Simulate Royle-Nichols occupancy data — simulate_royle

Latent abundance N_i ~ Poisson(lambda_i), log lambda = X beta_lambda, and per-visit detection y_ij ~ Bernoulli(1 - (1 - r_ij)^{N_i}) with per-individual detection logit r_ij = beta_r. Detection is site-level by default; supplying beta_r_visit draws a per-visit covariate w and makes logit r_ij = beta_r + beta_r_visit * w_ij visit-varying. Matches the royle_nichols() family; the response is an N x J 0/1 detection-history matrix.

Usage

simulate_royle_nichols(
  N = 200,
  J = 5,
  beta_lambda = c(0.3, 0.5),
  beta_r = -0.8,
  beta_r_visit = NULL,
  seed = NULL
)

Arguments

N: Number of sites (default 200).
J: Number of visits per site (default 5).
beta_lambda: Log-abundance coefficients c(intercept, slope_on_x).
beta_r: Per-individual detection logit (a scalar intercept).
beta_r_visit: Optional scalar slope on a drawn per-visit detection covariate w. When supplied, detection varies by visit and the returned list carries visits = list(w = ) (an N x J matrix) for tobs(..., visits =).
seed: Random seed.

Value

A list with y (N x J 0/1 matrix), data, truth (beta_lambda, beta_r, beta_r_visit, realised abundance N, lambda, per-site or per-visit r), and, when beta_r_visit is supplied, visits.

Examples

sim <- simulate_royle_nichols(N = 100, J = 4, seed = 1)
dim(sim$y)
# Visit-varying detection:
sv <- simulate_royle_nichols(N = 100, J = 4, beta_r_visit = 0.8, seed = 1)
names(sv$visits)