R help - Oct 2010 - glmer or not - glmer model specification

Hello,

I'm having some trouble figuring out the correct model specification for
my data. The system consists of multiple populations of an organism,
which have been genetically sampled for several years. The problem is
this: A minority of individuals are found in more than one sample,
either they have survived into the next sampling at the same location,
or have migrated to another another location and survived into the next
sampling there. This pseudoreplication constitutes about 12% of all
observations.

I have information on the number of heterozygous locus per individual,
its location, year, population size  and number of immigrants at that
location, and want to investigate the effect of population size and
migration on heterozygosity through time. Heterozygosity, the response
variable is a proportion, and I would like to account for the
pseudoreplication. Their heterozygosity score will be the same for each
sampling, and is constant. I have been trying with a mixed models
approach, more specifically glmer (lme4 package) with a binomial
distribution family and the response variable as an odds, eg.

glmer(heterozygosity~1+population size+number of immigrants+year+(1|
individual id)+(0+year|individual id)+(1+year|location),family=binomial)

To my very basic understanding, this poses the following question: Is
there a an effect of population size, number of immigrants and year,
while taking into account a non-correlated random effect of time and
individual, and a correlated random effect of year and location?
   
Does anyone have any advice as to wether a mixed model is the best
approach, and if so, what model specification to use?

-- 
-----------------------------------------------------

Vennlig hilsen/Kind regards,

Atle Torvik Kristiansen

M.Sc. student

Institutt for biologi (DU2-100) 
Realfagbygget
NTNU
7491 Trondheim
Norway

Email: atletorvik at gmail.com
       atletorv at stud.ntnu.no