Table 3.

Results of the tree-based analyses performed on the full molecular backbone of Rabosky et al. (2018) and two sampled trees with inferred mitochondrial and nuclear substitutions

Tree	Slope	Likelihood ratio	P value	$\delta$
Backbone	$2.4\times {10}^{-4}$	27.05	$1.98\times {10}^{-7}$	1.03
Mitochondrial	$2.4\times {10}^{-3}$	25.80	$3.79\times {10}^{-5}$	1.98
Nuclear	$1.2\times {10}^{-5}$	0.54	0.46	1.00

The results are shown with root-to-tip path length as the response variable. A significant likelihood ratio test (bold type) indicates a likely correlation between substitutions and node numbers. The combination of a significant likelihood ratio test and a $\delta$ curvilinearity value greater than one indicates the likely presence of the node-density artefact in the Actinopterygian tree. Trends are plotted in Figs. S13, S14 available as Supplementary Information