Table 1. Introduction success as predicted by the symbiosis trait.

Factor	All species			Non-native species Only
	Coefficient	Lower CI	Upper CI	Coefficient	Lower CI	Upper CI
Intercept	−8.055	—	—	−3.125	—	—
Symbiosis?	−0.523†	−0.960	−0.437	−0.587†	−0.832	−0.416
Latitude	0.142†	0.094	0.210	0.026	−0.013	0.063
Total native area	0.194†	0.119	0.211	−0.058†	−0.092	−0.038
Annual?	1.092†	0.915	1.267	0.198†	0.065	0.329
Woody?	0.353‡	0.084	0.404	−0.476†	−0.634	−0.408
Number of human uses	0.964†	0.865	0.981	0.323†	0.282	0.370
Area of introduced region	0.010	−0.049	0.046	0.019	−0.046	0.051
Symbiosis by human uses interaction	0.152†	0.104	0.223	0.080†	0.042	0.129

CI, 95% confidence interval.

Symbiosis is incorporated into the model as the presence or absence of the trait, with the inclusion of other factors found to predict introductions in our legume dataset. A negative coefficient indicates lower introduction success for symbiotic legumes compared with non-symbiotic legumes. We excluded non-significant interaction terms. 95% confidence intervals were obtained from parametric bootstrapping. Each response variable is modelled at the species by region level, and the model included a species and a region random effect to account for non-independence of observations within species or regions. The estimated variance for the species and region random effects respectively were 4.83 and 0.85 for the all species model, and 0.9 and 1.43 for the non-native species only model.

^†99.9% CI does not overlap zero.

^‡99% CI does not overlap zero.