Introduction
Seed provenance is an important consideration for restoration practitioners seeking to re-seed grassland ecosystems (Bischoff et al. 2006, Vander Mijnsbrugge et al. 2010, Bischoff et al. 2010, Breed et al. 2018). Seemingly minor differences in the fitness of seeds sourced from different populations can have profound effects on the establishment of focal plant populations at an ecosystem scale (Middleton et al. 2010, Seifert & Fischer 2010). In 2020, the United Nations declared 2021-2030 the “Decade on Ecosystem Restoration” (UNEP and FAO, 2020). Grasslands have high potential for restoration under this declaration, but careful planning is needed to ensure long term success (Dudley et al. 2020). To achieve ambitious global restoration targets for grassland ecosystems, research on the relationship between seed provenance and plant fitness is urgently needed (Breed et al. 2018).
Restoration practitioners must consider the degree of local adaptation - the superior fitness of local genotypes - for plant species used in their projects. Populations under intense selective pressure are more likely to show local adaptation, providing them with a distinct “home” advantage over nonlocal populations at a given location (Joshi et al. 2001, Breed et al. 2018). Although intuitive from an evolutionary perspective, local adaptation is certainly not universal (Bischoff et al. 2006, Leimu & Fischer 2008, DeMarche et al. 2019). Ecologists frequently use reciprocal transplant and common garden experiments to measure the degree to which local adaptation exists in plant populations (Hereford 2010). Approximately 70% of reciprocal transplant studies show local adaptation (Leimu & Fischer 2008, Hereford 2009), which likely depends upon three variables (Hereford 2009): the difference in selection pressure between local and nonlocal genotypes (Schluter & Grant 1984), the amount of gene flow between populations (Garcia-Ramos & Kirkpatrick 1997, Lenormand 2002, Kawecki & Ebert 2004), and the genetic structure of each population (Linhart & Grant 1996).
Differences in selection pressure between populations are likely linked to differences in site environmental characteristics, which is often closely related to geographic distance (Leimu & Fischer 2008, Hereford 2009). Theory predicts that as environmental and geographic distance increase between populations, so too should the magnitude of local adaptation (Garcia-Ramos & Kirkpatrick 1997, Joshi et al. 2001). The underlying logic is simple; genotypes proven to perform well in a site should continue to do so in the future, while genotypes sourced from elsewhere may not, especially as the differentiation between sites increases. The degree of local adaptation thus varies among populations and can be difficult to predict (Leimu & Fischer 2008, Gaillart et al. 2019). With germination being the critical first step in plant establishment, understanding how seeds germinate near and far from their maternal plants can help elucidate the degree of local adaptation in plant populations.
Another consideration for choosing seed sources for restoration is how to best maintain genetic diversity (McKay et al. 2005). Although most plant species produce a single type of seed, many exhibit seed heteromorphism – the production of multiple seed types. Nearly 700 angiosperm species exhibit cleistogamy, a breeding system that includes permanently closed, obligately self-pollinated flowers (Culley & Klooster 2007). The majority of these species are classified as dimorphically cleistogamous, producing seeds from both cleistogamous and chasmogamous (more typical, externally pollinated) flowers (Culley & Klooster 2007, Baskin & Baskin 2017). As such, cleistogamous seeds are likely to have less genetic diversity than their potentially outcrossed chasmogamous counterparts and could be more prone to inbreeding depression (Culley & Klooster 2007). Cleistogamous seeds also typically disperse much shorter distances than chasmogamous seeds (Schoen & Loyd 1984, Culley & Klooster 2007, Auld & de Casas 2013, Baskin & Baskin 2017), and average seed weight can differ substantially between the two types (Waller 1982). There are, however, several evolutionary advantages to cleistogamy, including insurance in the absence of external pollination, the reduced energy cost of production, and the retention of locally adapted gene complexes (Schoen & Loyd 1984, Culley & Klooster 2007, Baskin & Baskin 2017). Indeed, a review of field and lab studies comparing the germination of cleistogamous and chasmogamous seeds found that a higher proportion of cleistogamous seeds germinated in two-thirds of cases (Baskin & Baskin 2017). The inherent differences in genetic diversity and dispersal between these two seed types suggest that chasmogamous seeds might be better suited for success in novel environments, while cleistogamous seeds may perform better in the immediate vicinity of their maternal plant (Schoen & Lloyd 1984, Culley & Klooster 2007).
To date, researchers have mostly recommended the use of locally-sourced seeds for restoration (Bischoff et al. 2010, Vander Mijnsbrugge et al. 2010, Bucharova et al. 2017), despite approximately 30% of plant populations surveyed not showing local adaptation (Hereford 2009). In these cases, stringent seed sourcing restrictions likely inhibit the genetic diversity of the restored population, which may have negative effects on the population’s ability to respond to changing environmental conditions (Broadhurst et al. 2008, Miller et al. 2011). There is growing support for the use of nonlocal seeds sourced from populations that may be better adapted to future climatic conditions, as the use of climate-adapted genotypes could facilitate the maintenance of ecosystem services and critical habitat structure (Broadhurst et al. 2008, Bischoff et al. 2010, Kreyling et al. 2011, Ramalho et al. 2017). Climate-motivated translocation of seeds is controversial, however, as it relies on a series of assumptions that are difficult to test. These assumptions include that the seeds are sufficiently adapted to their local climate, that this climate adequately matches the future climate of the restoration site, that the nonlocal seeds will germinate and establish in a restored site under current conditions (Kreyling et al. 2011), and that climate is the most important driver of performance (DeMarche et al. 2019). While considering the long-term effects of introducing novel genotypes, the germination of nonlocal seeds in a novel environment needs further study to ensure such an approach is feasible in the first place (Bucharova et al. 2017, Breed et al. 2018).
Danthonia californica , a perennial bunchgrass native to western North America, is commonly used in the restoration of prairie ecosystems in the Pacific Northwest, USA (Buisson et al. 2006, Hayes & Holl 2011, Stanley et al. 2011, Pfeifer-Meister et al. 2012). Individuals produce both chasmogamous and cleistogamous seeds (Appendix A; A,B), the latter of which are enclosed within the stem. Cleistogamous seeds are difficult to remove manually, potentially contributing to their infrequent use inD. californica restoration (Hayes & Holl 2011). Although mating system generally does not influence the degree of local adaptation across species (Hereford 2010), many studies have shown different fitness and local adaptation patterns for conspecific seeds produced via different mating systems (Schmitt & Gamble 1990, Lovell et al. 2014, Rushworth et al. 2020). However, the applied aspects of mating-system dependent seed selection for restoration are relatively rare in the literature (Coulter 1914, Charlesworth 2007, Rushworth et al. 2020).Danthonia californica thus provides an excellent opportunity to study the impacts of sourcing distance and mating system on local adaptation in an ecosystem restoration context.
Here, we devised a common garden experiment using both chasmogamous and cleistogamous seeds collected from eight natural populations of D. californica across a latitudinal gradient in western Oregon and Washington, USA (Figure 1A). Our design allowed us to ask whether the effects of seed source origin (local vs. nonlocal) on germination are dependent on seed type, and whether there are other factors about seed source origin, such as the distance or direction (north or south) from the common garden, latitude, or average seed weight, that help explain germination patterns across source populations. Additionally, we experimented with seed processing techniques to facilitate cleistogamous seed preparation and decrease processing time for heteromorphic seed planting. This is necessary because cleistogamous seeds remain in the stem and are difficult to remove and separate.
We hypothesized the following: H1a: At each common garden, we predicted that both cleistogamous and chasmogamous seeds originating from that site (local seeds) would outperform seeds originating from other source populations (nonlocal seeds), regardless of whether the nonlocal seeds originated to the south or north of the common garden.H1b: However, we expected that the degree of local adaptation would depend on seed type. If inbreeding depression compromises local adaptation, then we would expect local chasmogamous seeds to outperform local cleistogamous seeds. Alternatively, if gene flow limits local adaptation, then we would expect local cleistogamous seeds to outperform local chasmogamous seeds. H2: Furthermore, we expected germination to decrease with increasing distance between source population and common garden (both geographic and environmental distance - the similarity in environmental conditions such as temperature, precipitation, etc.), considering that the magnitude of local adaptation between common garden and source sites should increase as distance does.H3: Finally, we predicted that nonlocal seed germination would decrease with increasing latitude, as seeds sourced from southern populations would outperform seeds sourced from northern populations due to recent climate warming. Demographic studies of natural D. californica populations, including most of the populations studied here, revealed that population growth rate decreases with increasing latitude and that locally, the population growth rate decreases under warmer and drier conditions (DeMarche et al. 2021). Thus, it follows that the higher-performing nonlocal seeds at the two common gardens should be those adapted to warmer and drier conditions (i.e., more southern populations).