unmarkedFrameOccuTTD {unmarked}

Create an unmarkedFrameOccuTTD object for the time-to-detection model fit by occuTTD

Description

Organizes time-to-detection occupancy data along with covariates. This S4 class is required by the data argument of occuTTD

Usage

  unmarkedFrameOccuTTD(y, surveyLength, siteCovs=NULL, obsCovs=NULL, 
                           numPrimary=1, yearlySiteCovs=NULL)

Arguments

y

An MxR matrix of time-to-detection data for a species, where M is the number of sites and R is the maximum number of observations per site (across all primary periods and observations, if you have multi-season data). Values in y should be positive.

surveyLength

The maximum length of a survey, in the same units as y. You can provide either a single value (if all surveys had the same max length), or a matrix matching the dimensions of y (if surveys had different max lengths).

siteCovs

A data.frame of covariates that vary at the site level. This should have M rows and one column per covariate

obsCovs

Either a named list of data.frames of covariates that vary within sites, or a data.frame with MxR rows in the ordered by site-observation (if single-season) or site-primary period-observation (if multi-season).

numPrimary

Number of primary time periods (e.g. seasons) for the dynamic or multi-season version of the model. There should be an equal number of secondary periods in each primary period.

yearlySiteCovs

A data frame with one column per covariate that varies among sites and primary periods (e.g. years). It should have MxT rows where M is the number of sites and T the number of primary periods, ordered by site-primary period. These covariates only used for dynamic (multi-season) models.

Details

unmarkedFrameOccuTTD is the S4 class that holds data to be passed to the occuTTD model-fitting function.

Value

an object of class unmarkedFrameOccuTTD

Note

If the time-to-detection values in y are very large (e.g., because they are expressed as numbers of seconds) you may have issues fitting models. An easy solution is to convert your units (e.g., from seconds to decimal minutes) to keep the values as close to 0 as possible.

Author(s)

Ken Kellner [e-mail]

Examples

  
  # For a single-season model
  N <- 100 #Number of sites
  psi <- 0.4 #Occupancy probability
  lam <- 7 #Parameter for exponential distribution of time to detection
  Tmax <- 10 #Maximum survey length

  z <- rbinom(N, 1, psi) #Simulate occupancy
  y <- rexp(N, 1/lam) #Simulate time to detection
  y[z==0] <- Tmax
  y[y>Tmax] <- Tmax
  
  sc <- as.data.frame(matrix(rnorm(N*2),ncol=2)) #Site covs
  oc <- as.data.frame(matrix(rnorm(N*2),ncol=2)) #obs covs

  umf <- unmarkedFrameOccuTTD(y=y, surveyLength=Tmax, siteCovs=sc, obsCovs=oc)