What kind of seed dormancy occurs in the legume genus Cassia?

Cassia is a diverse legume genus widespread in the (sub-)tropical zone of the world. Several studies have been done on this genus; however, significant changes have occurred at the taxonomic level over the years. This has led to inaccurate information about seed dormancy in Cassia since many species are no longer included in the genus. Thus, our work aims to investigate and update the information about the kind of dormancy that occurs in seeds of Cassia species and also look into two notorious species in this group (C. fistula and C. javanica) to compare myxospermous vs. non-myxospermous seeds regarding dormancy and germination traits. Seed dormancy reports were found for 53 Cassia species, and the only kind of seed dormancy found for these species was physical dormancy (PY). Non-dormancy was not found, and all seeds had a blockage to water uptake during the dormant state, that is, all have PY. Of these 53 species, only 18 are currently included in the genus Cassia. C. fistula and C. javanica have fully developed embryos, and dormancy is only conferred by the (water-impermeable) seed coat. The lens in the seed coat is the only structure that creates a water pathway to break PY in C. fistula. Myxospermous seeds came out of dormancy faster than non-myxospermous ones. PY seems to be the only kind of seed dormancy that has evolved in Cassia. The extent of this kind of dormancy in all subtribe Cassiinae is also discussed.

www.nature.com/scientificreports/ of this kind of dormancy [21][22][23][24][25] and an investigation on water-impermeable seeds in this genus may allow new advances in the evolution of this restrict coat-imposed dormancy. Thus, we addressed the following questions: (1) What is known about seed dormancy in Cassia? (2) How many of the six classes of seed dormancy [including non-dormancy (see Baskin and Baskin 7 )] have been reported in Cassia species? (3) If seeds of C. fistula L. and C. javanica L. are dormant, are the embryo fully developed and/or the seed coat water impermeable at seed maturity and dispersal? (4) If they have PY, what is(are) the water gap(s) in these seeds? (5) Since the genus Cassia has myxospermous seeds, could water-impermeable seeds take advantage of myxospermy?

Results
Literature review on seed dormancy in Cassia. Fifty-three Cassia species were found in the literature whose kind of seed dormancy was investigated. Thirty-five of these species are no longer included in Cassia. That is, these species have been transferred to Senna or Chamaecrista (Fig. 1). From all studies found, 18 species were currently classified as Cassia (see Fig. 1). All 53 species in Fig. 1 have seeds with PY. The detailed results for the seed dormancy reported for species currently included in Cassia are shown in Table 1. The studies found in the literature showed that the investigated seeds had dormancy, and the treatments used to break PY (i.e., mechanical and acid scarification, immersion in hot water) were efficient in triggering high percentages of seed germination. As examples, Jurado and Westoby 18 found that intact seeds of C. helmsii Symon and C. oligophylla F. Muell. germinated at low percentages (< 20%), reaching high germination after mechanical scarification (max. of 90%). Mechanical and acid scarification also were effective in breaking the dormancy of C. leptophylla Vogel 26 and C. renigera Benth. 27 seeds. For C. nemophila Vogel, nontreated seeds germinated to low percentages; however, immersion in hot water broke dormancy 17 .
In some studies, imbibition tests were conducted. Lopes et al. 28 showed that seeds of C. grandis L.f. do not absorb water without mechanical scarification. Also, De Paula et al. 10 found that intact C. leptophylla seeds do not absorb water, unlike those that were exposed to a wet heat treatment. Indeed, this thermal treatment acts on C. leptophylla seeds forming an opening in the micropyle, which was the water gap for this species 10 . The results for Cassia species indicated that PY is recurrent for this genus (Table 1).
Response to dormancy-breaking treatments. Seeds of C. fistula and C. javanica had distinct responses to the dormancy-breaking treatments (Fig. 2). The greater responsiveness of C. fistula seeds to the treatments compared to C. javanica was evident. Exposing C. fistula seeds to 70 °C (for 2 h) and sulphuric acid (15 min) increased the breaking of PY and consequently increased germination (Fig. 2a). However, an extension of the time seeds were exposed to each treatment increased mortality (Fig. 2a). For C. javanica, few seeds (1%) germinated without any treatment (19% for C. fistula), and thermal treatments were not efficient in breaking dormancy. Immersion of seeds in sulphuric acid for 45 min reached the highest germination (62%) for C. javanica (Fig. 2b). The mortality was not increased, and a high percentage of seeds was still dormant after being exposed to different treatments (Fig. 2b). Overall, C. fistula seeds germinated to higher percentages than those of C. javanica in all treatments (see solid red lines in Fig. 2). Additionally, the mean percentage of seeds that remained dormant following all treatments was low for C. fistula (37.6), while it was quite high for C. javanica (78.4) (see red dashed lines in Fig. 2). The high mortality for C. fistula seeds indicates that dormancy was broken, but the treatment damaged the embryo (Fig. 2a).
imbibition curve. Seeds of both Cassia species increased in mass when placed in germination conditions following scarification (Fig. 3a,b). At 72 h of imbibition, seed mass of C. fistula and C. javanica had increased 166.4% and 154.5%, respectively. Concerning intact seeds, a distinct pattern was observed for C. fistula. There was a definite increase in seed mass (25.6%) until 24 h of incubation on a moist substrate for these seeds, and then stabilization afterward (Fig. 3a). This increase in mass of intact seeds was due to the thick mucilage on the seed coat. The mucilage absorbed water when the seeds were placed in germination conditions, and even after the seeds were blotted dry, the moisture remained adhered to the mucilage. After full hydration, the increase in seed mass ceased (Fig. 3a). On the other hand, there was no increase in seed mass of intact C. javanica seeds (Fig. 3b), showing a typical pattern of seeds with PY.
Seed morphological traits. Seeds of C. fistula had a rounded shape and a thick mucilaginous layer flaking off (Fig. 4a). Under moist conditions, C. fistula seeds released sticky mucilage that expanded and increased seed volume (Fig. 4b). In this assay, dormant seeds did not absorb water (except the mucilage), and only the mucilage expanded (Fig. 4b). Longitudinal sections of the seeds showed that the species had a thick endosperm surrounding the fully developed embryo (Fig. 4c). C. javanica seeds also had a rounded shape but with no evident www.nature.com/scientificreports/ mucilaginous layer (Fig. 4d). Under the moist condition, this seed did not produce mucilage; thus, it is characterized as a non-myxospermous seed (Fig. 4e). C. javanica also had an albuminous seed, with a thick endosperm surrounding the fully developed embryo (Fig. 4f).
Cassia fistula seeds had a hilar region composed of a narrow lens, hilum, and a tiny micropyle (Fig. 5a,b). The whole seed coat had several cracks but only in the mucilaginous layer (Fig. 5c,d). Following the dormancy breaking treatment, the lens was uplifted, creating a gap below it (Fig. 5e,f). The other structures did not change (Fig. 5e), and the seed coat lost the prominent cracks, due to partial removal of mucilage (Fig. 5g,h). Seeds of C. javanica had a hilar region with the lens, forming a wide depression on the coat, hilum and micropyle (Fig. 5i,j). The lateral part of the lens was sealed by a mucilaginous layer (Fig. 5j). The seed coat of this species also had cracks (Fig. 5k,l) but not prominent as in C. fistula (Fig. 5c,d). No evident structural changes occurred following dormancy breaking treatments (Fig. 5m,n), only partial removal of the mucilaginous layer (Fig. 5o,p). The absence of changes may be explained by the high percentage of dormant seeds even following treatments. The results were similar for both acid treatments (30 and 45 min).   www.nature.com/scientificreports/ Discussion the kind of dormancy in the genus Cassia. The genus Cassia has changed drastically over time, from the largest genus in the Caesalpinioideae 3,4 to a genus with about 30 species 5 . Among the 53 Cassia species for which we found dormancy studies in the literature, 66% of them are no longer included in this genus. However, all these studies report that PY was only kind of seed dormancy found PY. It is worth noting that the reports for the kind of seed dormancy cover 56% of the species in the genus Cassia. The control of seed germination in this genus is caused exclusively by the water-impermeable seed coat. All species found in the literature germinate promptly following dormancy-breaking treatments, and the embryos are fully developed at seed maturation. That is, seed dormancy is not related to the embryo. This coat-imposing dormancy is related only to the water impermeability of the testa, and the embryo is ready to start germination when the water enters through the water gaps located on the seed coat 15,55 . Our results showed the lens is the water gap in C. fistula seeds. This seed structure is the most common water gap in Leguminosae 56 . This structure was uplifted when PY was broken as shown for other species 13,14,57 , thereby creating an entry point for water into the seed. For C. javanica, the seed structure functioning as a water gap remains unclear.
Since seeds with PY do not imbibe water, the imbibition test is an accurate way to identify this kind of dormancy in seeds 9 . However, specific seed traits such as the presence of mucilage can hinder this detection. The increase in mass of myxospermous seeds is due to absorption of water by the mucilage 58,59 . Thus, attention must be paid to imbibition tests performed using this type of seeds.
Physical dormancy and myxospermy. To release PY, seeds must experience specific environmental conditions 21 . For some species, moist conditions in the environment are required for the breaking of PY 60,61 . Thus, in this case, physically dormant seeds overcome dormancy in moist soil 24,62 . Since seeds with PY can germinate promptly when the dormancy is released 9 , the strategy to break PY under moist conditions for some species allows seeds to germinate in a suitable environment for seedling establishment 24,62 . Thus, soil humidity is a key factor modulating the germination timing of some water-impermeable seeds. In this way, studies have investigated this water-dependent mechanism to break PY and related it to the increasing internal vapour pressure in the seed caused by the association between humidity and high temperatures, which dislodge the weak regions in the seed coat forming the water gap (see Rodrigues-Junior et al. 25 and Jayasuriya et al. 63 ).
Hence, if the moisture surrounding the seeds is essential for completing the dormancy breaking process for some species, the capacity of seeds to retain water next to the seed coat would facilitate this process. Thus, myxospermy could have an ecological significance for seeds with PY. Myxospermous seeds release mucilage when in contact with water, and the chemical composition of the mucilage makes it hydrate quickly, forming a hydrogel component that holds significant amounts of water surrounding the seed 64 . Mucilage can also be considered a 'superabsorbent hydrogel' , due to its strong ability to absorb water 58 . The roles of myxospermy on seed dispersal and germination are detailed in the literature 59,[64][65][66][67] . In contrast, the role of mucilage on dormancy is poorly investigated; however, if moisture is a crucial factor on releasing PY, myxospermy may affect this process in water-impermeable seeds. As demonstrated for the myxospermous and non-myxospermous seeds of Cassia, the species that produces mucilage broke PY easily in all treatments they were subjected to when compared to the non-myxospermous species.
Our results may stimulate in-depth investigations about a new possible role of myxospermy. Detailed investigations need to be done, such as removing mucilage before incubating the seeds under conditions that promote dormancy release. Mucilage-producing seeds are mostly found in phylogenetically advanced groups 59 , with some of them also producing water-impermeable seeds (i.e., with PY), for example Guazuma ulmifolia Lam. 22 , Senna alata (L.) Roxb. 13 and the studied species C. fistula. Thus, this advanced evolutionary trait (mucilage), with ecological roles on several stages of the plant life cycle 59 , may also have a direct role on PY.
Seed dormancy in the subtribe cassiinae. The subtribe Cassiinae (comprising the genera Cassia, Chamaecrista, and Senna) has widespread members, and investigations about the kind of dormancy in their seeds have been done in several parts of the world. For Senna, the largest genus in this subtribe, all reports in literature showed that PY is the only kind of dormancy found in this group [9][10][11][12]14 . No results were found associating PY with another kind of dormancy, for example, with physiological dormancy (combinational dormancy). In the case of Chamaecrista, the second largest genus in this subtribe, the investigations also found PY as the only kind of dormancy occurring in these seeds [68][69][70][71][72] . The results found in the present study also corroborate with the evidence that impermeability to water is a prevailing trait in seeds of this subtribe, since PY was found in all Cassia species. All 53 species reported in our study, including those that are currently included in Chamaecrista or Senna, have PY.
All considerations about the evolutionary aspects of seed dormancy declare PY as the most recent germination constraint in the evolution of seed plants 7,73,74 . This advanced dormancy is restricted to certain groups of plants 73,74 and occurs in 18 angiosperm families 22 . Even with the high diversification presented by the species included in Cassiinae 1,75,76 , all information here leads us to consider this restricted kind of seed dormancy as the only one that has evolved in this legume subtribe. This subtribe with PY is then adapted to specific environmental cues that make the seeds overcome water impermeability and be able to germination as soon as the water is available.

Materials and methods
Literature review on seed dormancy in Cassia. The purpose of this investigation was to identify the kinds of seed dormancy reported for the genus Cassia. Literature data were retrieved from published studies on seed dormancy and germination of Cassia species. Information found in the seed book by Baskin -48.282410) were collected at Uberlândia (Southeastern Brazil) before dispersal, when the fruits were dry on the mother plant. Mature fruits of both species remain attached to the mother plant for several months, and the seeds are not released from them. These two species are trees and grow in Brazilian savanna (Cerrado), characterized by two well-defined seasons throughout the year: the dry season, during the autumn/winter; and rainy season, during the spring/summer 78 . Since the fruits are lignified and challenging to open manually, the seeds were removed using a hammer to break the fruits. To remove non-filled seeds, they were separated by flotation in the water, especially for C. fistula, whose seeds have a high predation percentage. Seeds were blotted dry and placed in plastic trays at ambient room conditions (25 ± 5 °C) for 24 h and then stored in paper bags until the beginning of the dormancy and germination experiments, one week later.
Response to dormancy-breaking treatments. Cassia fistula produces myxospermous seeds, while C.
javanica produces non-myxospermous seeds. Thus, the purpose of this experiment was to compare the response of both species to different treatments to break dormancy. The seeds were subjected to the following treatments: imbibition curve. Changes in mass were quantified for intact and scarified seeds of C. fistula and C. javanica to verify the presence (or not) of a water-impermeable seed coat. Thirty intact and manually-scarified seeds (using sandpaper) for each species were placed in gerboxes on germination paper moistened with distilled water. Seeds were placed in a sequential order in each Gerbox (30 seeds per Gerbox) and weighed individually. If the seeds are water-impermeable, scarification will remove the restriction to water uptake, resulting in increased seed mass when in contact with water. Seeds were incubated at 25 °C under constant light and weighed consecutively using a precision scale. Seeds were blotted dry before each weighing and weighed at distinct intervals for 72 h.
Seed morphological traits. Presence (or not) of mucilage, seed coat structures (water gap), and embryo development of C. fistula and C. javanica seeds were evaluated under a stereomicroscope and scanning electron microscope (SEM). First, seeds were evaluated under a stereomicroscope following exposure to dry and moist conditions. Under moist condition, seeds were placed on moistened paper in a Gerbox for 24 h, while under dry conditions the paper was not moistened. Mucilage in myxospermous seeds expands upon hydration, and the differences between the species were examined. Seeds were also manually scarified and imbibed for a few minutes and then longitudinally sectioned using a cutting blade to check if the embryo was fully developed at seed maturity. Additionally, intact and treated (according to the breaking dormancy methods used above) seeds of C. fistula and C. javanica were fixed in Trump fixative for 24 h 79 , dehydrated in an ethanol series and subjected to critical point drying. Seeds were then mounted on stubs using double-sided carbon tape and sputter-coated with gold 80 . The whole seed surface was examined under SEM to investigate the structural changes on the seed coat.
Statistical analyses. The experimental design was completely randomized. Germination data were analysed using a generalized linear model (GLM) followed by LSD test to compare the means at 5% probability 81 . The graphs were designed using the software Sigmaplot (Systat Software, San José, CA, USA).

Data availability
All data generated in this study are included in the main text. www.nature.com/scientificreports/