In arid and semi-arid regions of the world, soil salinity is an environmental stress factor to plants, partly because soil conditions are strongly heterogeneous in space and time (Bernstein, 1975). Much work on halophytic species has focused on their morphology, photosynthesis, physiology and molecular biology in an attempt to understand their distinctive adaptations to saline environments (Bernstein, 1975; Zhu, 2001; Redondo-Gómez et al., 2006; Pandolfi et al., 2012). Fluctuations in soil salinity potentially can result in high mortality in the germinating seed cohorts, and therefore high salt tolerance of seeds of halophytes during the germination stage has been studied (Ungar, 1978; Khan and Gul, 2006; Ameixa et al., 2016). As a link between two generations, seeds traits have a direct impact on population regeneration and dynamics (Harper, 1977; Donohue, 2009). Seed dormancy and germination responses enable seedling populations to emerge at the right time under relatively favorable environmental conditions that ensure establishment and completion of the life cycle (Hoyle et al., 2015; Huang et al., 2016). Therefore, seed traits and subsequent growth strategy are important for seedling survival and reproductive success in arid and saline habitats (Ungar, 1987; Gutterman, 2002; Yang et al., 2012).

Seed heteromorphism, i.e., the production by a single plant of two or more seed morphs that differ in mass, morphology, and ecology, is prevalent in arid habitats, and it is particularly common in halophytes (Khan and Ungar, 2001; Imbert, 2002). Harper (1977) hypothesized that seed heteromorphism confers a selective advantage to plants that grow in extreme, fluctuating environments, for example because different seed morphs may germinate at different times due to differences in dormancy and seed coat properties (Mohamed-Yasseen et al., 1994; Baskin and Baskin, 2014). A prolonged overall germination period enabled by the combination of different seed morphs may provide more opportunities for establishment of seedling cohorts than monomorphic morphs by spreading the risks of survival in variable environments (Venable, 1985a; Venable et al., 1987). Heteromorphic seeds not only have distinct dormancy/germination behaviors (Mandák and Pyšek, 1999; Wang et al., 2008; Yao et al., 2010; Cao et al., 2012; Baskin and Baskin, 2014), but plants growing from them can differ in various life history traits such as survival and biomass allocation (Sorensen, 1978; Mandák and Pyšek, 2005; Ruiz de Clavijo, 2005; Wang et al., 2012; Baskin and Baskin, 2014; Yang et al., 2015a). Thus, the relative production of different seed morphs can be used to predict germination behavior and adaptive strategy of a heteromorphic population (Venable et al., 1995).

Variation in seed (and seedling) traits of seed-heteromorphic halophytes occurs among populations (Ladiges et al., 1981; Song et al., 2008). However, evidence for geographic variation (reflecting local adaptation) is almost nonexistent probably because the carryover effects of field environments have not been ruled out in most studies. By investigating natural populations, any phenotypic differences between them can be distinguished. Further, by rearing plants from different habitats in a common environment under different levels of salinity, we can remove the effects of the natural (maternal) environment and test the phenotypic plasticity response of heteromorphic seeds to soil salinity. We previously reported that maternal soil salinity can affect heteromorphic seed germination and seedling growth (Yang et al., 2015b), but our knowledge of seed morph production is much lower than that of other life history traits influenced by soil salinity. Here, we hypothesized that (1) phenotypic differences occur between populations in the field; (2) these differences in plants and seed traits between populations are genetically based; and (3) seed morph production and germination behavior of heteromorphic seeds are influenced by the magnitude of soil salinity. We tested our hypothesis using two distant populations (about 1700 km away from each other in northern China) of the halophyte Suaeda corniculata (C. A. Meyer) Bunge subsp. mongolica Lomon. & Freitag (Amaranthaceae) (hereafter Suaeda corniculata), which differ strongly in habitat soil salinity regime.

Suaeda corniculata is a succulent summer annual herb that grows in saline-alkaline soils (Zhu et al., 2003) and is widely distributed in steppes and semiarid and arid zones of northern China, Mongolia and south-central Siberia (Lomonosova et al., 2008). It produces two diaspore morphs: brown and black seeds. The brown seeds are non-dormant, whereas the black seeds are dormant and can exhibit an annual dormancy/non-dormancy cycling in the soil seed bank (Cao et al., 2012). Dimorphic seeds differ in germination timing (seedlings derived from brown seeds emerged in spring and those from black seeds in summer), but seedling recruitment and survival percentages were almost the same (Cao et al., 2012; Yang et al., 2015a).

The present study asked if there are phenotypic differences in life history traits between two populations of Suaeda corniculata. To minimize carryover effects from field environments, plants from the two seed morphs derived from field populations were grown in a common environment. Distinct phenotypic differences between populations in the field were retained in the common environment. Indeed, seed morph ratio and germination percentage of dimorphic seeds differed greatly between magnitudes of soil salinity. Although many studies have tested the effects of soil salinity on a limited number of plant life history traits (Yao et al., 2010; Wang et al., 2012, 2015), our study appears to be the first to clearly detail the phenotypically plastic response of heteromorphic seeds to different levels of salinity by cultivating plants from different geographic populations in a common environment.

Soil salinity in arid and semi-arid zones has been considered both an extreme environmental stress and a source of disturbance to plants (Ungar, 1978; Khan and Ungar, 2001; Imbert, 2002), and halophytes have morphological and physiological adaptations that facilitate their survival and growth under such harsh conditions (Debez et al., 2004; Yang et al., 2010). Plants of the seed-dimorphic species Suaeda splendens grew better at 0.2 mol⋅L^-1 NaCl than in pure water (Redondo-Gómez et al., 2008), and the optimum growth of Salicornia dolichostachya was at 0.3 mol⋅L^-1 NaCl (Katschnig et al., 2013). Likewise, in the present study F₁ plants of S. corniculata derived from both seed morphs performed better at low salinity than in salt-free tap water. And although a high level of soil salinity had adverse effects on plant growth and seed production, the plants allocated a higher proportion of biomass to seeds under salinity stress. Stress can cause a shift in the proportion of biomass allocated to vegetative and reproductive components (Lu et al., 2012). These plastic responses in life history traits to stress can be interpreted as to be adaptive because they increase the seed production component of fitness (Primack and Kang, 2005; Lu et al., 2016).

Studies have shown that growth conditions of the mother plant can alter the proportion of different seed morphs produced by an individual plant (reviewed in Baskin and Baskin, 2014), but the results are inconsistent. For example, seed morph ratio was not changed by growth of maternal plants of Suaeda aralocaspica (Amaranthaceae) in different salinities (Wang et al., 2012), whereas high salt stress increased the proportion of (non-dormant) brown seeds of Chenopodium album (Amaranthaceae) (Yao et al., 2010) and of Suaeda salsa (Amaranthaceae) (Wang et al., 2015). In our study, the proportion of brown seeds increased with increased salinity levels in the two populations of S. corniculata, thus agreeing with the results of Yao et al. (2010) and Wang et al. (2015). Because they germinate to high percentages in a wide range of salinities, brown seeds of Amaranthaceae species have been considered to be more fit than black seeds in harsh saline habitats (Khan et al., 2001; Song et al., 2008; Yao et al., 2010; Wang et al., 2015). However, Cao et al. (2012) showed that brown seeds of S. corniculata lost viability easily after exposure to salt stress, whereas black seeds did not. Thus, the ecological significance of brown versus black seeds is highly complex and should not be determined only by germination percentage. Since brown seeds are larger and growth of seedlings derived from them is faster than for black seeds, seedlings may benefit from more resources stored in large brown seeds (Grieve and Francois, 1992; Jakobsson and Eriksson, 2000; Easton and Kleindorfer, 2009), which would then lead to faster establishment. Thus, brown seed cohorts may benefit from exploiting temporarily favorable conditions, for example high soil moisture (and thus lower salinity) after rainfall. In which case, they may establish populations rapidly in saline habitats. On the other hand, black seeds of S. corniculata cannot germinate at high salinity, but they can form a persistent seed bank, which enables them to persist in the habitat should the entire population die before reproducing in a particular year. Black seed cohorts mainly germinate in the summer rainy season (Cao et al., 2012), and consequently they may be exposed to less harsh conditions for establishment than brown seed cohorts.

Variation in morph ratio and seed size has been found among populations of diaspore-heteromorphic species. For the annual halophyte, Suaeda salsa (Amaranthaceae), brown/black seed ratio and 100-seed mass were significantly higher in the intertidal zone (4.5 and 3.3 g⋅kg^-1, Na⁺ and Cl^-, respectively) than in an inland saline habitat (2.4 and 2.0 g⋅kg^-1, Na⁺ and Cl^-, respectively) (Song et al., 2008). The proportion of brown seeds and reserve mass (embryo plus perisperm) of black seeds of Chenopodium album were higher in populations on flat ground than that of a population growing on a slope (Yao et al., 2010). However, whether these differences between populations were due to genetics (G), environment (E) or G x E interactions was not determined. We conducted a common environment experiment in which F₁ plants were grown from F₁ seeds under different salinity levels to distinguish possible genetic differentiation between populations and to demonstrate phenotypic plastic response to different salt concentrations. For S. corniculata, the field investigation and common environment experiments revealed that the differences between populations were significant in both morph ratio and seed size, as well as high salinity increased brown seed proportion but had limited impact on seed size. Quinn and Colosi (1977) suggested that one generation of seed production under the same conditions is required to demonstrate that differences in germination are genetically based. However, we used F₁ plants for testing post-germination life history traits, and thus the possibility exists that differences among the traits could have been due to some carryover effects on plants grown from F₁ seeds (Whittle et al., 2009; Herman et al., 2012; Yang et al., 2015b).

Seed dimorphism in S. corniculata may be a bet-hedging strategy since the offspring have two distinct life histories to cope with a variable environment (Venable, 1985a; Simons, 2011). As such, then, it seems that population persistence is more likely for plants producing both brown and black seeds than it would be for them producing either one or the other seed morph. However, to document bet-hedging and thus to understand the ecological consequences of production of dimorphic seeds, a comparative life history/demographic analysis of each of the two morphs is required (Venable, 1985a,b; Venable and Levin, 1985). It must be shown that the two morphs maximize the geometric mean of the number of offspring (Ro, a measure of fitness) across generations (Simons, 2011) in order to prove that the seed heteromorphism in S. corniculata is a bet-hedging strategy.

According to the field investigation and common environment experiments, phenotypic differences between S. corniculata populations are genetically based. Adverse effects of salinity stress are alleviated by the ability of plants to allocate a high proportion of resources to reproduction. Morph production and germination behavior of dimorphic seeds are influenced by the level of soil salinity, suggesting that soil salinity plays an ecologically important role in population regeneration of S. corniculata in the natural habitat.

ZH, FY, CB, JB, and XY conceived and designed the experiments. FY, DC, and XY performed the experiments and analyzed the data. FY, ZH, CB, and JB wrote the manuscript.

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.