Introduction
Water-impermeable seed- or fruit-coats give seeds with physical dormancy (PY). PY is absent in gymnosperms but occurs in 18 angiosperm families that are distributed throughout several clades across all of the biogeographical regions on Earth, and is particularly important in tropical savannas (Baskin and Baskin, Reference Baskin and Baskin2014). The phylogenetic distribution of PY strongly suggests its independent evolution during angiosperm history (Willis et al., Reference Willis, Baskin, Baskin, Auld, Venable, Cavender-Bares, Donohue and de Casas2014). The recent discovery of water-impermeable fruits in basal angiosperms (Mahadevan and Jayasuriya, Reference Mahadevan and Jayasuriya2013) indicates that PY has been overlooked in angiosperms and suggests that our understanding of the phylogenetic and biogeographical distribution of seed dormancy increases as additional taxa are studied (see a complete list of taxa with PY seeds in Baskin and Baskin, Reference Baskin and Baskin2014).
PY is caused by a multitude of seed and fruit structures in different plant lineages. Water-repellent substances can be found throughout the pericarp, only in the endocarp, the exotestal or exotegmic palisade cells, or endotestal cells of seeds across different botanical families (Baskin et al., Reference Baskin, Baskin and Xiaojie2000; Baskin and Baskin, Reference Baskin and Baskin2014). In addition to physical dormancy, seeds of many species may also have physiologically dormant embryos (PD). The combination of PY and PD is termed physiophysical dormancy and is the rarest seed dormancy type (Willis et al., Reference Willis, Baskin, Baskin, Auld, Venable, Cavender-Bares, Donohue and de Casas2014). Physiophysical dormancy evolved only in eight families, most of which are restricted to Rosids (Baskin and Baskin, Reference Baskin and Baskin2014).
Based on seed anatomy, Baskin et al. (Reference Baskin, Baskin and Xiaojie2000) predicted the evolution of water impermeability in several eudicot lineages, including Melastomataceae, but empirical data on germination and seed coat permeability are still lacking for this family. Melastomataceae (Myrtales) comprises nearly 4500 species in c. 150 genera (Renner, Reference Renner2004) and attains its highest diversity in the Neotropics (Clausing and Renner, Reference Clausing and Renner2001). This pantropical family comprises shrubs, climbers, herbs, epiphytes and trees that dominate montane to lowland forests, savannas, grasslands and disturbed areas throughout Central and South America (Silveira et al., Reference Silveira, Fernandes and Lemos-Filho2013). The seeds of Melastomataceae are morphologically diverse (Corner, Reference Corner1976; Groenendijk et al., Reference Groenendijk, Bouman and Cleef1996), with six basic forms – cochlear or subcochlear, obpyramidal, winged, ovate, oblong-obovate, orbicular plano-convex and clavate (Baumgratz, Reference Baumgratz1985) – and have a wide range of cell structures, forms and appendages (Ocampo and Almeda, Reference Ocampo and Almeda2013).
Seeds of Melastomataceae are either non-dormant or physiologically dormant (PD), but there are no reports on water-impermeable seeds in plants from this family (Willis et al., Reference Willis, Baskin, Baskin, Auld, Venable, Cavender-Bares, Donohue and de Casas2014). PD has been described in species from temperate areas (Baskin et al., Reference Baskin, Baskin and Chester-Edward1999), tropical seasonal vegetation (Silveira et al., Reference Silveira, Fernandes and Lemos-Filho2013) and tropical rainforests (Garwood, 1983). Although we have previously demonstrated PD in melastomes from tropical montane grasslands (Silveira et al., Reference Silveira, Ribeiro, Oliveira, Fernandes and Lemos-Filho2012), preliminary seed anatomy studies suggest a new mechanism determining water-impermeable seed coats in at least one species. In this study, we evaluated the seeds of Chaetostoma armatum using dye tracking, germination experiments, embryo imbibition and detailed seed coat development and anatomical data. We provide evidence that C. armatum produces seeds with a water-impermeable seed coat and dormant embryo, and discuss the implications of our findings to the understanding of the biogeographical and phylogenetic distribution of physiophysical dormancy (PD+PY).
Materials and methods
Study site and species
Seeds of C. armatum D.C. (Microlicieae: Melastomataceae) were collected in 2010 from 25 individuals from rupestrian grasslands (campos rupestres), Serra do Cipó, south-eastern Brazil (19°16′53.3″S, 43°35′31.8″W; 1200 m above sea level). The campos rupestres are fire-prone seasonal grasslands in which sclerophyllous, ericoid-like plants establish on quartzite-derived, nutrient-poor, acidic and shallow soils (Giulietti et al., Reference Giulietti, Pirani, Harley, Davis, Heywood, Herrera-MacBryde, Villa-Lobos and Hamilton1997; Benites et al., Reference Benites, Schaefer, Simas and Santos2007). The climate in Serra do Cipó is mesothermic, with rainy summers from October to March and dry winters from April to September. The mean monthly temperatures range from 7.8°C in July to 32°C in February (Madeira and Fernandes, Reference Madeira and Fernandes1999).
C. armatum is a subshrub that is distributed in rupestrian grasslands and open savannas from Bahia (north-eastern Brazil) to Paraná (southern Brazil) (Koschnitzke and Martins, Reference Koschnitzke and Martins2006). The sampled population of C. armatum occurs in sandy grasslands and produces dry capsules that open at the beginning of the dry season, when the soil moisture is still high (Silveira, Reference Silveira2011). The seeds of C. armatum are small (average 0.63 mm in length and 0.0013 mg in weight), and nearly half of them are embryoless (Silveira et al., Reference Silveira, Ribeiro, Oliveira, Fernandes and Lemos-Filho2012). Increased seed weight after soaking, and well-developed and differentiated embryos, has led to the conclusion that seeds of C. armatum are physiologically dormant (Silveira et al., Reference Silveira, Ribeiro, Oliveira, Fernandes and Lemos-Filho2012). Vouchers were deposited at the Herbarium of the Federal University of Minas Gerais (BHCB), accession number 154026.
Seed coat anatomy, development and histochemistry
The data on seed coat anatomy and development were assessed in individuals under natural field conditions. We tagged flower buds of 20 randomly chosen individuals at Serra do Cipó and sampled ovules and developing seeds at 15, 30 and 50 d after anthesis (DAA), and immediately fixed these in formalin–acetic acid–50% ethanol (Johansen, Reference Johansen1940) for 48 h, including 24 h under vacuum. The samples were dehydrated in an increasing ethanol series (50, 60 and 70%, plus 10% glycerin), stored for 3 d and dehydrated once more in a subsequent ethanol series (80, 90 and 95%). The samples were infiltrated in the refrigerator for 1 week and embedded in (2-hydroxyethyl)-methacrylate (Leica™) according to Paiva et al. (Reference Paiva, Pinho, Oliveira, Chiarini-Garcia and Melo2011). The mature seeds were fixed and softened according to Ribeiro and Oliveira (Reference Ribeiro and Oliveira2014) using Franklin solution (Franklin, Reference Franklin1945), glycerin and heating in a water bath; the embedding was performed as previously described (Paiva et al., Reference Paiva, Pinho, Oliveira, Chiarini-Garcia and Melo2011). Longitudinal and transverse sections (6 μm thickness) were stained with toluidine blue 0.05%, pH 4.7 (modified from O'Brien et al., Reference O'Brien, Feder and McCully1964) and mounted in Entellan™. Toluidine blue is a metachromatic dye that stains phenolic compounds with a blue to green colour and pecto-cellulosic substances with red to purple colours (O'Brien and McCully, Reference O'Brien and McCully1981).
Histochemical tests were performed in 8-μm-thick sections of ovules and seeds (15 DAA, 30 DAA and mature). We used 10% aqueous ferric chloride plus sodium carbonate to detect phenolic compounds (Johansen, Reference Johansen1940); phloroglucinol in HCl for lignin; Sudan IV in 70% ethanol for lipids (Jensen, Reference Jensen1962); lugol for starch; ruthenium red for pectic substances (Jensen, Reference Jensen1962); and mercuric bromphenol blue for proteins (Mazia et al., Reference Mazia, Brewer and Alfert1953). The sections were analysed under a Zeiss light microscope (Zeiss, Oberkochen, Germany) and described using the terminology of Corner (Reference Corner1976) and Werker (Reference Werker1997).
Dye tracking and embryo imbibition
To determine the seed coat permeability, some seeds were imbibed in 1% methylene blue (modified from Johansen, Reference Johansen1940) and others in 0.01% Lucifer yellow (modified from Briggs et al., Reference Briggs, Morris and Ashford2005) for 12 h. Due to the small size of C. armatum seeds, scarification was performed through puncturing the radicular lobe; the punctured seeds were also exposed to the methylene blue solution. Methylene blue is a polar compound, and Lucifer yellow is a double-negatively charged molecule, both of which are frequently used as an apoplastic tracker in seeds (Egerton-Warburton, Reference Egerton-Warburton1998; Briggs et al., Reference Briggs, Morris and Ashford2005; Mahadevan and Jayasuriya, Reference Mahadevan and Jayasuriya2013). Sections were obtained by cryomicrotomy using a 5030 Bright cryomicrotome (Bright Instruments Co. Ltd, Huntingdon, United Kingdom) and observed under a Zeiss light microscope (methylene blue) and under an epifluorescent Olympus microscope (Olympus Optical Co. Ltd, Southall, United Kingdom) with blue excitation (Lucifer yellow). Photomicrographs were taken using a Nikon digital camera (Nikon Corporation, Tokyo, Japan) coupled to the microscope.
Seed coat permeability tests were performed in control and punctured seeds in the micropylar region. Six replicates of 50 seeds were used for each treatment. We ensured that only those seeds containing embryos were used in this experiment, by separating brown and dark-yellow seeds (seeds containing embryos) from the pink or light-yellow seeds, which were embryoless. The seeds were weighed on a digital balance, soaked in tap water for 4, 12 and 24 h at room temperature and reweighed. The seed permeability was determined by the seed mass increase, and differences in the percentage increase between seed mass of dried and soaked seeds were determined by a paired t-test or Wilcoxon test (Silveira et al., Reference Silveira, Ribeiro, Oliveira, Fernandes and Lemos-Filho2012).
Germination experiments
The seeds were checked for embryo presence under a Zeiss stereomicroscope, and embryoless seeds were excluded from the germination experiments. Six experimental treatments were created: control, punctured seeds, punctured +500 mM GA3, and control + GA3 in three standard concentrations – 250, 500 and 1000 mM. These concentrations cover a wide range of germination responses to gibberellins. Six replicates of 25 fresh seeds were used for each treatment. The seeds were set to germinate in Petri dishes that were layered with a double sheet of filter paper, moistened with 1% Nistatin solution to prevent fungal growth. For gibberellin treatments, the seeds were exposed to gibberellin for 3 d and then transferred to Petri dishes. The germination was monitored at daily intervals for 56 d, and the germinated seeds that showed root protrusion were removed from the dishes. The dishes were incubated at 25°C under a 12/12-h photoperiod (Silveira et al., Reference Silveira, Fernandes and Lemos-Filho2013). We calculated final germination percentage and the mean germination time (MGT) for each replicate, following Ranal and Santana (Reference Ranal and Santana2006):
$$\begin{eqnarray} MGT = { \sum _{ i = 1}^{ k } }\, n _{ i } t _{ i }/{ \sum _{ i = 1}^{ k } }\, n _{ i } \end{eqnarray}$$
where n i is the number of seeds that were germinated in the time i, t i is the time from the start of the experiment to the ith observation, and k is the time of the last germination.
At the end of the germination trials, the seed viability was estimated for four replicates of 25 seeds, using the tetrazolium test (Delouche et al., Reference Delouche, Sill, Raspet and Lienhard1962). The seeds were punctured at the chalazal end, imbibed in distilled water for 24 h to soften the seed coat, and then incubated in 1% tetrazolium solution for 24 h at 30°C under dark conditions (Silveira et al., Reference Silveira, Ribeiro, Oliveira, Fernandes and Lemos-Filho2012).
The data were checked for the assumptions of parametric analyses (Sileshi, Reference Sileshi2012), and because the residual normal distribution was not met, we ran the non-parametric Kruskal–Wallis test. To determine the statistical significance among the groups, the Mann–Whitney test with the Bonferroni correction was used.
Results
Seed coat anatomy, development and histochemistry
The ovules of C. armatum are anatropous and bitegmic (Fig. 1A). The outer integument is two-layered and thicker at the micropyle end (Fig. 1B). The inner integument is single-layered and slightly thicker in the micropyle. In the micropyle, the exostome (Fig. 1B) and endostome are not in the same line, compounding a zigzag.
Longitudinal sections of Chaetostoma armatum. (A) Ovule of a pre-anthesis floral bud. (B) Detail of the previous figure at the micropylar end. (C) Young seed in the immediate post-anthesis stage. (D) Seed, 15 DAA; note the globular embryo. (E) Detail of the micropylar end at 15 DAA; note the cell elongation. (F) Seed, 30 DAA; still showing remnants of nucellar cells in the chalazal end. (G) Seed, 30 DAA in a transmedian section, showing a cordate embryo. (H) Seed, 50 DAA, with embryo; note the lignified cells in the xylem of the rapheal bundle (arrow). (I) Detail of the seed coat of a seed 50 DAA, showing the exotesta. (J) Detail of the micropylar end of a seed at 50 DAA. (K) Mature seed. (L) Detail of the testa and tegmen of a mature seed. (M) Detail of the micropylar end of a mature seed, showing the presence of calcium oxalate crystals in the testa and tegmen (arrowheads). co, Cotyledon; ea, embryo axis; em, embryo; en, endosperm; ex, exostome; hy, hypostase; ii, inner integument; nu, nucellus; oi, outer integument; ra, raphe; te, testa; tg, tegmen.
Few changes were observed in seed size during development (compare Fig. 1A with Fig. 1K). The structure remains bitegmic (Fig. 1C–L), and the formation of the endosperm can be observed (Fig. 1D). The testa is non-multiplicative, and only the exotesta differentiates during development (Fig. 1C–D, F–G); the endotesta is reabsorbed, remaining in the micropylar end (Fig. 1E) and in the raphe as lignified cells (Fig. 1H). The exotestal cells have a thin and pectic–cellulosic outer periclinal face and thick secondary walls in a U-shaped pattern in the other faces; the secondary walls are impregnated with phenolic compounds (Fig. 1H–J).
The tegmen is also non-multiplicative, remaining one-layered during seed development (Fig. 1E, I–L). At 15 and 30 DAA, the seeds have only a few cells in the embryo, changing from a globular (Fig. 1D) to a cordate state (Fig. 1G); at 50 DAA, the embryo is nearly cylindrical (Fig. 1H).
In mature seeds, the phenolic deposition on the inner periclinal face and thickened anticlinal faces is conspicuous (Fig. 1L). The testa and tegmen have calcium oxalate crystals at the micropylar end, which is multistratified (Fig. 1M). Except in the raphe (Fig. 1K), all of the other regions of the seed coat are one-layered. The embryo is well developed and differentiated, with fleshy, slightly curved cotyledons filling most of the seed (Fig. 1K). The mature seed is exalbuminous (Fig. 1K), and the embryo is rich in protein. A summary of the results of the histochemical tests is shown in Table 1 and in the supplementary Fig. S2.
Results of histochemical tests on the ovule of a flower during anthesis, a young seed immediately post-anthesis, a seed 30 DAA and a mature seed of Chaetostoma armatum (see also supplementary Fig. S2). –, Reaction not detected; +, detected reaction
Dye tracking and embryo imbibition
We found a significant increase in seed weight, after soaking, in both control (W= 21; P= 0.036) and punctured seeds (t= 4.7; P= 0.018). The percentage mass increase was 14.3 and 15.7%, respectively, for control and punctured seeds. Both of the dyes penetrated the seed coat of C. armatum in the control and punctured seeds. However, in the control seeds, the dyes penetrated until the inner periclinal face of the exotestal cells, but never beyond it (Figs 2A). The dyes remained confined within the lumen of the exotestal cells (Fig. 2C, E) and never reached the embryo (Fig. 2B, D; supplementary Fig. S1A). In contrast, dye staining was observed in the embryos and inner seed coat layers in the raphe of the punctured seeds (supplementary Fig. S1B).
Permeability of the seeds of Chaetostoma armatum with an intact seed coat (A, B, D, cross-sections; C, E, paradermic sections). (A–C) Permeability test using epifluorescence microscopy. (A) Control (white) of the permeability test; note the absence of autofluorescence. (B) Penetration of Lucifer yellow into the testa; note the barrier provided by the inner periclinal face of exotesta. (C) Lucifer yellow retained in the lumen of the exotestal cells; note the secondary walls with phenolic compounds acting as barriers. (D–E) Permeability test with methylene blue using light microscopy. (D) Penetration of methylene blue into the exotestal cells; note the barrier provided by the inner periclinal face of the exotesta. (E) Methylene blue retained in the lumen of the exotestal cells; as in (C), note the secondary walls with phenolic compounds acting as barriers. Asterisks, secondary walls impregnated by phenolic compounds; cr, crystal; em, embryo; et, exotesta; le, Lucifer yellow; mb, methylene blue.
Mean ( ± SE) germinability (%, bars) of fresh mature seeds of Chaetostoma armatum after 56 d of germination. Pre-germination treatments include control vs. punctured seeds, and GA3 concentrations 250 mM, 500 mM and 1000 mM. The same letters indicate no significant differences.
Germination experiments
We found statistically significant differences in the germination of C. armatum seeds after several treatments (using the Kruskal–Wallis test, H= 23.74, P< 0.0001; Fig. 3). Despite being viable, the control seeds germinated less than 5%. Puncturing resulted in a more than tenfold increase in the germination percentage (52 ± 5.3%). Adding gibberellins did not increase the germination of the control or punctured seeds (Fig. 3). The punctured seeds took on average 13.8 ( ± 0.45) d to germinate. The MGT of the gibberellin-exposed seeds ranged from 25.3 d to 51.8 d, depending on the GA3 concentration.
Discussion
This work provides the first unequivocal report of the occurrence of PY in Melastomataceae and Myrtales. In the dye tracking and embryo imbibition experiment, our data indicate that both of the dyes stained the lumen of the exotestal cells but not the embryo, which remained in a dry state (dormant) even after a significant increase in the total seed weight after soaking in distilled water. After puncturing, both of the dyes penetrated the seed coat and stained the well-developed and differentiated embryo. Nearly half of the punctured seeds germinated, suggesting that the non-germinating seeds are also physiologically dormant (Willis et al., Reference Willis, Baskin, Baskin, Auld, Venable, Cavender-Bares, Donohue and de Casas2014). Therefore, the seeds of C. armatum have physiophysical dormancy (PY+PD), increasing to nine the number of families with this dormancy type.
The report of a water-impermeable seed coat in C. armatum increases the number of families with PY to 19 (for the other 18 families, see Baskin and Baskin, Reference Baskin and Baskin2014). As far as we are aware, C. armatum is the only species in the Myrtales with a water-impermeable seed coat; the closest clade to Myrtales (sensu APG III; Angiosperm Phylogeny Group, 2009) with PY (and also PY+PD) is Geraniaceae (Geraniales) (Willis et al., Reference Willis, Baskin, Baskin, Auld, Venable, Cavender-Bares, Donohue and de Casas2014). The phylogenetic distribution of the seed dormancy types (Baskin and Baskin, Reference Baskin and Baskin2014; Willis et al., Reference Willis, Baskin, Baskin, Auld, Venable, Cavender-Bares, Donohue and de Casas2014) suggests that PD+PY in C. armatum is likely derived from PD, because PD is common in melastomes (Silveira et al., Reference Silveira, Fernandes and Lemos-Filho2013) and PY was absent in the Myrtales (Willis et al., Reference Willis, Baskin, Baskin, Auld, Venable, Cavender-Bares, Donohue and de Casas2014) until this study. Physiophysical dormancy is a derived state in angiosperms and is the less common type of dormancy across all vegetation; our results represent the first report of physiophysical dormancy in a shrub from tropical montane vegetation (see Baskin and Baskin, Reference Baskin and Baskin2014). Further investigation may reveal other species with physiophysical dormancy, particularly those with higher phylogenetic affinities with C. armatum (Fritsch et al., Reference Fritsch, Almeda, Renner, Martins and Cruz2004).
The seeds of C. armatum are dispersed abiotically without the assistance of dispersal agents. Preliminary data indicate that dispersal distance for this species is within the range of a few centimetres (R.C. Ribeiro et al., unpublished data). Thus, a likely adaptive significance of PY+PD is to prevent germination below the mother plant. A horizontal run-off-mediated seed displacement would increase the seed dispersal distance, spreading germination in space and time and, therefore, decreasing the density-dependent mortality and sibling competition. In addition, the studied population of C. armatum shed seeds at the beginning of the dry season, when the conditions for establishment are unfavourable (Silveira et al., Reference Silveira, Ribeiro, Oliveira, Fernandes and Lemos-Filho2012). Physiophysical dormancy would also postpone germination until the next rainy season and maximize fitness when the conditions for establishment become more favourable. PY species have a specific germination season (Baskin and Baskin, Reference Baskin and Baskin2014), but a dormancy break in C. armatum under natural conditions is currently unknown.
PY appears to have originated during periods of increased aridity (Baskin and Baskin, Reference Baskin and Baskin2014). The core Microlicieae (Melastomataceae) originated in the Miocene, nearly 8 million years ago (Simon et al., Reference Simon, Grether, Queiroz, Skema, Pennington and Hughes2009), during the expansion of C4 grasses and, therefore, savannas and grasslands (Beerling and Osborne, Reference Beerling and Osborne2006). PY attains its greater geographical importance in tropical savannas, where nearly 34% of the species have this type of dormancy (Baskin and Baskin, Reference Baskin and Baskin2014). Although fire-triggered germination and dormancy are assumed to be important in seeds that have a physical seed coat barrier to uptake of water in fire-prone environments (Bond and Keeley, Reference Bond and Keeley2005), caution should be taken in sustaining a fire-induced argument. Fire-triggered germination data are scarce for tropical shrubs and herbs (for trees, see Ribeiro and Borghetti, Reference Ribeiro and Borghetti2014), and when they are available, they suggest otherwise (Le Stradic, Reference Le Stradic2012). Second, PY can be broken by fluctuating temperatures without fire (Baskin and Baskin, Reference Baskin and Baskin2000). PD, in turn, may be broken by cold stratification during dry winters, when temperatures can drop to 7°C (Brito et al., unpublished data).
In the water-permeable seeds of Miconia ferruginata (Miconieae, Melastomataceae), water uptake occurs through the raphe (Mendes-Rodrigues et al., Reference Mendes-Rodrigues, Araújo, Barbosa-Souza, Barbosa-Souza, Ranal, Santana and Oliveira2010). Unlike other regions, the micropyle of C. armatum has outer cell layers with pectic–cellulosic walls and an inner layer impregnated by phenolic compounds, suggesting that it is the water gap. The pore is evident, and anatomical analyses also indicate the absence of obstructions to the water, but further studies are necessary to locate the water gap precisely in C. armatum.
In species with PY, it is common to find water-impermeable macrosclereids, because these structures are impregnated with water-repellent substances. Although macrosclereids are absent in the seed coat of Melastomataceae species, the exotestal cells of C. armatum seeds have secondary walls with phenolic compounds in certain faces that are thought to cause water impermeability, as described for numerous species (Baskin and Baskin, Reference Baskin and Baskin2014). However, the outer periclinal faces of these cells are thin and not impregnated, i.e. they are permeable. The U-shaped impregnation pattern explains why C. armatum seeds can acquire water and remain dormant: the accumulation of water is restricted to the lumen of the exotestal cells, but water passage to the embryo is prevented by the phenolic compounds of the inner face of the exotestal cells. If this new reported mechanism occurs in other species, simple imbibition experiments may not be sufficient to determine the seed dormancy types. To accurately determine the mechanisms accounting for dormancy, it may be necessary to perform dye-tracking analyses and to determine whether embryos imbibe water rather than determining whether seeds acquire water. In this sense, ontogenetic and anatomical studies are necessary to understand the seed coat and to interpret unexpected results in germination tests.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0960258515000070
Acknowledgements
We thank L. Bentsink for the invitation to participate in this special issue. We profoundly thank C.C. Baskin and J.M. Baskin for their continuous encouragement and suggestions. The comments by two anonymous reviewers improved early versions of this manuscript.
Financial support
Financial support was provided by the Brazilian Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, process 482720/12), the Fundação de Amparo à Pesquisa de Minas Gerais (FAPEMIG, process APQ-02231/12) and the Pró-Reitoria de Pesquisa da Universidade Federal de Minas Gerais. This study was in partial fulfilment of the requirements for an MSc in Plant Biology by R.C.R., who received a scholarship from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES). D.M.T.O. received a research grant from the CNPq (process 309416/2011-6).
Conflicts of interest
None.
