Hurdle gamma gams summary interpretation

[Rubén García]

Hello.

I am attempting to evaluate the effect of some covariates in a zero- inflated continuous data.

I am also interested in the spatial prediction so that's why I used INLA.

I followed the examples from beginners guide to spatiotemporal books from Dr. A. Zuur, but I am a little too much lost. 

Instead of using the mgcv's smoother function, I simply used a basic cubic spline (with 3 knots) for all of my covariates and a SPDE for the spatial .

I tried a hurdle gam with both binomial and gamma families. I used the link=1 control predictor option to obtain the mu parameter for the Pearson residuals estimation. 

I also tried a joined hurdle with a stacked inla.stack (each stack for the zero and density parts) and everything , as far I checked, according to the examples.

Here are my questions:

How can I evaluate the variation/effect of my covariates in both the zero inflated and the density/continuous  models? 

Is there a game.check version that works with these INLA-gam models?

It is correct that I run a gam on the dispersion (density values - mu divided by square root (mu)) from each model or do I have to run it on the  Pearson residuals? (I've tried both but I'm not sure which one was correct)

I hope some of you can enlight me or guide me where I need to.

Thank your in advance.

Rubén.