Skip to contents

Sets the release cohorts, release platform, tagging timing, tag-induced mortality, chronic shedding, the recapture likelihood and reporting rates, and allocates the recapture containers. Call after Setup_Sim_Dim.

Usage

Setup_Sim_Tagging(
  sim_list,
  n_tags = NULL,
  n_tags_rel_input = NULL,
  use_conv_fish_tagging = 0,
  conv_tag_max_liberty = sim_list$n_ages/2,
  conv_tag_release_indicator = expand.grid(regions = 1:sim_list$n_regions, tag_years =
    1:sim_list$n_yrs, tag_seas = 1:sim_list$n_seas),
  conv_tag_release_platform = default_tag_release_platform(conv_tag_release_indicator),
  conv_tag_t_tagging = 1,
  ln_init_conv_tag_mort = -1000,
  ln_conv_tag_shed = -1000,
  conv_fish_tag_attr = "p_a_s",
  conv_tag_fish_reporting_input = array(0.5, dim = c(sim_list$n_regions, sim_list$n_yrs,
    sim_list$n_fish_fleets, sim_list$n_sims)),
  conv_fish_tag_like = 0,
  ln_conv_fish_tag_theta = log(1),
  conv_tag_pop_pool = NULL,
  conv_tag_age_pool = NULL,
  conv_tag_sex_pool = NULL,
  conv_tagged_fish_input = NULL
)

Arguments

sim_list

Simulation list returned by Setup_Sim_Dim.

n_tags

Constant number of tags released per release event. Cannot be given alongside n_tags_rel_input. Default NULL.

n_tags_rel_input

Array of cohort-specific release numbers, overriding n_tags and not given alongside it. Default NULL.

use_conv_fish_tagging

Integer (0/1) for whether fishery tag recaptures are simulated. Default 0.

conv_tag_max_liberty

Integer maximum years at liberty tracked per cohort, which sizes the recapture containers. Default n_ages / 2.

conv_tag_release_indicator

Data frame of release cohorts with columns regions, tag_years and tag_seas. Default every combination of the region, year and season in sim_list.

conv_tag_release_platform

Character matrix [n_conv_tag_cohorts × 2] of the release platform and fleet index per cohort, aligned row by row with conv_tag_release_indicator. "population" releases tags into the population independently of any sampling process, with fleet set to NA; "fishery" and "survey" release them through the fleet the fleet column names. Default "survey" with fleet 1.

conv_tag_t_tagging

Numeric scalar or vector [n_tag_rel_events] in \([0, 1]\), the fraction of the season remaining at release: 1 the start of the season, 0.5 mid-season, 0 the end. A scalar is recycled. Default 1.

ln_init_conv_tag_mort

Log-scale tag-induced mortality applied at release, a scalar or a vector [n_tag_rel_events] with a scalar recycled. Default -1000, effectively none.

ln_conv_tag_shed

Log-scale annual chronic shedding rate, a scalar or a vector [n_tag_rel_events] with a scalar recycled. Default -1000, effectively none.

conv_fish_tag_attr

Character scalar or vector [n_tag_rel_events] naming which dims are resolved at release and so kept at recapture, built from "p" (population), "a" (age) and "s" (sex) joined by underscores: "p_a_s" (default), "a_s", "p_a", "p_s", "a", "s", "p" or "none". A scalar is recycled; a vector lets events resolve different dims. Region and fleet are always kept.

conv_tag_fish_reporting_input

Fishery reporting rate array [n_regions × n_yrs × n_fish_fleets × n_sims] in \([0, 1]\), the probability a recaptured tag is reported. Default 0.5.

conv_fish_tag_like

Tag recapture likelihood: 0/"Poisson" (default), 1/"NegBin", 2/"Multinomial_Release", 3/"Multinomial_Recapture", 4/"Dirichlet-Multinomial_Release" or 5/"Dirichlet-Multinomial_Recapture".

ln_conv_fish_tag_theta

Log-scale overdispersion for the negative binomial and the two Dirichlet-multinomial likelihoods, ignored by the others. Default log(1).

conv_tag_pop_pool, conv_tag_age_pool, conv_tag_sex_pool

Lists of integer vectors grouping populations, ages and sexes whose recaptures the estimation model fits as one count, as in Setup_Mod_Tagging. A negative binomial is drawn once per group, since a sum of negative binomials is not the negative binomial the pooled count is fit with. NULL (default) is one group per level.

conv_tagged_fish_input

Released tags by event, population, age and sex [n_tag_rel_events × n_pop × n_ages × n_sexes], the estimation model's conv_tagged_fish. An event given here is released exactly so, and one left NA is spread over populations, ages and sexes by the release platform's selected abundance as before. simulation_self_test and condition_closed_loop_simulations pass the fit's releases, so the tags at liberty are the ones the fit followed. Default NULL.

Value

sim_list with the tagging fields appended: $n_tags or $n_tags_rel_input, $conv_tag_max_liberty, $conv_tag_t_tagging, $ln_init_conv_tag_mort and $ln_conv_tag_shed (each of length n_tag_rel_events), $conv_tag_release_indicator, $conv_tag_release_platform, $n_tag_rel_events, $use_conv_fish_tagging, $conv_fish_tag_like, $conv_fish_tag_attr, $ln_conv_fish_tag_theta and $conv_tag_fish_reporting, plus the zero-filled $conv_tagged_fish, $conv_tagged_fish_attr, $conv_tag_fish_avail, $obs_conv_tag_fish_recap and $pred_conv_tag_fish_recap.