DISCUSSION
This study examined the effect of soil warming on a whole-of-life array of traits (vegetative growth, reproductive output and phenological response traits including senescence), as well as the germination traits of the next generation, for populations of O. eriopoda with four different germination strategies. Our findings suggested that responses to soil warming and germination strategy are complex, indeed not as straightforward as we hypothised. In many instances, the responses were trait-specific. Since we followed the whole life cycle, we were able to observe that the effect of soil warming on traits changed through ontogeny and was mostly apparent in the respective active growth stages of the plant life cycle, demonstrating that care must be taken in extrapolating from responses of a given trait at any point in time to a whole-of-life conclusion. We also confirmed that the pattern and direction of warming effects varied depending on the germination strategy of the populations of O. eriopoda in question. We consider the diversity of these responses to warming throughout the life cycle, and among germination strategies within a species, to be extraordinary as they highlight the complexity of linkages between the maternal and offspring environment and make evident that without a whole-of-life perspective we will struggle to predict impacts of global change on species. Here we interpret these elements in the context of impacts of warming on persistence of O. eriopoda and other alpine species in a novel future climate.
Under rapid climate change, plasticity in phenology and reproductive traits is likely to have strong fitness consequences (Kozłowski, 1992; Stinson, 2004), and it is possible that different germination strategies will have different selective advantages (Hoyle, Cordiner, Good, & Nicotra, 2014; Willis et al., 2014). We found that variation in reproductive phenology (Fig. 4) and seed mass (Fig. 3) were inherent to the germination strategy, whereas adult vegetative traits and reproductive outputs were more strongly affected by warming than germination strategy. The ability of this species to adjust reproductive phenology by shortening seed maturation while maintaining seed quality (seed mass) may be advantageous for persistence under in changing climate (Bonser, 2013; Visser & Both, 2005; Willis, Ruhfel, Primack, Miller-Rushing, & Davis, 2008); even if the total seed number is reduced the seed are spread through the season. Although warming substantially reduced lifespan, it was apparent that O. eriopodaindividuals could complete their life cycle and produce healthy, full-size seed with adequate reserves for early establishment success. Nearly all of the seeds produced were viable regardless of germination temperature, maternal conditions, or germination strategy indicating potential to maintain population regeneration.
The interactions of soil warming and germination strategy were mainly evident for seedlings or early vegetative traits and not vegetative traits during the transition to reproductive stage; this corroborates the finding of a previous study by Hoyle et al. (2015) that germination strategy of Australian alpine plants does not correlate with adult vegetative traits. In particular, populations of the immediate germination strategy (mainly occupy lower <1520 m elevation sites) exhibited greater plasticity in early leaf increment in response to warming compared to the staggered, postponed, and postponed deep strategies (elevations ranging from 1600 m to 2200 m, Supplement Table 1). Previous research on Wahlenbergia ceracea , an alpine herb that shares the same habitat with O. eriopoda , also found individuals from higher elevations were less plastic and less likely to express adaptive plasticity in growth-response to warming (Nicotra et al., 2015). The positive response of O. eriopoda vegetative traits to warming that was mainly pronounced during earlier ontogeny, however, did not lead to greater growth accumulation and earlier reproductive timing (flowering). The results indicate that there might be an internal constraint for vegetative growth and maintenance (Starr, Oberbauer, & Pop, 2000) and thus individuals with an immediate germination strategy grew and reached the reproductive stage more quickly but also died earlier than individuals displaying the other strategies.
By germinating the seeds produced in the warming experiment we confirmed that the source of variation in timing of germination in the staggered populations lies within individual plants, indicating a potential bet-hedging strategy (Starrfelt & Kokko, 2012), which has not been verified in this species before (Hoyle et al., 2015; Satyanti et al., 2019). Interestingly, individual plants produced both non-dormant and dormant seed of varying proportions (Figure S3). This leads to asynchronous germination within the population, and can reduce the risk against recruitment failure (Brown & Venable, 1986; Simons Andrew, 2009; Stevens, Seal, Archibald, & Bond, 2014; Venable & Lawlor, 1980). Should variation in snowmelt patterns compromise recruitment in either season, populations with a staggered strategy could be advantaged and this raises a question about how germination strategy is established and controlled in the species; its high variability indicates it is highly labile on some timeframe.
We examined the transgenerational effect of different seed development (maternal) conditions on germination traits since this has been proposed as a mechanism that may help species to tolerate future climates (Herman & Sultan, 2011). Given how variable germination strategy is we hypothesized that it may be highly plastic and reflect developmental conditions. But, contrary to expectation, we found no evidence of phenotypic plasticity in seed dormancy, i.e. germination strategy acrossO. eriopoda populations was constant regardless of seed development temperature. The warming impact was imposed when the plants were 18-20 weeks old, not the earliest seedling stage, but given that warming was imposed before the transition to reproductive meristems, this delay seems unlikely to have impeded the response. Our results suggest that seed dormancy variation is not dependent on seed maturation environment. Seed development temperatures have been shown to control seed dormancy induction and cycling in other species (Bernareggi et al., 2016; Donohue et al., 2005; Footitt & Finch-Savage, 2017; Steadman, Ellery, Chapman, Moore, & Turner, 2004) but we do not know the relative contributions of air or soil temperature. Further, the degree of seed dormancy, the important determinant of germination strategy, and dormancy cycling may indeed depend on the variation not just the mean of temperature (Satyanti et al., 2019; Topham et al., 2017).
Alternatively, it may be the case that germination strategy is plastic but has a ‘half-life’. It is possible that more than one cycle of warming is required to change the dormancy degree or fraction of non-dormant seed of the respective population. Previous studies suggest that the warming effects on seed dormancy and germination traits may be gradual rather than instant, just like phenological trait responses to warming (Franks, Sim, & Weis, 2007; Hoffmann et al., 2010). Such a delay effect may indicate that the mechanism underlying this shift is epigenetic. Dormancy patterns have been shown to be under epigenetic control in other systems (Nonogaki, 2017; Richards et al., 2017), but further research would be needed to demonstrate that here.
Diversification of germination strategies across population,s as exhibited by O. eriopoda , could still be an advantage that assists species’ persistence (Cochrane, Yates, Hoyle, & Nicotra, 2015). The potential for plants to respond to warming not only plastically but also as a function of genetic (or epigenetic) variation within the species demonstrates that species’ response to warming will often be manifested as a combination of rapid plastic responses and long-term evolutionary responses (Valdés et al., 2019). Considering the spatial and temporal heterogeneity of the alpine environment it is intuitive to expect variations in seed trait, particularly in dormancy, and hence we suggest that variation in germination phenology may be quite common within alpine species (Venable & Brown, 1988). However, very few studies have documented intraspecific variations in seed dormancy and germination strategy at the level exhibited by O. eriopoda (from populations with non-dormant, dormant, to deeply dormant seeds) and hence, we could not really conclude whether other alpine species would be as resilient under future warming.
An important implication of the results of our study is that predicting species’ responses and fate under global warming as either positive, negative or neutral could be a gross oversimplification when such assessments consider only one or few populations or are based only on a limited number of life-stages or traits (Saatkamp et al., 2018). The effects of warming vary not only among populations and individuals but as a function of ontogeny and hence, when assessing response to climate change at both species and community levels, within-species variations in germination strategy should be considered as important as between-species variation, and impacts must be assessed on a whole-of-life scale, not just at a single life stage. Our results imply that variability in seed dormancy and germination strategy could be key to assisting a species to persist under unpredictable conditions. Thus, further studies on how common these extraordinary intraspecific variations in seed and germination traits are across species, and the source of variation among individuals, are fundamental to predict the risk of species’ survival in a warmer world.