Seed-type dependent geographic distance effect
Even after controlling for seed mass, we saw a significant geographic distance effect on cleistogamous germination at the HD common garden. We found that cleistogamous germination peaked at approximately 125km from the HD garden, while chasmogamous germination declined slightly with increasing distance from the source population, although this effect was not significant. This corroborates our prior finding that non-local cleistogamous seeds germinate at a higher rate than non-local chasmogamous seeds at the HD garden, suggesting this pattern may be biologically significant. These results are surprising for the genetic and seed weight differences discussed previously, in addition to differences in the dispersal ability of the two seeds. In nature, cleistogamous seeds germinate almost exclusively within the length of a single stem of the maternal plant (Holsinger 2000), making them extremely unlikely colonizing mechanisms for D. californica . By contrast, chasmogamous seeds can disperse farther (Schmitt et al. 1985), making them better candidates to colonize novel environments (Culley & Klooster 2007, but see Masuda & Yahara 1992).
Contrary to prediction, the observed seed type and geographic distance effect cannot be explained by genetic differences between cleistogamous and chasmogamous seeds. Indeed, our observation that cleistogamous germination success is improved, while chasmogamous germination success decreases away from the source population fundamentally dissatisfies the conditions for dimorphic cleistogamy maintenance at the population level as outlined by Culley & Kooster (2007). Mating system evolution theory predicts that cleistogamous and chasmogamous seeds with different dispersal abilities can be jointly maintained if their dispersal abilities differ, as they do for D. californica (Schoen & Loyd 1984). The near and far dispersal model for the selection of cleistogamy predicts that resource allocation will first be spent on the production of cleistogamous seeds - a so-called pessimistic reproductive strategy - followed by chasmogamous seed production - anoptimistic reproductive strategy (Zeide 1978, Schoen & Loyd). Our finding that cleistogamous seeds are generally more vigorous than their chasmogamous counterparts supports theory predicting that their investment should be prioritized and aligns with other studies that have found cleistogamous seeds to be generally superior (Dyksterhuis 1945, Cheplick & Quinn 1982, Schoen & Loyd 1984, Baskin & Baskin 2017). It is possible that our experimental design, which planted 25 seeds from the same maternal plant within close proximity, may have unintentionally favored cleistogamous germination as these seeds are more likely to be subjected to sibling competition than farther dispersing chasmogamous seeds (Schoen & Loyd 1984). This may partially explain our finding that cleistogamous seeds were more vigorous than chasmogamous seeds. Although density-dependent processes such as intraspecific competition could influence the expression of local adaptation, this is rarely addressed in local adaptation studies (Siepielski et al. 2016).