Table 2.

Performance of alternative hypotheses for geographic range evolution. Model selection frequency for hypothesis j: Offshore w/ Nearest Mainland was 100% across 100 alternative phylogenies

Hypothesis	Subset		ΔAIC	D	E
j: Offshore w/ Nearest Mainland	b+island	996	0	0.0094	0.021
i: Offshore w/ EPCT	b+island	1002	6	0.0086	0.021
h: Historical Owen Stanleys	b+island	1006	10	0.0096	0.020
Unconstrained DEC		1019	23	0.0046	0.018
g: Owen Stanleys	b+island	1072	76	0.011	0.021
l: Island Distance	b+distance	1114	118	0.018	0.026
k: Offshore Islands w/ Accreted	b+island	1184	188	0.019	0.034
f: Woodlark w/ Louisiades	b+island	1285	289	0.027	0.025
e: Louisiade Archipelago	b+island	1334	338	0.028	0.023
b: Slow and Steady	mainland	1379	383	0.030	0.023
c: Mobile Belt	mainland	1398	402	0.011	0.026
a: Current Connectivity	mainland	1432	436	0.017	0.020
d: Recent Emergence	mainland	1686	691	0.025	0.058

We indicate whether hypotheses belong to the subset of mainland assembly hypotheses (mainland), or whether they build from the best-fit mainland model and explore island histories (b+island) or island distance (b+distance). For each hypothesis, the likelihood values (−2log) and difference in AIC from the best fitting model (ΔAIC) are given along with parameter estimates for the rate of dispersal (D) and extinction (E). Degrees of freedom for all models is 2.