Unestimable month in Closed Heterogeneity Model

posts related to the RMark library, which may not be of general interest to users of 'classic' MARK

Unestimable month in Closed Heterogeneity Model

Postby catharina.k » Tue May 09, 2017 9:29 pm

Hello,
I am working with camera trap data of the rodent Cuniculus paca. I have five sampling months, each one divided in five secondary intervals. I am using the closed population model with heterogeneity (FullHet), through RMark. I am considering each sampling month a separate group.
When I run the simplest model (p and c are equal and constant, f0 varies between groups), I get bad estimates for the first group/sampling month. Which is strange, because it is the interval with most recaptures. If it was the other way around (the first month was the only one with an estimable N) I would understand, but this got me really confused. Does anyone know what could be happening?

Input data
Code: Select all
ch group freq
00010     1    1
01000     1    4
01001     1    2
11010     1    2
11111     1    2
00010     2    2
01000     2    2
10110     2    1
00010     3    2
00100     3    2
01000     3    3
10000     3    1
10111     3    1
11011     3    1
00111     4    1
01000     4    1
10000     4    4
11110     4    1
00001     5    2
00010     5    2
00100     5    1
01000     5    1
01101     5    1
10000     5    1


Code
Code: Select all
test<-RMark::mark (paca.data,paca.ddl,
                       model.parameters=list(
pi=list(formula=~1),
                                             p=list(formula=~1,share=T),
                                             f0=list(formula=~group)
                       ),
                       options="SIMANNEAL",
                       threads=-1)

Results:
Code: Select all
> test $results$derived
$`N Population Size`
   estimate        se       lcl      ucl
1 11.000003 0.0000000 11.000003 11.00000
2  5.091734 0.9248441  5.001351 11.22845
3 11.937425 1.9811715 10.369055 20.17090
4  9.478182 2.5067593  7.478165 19.84365
5 13.112574 4.5145513  9.154381 30.64280

> test$results$beta
                 estimate        se         lcl         ucl
pi:(Intercept) -16.543780 0.0000000 -16.5437800 -16.5437800
p:(Intercept)    0.067066 0.3567744  -0.6322119   0.7663440
p:group         -0.341798 0.1364858  -0.6093101  -0.0742859
f0:(Intercept) -12.603484 0.0000000 -12.6034840 -12.6034840
f0:group2       10.214624 0.0000000  10.2146240  10.2146240
f0:group3       13.264843 0.0000000  13.2648430  13.2648430
f0:group4       13.511009 0.0000000  13.5110090  13.5110090
f0:group5       14.235187 0.0000000  14.2351870  14.2351870


Thank you in advance!
catharina.k
 
Posts: 1
Joined: Tue May 09, 2017 9:14 pm

Re: Unestimable month in Closed Heterogeneity Model

Postby cooch » Tue May 09, 2017 10:10 pm

catharina.k wrote:Hello,
I am working with camera trap data of the rodent Cuniculus paca. I have five sampling months, each one divided in five secondary intervals. I am using the closed population model with heterogeneity (FullHet), ...


Fitting heterogeneity models (meaning, n-point mixture models) with <7 occasions of data will not work particularly well (verging on a near waste of time). Recommendation is not to.
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University


Return to RMark

Who is online

Users browsing this forum: No registered users and 10 guests