I am modeling a population of frogs using a Huggins Robust Design model in RMark. I have 15 years of capture data. I'm looking at the effects of different (annual) climate variables on annual survival of males and females separately, but I would eventually like to be able to predict survival using a new set of climate variables representing the 2080's.
Originally, I coded the climate variables as group covariates, but my understanding after sifting through the forum and revisiting the workshop notes, is that in order to use covariate predictions (which will allow me to predict using a new set of covariate values), I need to code the climate variables as individual covariates.
After doing so and looking at the AIC tables for models in which covariates were included at the group level vs. individual level, the models are ranked much differently. Should this be the case?
Thank you for any insight you might be able to provide. Apologies if I've missed this explanation in a previous post.
-Amanda
Here is my ddl in which I've coded the climate variables as group covariates (I've only included one here as an example, mswe=max snow water equivalent):
- Code: Select all
par.index model.index group cohort age time occ.cohort Cohort Age Time sex mswe
1 1 1 0 2002 0 2002 1 0 0 0 0 53.1
2 2 2 0 2002 1 2003 1 0 1 1 0 37.1
3 3 3 0 2002 2 2004 1 0 2 2 0 31.4
4 4 4 0 2002 3 2005 1 0 3 3 0 12.7
5 5 5 0 2002 4 2006 1 0 4 4 0 42.6
6 6 6 0 2002 5 2007 1 0 5 5 0 48.5
7 7 7 0 2002 6 2008 1 0 6 6 0 54.9
8 8 8 0 2002 7 2009 1 0 7 7 0 30.4
9 9 9 0 2002 8 2010 1 0 8 8 0 54.0
10 10 10 0 2002 9 2011 1 0 9 9 0 65.9
11 11 11 0 2002 10 2012 1 0 10 10 0 57.6
12 12 12 0 2002 11 2013 1 0 11 11 0 52.1
13 13 13 0 2002 12 2014 1 0 12 12 0 38.5
14 14 14 0 2002 13 2015 1 0 13 13 0 7.4
15 15 15 0 2003 0 2003 2 1 0 1 0 37.1
16 16 16 0 2003 1 2004 2 1 1 2 0 31.4
17 17 17 0 2003 2 2005 2 1 2 3 0 12.7
18 18 18 0 2003 3 2006 2 1 3 4 0 42.6
19 19 19 0 2003 4 2007 2 1 4 5 0 48.5
20 20 20 0 2003 5 2008 2 1 5 6 0 54.9
21 21 21 0 2003 6 2009 2 1 6 7 0 30.4
22 22 22 0 2003 7 2010 2 1 7 8 0 54.0
23 23 23 0 2003 8 2011 2 1 8 9 0 65.9
24 24 24 0 2003 9 2012 2 1 9 10 0 57.6
25 25 25 0 2003 10 2013 2 1 10 11 0 52.1
26 26 26 0 2003 11 2014 2 1 11 12 0 38.5
27 27 27 0 2003 12 2015 2 1 12 13 0 7.4
And here is (part of) my design matrix when I've included mswe as an individual covariate:
- Code: Select all
>results[[18]]$design.matrix[1:28, 1:3]
S:(Intercept) S:mswe S:sex1
S g0 c2002 a0 t2002 "1" "mswe2002" "0"
S g0 c2002 a1 t2003 "1" "mswe2003" "0"
S g0 c2002 a2 t2004 "1" "mswe2004" "0"
S g0 c2002 a3 t2005 "1" "mswe2005" "0"
S g0 c2002 a4 t2006 "1" "mswe2006" "0"
S g0 c2002 a5 t2007 "1" "mswe2007" "0"
S g0 c2002 a6 t2008 "1" "mswe2008" "0"
S g0 c2002 a7 t2009 "1" "mswe2009" "0"
S g0 c2002 a8 t2010 "1" "mswe2010" "0"
S g0 c2002 a9 t2011 "1" "mswe2011" "0"
S g0 c2002 a10 t2012 "1" "mswe2012" "0"
S g0 c2002 a11 t2013 "1" "mswe2013" "0"
S g0 c2002 a12 t2014 "1" "mswe2014" "0"
S g0 c2002 a13 t2015 "1" "mswe2015" "0"
S g1 c2002 a0 t2002 "1" "mswe2002" "1"
S g1 c2002 a1 t2003 "1" "mswe2003" "1"
S g1 c2002 a2 t2004 "1" "mswe2004" "1"
S g1 c2002 a3 t2005 "1" "mswe2005" "1"
S g1 c2002 a4 t2006 "1" "mswe2006" "1"
S g1 c2002 a5 t2007 "1" "mswe2007" "1"
S g1 c2002 a6 t2008 "1" "mswe2008" "1"
S g1 c2002 a7 t2009 "1" "mswe2009" "1"
S g1 c2002 a8 t2010 "1" "mswe2010" "1"
S g1 c2002 a9 t2011 "1" "mswe2011" "1"
S g1 c2002 a10 t2012 "1" "mswe2012" "1"
S g1 c2002 a11 t2013 "1" "mswe2013" "1"
S g1 c2002 a12 t2014 "1" "mswe2014" "1"
S g1 c2002 a13 t2015 "1" "mswe2015" "1"