LP_TL_simulate {Petersen}

Simulate data from a Lincoln-Petersen Model with Tag Loss

Description

This function creates simulated capture histories for the Lincoln-Petersen model with tag loss.

Usage

LP_TL_simulate(
  dt_type = NULL,
  N = 1000,
  cov1 = function(N) {
     rep(1, N)
 },
  cov2 = function(cov1) {
     rep(1, length(cov1))
 },
  p1 = function(cov1, cov2) {
     rep(0.1, length(cov1))
 },
  pST = function(cov1, cov2) {
     rep(0.5, length(cov1))
 },
  pST.1 = function(cov1, cov2) {
     rep(1, length(cov1))
 },
  rho1 = function(cov1, cov2) {
     rep(0.8, length(cov1))
 },
  rho2 = function(cov1, cov2) {
     rep(0.8, length(cov1))
 },
  p2 = function(cov1, cov2) {
     rep(0.1, length(cov1))
 },
  seed = round(1e+08 * runif(1)),
  trace = FALSE
)

Arguments

dt_type

Double Tag type. Valid values are notD, twoD, and t2perm for two indistinguishable tags; two distinguishable tags, when the second tag is a permanent tag and cannot be lost, respectively.

N

Population size

cov1

Function to generate first covariate for each member of population as function of N

cov2

Function to generate second covariate for each member of population as function of cov1.

p1

Function to generate P(capture) at event 1 for each member of population as function of cov1,cov2.

pST

Function to generate P(single tag) if captured at event 1 as function of cov1,cov2.

pST.1

Function to generate p(apply single tag to first position at event 1) as function of cov1,cov2.

rho1

Function to generate P(tag1 retained) as function of cov1,cov2.

rho2

Function to generate P(tag2 retained) as function of cov1,cov2.

p2

Function to generate P(capture) at event 2 for each member of population as function of cov1,cov2.

seed

Initial value of random seed

trace

Trace flag to help debug if things fail.

Details

The cov1 function takes the value N and returns N covariate values. For example these could be simulated length, or sex of each fish. The cov2 function takes the cov1 values and generates a second covariate. Two covariates should be sufficient for most capture-recapture simulations. If generating continuous covariates, you should round the covariate to about 100 distinct values to speed up your simulation.

The remaining functions take the two covariate values and generate capture probabilities, single tag probabilities, placing tags on fish, and tag retention probabilities. These should all be in the range of 0 to 1.

After generating capture histories for the entire population, animals never seen are "discarded" and the data set is compress to unique combinations of the two covariates and the capture history with the frequency variable set accordingly.

Value

Data frame with observed capture histories

Examples

sim_data <-LP_TL_simulate(
      dt_type="t2perm",  # permanent tag
      N=1000,
      cov1=function(N)         {rep(1,N)},
      cov2=function(cov1)      {rep(1,  length(cov1))},
      p1  =function(cov1, cov2){rep(.1, length(cov1))},
      pST =function(cov1, cov2){rep(.25,length(cov1))},
      rho1=function(cov1, cov2){rep(.70,length(cov1))},
      rho2=function(cov1, cov2){rep(1,  length(cov1))},  # permanent second tag
      p2  =function(cov1, cov2){rep(.1, length(cov1))},
      seed=round(1000000*runif(1)))
sim_data