Introduction

Seeds of many genera belonging to 18 plant families develop impermeable seed (or fruit) coats, i.e., physical dormancy (PY) during the maturation drying phase of seed development (Baskin and Baskin 2014). At the time of PY acquisition, depending on the species, one or more palisade cell layers present in the seed/fruit coat contain water-repellent substances that prevent the penetration of water into the seed (Rolston 1978). During the final stage of PY acquisition, other structures such as the micropyle, hilum, and chalaza also close, but these structures serve as detectors of the external environment that guide the timing of dormancy break in PY seeds (Baskin et al. 2000). After dispersal, the presence of an impermeable seed coat precludes immediate germination, and these seeds persist in soil until suitable dormancy-breaking signal cues, such as seasonal temperature variation, high or low temperature, and/or other factors, like fire, open a ‘water gap’ marking the termination of dormancy (Baskin and Baskin 2014; Jaganathan 2015). Water first enters through this gap, hydrates the embryo, and triggers immediate germination if the embryo does not have physiological dormancy (PD), i.e., combinational dormancy (PY + PD) (Baskin and Baskin 2004). Such a trait noticeably increases fitness for colonization and maintains species in highly variable environments (Jaganathan and Liu 2014b; Van Assche et al. 2003).

The broadly accepted notion is that the seeds of PY species, collected before the maturation drying stage, have permeable seed coats and germinate when incubated at appropriate temperature, which is species specific (Baskin and Baskin 2014; Egley 1979; Gama-Arachchige et al. 2011; Helgeson 1932; Hills 1942; Quinlivan 1971). These seeds become impermeable only at the end of the maturation drying stage (Baskin and Baskin 2014; Baskin et al. 2000). Although seeds collected early in the maturation phase and tested for dormancy contributed to some misconception about the presence or absence of PY in a particular species, the number of PY seeds, and the level of hardseededness (collected even after maturation drying) produced by any given plant varies significantly between the sites, between years or even on the same plant depending on the location of seeds (Baskin and Baskin 2014; Smith et al. 1998; Hudson et al. 2015). This raises important questions: (1) why does the number of PY seeds vary between plants, sites, or years, (2) what factors control this variation, and (3) is there any ecological significance behind this? To date, most of the PY literature has had a primary interest in determining how PY is broken by any given set of artificial and/or natural conditions. Despite detailed research over the past several decades, knowledge of the factors regulating development of PY is still inconclusive. Consequently, addressing these questions may provide an opportunity to understand the ecological significance of PY.

Influence of environmental conditions on PY development

Although the trait of PY is mostly heritable, the maternal environment of the plant during seed development plays an important role in regulating its development. Environmental factors, particularly temperature, relative humidity, water stress, and length of growing season, contribute to the development of PY that can vary from no impermeable seeds to all seeds having an impermeable seed coat (D’hondt et al. 2010; Tozer and Ooi 2014). Nevertheless, which of these factors plays a more dominant role is often difficult to disentangle due to the species-specific variation that interacts with these factors (Argel and Paton 1999; Souza and Marcos-Filho 2001).

Numerous studies have shown that seeds maturing at higher temperature produce more impermeable seeds compared with the low-temperature environment (Gladstones 1958; Llorens et al. 2008; Jaganathan et al. submitted). In Stylosanthes hamate cv. Verano, for example, seeds that matured at 27 °C produced more impermeable seeds compared with those maturing at 24 °C (Argel and Humphreys 1983a). Based on the microscopic examination, these authors revealed that in the seeds matured under warmer temperature, palisade cells were shorter and contained less cellulose but increased levels of hemicellulose and lignin that contributed to the impermeable seed coat. However, given the difficulty in measuring temperature throughout the seed maturation period, direct information on this subject is mostly incomplete. Further, seasonal or location-specific variation of temperature around the parent plant may partly explain the development of both permeable and impermeable seed coat in the same species maturing at different times or in different places. Piano et al. (1996) reported that the number of impermeable seeds produced by Trifolium subterraneum decreased with increasing altitude indicating that the high temperature prevalent at lower altitudes contributes to the production of more impermeable seeds. Thus, it is reasonable to predict that seeds maturing at higher temperature are likely to lose more water and produce higher proportion of PY seeds than the seeds developing in a low-temperature environment.

The development of an impermeable testa is initiated only when the moisture content of the seeds falls below species-specific threshold levels. In Trifolium repens, T. pratense, and Lupinus arboreus during maturation drying, moisture content fell rapidly to approximately 25 % (dry weight basis), but this drop did not lead to the development of impermeable seed coats (Hyde 1954). However, towards the end of maturation drying, the moisture loss was slow and when the seeds reached 14 % moisture content, the epidermis became impermeable. Thereafter, water present in the seeds could only diffuse out through hilum when the external relative humidity was less, but at higher relative humidity the fissure closes preventing the movement of water in the form of vapor into the seeds (Hyde 1954).

Many studies established that the relative humidity (RH) of the seed maturation environment can affect the drying capacity of seeds, and that RH consequently plays a critical role in the development of impermeable seed coats (Barrett-Lennard and Gladstones 1964; Corner 1951). In Western Australia, seeds of Ornithopus compressus collected before maturation and stored at 76 % RH and above did not develop impermeable seeds, while storage at 44 % RH or below resulted in all seeds being impermeable (Barrett-Lennard and Gladstones 1964). By growing plants of Trifolium repens at three different relative humidities, D’hondt et al. (2010) found that seeds matured under wet conditions (relative humidity between 89 and 99.9 %) developed permeable seeds, compared with dry conditions (RH between 47 and 98 %) where few permeable seeds developed. At RH levels between 61 and 98 %, there were almost equal numbers of seed with permeable and impermeable coats. Similarly, Tozer and Ooi (2014) reported that seeds of Acacia saliga matured in a dry year produced more impermeable seeds, than in a year with higher relative humidity.

The development of an impermeable seed coat is affected by the temperature and RH prevailing at the time of seed maturation, but these factors continue to act on seeds and dry them even after they are shed from parent plant to the soil (Quinlivan and Millington 1962). While measuring these effects in the field is challenging, laboratory experiments storing seeds at different RH and temperatures provided support for this suggestion. In the case of Melilotus officinalis, seeds collected before maturation drying and stored at dry conditions became impermeable, but storing these seeds in a moist environment did not develop impermeability (Helgeson 1932). Seeds of Lupinus luteus maturing in Israel remained permeable to water during the maturation drying stage, but after removal from the mother plant, seeds stored at lower humidity became impermeable, whereas those under higher humidity conditions remained permeable (Hurwitz and Gabrielith-Gelmond 1952). In Lupinus varius, the moisture content of the seeds before shedding was 18–20 %, but after dispersal on to the ground, the moisture content of the seeds decreased to 10–12 % within 2–3 days. This loss of moisture was accompanied by the development of an impermeable seed coat (Quinlivan 1967). Working with nine varieties of garden bean seeds, Nutile and Nutile (1947) showed that no seeds collected before completing maturation drying and held at higher RH developed impermeable seed coats following 60 days of storage; however, the ability of seeds to develop impermeable seed coats increased with a decrease in RH, with some cultivar-specific variation. Nevertheless, there appears to be a threshold moisture content below which, irrespective of maternal environment, a high percentage of PY develops in the seed population (Hyde 1954). For example, although seeds of Stylosanthes hamate cv. Verano had various levels of PY at different temperatures, the effect of temperature became irrelevant, if the moisture content of the seeds fell below 7 %. Below this moisture all seeds developed impermeable seed coats, irrespective of ambient temperature (Argel and Humphreys 1983a).

It has been suggested that there is an inverse relationship between temperature and relative humidity in the development of impermeable seed coat. In general, seeds maturing at a low relative humidity and high temperature environment are more likely to form impermeable seed coats, whereas higher humidity combined with low temperature favors the development of a permeable seed coat (Biasutti Owen 1956). Hopkins et al. (1947) showed that storing freshly collected seed of Phaseolus vulgaris at 4.4 °C for 6 months at 0 and 80 % RH resulted in 61 and 0 % of impermeable seeds compared with 14.2 % impermeable seeds recorded before storage. However, storing the same batch of seeds at a higher temperature (26.6 °C) resulted in 85 % impermeable seeds at 0 % RH, but no impermeable seeds at 80 % RH.

In addition to the relative humidity and temperature, gases present in the atmosphere affect the development of an impermeable seed coat. For example, Marbach and Mayer (1974) found that seeds of Pisum elatius collected prior to maturation drying and subsequently dried over CaCl2 became impermeable only if they were dried in air or oxygen, but all the seeds remained permeable if dried in a vacuum with nitrogen. In contrast, all Robinia pseudoacacia seeds dried in any of the above conditions developed permeable seed coats when collected early in the maturation drying stage, but seeds collected late developed impermeable seed coats when dried in air and oxygen, but only a few in vacuum with nitrogen.

The season in which the PY seeds mature greatly affects the proportion of impermeable seeds developed. In general, seeds maturing in wet seasons are less impermeable than those that mature during dry seasons. For instance, Quinlivan (1965) found that the seeds of Trifolium subterraneum growing in environments with a long spring growing period developed more impermeable seeds than those matured where there was a short growing period. Similarly, seeds of Centrosema virginianum and Macroptilium atropurpureum produced higher number of impermeable seeds in the spring than in autumn (Clements 1977). In the case of Prosopis juliflora, there were significantly fewer impermeable seeds when seeds matured in autumn compared with winter and spring (El-Keblawy and Al-Rawai 2006). The number of dormant seeds produced by Fumana ericoides matured in winter was 1.7 % compared with 100 % dormant seeds matured in late spring/early summer (Llorens et al. 2008). In addition, the variation in the length of the growing period during seed maturation also seems to be a crucial factor in the development of various levels of hardseededness (Collins 1981; Lodge et al. 1990; Revell et al. 1999).

Evidence also indicates that the amount of rainfall received by the parent plant site plays a vital role in whether an impermeable seed coat develops. One possible reason for the variation in development of impermeable seed coats can be ascribed to water stress resulting from varied rainfall (Argel and Humphreys 1983b; Cameron 1967; Hill et al. 1986; Smith et al. 1998). Less rainfall results in exposure to higher water stress, which leads to development of a high proportion of water impermeable seed, but in the years with high rainfall, the proportion impermeable seed is low. Even within the same site, variation in rainfall can produce various numbers of permeable and non-permeable seed coat within the same plant in seeds maturing at different years (Nichols et al. 2009).

Many detailed investigations agree that species with a known history of PY can maintain or lose the dormancy state depending upon the environmental conditions under which the descendent populations are grown. Development of an impermeable seed coat in Phaseolus vulgaris is heritable from the parent plant, but the next generation plants growing in a different set of environments compared with that of the mother plant can result in different proportions seeds with impermeable coats. Generally, there are more impermeable seeds in dry environments (Lebedeff 1947). The only detailed study that has shown environmental conditions as selective force in the development of an impermeable seed coat (Nichols et al. 2009) revealed that plants that migrated to where the growing season was short generally developed seeds with high proportion of impermeable coats, whereas in sites with a long growing season, the proportion of impermeable seed was low. In addition to the length of growing season, other factors, such as low rainfall and harsh environmental conditions prevalent during a short growing season, act as driving forces for developing high proportions of impermeable seeds (Nichols et al. 2009). Because this change in the proportion of impermeability in response to the prevailing environment occurred within 2 years, climate might possibly be the important driving force contributed to the evolution of PY in seeds. Thus, as proposed by Baskin et al. (2000), climatic drying could be the important selective force for the evolution of PY, and natural selection fine-tunes the species with PY to fit the environment.

The literature available on environmental factors controlling the extent of PY development has been reviewed thus far. However, it is worth noting that apart from the external environmental factors, there are some maternal effects such as age of the mother plant (reviewed by Baskin and Baskin 2014), seed size produced (Calero et al. 1981; Jaganathan et al. submitted; Yaklich et al. 1986), composition of seed coat (Egley et al. 1983; Saio 1976; Stabell et al. 1996), and position of seed on plant and inflorescence (Pérez-García 1997; Taylor and Palmer 1979) that can also potentially affect the PY status of seeds. Nevertheless, almost nothing is known about the relationship (if any) between the environmental factors and maternal effects. The studies concluding that maternal effects play an important role in determining the development of PY did not consider the effect of maternal environment at different maturation sites. Given the fact that environmental conditions around the mother plant varies significantly at different positions on the mother plant, it is not clear whether maternal effects per se influence the PY development, or the influence is from the environmental conditions or a combination of both. For example, seeds in the inflorescence exposed to direct sunlight presumably dry to lower moisture level compared with the seeds developing in shade, and this scenario is particularly true for plants in wild (Hay et al. 2010). Thus, this area warrants detailed studies.

Ecological considerations

The ability of plants to produce seeds with different levels of PY offers extensive ecological benefit for species colonization (Venable 1985). Indeed, variation in the number of PY seeds produced by plants growing in different environmental conditions is likely to be advantageous for persistence in diverse ecosystems. Given that species producing PY seeds occur in many types of the ecosystems from alpine to warm deserts (Baskin and Baskin 2014; Jaganathan et al. 2015), the environmental control of PY development has several ecological implications. Based on the evidence presented above, it is safe to assume that species with a known history of developing an impermeable seed coat growing in arid and dry environments could produce higher percentage of PY seeds. All the environmental conditions reviewed above affect not only the number of impermeable seeds that develop but also the level of hardseededness within impermeable proportion of the seed population. Under conditions of low relative humidity, high temperature, and low rainfall during seed maturation, seeds persist in soil for a long time, forming persistent seed banks, and spread the risk of germination over many years. High summer temperatures in these environments favor dormancy loss and synchronize germination with the wet season (Jaganathan and Liu 2014b). Even seeds that break dormancy before the wet season do not germinate, as the seedling mortality would be high (Baskin et al. 2004; Jaganathan and Liu 2014a; Leck et al. 2008).

Although strong empirical evidence to validate that maternal environmental conditions affect the seed persistence and dormancy-breaking requirements of PY seeds is scarce, several studies provide evidence (Bolland 1985; Taylor 1996). Taylor (1996) found that seeds of Medicago polymorpha and M. truncatula, collected from three sites, showed different dormancy loss responses when buried in four seed softening sites. Thus, Taylor (1996) unequivocally concluded that ‘seed growing environments were more important than seed softening environments in determining rates of seed softening in all cultivars.’ Quinlivan (1967) also presented evidence to show that PY seeds maturing in different environments would have different dormancy-breaking requirements, and this would essentially determine the loss of seeds from soil seed banks. It was shown that L. varius plants grown in a low moisture environment persisted in soil for 5 years, with 15–20 % of propagules broke dormancy every summer and germinated during the rain in autumn. However, seeds from drier site germinated within 2 years. The difference in the level of hardseededness in seeds matured at different sites along with different environmental conditions (moist vs. dry) can explain this variation in persistence capacity. Similarly, Michael et al. (2006) reported that more than 80 % of Malva parviflora seeds collected at higher rainfall site lost dormancy in 3 months (100 % in 6 months) when incubated at 50/20 °C, but dormancy in seeds collected from a lower rainfall area only germinated to about 60 % even after 2 years of incubation at 50/20 °C. In addition, when batches of seeds from four different sites (two with high rainfall, two with low rainfall) were buried in a site different from the seed maturation site, about 50 % of the seeds from lower rainfall sites broke dormancy in 3 months compared with <5 % germination from seeds matured in the higher rainfall sites. Various levels of coat impermeability and different dormancy-breaking requirements have also been reported for Fabaceae (Taylor et al. 1991), indicating that this pattern might be common to all PY species. Norman et al. (2002) showed the level of hardseededness in Trifolium (Fabaceae) seeds produced in the same year differs between sites and also demonstrated that the softening pattern varies in accordance with the level of coat impermeability.

Laboratory studies have shown that seed coat of PY species becomes impermeable to water only when the moisture content of the seeds drop below particular species-specific threshold value, e.g., 15 % in Peltophorum pterocarpum (Mai-Hong et al. 2003) and Lupinus varius (Quinlivan 1968); 12 % in Gleditsia triacanthos (Geneve 2009), G. aquatica (Geneve 2009), Gymnocladus dioicus (Geneve 2009), L. arboreus (Hyde 1954), Pisum sativum (Ellis and Roberts 1982), Trifolium ambiguum (Hay et al. 2010), T. pretense (Hyde 1954), and T. repens (Hyde 1954); 11 % in Lupinus digitatus (Gladstones 1958). However, in the field seeds are likely to lose more moisture (Jaganathan and Liu 2014b; Jayasuriya et al. 2013). Thus, based on the available evidence, it is proposed that two levels of PY might result from the moisture content attained during maturation drying or continued drying after seed dispersal: (1) shallow and (2) absolute. Seeds with ‘shallow’ PY have higher moisture content, whereas those with ‘absolute’ PY have much lower water content. It is not clear from the previous studies whether there is also an ‘intermediate’ level of PY. In addition, the reversing of PY in seeds with shallow dormancy reported in Lupinus digitatus (Gladstones 1958) and Trifolium ambiguum (Hay et al. 2010) must be investigated in a wide range of species.

Interestingly, Quinlivan (1968) found that seeds of Lupinus varius with moisture content above 10 % (but below 15 %) had an impermeable seed coat and that dormancy was broken in moist or high humidity atmosphere by diffusion of water in the form of vapor only through testa. On the other hand, for the seeds with a moisture content below 10 %, moist conditions did not break dormancy, but temperature fluctuations between 15 and 65 °C had fractured seed coat at the strophiole region and allowed seeds to imbibe water. This study confirmed that the germination pattern of PY species largely depends on the degree to which the summer temperature and/or ambient RH desiccates the seeds. Thus, if the moisture content remains above 10 %, a large proportion of seeds will germinate over several months, and only a smaller proportion remains as part of the seed bank, but with moisture content below 8.5 %, only appropriate summer temperature fluctuations can break dormancy and these seeds would begin to germinate at the onset of the wet season.

Many studies have documented that under natural conditions, the dormancy break in PY species occurs in two steps (1) a pre-conditioning step that softens and makes the seeds sensitive to dormancy breaking and (2) the actual dormancy-breaking step, during which seeds imbibe water (Jayasuriya et al. 2008a, b, 2009; Taylor 2005; Taylor and Revell 1999; Van Assche et al. 2003). These studies also displayed a relationship between the RH and dormancy breaking. In Ipomoea lacunose, for example, seeds generally became sensitive to dormancy break under wet conditions (Jayasuriya et al. 2008a). However, dormancy is only alleviated in sensitive seeds when the seeds experience higher temperature (35 °C) in moist soil. In contrast, in sensitive seeds exposed to dry conditions at ≤5 °C or ≥30 °C, the sensitive stage was reversed, i.e., seeds became insensitive to dormancy break. Interestingly, some samples collected at a different time showed alleviation of dormancy when subjected to 35 °C wet treatment even in the absence of prior dry storage to make seeds sensitive. In contrast, seeds of Cuscuta australis required dry storage to become sensitive and then wet conditions combined with high temperature to break dormancy (Jayasuriya et al. 2008b). Unfortunately, the moisture content of the seeds was not measured at the time of maturity and it is likely that variation in lower level of moisture content attained during seed maturation could have led to production of ‘shallow’ and ‘absolutely’ dormant seeds matured at different times. As such, this would have affected the dormancy release process.

Seeds with PY generally persist in soil for a long time before they germinate (Baskin and Baskin 2014). In their natural environment, the ability of seeds to respond to the humid and dry conditions during dormancy break can also be influenced by the position of seeds in the soil. In general, PY seeds present in the surface soil layers are released from dormancy more quickly than seeds buried deeper. This is presumably due to temperature fluctuations at the surface (Loi et al. 1999; Taylor and Ewing 1988). However, Hu et al. (2009) found that seeds of Sophora alopecuroides matured at higher and lower rainfall site differed in the initial level of hardseededness, and this difference contributed to two distinct patterns of dormancy loss when buried at two different depths. Dormancy release rate in seeds from the lower rainfall site decreased as burial depth increased, but only 9 % of the seeds matured from higher rainfall site showed alleviation of dormancy when left on the soil surface (compared with 73 % of seeds from lower rainfall site and left on soil surface), although 71 % of the seeds from the high rainfall site became non-dormant when buried at 2 cm depth.

None of the studies that illustrated a two-step dormancy break in PY seeds (see references above) measured the initial moisture content of seeds. There are at least three difficulties in understanding the dormancy break of PY seeds without this information. Firstly, the pre-conditioning step in seeds dried to lower moisture content (absolute dormant seeds) may be different from that of seeds maintaining higher water content as an equilibration to surrounding RH and temperature (shallow dormant seeds). It is not clearly known, if this difference is the reason for mixed results with some studies claiming wet heat is effective in breaking dormancy compared with dry heat and others concluding the opposite (Martin et al. 1975; Van Klinken et al. 2006). Nevertheless, the findings reported by Hu et al. (2009) suggest that variation in dormancy levels resulting from different maturation sites could be the important factor in this disparity. Those authors reported wet heat released dormancy in S. alopecuroides seeds matured in both high and low rainfall sites; however, dry heat was less effective in breaking dormancy of seeds collected from the high rainfall site. Perhaps, this difference determines the time and duration required by high temperature, e.g., during fire to break dormancy (Jaganathan 2015). Secondly, it remains unclear whether shallow and absolute dormant seeds have any specific pre-conditioning and dormancy-breaking requirements. Thirdly, the rate of completion of each step could also vary between seeds with different moisture levels. Moist or dry conditions and the temperature at which these treatments are applied might affect dormancy loss mechanisms. Thus, future studies should place special emphasis on understanding the dormancy-breaking mechanisms in PY seeds with different moisture contents, i.e., shallow and absolute dormant seeds.

Conclusions

Environmental factors play an important role in the development of the number of impermeable seeds and various levels of hardseededness within impermeable seeds. Available literature strongly suggests that maternal environment during seed development affects the level of hardseededness, resulting in two levels of PY, i.e., shallow and absolute. This notion helps explain the year-to-year variation in dormancy of PY seeds growing in the same site or within year variation of seeds from different sites and the difference in pattern of dormancy loss. This also partly supports the proposal that climate drying would have acted as the selective force for the development of hard seed coat in seeds. Thus, depending on environmental conditions, PY seeds dispersed to different locations could develop all or no dormant seeds. Although the data are not conclusive, it is believed that the variation in moisture level reached during seed maturation and subsequent soil drying can affect the dormancy loss mechanism. Nevertheless, almost nothing is known about the ecological significance of this process. Understanding how PY seeds having shallow or absolute dormancy germinate in natural environments will greatly increase our knowledge of the mechanism(s) underlying large-scale ecological patterns of PY species.