Table 2. AICM estimates for the fit of different discrete trait models.

Joint Discrete trait models	AICMb	S.E.c	Mod1d	Mod2	Mod3	Mod4	Mod5	Mod6	Mod7	Mod8
Mod1 Reduce_BSSVS_sym a	111031	+/−7.5	−	107	490	712	863	1102	4084	6582
Mod2 Reduce_nonBSSVS_sym	111137	+/−9.449	−107	−	384	605	756	995	3977	6475
Mod3 Reduce_BSSVS_asym	111521	+/−11.798	−490	−384	−	221	372	611	3593	6091
Mod4 Reduce_nonBSSVS_asym	111742	+/−7.822	−712	−605	−221	−	151	390	3372	5870
Mod5 Original_BSSVS_sym	111893	+/−6.668	−863	−756	−372	−151	−	239	3221	5719
Mod6 Original_nonBSSVS_sym	112132	+/−13.455	−1102	−995	−611	−390	−239	−	2982	5480
Mod7 Original_BSSVS_asym	115115	+/−14.662	−4084	−3977	−3593	−3372	−3221	−2982	−	2498
Mod8 Original_nonBSSVS_asym	117612	+/−26.584	−6582	−6475	−6091	−5870	−5719	−5480	−2498	−

^aThe names of the 8 models (Mod1-8) in the comparison test.

^bThe estimated AICM score of the posterior: lower values of marginal likelihood indicate a better fit to the data. The model with the best performance is indicated in bold.

^cThe standard error of the AICM estimated using 1000 bootstrap replicates.

^dThe AICM comparisons are shown in the matrix composed of columns 4 to 11. In each row of the matrix, the positive value in a cell represents the support for one model (in column 1) over the other (indicated in the column titles). A difference of AICM = 10 is considered to indicate a strong preference for one model over another.