Occupancy Estimation Multiple States Single Season

Contents – Index

---

Occupancy Estimation Multiple States Single Season

Estimation of occupancy rates generally assume that a site or plot is occupied or not occupied.  This model extends this concept to not occupied, and 2 levels of occupancy.  For example, a plot may be occupied by a pair of adults that either reproduce, or do not reproduce.  Given that the occupancy states are hierarchial, a “1” is coded for the lower level of occupancy (e.g., adults, but no young), and a “2” is coded for the higher level of occupancy (adults plus young).  Nichols et al. (2007) developed this model with an example of California spotted owls (Strix occidentalis occidentalis) where sites were occupied by 1 or more adults, and or occupied by both adults and young.  I will use their example to explain the 5 parameters of this model.

The real parameters of this model are:

psi1(i) = probability that a site i is occupied regardless of reproductive state, Pr(true state = 1 or 2);

psi2(i) = probability that young occurred, given that the site i is occupied, Pr(true state = 2 | true state = 1 or 2);

p1(i, t) = probability that occupancy is detected for site i, period t, given that true state = 1, Pr(detection | true state = 1);

p2(i, t) = probability that occupancy is detected for site i, period t, given that true state = 2, Pr(detection | true state = 2);

delta(i, t) = probability that evidence of successful reproduction is found, given detection of occupancy at site i, period t, with successful reproduction, Pr(classified state 2|true state = 2).

Thus, psi1 and psi2 are single parameters relating to a site, whereas p1, p2, and delta are time-specific parameters, and thus the number of values in the PIM for each of these parameters equals the number of occasions.

The product psi1*psi2 is the unconditional probability that a site is occupied by successful breeders, and is provided as a derived parameter in MARK.

The encounter histories file for this data type consists of only the L columns, and the example from Nichols et al. (2007) is shown here.

/* occasions=5, groups=1 */
1.221  1;  
011..  1;  
1.1..  2;  
.1212  1;  
1111.  1;  
1022.  1;  
111.1  1;  
0.20.  1;  
11100  1;  
.11..  1;  
.011.  1;  
0110.  1;  
11111  3;  
.0.11  1;  
1100.  1;  
101..  1;  
…0.  2;  
.111.  2;  
0.0.0  1;  
0.210  1;  
01110  1;  
0.0.1  1;  
…1.  2;  
0….  3;  
1.2..  3;  
.1101  1;  
.12..  2;  
1122.  3;  
0120.  1;  
.1111  1;  
11202  1;  
11220  1;  
10110  1;  
.10..  1;  
1.200  1;  
00…  1;  
0122.  2;  
0.11.  1;  
0.100  1;
0..11  1;
  
Notice that “.” are used to indicate that a site was not surveyed on a particular occasion, and “0”, “1” and “2” indicate the level of occupancy detected.

To accommodate this data type’s use of “2”, the other occupancy models, including the Royle/Nichols occupancy models, treat the “2” encounter code as a “1”, and do not distinguish between them as does this data type.