R package unmarked

unmarked is an R package for analyzing ecological data arising from several popular sampling techniques. The sampling methods include point counts, occurrence sampling, distance sampling, removal, double observer, and many others. unmarked uses hierarchical models to incorporate covariates of the latent abundance (or occupancy) and imperfect detection processes.

Installation

The latest stable version of unmarked can be downloaded from CRAN:

install.packages("unmarked")

The latest development version can be installed from Github:

install.packages("remotes")
remotes::install_github("rbchan/unmarked")

Support

Support is provided through the unmarked Google group. The package website has more information. You can report bugs here, by posting to the Google group, or by emailing the current maintainer.

Example analysis

Below we demonstrate a simple single-season occupancy analysis using unmarked. First, load in a dataset from a CSV file and format:

library(unmarked)
wt <- read.csv(system.file("csv","widewt.csv", package="unmarked"))

# Presence/absence matrix
y <- wt[,2:4]

# Site and observation covariates
siteCovs <-  wt[,c("elev", "forest", "length")]
obsCovs <- list(date=wt[,c("date.1", "date.2", "date.3")]) 

Create an unmarkedFrame, a special type of data.frame for unmarked analyses:

umf <- unmarkedFrameOccu(y = y, siteCovs = siteCovs, obsCovs = obsCovs)
summary(umf)

## unmarkedFrame Object
## 
## 237 sites
## Maximum number of observations per site: 3 
## Mean number of observations per site: 2.81 
## Sites with at least one detection: 79 
## 
## Tabulation of y observations:
##    0    1 <NA> 
##  483  182   46 
## 
## Site-level covariates:
##       elev               forest              length      
##  Min.   :-1.436125   Min.   :-1.265352   Min.   :0.1823  
##  1st Qu.:-0.940726   1st Qu.:-0.974355   1st Qu.:1.4351  
##  Median :-0.166666   Median :-0.064987   Median :1.6094  
##  Mean   : 0.007612   Mean   : 0.000088   Mean   :1.5924  
##  3rd Qu.: 0.994425   3rd Qu.: 0.808005   3rd Qu.:1.7750  
##  Max.   : 2.434177   Max.   : 2.299367   Max.   :2.2407  
## 
## Observation-level covariates:
##       date         
##  Min.   :-2.90434  
##  1st Qu.:-1.11862  
##  Median :-0.11862  
##  Mean   :-0.00022  
##  3rd Qu.: 1.30995  
##  Max.   : 3.80995  
##  NA's   :42

Fit a null occupancy model and a model with covariates, using the occu function:

(mod_null <- occu(~1~1, data=umf))

## 
## Call:
## occu(formula = ~1 ~ 1, data = umf)
## 
## Occupancy:
##  Estimate    SE     z  P(>|z|)
##    -0.665 0.139 -4.77 1.82e-06
## 
## Detection:
##  Estimate    SE    z  P(>|z|)
##      1.32 0.174 7.61 2.82e-14
## 
## AIC: 528.987

(mod_covs <- occu(~date~elev, data=umf))

## 
## Call:
## occu(formula = ~date ~ elev, data = umf)
## 
## Occupancy:
##             Estimate    SE     z  P(>|z|)
## (Intercept)   -0.738 0.157 -4.71 2.45e-06
## elev           0.885 0.174  5.10 3.49e-07
## 
## Detection:
##             Estimate    SE     z  P(>|z|)
## (Intercept)   1.2380 0.180 6.869 6.47e-12
## date          0.0603 0.121 0.497 6.19e-01
## 
## AIC: 498.158

Rank them using AIC:

fl <- fitList(null=mod_null, covs=mod_covs)
modSel(fl)

##      nPars    AIC delta AICwt cumltvWt
## covs     4 498.16  0.00 1e+00     1.00
## null     2 528.99 30.83 2e-07     1.00

Estimate occupancy probability using the top-ranked model at the first six sites:

head(predict(mod_covs, type='state'))

##   Predicted         SE      lower     upper
## 1 0.1448314 0.03337079 0.09080802 0.2231076
## 2 0.1499962 0.03351815 0.09535878 0.2280473
## 3 0.2864494 0.03346270 0.22555773 0.3562182
## 4 0.3035399 0.03371489 0.24175619 0.3733387
## 5 0.1607798 0.03374307 0.10502635 0.2382512
## 6 0.1842147 0.03392277 0.12669813 0.2600662

Predict occupancy probability at a new site with given covariate values:

nd <- data.frame(elev = 1.2)
predict(mod_covs, type="state", newdata=nd)

##   Predicted         SE     lower     upper
## 1 0.5803085 0.06026002 0.4598615 0.6918922