AEC 501

OCCUPANCY EXERCISE

HOMEWORK EXERCISE

The exercise below works through an analysis of Willow Flycatcher occupancy data. Please include your R script (just the script file) so we can help address any difficulties. The in-lab pdf file includes all the necessary formats for this exercise. You will need to update/edit the in-lab code to complete this exercise.

Detection/non-detection data were collected during Willow Flycatcher surveys at 50 sites that were visited 3 times during the 2012 breeding season. These data are located in “flycatch.csv”. Data on percent woody habitat and survey date were also collected and are included in the file (survey date is formatted so the very first survey occurred on day 1).

1.  Download and read in the data.

2.  Calculate the naïve occupancy rate for the first survey.

3.  Properly format the data for occupancy analysis in unmarked – notice there is a site level covariate (woody) and an observation level covariate (date). The last slide in the lab exercise provides an example of formatting covariates for unmarked.

4.  Think about hypotheses for date and woody habitat. Describe why these covariates may influence occupancy and/or detection?

5.  Run the following 5 models -

a.  The null model (constant occupancy and constant detection)

b.  The model with woody on occupancy and constant detection

c.  The model with woody on detection and constant occupancy

d.  The model with date on detection and constant occupancy

e.  The model with woody on occupancy and date on detection

6.  Examine the results of your models.

a.  Use backTransform() to get the estimate of occupancy and detection from the null model. How does the occupancy estimate compare with the naïve estimate above, why?

b.  Use AIC to compare all 5 models – show the AIC table.

c.  What was the top model, explain why?

d.  Look at the top model. Are woody habitat and/or date associated with Willow Flycatcher detection and/or occupancy?

e.  Describe possible biological reasons for these results. Did the results support your hypotheses in #4?