N estimation by POPAN with two groups

[Luisdom]

Hello all:

I would greatly appreciate if someone could help me with the following questions:

We have been working with insects borers for ten years and we are trying to estimate the population size of several plots by using JS models (POPAN protocol). Sampling has been developed in an extensive way in one plot (ten years) and punctually in three other plots (one year). We pretend, in one hand, not a very accurate figure of population size but get population ranges in order to categorize the different farms according their correspondence between damage and population. On the other hand, we want to know if the apparent population trends observed through indirect indicators in the same plot over the ten sampling years are reflected in the estimates obtained from the models.

In the estimation of population size for each year of sampling, the same problem (of which I present an example below) always arises:
Capture-Recapture data were obtained over 28 days of sampling. The best models are those with pent, time dependent. The best model of all is the phi(g * t)p(t)pent(t)N(.). This model is not significantly better than that in which N () is dependent on the group –male, female- (phi (g * t) p(t) pent(t)N(g)). The initial value of N for the best model is N(.)=2200 and for the equivalent model N(m)=1800 for males and N(f)=1500 for females (total N, 3300), ie there is a big difference compared to the first.
In the set of models with nonzero value in the AICc Weight there is both N (g) and N (.) models. When calculating the average value of the parameters for these models, the program uses the same value N (.) in calculation of average value both for males and for females. [N(f av)= (N(f)+N(.))/2; N(m av)=(N(m)+N(.))/2] (program actually doubled the value of N (.)).
How can we interpret this? What N value should we consider?

Thanks in advance