Geographic distance and latitude, but not environmental distance, affect germination rate:
There was a significant, nonlinear effect of geographic distance between source population and common garden at the HD garden, but this effect depended on seed type (distance2 x seed type: P = 0.001; Fig. 4; Appendix B). Chasmogamous seed germination decreased weakly as distance from garden increased, whereas cleistogamous seed germination was greatest when sourced somewhat close to (~125 km away) but not directly from the HD garden (Fig. 4). When modeled separately by seed type, however, the weak distance effect for chasmogamous seeds became nonsignificant (distance: P = 0.30; distance2: P = 0.80), whereas the effect remained significant for cleistogamous seeds (distance2: P = 0.002). The overall trend appears to be similar at the HR garden, although there were no significant effects of distance or distance2 at that garden (Fig. 4; Appendix H, Table 2).
The first two principal components explained 74.1% of the variation in the environmental data from source populations (Appendix F). However, there was no significant effect of environmental distance between source population and common garden on germination at either common garden (Fig. 5; Appendix H, Table 3). Furthermore, there was no significant correlation between geographic and environmental distances of source populations (Appendix G).
The effect of source population latitude on germination depended on seed type at HD (P = 0.023) but was not significant at HR (P = 0.267; Fig. 6; Appendix H, Table 4). Higher latitude exhibited a slight positive effect on germination for chasmogamous seeds but a negligible effect for cleistogamous seeds at HD (Fig. 6).