Widespread and persistent invasions of terrestrial habitats coincident with larval feeding behavior transitions during snail-killing fly evolution (Diptera: Sciomyzidae)

Table 2 Results of the likelihood-based approximately unbiased (AU), Shimodiara-Hasegawa (SH), weighted Kishino-Hasegawa (WKH), and weighted Shimodiara-Hasegawa (WSH) tests calculated using CONSEL

			Test
Constraint	-ln L	Difference	AU	SH	WKH	WSH
Tetanocera feeding group analysis (59-taxon data set)
Unconstrained	−38932.126	(Best)
Aquatic snail predators1*	−39212.682	280.556	p = 4e-06	p < 1e-100	p < 1e-100	p < 1e-100
Aquatic snail predators2*	−39252.447	320.321	p = 2e-41	p < 1e-100	p < 1e-100	p < 1e-100
Shoreline snail predators1	−38971.538	39.412	p = 0.001	p = 0.002	p = 0.002	p = 0.002
Shoreline snail predators2*	−39004.527	72.401	p = 1e-07	p = 0.005	p < 1e-100	p = 4e-05
Slug parasitoids	−38941.015	8.889	p = 0.038	p = 0.631	p = 0.062	p = 0.193
Terrestrial snail predators*	−39051.229	119.103	p = 1e-08	p < 1e-100	p < 1e-100	p < 1e-100
Renocerinae monophyly analysis (entire 115-taxon data set)
Unconstrained	−73022.049	(Best)
Renocerinae	−73056.937	34.89	p = 0.055	p = 0.059	p = 0.059	p = 0.059
Comparison of Bayes MAP (Figure 1) and best ML (Additional file 1: Figure S1) trees (entire data set)
ML tree	−72999.441	(Best)
Bayes MAP tree	−73005.315	5.875	p = 0.377	p = 0.388	p = 0.388	p = 0.388

Knutson and Vala [17] feeding group constraints were done with an abbreviated data set containing 59 terminal taxa (all Tetanocera plus 4 outgroups). Trees compared were the best topology from unconstrained analysis versus an analysis where the feeding groups (see Table 1) were constrained to be monophyletic. Tetanocera plumosa, which can either live in the water or on the shoreline was coded both ways (Aquatic2 & Shoreline2 = T. plumosa considered a shoreline snail predator). The monophyly of Anticheta + Renocera, proposed as subfamily Renocerinae by Verbeke [74], was tested by constraining them to be outside of the Tetanocerini and Sciomyzini. The Bayesian MAP tree and ML tree were tested to see if they were significantly different from one another. P-values in bold are significant. Constraints with an asterisk (*) were constrained trees that were significantly worse than the unconstrained tree in all statistical tests.

ISSN: 2730-7182