Hi Thom,
You're right that in general option 1 should have female selectivity
peak at 1.0 and male selectivity be relative to female. However,
looking at the description in the User Manual under "Male
Selectivity", which starts on page 130 of the 3.30.14 manual), I
see it notes that "If selectivity for the dependent sex is
greater than the selectivity for the first sex (which always peaks at
1.0), then the male-female selectivity matrix is rescaled to have a
maximum of 1.0." So in this case, if your offset parameters are
such that ratio of male to female selectivity is greater than 1 at a
point where female selectivity is high enough to make the male
selectivity exceed 1, then rescaling would occur. I suppose the
"always peaks at 1.0" phrase should be revised.
Your question also helps me realize that it would be useful to add
some figure showing the estimated dogleg function used in the male
selectivity offset. You can create one yourself by looking at the
ratio of the male to female selectivity on a log scale. The code in
the attached file resulting example plot also attached is a quick
attempt to do so, where the linear function in the last panel is
consistent with the estimated parameters for the the model (pasted
below) except that it used male selectivity option 2 (females relative
to males), so the 1.04 and -1.21 are the negative of the values in the
figure at 0 and at the dogleg point of 111.15.
Value
SzSel_MaleDogleg_Fishery_current(1) 111.15800
SzSel_MaleatZero_Fishery_current(1) 1.04781
SzSel_MaleatDogleg_Fishery_current(1) -1.21371
SzSel_MaleatMaxage_Fishery_current(1) 0.00000