The occupancy estimation problem and data structure are both very closely allied to capture-mark- recapture (Week 6). The array of sites in an occupancy problem is conceptually the same as the list of (unknown) animals in a CMR problem. In fact if we confined ourselves to the animals caught at least once, then a capture history and an occupancy detection history are the same:
E.g., if we had 3 animals captured over 5 occasions we might have capture histories like this:
10010
01010
10100
Likewise these could just as easily have been 3 sites
10010
01010
10100
In fact in this example both problems come down to just modeling the detections (1 or 0) for animals we know are present in the population, and sites we know area occupied, since we detected them at least once. The modeling of the 2 data structures diverges somewhat from that point, since in general 1) we don't know how many sites are occupied (but we know how many sites there are ) 2) for CMR we potentially don't know how many "sites" (i.e., animals there are).. Both problems in fact can be tackled by variations on the same general multinomial modeling framework, so maybe it should be of no surprise that there is overlap in computational approaches.
First, program PRESENCE written by Jim Hines was actually specifically designed to work with occupancy data structures and is very well supported for this class of problems. I tend not to use it because 1) I don't need another software tool to learn and 2) it requires use of an interface and data entry formatting that I find inconvenient to use (read: does not play well with R).
MARK on the other hand will (in addition to performing analysis of CMR and many other types of binomial, multinomial and count data, perform many of the same analyses that the unmarked package in R does, and can interact with R nicely via RMark. I chose to present occupancy analysis strictly using unmarked, but in fact could have done everything we did using MARK/RMark.