discrepancies between CJS and JS models

[Gloria]

Hello to all: we are working with Cerambyx welensii (insect xilphagous of cork oak (Quercus suber) and oak (Quercus ilex)) from 2002. We have four campaigns of data and in each one of the campaigns we have sampled 39 occasions. We have selected the models that better adjust to the behavior of the species. In principle we have left from CJS model and from the best obtained model the rate of recapture and the rate of survival have been considered. In order to calculate the size of the population we have used JS model. In addition to this parameter the rate of recapture and the rate of survival have been considered also. Comparing the values obtained from CJS model and the obtained ones with model JS, we have seen that they are very different. It is normal that this happens? and if it is thus to what must?

Thanks for your attention

[cschwarz@stat.sfu.ca]

The capture and survival component of the JS likelihood is exactly the same as the CJS model. Consequently, you should get similar answers.

Depending on what model you fit for recruitment, there is some information about survival and catachability there, but the amount of information is generally neglible.

The problems you see may be an artefact of non convergence - have you started the JS model with the phi and p terms at the CJS values?

Carl Schwarz.

[Gloria]

I greatly appreciated your suggestion, I will consider it to estimate population size. However, it even worries to me that estimations of both models do not agree. Could it be due to existence of heterogeneity in captures?
thanks in advance for your attentions

[cschwarz@stat.sfu.ca]

It is true that heterogeneity has an impact on estimates, especially abundance, but it should affect estimates of survival and catchability equally in JS and CJS models.

Something else is going on. I suspect it is a simple convergce issue.