Table 2.

Effect of host breadth and phylogenetic distance in the likelihood the two hosts share a pest

	Coefficients			95% confidence intervals
Pest group	β0	β1	β2	β0	β1	β2
Bacteria	3.7354	−3.5794	0.0664	1.7502–5.2889	−4.2790 to −2.7080	0.0634–0.0711
Fungi	6.0593	−4.5293	0.0466	5.0467–6.7900	−4.8649 to −4.0673	0.0455–0.0483
Oomycetes	3.2131	−3.5290	0.0818	0.3439–4.9863	−4.3281 to −2.2792	0.0788–0.0918
Insects	3.4837	−3.2188	0.0532	2.9285–3.9645	−3.4273 to −2.9736	0.0526–0.0539
Mites	2.6463	−2.8747	0.0447	1.1852–3.6436	−3.3190 to −2.2357	0.0438–0.0462
Mollusks	−1.2019	−1.1576	0.0474	−3.7591–0.8348	−2.0378 to −0.0713	0.0467–0.0495
Nematodes	2.2984	−2.9162	0.0884	1.3244–3.1807	−3.3146 to −2.4921	0.0857–0.0932
Viruses	8.3535	−5.6544	0.0967	7.5721–9.0395	−5.9872 to −5.2985	0.0904–0.1029
Plants	2.2164	−2.7873	0.0727	1.2404–2.8660	−3.0934 to −2.3506	0.0704–0.0782

Given are the coefficients of logistic regressions (median and 95% confidence intervals) of phylogenetic signal in host sharing with two independent variables: phylogenetic distance between source host genus and target host genus (coefficient β₁) and the number of known hosts for the pest (coefficient β₂). The dependent variable was whether the target host genus was known to be susceptible (S) to a pest from the source host genus. The regression takes the form of logit(S) = β₀ + β₁^*log₁₀(PD + 1) + β₂^*(Number of known hosts), where PD is the phylogenetic distance (time of independent evolution in My) between the source and target host genera. Note that interaction term was not significant (S6), so the models presented here were run with main effects only. The number of pests in each group is the same as in Table 1. Coefficient β₁ was significantly negative and β₂ was significantly positive for all nine pest and pathogen groups (95% CI did not overlap zero).