[Bug]: rho parameters not impacting semi-parametric (2D-AR) selectivity when multiple sigma values are used

[iantaylor-NOAA]

Describe the bug

This issue was found in discussion with @rosana-ourens who successfully set up a model with semi-parametric selectivity where there were multiple sigma values to allow more flexible deviations for the youngest ages. However, the rho parameters did not seem to have any impact on the results.

I replicated the issue in the attached models, modified from https://github.com/nmfs-ost/ss3-test-models/tree/main/models/Hake_2019_semi-parametric_selex, which have a similar set up where sigma_amax = 3 and rho_year and rho_AGE are both fixed at either 0 or 0.5: sigma_3_rho_0.zip sigma_3_rho_0.5.zip

The likelihoods differ between models but the difference is all due to the Parm_devs component as shown in the comparison below.

Changing sigma_amax from 3 to 0 and reducing from 3 to 1 sigma_sel parameters, allows the rho parameter to have a modest impact on the selectivity parameters and all other aspects of the dynamics.

This may be due to my misunderstanding of how the rho parameters work, but I was assuming that they apply across all ages and years of deviations regardless of the how the sigmas are configured.

To Reproduce

Run the models attached above.

Expected behavior

Different rho values impact the model results beyond just the parameter deviations.

Screenshots

No response

Which OS are you seeing the problem on?

No response

Which version of SS3 are you seeing the problem on?

3.30.22.1

Additional Context

No response