[R-sig-ME] Seeking Advice on Structural Equation Modeling with GLMMs

[杨李舒男]

Dear interest group,
As a newcomer to R and statistics, I now have an unsolvable problem.
I are investigating how neighborhood and topography affect the intensity of seed predation with two treatments. Using GLMMs for neighborhood and topography, I observed complete separation in the models, indicated by coefficients greater than 10. Following Ben Bolker's guidance, I employed the bglmer function from the blme package. These are the two models we constructed:
treatment * elev_lg_sd + slope_lg_sd + (1|obs1) + (1|plot)
shannon_lg_sd * treatment + A.het_lg_sd + (1|obs1) + (1|plot)
Assuming I used the bglmer function correctly, we then attempted to establish a cause-and-effect relationship using the psem function from the piecewiseSEM package. Unfortunately, it seems that psem function does not support models of class bglmerMod.
I now understand the interrelationships between topography, neighborhood, and seed predation, but I'm unable to use structural equation modeling to determine the cause-and-effect relationship between them. 
I have attached my code and partial data.
Could you please advise me on how to proceed? 
Thank you for your time and assistance. 
Best regards, 
Lishunan Yang