Set up parameters for a trait-based model

Edit this page.   Source: R/wrapper_functions.R

This functions creates a MizerParams object so that trait-based models can be easily set up and run. A trait-based size spectrum model is a simplification of the general size-based model used in mizer. The species-specific parameters are the same for all species, except for the asymptotic size, which is considered the most important trait characterizing a species. Other parameters are related to the asymptotic size. For example, the size at maturity is given by w_inf * eta, where eta is the same for all species. For the trait-based model the number of species is not important. For applications of the trait-based model see Andersen & Pedersen (2010). See the mizer vignette for more details and examples of the trait-based model.

newTraitParams(
  no_sp = 11,
  min_w_inf = 10,
  max_w_inf = 10^3,
  min_w = 10^(-4),
  max_w = max_w_inf,
  eta = 10^(-0.6),
  min_w_mat = min_w_inf * eta,
  no_w = log10(max_w_inf/min_w) * 50 + 1,
  min_w_pp = NA,
  w_pp_cutoff = min_w_inf,
  n = 2/3,
  p = n,
  lambda = 2.05,
  r_pp = 0.1,
  kappa = 0.005,
  alpha = 0.4,
  ks = 4,
  h = 30,
  beta = 100,
  sigma = 1.3,
  f0 = 0.6,
  gamma = NA,
  bmort_prop = 0,
  starv_coef = 10,
  rfac = 4,
  knife_edge_size = 1000,
  gear_names = "knife_edge_gear",
  egg_size_scaling = FALSE,
  perfect_scaling = FALSE,
  ...
)

Arguments

no_sp

The number of species in the model. The default value is 11.
min_w_inf

The asymptotic size of the smallest species in the community. Default value is 10. This will be rounded to lie on a grid point.
max_w_inf

The asymptotic size of the largest species in the community. Default value is 1000. This will be rounded to lie on a grid point.
min_w

The size of the the egg of the smallest species. Default value is 10^(-4). This also defines the start of the community size spectrum.
max_w

The largest size in the model. By default this is set to the largest asymptotic size max_w_inf. Setting it to something larger only makes sense if you plan to add larger species to the model later.
eta

Ratio between maturity size and asymptotic size of a species. Ignored if min_w_mat is supplied. Default is 10^(-0.6), approximately 1/4.
min_w_mat

The maturity size of the smallest species. Default value is eta * min_w_inf. This will be rounded to lie on a grid point.
no_w

The number of size bins in the community spectrum. These bins will be equally spaced on a logarithmic scale. Default value is such that there are 50 bins for each factor of 10 in weight.
min_w_pp

The smallest size of the plankton spectrum. By default this is set to the smallest value at which any of the consumers can feed.
w_pp_cutoff

The largest size of the plankton spectrum. Default value is max_w_inf unless perfect_scaling = TRUE when it is Inf.
n

Scaling exponent of the maximum intake rate. Default value is 2/3.
p

Scaling exponent of the standard metabolic rate. By default this is equal to the exponent n.
lambda

Exponent of the abundance power law.
r_pp

Growth rate parameter for the plankton spectrum. Default value is 0.1.
kappa

Coefficient in abundance power law. Default value is 0.005.
alpha

The assimilation efficiency of the community. The default value is 0.4.
ks

Standard metabolism coefficient. Default value is 4.
h

Maximum food intake rate. Default value is 30.
beta

Preferred predator prey mass ratio. Default value is 100.
sigma

Width of prey size preference. Default value is 1.3.
f0

Expected average feeding level. Used to set gamma, the coefficient in the search rate. The default value is 0.6. Ignored if gamma is given explicitly.
gamma

Volumetric search rate. If not provided, default is determined by get_gamma_default using the value of f0.
bmort_prop

The proportion of the total mortality that comes from background mortality, i.e., from sources other than predation or fishing. A number in the interval [0, 1). Default 0.
starv_coef

Proportionality constant for starvation mortality. When starv_coef is equal to 10 the instantaneous starvation mortality (1/year) is 1 when energy deficit is 10 is no starvation mortality
rfac

The factor such that R_max = rfac * R, where R_max is the maximum recruitment allowed and R is the steady-state recruitment. Thus the larger rfac the less the impact of the non-linear stock-recruitment curve. The default is 4.
knife_edge_size

The minimum size at which the gear or gears select fish. A vector with the length equal to the number of gears.
gear_names

The names of the fishing gears. A character vector, the same length as the number of gears. Default is "knife_edge_gear".
egg_size_scaling

Boolean. Default FALSE. If TRUE, the egg size is a constant fraction of the maximum size of each species. This fraction is min_w / min_w_inf. If FALSE, all species have the egg size w_min.
perfect_scaling

Boolean. Default FALSE. If TRUE then parameters are set so that the community abundance, growth before reproduction and death are perfect power laws.
...

Other arguments to pass to the MizerParams constructor.

Value

An object of type MizerParams

Details

The function has many arguments, all of which have default values. Of particular interest to the user are the number of species in the model and the minimum and maximum asymptotic sizes.

The characteristic weights of the smallest species are defined by min_w (egg size), min_w_mat (maturity size) and min_w_inf (asymptotic size). The asymptotic sizes of the no_sp species are logarithmically evenly spaced, ranging from min_w_inf to max_w_inf. Similarly the maturity sizes of the species are logarithmically evenly spaced, so that the ratio eta between maturity size and asymptotic size is the same for all species. If egg_size_scaling = TRUE then also the ratio between asymptotic size and egg size is the same for all species. Otherwise all species have the same egg size.

In addition to setting up the parameters, this function also sets up an initial condition that is close to steady state.

Although the trait based model's steady state is often stable without imposing a stock recruitment relationship, the function can set a Beverton-Holt type stock recruitment relationship that imposes a maximal reproduction rate that is a multiple of the recruitment rate at steady state. That multiple is set by the argument rfac.

The search rate coefficient gamma is calculated using the expected feeding level, f0.

The option of including fishing is given, but the steady state may lose its natural stability if too much fishing is included. In such a case the user may wish to include stabilising effects (like rfac) to ensure the steady state is stable. Fishing selectivity is modelled as a knife-edge function with one parameter, knife_edge_size, which is the size at which species are selected. Each species can either be fished by the same gear (knife_edge_size has a length of 1) or by a different gear (the length of knife_edge_size has the same length as the number of species and the order of selectivity size is that of the asymptotic size).

The resulting MizerParams object can be projected forward using project() like any other MizerParams object. When projecting the model it may be necessary to reduce dt below 0.1 to avoid any instabilities with the solver. You can check this by plotting the biomass or abundance through time after the projection.

See also

Other functions for setting up models: newCommunityParams(), newMultispeciesParams(), newSheldonParams()

Examples

if (FALSE) {
params <- newTraitParams()
sim <- project(params, t_max = 5, effort = 0)
plotSpectra(sim)
}