LCI negative for N?

[sea-shepherd]

I used zero-truncated Poisson-log normal to run {alpha(g)sigma(g)U(g)};
{alpha(.)sigma(g)U(g)}; {alpha(g)sigma(.)U(g)}; {alpha(g)sigma(0)U(g)};
{alpha(.)sigma(0)U(g)}; {alpha(.)sigma(.)U(g)}.

After model-averaging for N I received negative values for LCI. Since a population size cannot be negative, I was wondering where I went wrong?

Harry

[cooch]

I used zero-truncated Poisson-log normal to run {alpha(g)sigma(g)U(g)};
{alpha(.)sigma(g)U(g)}; {alpha(g)sigma(.)U(g)}; {alpha(g)sigma(0)U(g)};
{alpha(.)sigma(0)U(g)}; {alpha(.)sigma(.)U(g)}.

After model-averaging for N I received negative values for LCI. Since a population size cannot be negative, I was wondering where I went wrong?

Harry

Have a look at the CI section (14.9) in Chapter 14.

[sea-shepherd]

Okay, got you and I found another error in my data design which actually produced the negative values.

Another question: since I have only the option to select for alpha, sigma and/or U when I open the "Model Averaging Parameter Selection - Real" or select "N Population Size" when opening the "Select the Derived Parameter VC matrix"logically I do receive different results when running the calculation.

I do get estimates for N population size and estimates for U the unmarked population size.

Question: which number does resemble the estimate for the superpopulation or better, does it make sense to use both values to describe a population?

Harry

[bmcclintock]

Harry,

N=U+n, where n is the number of marked individuals. N is therefore the population size, but this would be interpreted as the "super population" size if there is a lack of geographic closure.

Cheers,
Brett