dw_p.$

Here $N_j(w)$ is the abundance density of species $j$ and $N_R(w)$ is the abundance density of resource. The overall prefactor $\gamma_i(w)$ determines the predation power of the predator. It could be interpreted as a search volume and is set with the setSearchVolume() function. The predation kernel $\phi(w,w_p)$ is set with the setPredKernel() function. The species interaction matrix $\theta_{ij}$ is set with setInteraction() and the resource interaction vector $\theta_{ip}$ is taken from the interaction_resource column in species_params(). $f(w)$ is the feeding level calculated with getFeedingLevel(). $\psi(w)$ is the proportion of the available energy that is invested in reproduction instead of growth, obtained with psi().

The diffusion integral is normally evaluated efficiently with a fast Fourier transform, which assumes that the predation kernel depends only on the ratio of predator to prey size. If a custom predation kernel that depends on predator and prey size separately has been set with setPredKernel(), the integral is instead evaluated by direct summation over the full predation kernel, as in getEncounter().

Value

• MizerParams: An ArraySpeciesBySize object (predator species x predator size) with the diffusion rates.

• MizerSim: An ArrayTimeBySpeciesBySize object (time step x predator species x predator size) with the diffusion rates at every time step. If drop = TRUE then dimensions of length 1 will be removed.

References

Datta, S., Delius, G. W. and Law, R. (2010). A jump-growth model for predator-prey dynamics: derivation and application to marine ecosystems. Bulletin of Mathematical Biology, 72(6):1361–1382

See Also

Other rate functions: getEGrowth(), getERepro(), getEReproAndGrowth(), getEncounter(), getFMort(), getFMortGear(), getFeedingLevel(), getFlux(), getFluxGradient(), getMort(), getPredMort(), getPredRate(), getRDD(), getRDI(), getRates(), getResourceMort()

Description

Computes the eigenvalues $\mu_i$ of the linearised one-step-ahead map at the steady state stored in params@initial_n, for a given step size dt. These describe how mizer's numerical scheme, rather than the model, behaves near the steady state: the map does not amplify perturbations when the spectral radius $\max_i|\mu_i|$ is less than 1.

Usage

getDiscreteStability(params, effort = params@initial_effort, h = 1e-04, dt = 1)

Arguments

Details

This is the numerical counterpart of getStability(), which analyses the model itself and involves no time step at all. Use getStability() to ask whether the steady state of the model is stable, and this function to ask what mizer's solver does at a particular dt. The two can disagree, and that disagreement is the point: the implicit transport solve damps oscillations artificially, so a physically unstable steady state can have a spectral radius below 1 at a large dt, and the simulation then sits at a state the model does not actually hold.

The map that is linearised

One step is what project() takes with method = "euler": the rates are evaluated at the state at the start of the step, and the resulting transport problem is solved implicitly,

$A(N^t, n_{pp}^t)\,N^{t+1} = S(N^t, n_{pp}^t),$

with the same project_n_loop() C++ Thomas solver and the same spatial scheme (second_order_w()) as the regular dynamics.

Because the rates are evaluated at the input state, the step is not fully implicit, and the discrete eigenvalues therefore cannot be converted into continuous-time eigenvalues by any exact algebraic relation. That conversion is what getStability() avoids by differentiating the rates of change themselves.

The resource is advanced by the model's own resource_dynamics function, the one project() calls. Nothing is substituted for it: the map that is differentiated here reproduces a single project(method = "euler") step exactly, which is what makes the spectral radius a statement about mizer's solver rather than about a nearby scheme.

Value

A named list with the following components:

discrete_eigenvalues

Complex vector of the eigenvalues $\mu_i$ of the one-step map, sorted by decreasing modulus.

spectral_radius

$\max_i|\mu_i|$. Less than 1 means the numerical scheme is stable at this dt.

stable

Logical: TRUE when spectral_radius < 1.

dt

The step size the map was evaluated at.

n_active

Dimension of the Jacobian.

leading_eigenvectors

The eigenvectors of the two largest-modulus eigenvalues, in the same shape as for getStability().

params

The validated params object the analysis was made at.

Requires smooth dynamics

The finite-difference Jacobian is only meaningful if the rates of change are differentiable at $N^*$. A custom rate function registered with setRateFunction() that jumps as a function of the abundances breaks this in two ways. If the state sits on the switching threshold, some perturbations straddle it and pick up the jump, and the reported eigenvalues then vary wildly with h. If the state is near but not on the threshold, no perturbation crosses it, and the function silently returns the stability of the single branch the state happens to lie on — which can read as stable for a model whose simulations never settle.

Re-running with a different h is the cheapest check: if the answer moves, do not trust it. See Discontinuous rate functions.

See Also

getStability(), findSteadyState()

Description

Note that the array returned may not be exactly the same as the effort argument that was passed in to project(). This is because only the saved effort is stored (the frequency of saving is determined by the argument t_save).

Usage

getEffort(sim)

Arguments

Value

An array (time x gear) that contains the fishing effort by time and gear.

Examples

str(getEffort(NS_sim))

Description

Calculates the energy rate $g_i(w)$ (grams/year) available by species and size for growth after metabolism, movement and reproduction have been accounted for.

Usage

getEGrowth(object, ...)

Arguments

Details

The growth rate is calculated as the difference between the energy available for reproduction and growth (obtainable with getEReproAndGrowth()) and the energy used for reproduction (obtainable with getERepro()), but is set to 0 if the result would be negative.

Value

• MizerParams: An ArraySpeciesBySize object (species x size) with the somatic growth rates (grams/year).

• MizerSim: An ArrayTimeBySpeciesBySize object (time step x species x size) with the growth rates at every time step. If drop = TRUE then dimensions of length 1 will be removed.

Your own growth rate function

By default getEGrowth() calls mizerEGrowth(). However you can replace this with your own alternative growth rate function. If your function is called "myEGrowth" then you register it in a MizerParams object params with

params <- setRateFunction(params, "EGrowth", "myEGrowth")

Your function will then be called instead of mizerEGrowth(), with the same arguments.

See Also

getERepro(), getEReproAndGrowth()

Other rate functions: getDiffusion(), getERepro(), getEReproAndGrowth(), getEncounter(), getFMort(), getFMortGear(), getFeedingLevel(), getFlux(), getFluxGradient(), getMort(), getPredMort(), getPredRate(), getRDD(), getRDI(), getRates(), getResourceMort()

Examples

params <- NS_params
# Project with constant fishing effort for all gears for 20 time steps
sim <- project(params, t_max = 20, effort = 0.5)
# Get the energy at a particular time step
growth <- getEGrowth(params, n = N(sim)[15, , ], n_pp = NResource(sim)[15, ], t = 15)
# Growth rate at this time for Sprat of size 2g
growth["Sprat", "2"]

Description

Returns the rate at which a predator of species $i$ and weight $w$ encounters food (grams/year).

Usage

getEncounter(object, ...)

Arguments

Value

• MizerParams: An ArraySpeciesBySize object (predator species x predator size) with the encounter rates.

• MizerSim: An ArrayTimeBySpeciesBySize object (time step x predator species x predator size) with the encounter rates at every time step. If drop = TRUE then dimensions of length 1 will be removed.

Predation encounter

The encounter rate $E_i(w)$ at which a predator of species $i$ and weight $w$ encounters food has contributions from the encounter of fish prey and of resource. This is determined by summing over all prey species and the resource spectrum and then integrating over all prey sizes $w_p$, weighted by predation kernel $\phi(w,w_p)$:

$E_i(w) = \gamma_i(w) \int \left( \theta_{ip} N_R(w_p) + \sum_{j} \theta_{ij} N_j(w_p) \right) \phi_i(w,w_p) w_p \, dw_p.$