Appendix A. Full statistical model results for aphid and lab experiments (I–IV).

The following tables present the full statistical models for experiments I–IV. For model details, see the *Methods* section. For most models, we performed orthogonal contrasts to examine the independent effects of life history (annual vs. perennial) and host provenance (native perennial vs. exotic perennial). (Significance codes for all tables: *P* < 0.001 '***'; 0.001 < *P* < 0.01 '**'; 0.01 < *P* < 0.05 '*'; 0.05 < *P* < 0.1 '.')

TABLE A1.

R. padifield fecundity estimates (experiment I). The statistical model shown here examines fecundity (# babies produced per experimental unit) as a function of host life history, provenance, and phylogenetic group.

TABLE A2.

R. padigreenhouse fecundity estimates (experiment II). The statistical model shown here examines fecundity (# babies produced per experimental unit) as a function of nitrogen and phosphorus fertilization, host life history, provenance, and phylogenetic group.

TABLE A3. Model of host mass from the greenhouse fecundity experiment (II). The statistical model shown here examines plant aboveground biomass as a function of nitrogen and phosphorus fertilization, host life history, provenance, and phylogenetic group.

TABLE A4. Host tissue chemistry models from the greenhouse fecundity experiment (II). The statistical model shown here examines tissue chemistry in aboveground biomass as a function of nitrogen and phosphorus fertilization, host life history, provenance, and phylogenetic group.

TABLE A5.

R. padipreference estimates (experiment III). This statistical model examines preference ofR. padiaphids for hosts of differing life history, provenance, and phylogeny (estimated using aphid density on each plant after 24 hours, or # aphids/ aboveground host mass)

TABLE A6.

R. maidis, R. padi, and S. avenaegreenhouse fecundity estimates (experiment IV). This model examines fecundity (# babies produced per experimental unit) as a function of aphid species, nitrogen fertilization, host life history, provenance, and phylogenetic group. Because of overdispersion in the data, we fit a quasi-Poisson model.

[Back to A019-045]