Seagrass meadows are key ecosystems, and they are among the most threatened habitats on the planet. Seagrasses are the only submerged marine higher plants with a system of underground root and rhizome (Short et al., 2007). While colonizing the sedimentary shorelines of the world’s oceans, seagrasses have experienced genomic losses and gains to achieve the structural and physiological adaptations required for its marine lifestyle, arguably the most severe habitat shift ever accomplished by flowering plants (Olsen et al., 2016). As a functional group, they provide the basis for productive ecosystems along the coasts of all continents except Antarctica, and the ecosystem services they provide are on a par with tropical rainforests and coral reefs (Costanza et al., 1997; Fourqurean et al., 2012; Olsen et al., 2016). Seagrass species colonize the sea through both sexual (via seeds) and asexual (via clonal growth of rhizomes) reproduction (Phillips and Menez, 1988; Darnell et al., 2015; Zhang et al., 2020). Low pollination success, restricted dispersal of pollen and seeds, and low survival of seeds and seedlings, means that successful recruitment via sexual reproduction is restricted (Les, 1988; Laushman, 1993; Reusch, 2003); therefore, asexual reproduction is the main method of recruitment (Duarte and Sand-Jensen, 1990; Rasheed, 2004). Despite this, sexual reproduction is the only way to maintain the genetic diversity of the population, which improves resistance to adverse environments and resilience to disturbances (Ehlers et al., 2008; Cabaco and Santos, 2012; Xu et al., 2018). Furthermore, sexual reproduction also plays an important role in colonizing new habitats and recolonizing severely declined seagrass meadows by dispersing seeds from parental meadows (Harwell and Orth, 1999; Jarvis and Moore, 2010; Kendrick et al., 2012).

The seagrass Zostera marina L., or eelgrass, is the most widespread marine angiosperm species throughout the temperate northern hemisphere of the Pacific and Atlantic (Green et al., 2003; Olsen et al., 2016). Many studies have reported that sexual reproduction is crucial for the maintenance of within-species genetic diversity of eelgrass populations (Larkum et al., 2005; Kendrick et al., 2012; Ort et al., 2012). During sexual reproduction, eelgrass seeds are produced after flowering (Ort et al., 2012). No regional studies have directly documented long-distance dispersal, but genetic isolation by distance studies have suggested 150 km as an upper limit for the natural dispersal distance of Z. marina (Olsen et al., 2004; Muñiz-Salazar et al., 2005), which provides evidence for the colonization of new habitats with seeds. In most cases, once eelgrass seeds are released from the flowering shoots, they are deposited on or in the sediment, forming the seed bank (Fenner et al., 2005). A certain proportion of the seeds in the seed bank are maintained in a suitable physiological state conducive for germination to occur at an ecologically advantageous time. This reproductive strategy is vital for enhancing genetic diversity and avoiding the catastrophic loss of a plant population (Venable and Brown, 1988; Reynolds et al., 2013). Although much of the practical work on restoration of seagrass beds is carried out using adult seagrass plants, this approach has its limits. Apart from the huge labor and financial support required, the main problem is the potential loss of genetic diversity when adult plants are used to rebuild populations (Williams and Davis, 1996; Williams, 2001). In contrast, restoration of seagrass beds by seeding can avoid these problems, especially the loss of genetic diversity. Several restoration efforts have already shown that large areas of seagrass beds can be successfully restored by seeding (McGlathery et al., 2012; Orth et al., 2012, 2020; Orth and McGlathery, 2012), and the high genetic diversity in seagrasses restored using seeds rather than adult plants confers a greater level of ecosystem resilience to the restored seagrass meadows (Reynolds et al., 2012).

The transition between seed dormancy and germination is an important ecological trait and key stage in the life cycle of higher plants, which ensures species survival (Cadman et al., 2006; Holdsworth et al., 2008). Most seeds can survive long periods of unfavorable conditions and only germinate and grow into plants under the most favorable conditions. This ability stems, in part, from the existence of a dormancy mechanism (Cadman et al., 2006). Whether seeds are dormant or non-dormant and the type and depth of dormancy varies between species, or even between individuals of the same species (Baskin and Baskin, 2004). Seed dormancy and germination have been studied for decades in terrestrial higher plants (Arabidopsis thaliana, Oryza sativa, Triticum aestivum, Hordeum vulgare, etc.) at morphological, physiological, genetic, and molecular levels (Barrôco et al., 2005; Fait et al., 2006; Sreenivasulu et al., 2008; Howell et al., 2009; Galland et al., 2014; Han et al., 2017; He et al., 2020). As the only marine higher plant, the seed ecology of eelgrass has been studied to a certain extent (Churchill, 1992; Moore et al., 1993; Sugiura et al., 2009; Wang et al., 2016, 2017; Xu et al., 2016), but little is known about the molecular mechanisms of eelgrass seed germination.

We have observed that the seed germination time of Z. marina differed between Swan Lake (SL) and Qingdao Bay (QB) in northern China over a number of years. The time of seed maturation (summer time) is same in both places, but the time of germination is totally different. Seed germination in QB began in autumn of the same year as seed maturation, and most seeds germinated in October and November. In contrast, the seed germination period in SL is in spring of the year following seed maturation, from March to the end of May. Globally, a number of studies have found that the germination seasons of different geographical populations of eelgrass differ. For example, the seeds of eelgrass populations in the northeastern Pacific Ocean mainly germinate in spring and autumn (Kentula, 1982; Phillips and Backman, 1983), while populations in the northwestern Atlantic mainly germinate in autumn (Orth and Moore, 1986; Jarvis and Moore, 2010). Populations on European coasts mainly germinate in spring, but some can germinate in winter (Harrison, 1993; van Lent and Verschuure, 1994). Based on this phenomenon, we hypothesize that inherent genetic traits rather than environment determine the germination time of seeds, and the germination time does not change with the environment in which the seeds are planted. Therefore, we designed a field experiment in which seeds collected from QB and SL were reciprocally transplanted and their germination time was recorded. Comparative transcriptome analysis was then carried out to analyze the different events in eelgrass seed germination between SL and QB, which provides in-depth and comprehensive information for researchers relating to the molecular mechanism of seagrass seed germination.

According to the results of the seed germination time of the natural population of Z. marina observed for many years, the seeds from QB mainly germinate in autumn of the same year as seed maturation, and the seeds from SL mainly germinated in spring of the following year. In this study, we carried out a field experiment transplanting seeds from QB to SL and from SL to QB. The results showed that seeds from QB still germinated in autumn regardless of whether they were planted in QB or SL, and seeds from SL, whether planted in QB or SL, still germinated in the following spring. We also conducted comparative transcriptome analysis in autumn (germination stage of seeds in QB and dormant stage of seeds in SL). The results showed that genes involved in some metabolic pathways, such as energy metabolism, lipid metabolism, amino acid metabolism, and biosynthesis of secondary metabolites, were significantly upregulated in the seeds of QB compared with those of SL, which indicated that the seeds in QB were gradually emerging from dormancy and starting the germination process, while the seeds in SL were still dormant. Therefore, we believe that the factors that determine the seed germination time at the two sites are not environmental factors, but their different internal molecular mechanisms.

Comparative transcriptome analysis on seeds at the early germination stage in both places were conducted, mainly from four aspects: the overall comparison of DEGs in germination, the comparison of energy metabolism-related pathways, the comparison of major hormone-related genes controlling germination and the comparison of genes related to cell changes. Seed dormancy is a common phenomenon in the plant kingdom, which allows seeds to survive unfavorable conditions for a long time before germination and to establish plants under the most favorable conditions (Cadman et al., 2006). In many cases, seeds acquire primary dormancy during seed maturation and may enter a state of secondary dormancy if seeds are exposed to unsuitable temperature or lack adequate light (Finkelstein et al., 2008). Presumably, each seed is in a state somewhere along the continuum from deeply dormant to non-dormant, and the transition between different dormancy states is an active process involving changes in gene expression (Finkelstein et al., 2008). To investigate whether the dormancy states of the seeds in the two regions were the same, transcriptome analysis was performed to characterize gene expression in different physiological states. Principal component analysis was applied to the expression of all RNA-seq genes between two different physiological states at two sites. The analysis found that the first dimension grouped the mature seeds of the two sites together and separated them from the seeds of the germination stage of the two sites. The second dimension showed that the seeds in the same location had a certain similarity, and the seeds in different locations had certain differences, especially the seeds in the germination stage, SL and QB had great separation. Cluster analysis (heat map) also showed that the gene expression patterns of SL-G and QB-G were similar on the whole, but the number of upregulated genes of early germination seeds in SL was much higher than that in QB. In a review of molecular studies on seed dormancy, Finkelstein concluded that the expression level of dormancy related genes was to some extent correlated with dormancy depth (Finkelstein et al., 2008). Cadman, in his study of Arabidopsis seed germination, found that the number of genes involved in breaking dormancy or inducing germination was much higher at the post-ripening stage than at the dormant stage (Cadman et al., 2006). Based on this, we speculated that the dormancy depth of the seeds in SL was deeper than that in QB.

Embryos must mobilize carbon and energy sources to germinate and grow (Finkelstein et al., 2008). KEGG enrichment analysis of upregulated DEGs sets in early germination seeds from SL and QB revealed that pyruvate metabolism and glycolysis were both common and significantly enriched pathways, indicating that seeds in this period had started to mobilize reserves to prepare energy supply for the subsequent germination processes. Previous studies have shown that the photosynthesis and mineral uptake systems are not active during seed germination, and the energy for seed physiological activities is provided by reserve mobilization (Yu et al., 2014). In the early stages of germination, the energy generated by molecule degradation (e.g., glycolysis, glyoxylate cycle, and tricarboxylic acid cycle) is presumed to be a key determinant of germination vigor (Galland et al., 2014). Early activation of transcripts of starch and lipid reserve mobilization pathways can provide sucrose and hexose, which provide energy until the cotyledons become photoautotrophic plants (Sreenivasulu et al., 2008). Because of the limited penetration of oxygen into dense seed tissue, the large amount of energy required for seed development seems to depend primarily on glycolysis (Yang et al., 2007). Han et al. (2017) also hypothesized that glycolysis provides the embryo with a large amount of energy for the biosynthesis of new compounds and other metabolic requirements. In addition, our KEGG enrichment results also revealed that five other pathways related to energy metabolism were significantly enriched in seeds from SL, while only two pathways, pyruvate metabolism and glycolysis, were significantly enriched in seeds from QB. This demonstrated that the energy pathway mobilized in the early germination stage of Z. marina seeds in SL was more active than that in QB. We hypothesized that the amount of energy required to break dormancy to initiate germination was also different, because of the different depths of seed dormancy at different sites.

A large number of studies have shown that plant hormones are key regulators of seed dormancy and germination in terrestrial plants, including ABA, GA, ethylene, auxin, and brassinosteriods, with ABA and GA being the most important regulatory factors that play antagonistic roles in seed germination (Holdsworth et al., 2008; Nambara et al., 2010; Han and Yang, 2015). ABA is a positive regulator of induction and maintenance of dormancy, and a negative regulator of germination, while GA can counteract the effect of ABA, releasing dormancy and promoting seed germination (Kucera et al., 2005). ABA deficiency is associated with the loss of primary dormancy in mature seeds during seed development in many plants, and the overexpression of ABA biosynthetic genes can increase ABA content, enhance seed dormancy, or delay seed germination (Finkelstein et al., 2002; Nambara and Marion-Poll, 2003; Kushiro et al., 2004). Studies have shown that ABA restricts embryo growth potential by inhibiting water absorption (imbibition) and cell wall loosening, which are critical steps in the initiation of germination (Schopfer and Plachy, 1984; Gimeno-Gilles et al., 2009). GA, on the other hand, weakens the restriction of barrier tissues such as the endosperm or seed coat by inducing hydrolase, induces the mobilization of seed storage materials, and stimulates the expansion of the embryo to stimulate seed germination (Bewley, 1997). Studies have found that mutants with defects in the gene encoding GA biosynthetase in some species cannot germinate (Mitchum et al., 2006; Steber, 2008). In this study, it was found that the number of genes leading to the decrease of ABA content in SL during the early germination stage was higher than that in QB, with the number in SL being twice that in QB (SL:28 vs. QB:13). Among these DEGs, those related to ABA degradation, MYB transcription factors, and carbohydrate metabolism were significantly upregulated, while those related to ABA response or ABA signal transduction were significantly downregulated. Similarly, the number of genes responsible for the increase of GA content in SL was twice as high as that in QB (SL:12 vs. QB:6). In these DEGs, those related to the synthesis, perception, and regulation of GA were significantly upregulated, while those related to the decomposition of GA were significantly downregulated. As previously mentioned, ABA and GA are positive regulators of dormancy maintenance and dormancy breaking, respectively. Our results demonstrated that there were more genes related to emergence from dormancy when the seeds were about to germinate in SL seeds compared with those from QB. Therefore, it was speculated that the dormancy depth of the seeds of SL was deeper than QB.

Seed germination is a series of physiological and morphogenetic processes, where the plant embryo resumes growth after a period of quiescence (Barrôco et al., 2005). Considerable experimental evidence compels people to believe that the resumption of cell cycle activity is a specific feature of early germination (de Castro et al., 2000; Gallardo et al., 2002; Vázquez-Ramos and De la Paz Sánchez, 2003). In terrestrial dicotyledon seeds, the embryonic root first emerges from the seed, and then the embryonic shoot emerges from the seed after absorbing water. In terrestrial monocotyledon seeds, the cotyledon and radicle are covered by a coleoptile and coleorhiza, respectively, and the coleorhiza and radicle grow out of the seed in sequence. Following this, the coleoptile is pushed up until it reaches the surface (Nonogaki et al., 2010; Rajjou et al., 2012; Xue et al., 2021). However, eelgrass is a marine monocotyledon, whose seed germination process is slightly different to that of terrestrial seeds (Taylor, 1957). When the seeds germinate, the cotyledon sheath elongates first, then the first true leaf elongates from inside of the cotyledon sheath. Following this, the second true leaf and the adventitious roots emerge from inside of the cotyledon sheath (Sugiura et al., 2009). The radicle of eelgrass seeds never develops, so no primary root is formed and the adventitious root is the only true root of eelgrass seeds (Taylor, 1957). The rapid growth of the embryos, which eventually leads to the rupture of the covering layers and the emergence of the radicle, is considered to be the completion of seed germination in terrestrial plants (Barrôco et al., 2005). At this stage, the decision of individual embryo cells reenter the cell cycle or continue to be arrested is crucial for seedling formation. The establishment of plant shape and function depends on the ability of embryo cells to resume division and differentiation (Barrôco et al., 2005). In this study, we identified genes related to cell cycle and cell changes in SL and QB gene sets, which revealed that the number of upregulated DEGs in SL was 2.5 times that in QB at the early stage of germination. Therefore, we speculate that the dormancy depth of SL seeds is deeper, so more genes related to the cell cycle and cell changes need to be mobilized in the early stages of germination to initiate the subsequent germination process.

Germination is usually controlled by both internal factors, such as genetic control, and external factors, such as water temperature, which was very important in the field (Orth et al., 2000). According to Xu et al. (2018) and our observations over the years, although the latitude of SL is only 1°higher than that of QB, the temperature of SL is much lower than that of QB in winter. The intertidal zone of the QB have little chance to be covered by ice, while most of the intertidal zone in SL is covered by ice in winter, even the entire lagoon is covered by ice in some years. Phillips et al. (1983) reported that ice scouring in winter is an important source of eelgrass mortality at the northern margin of the range in the Bering Sea. Thus, eelgrass seeds from the SL population might be not acclimated to an autumn germination, because their seedlings cannot survive the severe winter. By contrast, seeds from the QB population can germinate in autumn and their seedlings can survive the relatively warm winter. Therefore, in autumn, seeds from SL would choose to go into a deeper level of dormancy to survive the cold winter, while seeds from QB would choose to germinate, which suggests that long-term differences in winter temperatures may be an important factor for the genetic differences.

The results of field observations, the reciprocal transplant experiment, and molecular analysis were consistent, which supports our hypothesis that differences do exist in the seed germination period at two sites, and that these differences are not controlled by environmental factors but are determined by internal molecular mechanisms. Therefore, when using seed for seagrass bed restoration, it is necessary to take into account the germination characteristics of the donor seeds and choose the correct time for sowing. In addition, the comparative transcriptome analysis of seeds at the early germination stage at both sites from the aspects of overall germination DEGs, energy metabolism related pathways, major hormone genes controlling germination and genes related to cell changes, revealed that the dormancy depth of the seeds in SL was deeper than that in QB, and the long-term differences in winter temperature between the two sites might be the important factor for the genetic differences in seed germination time or dormancy depth between the two populations from different geographic regions.

All sequencing data are available through the NCBI Sequence Read Archive under the accession number PRJNA770030, and other data that support the findings of this study are available from the corresponding author, without undue reservation.

YZa: writing – original draft, conceptualization, methodology, and investigation. SX: writing – review and editing, conceptualization, and investigation (especially reciprocal seed transplant experiment). SY, XZ, YQ, and ML: investigation. YZo: funding acquisition, supervision, methodology, investigation, and writing – review and editing. All authors contributed to the article and approved the submitted version.

This research was supported by the National Key R&D Program of China (2019YFD0901300), the National Science & Technology Basic Work Program (2015FY110600), the National Natural Science Foundation of China (No. 32000269), the Natural Science Foundation of Shandong Province (ZR2020QD106), the Key Research Project of Frontier Sciences of CAS (QYZDB-SSW-DQC041-1), and the Taishan Scholars Program (Distinguished Taishan Scholars).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.