Pollen morphology of the genus Eremanthus Less . ( Vernonieae , Asteraceae )

(Pollen morphology of the genus Eremanthus Less. (Vernonieae, Asteraceae)). In order to evaluate the signifi cance of the pollen morphology for generic and infrageneric taxonomy of the genus Eremanthus (Vernonieae, Asteraceae), and to provide additional data for its phylogenetic reconstruction, the pollen of 20 of the 23 species of the genus was examined using light and scanning electron microscopy. Acetolysed pollen grains were measured, described, and illustrated using light microscopy, while non-acetolysed pollen grains were analyzed using scanning electron microscopy. Pollen grains of these species are isopolar, oblate-spheroidal in most of the species, more rarely prolate spheroidal or suboblate, subtriangular amb, tricolporate and subechinolophate. Th e variation among quantitative characters does not correlate with the macromorphological subdivision of the genus or with the generic or specifi c limits.


Introduction
Th e tribe Vernonieae Cass.(Asteraceae) contains c. 1500 species distributed mostly in the tropical parts of the world with two major centers of diversifi cation: southeastern Brazil and the central region of Africa (Robinson 1999(Robinson , 2006(Robinson /2007;;Keeley & Robinson 2009).Th e 129 genera are placed into 21 subtribes mainly defi ned by their infl orescence pattern, pollen morphology, chemical composition and chromosome number (Keeley & Robinson 2009).Lychnophorinae is one of the biggest subtribes of Vernonieae, and includes ca. 100 species (11 genera) from the cerrado and campos rupestres of Brazil.Th is subtribe is defi ned by the lack of enlarged nodes or sclerifi ed cells at the base of the styles, usually the extensive presence of a pubescence of T-shaped hairs, presence of sclerifi ed cells and lack of glands on the anther appendages (Robinson 1992(Robinson , 1999(Robinson , 2007;;Keeley & Robinson 2009).
Pollen characters of Asteraceae have been shown to be particularly variable and form useful patterns in the context of phylogenies.Such patterns could be used to support hypotheses of relationships, or provide diagnostic characters for groups at diff erent levels (Wortley et al. 2007, Wortley et al. 2008, Blackmore et al. 2009).Blackmore et al. (2009) noticed that Vernonieae provides an example where a large number of pollen characters are congruent with the relationships presented in the Asteraceae supertree (Funk et al. 2005).Blackmore et al. (2009) provide putative synapomorphies for groups, most of them concerning African clades.Vernonieae pollen grains are usually lophate or sublophate, tricolporate or triporate, with either a continuous or discontinuous punctate or micropunctate tectum, and are with or without spines (Keeley & Robinson 2009; for a detailed description see Blackmore et al. 2009).Th e large pollen diversity of the tribe has been used extensively in taxonomic delimitations at generic and subtribal levels (Jones 1979, 1981, Bolick & Keeley 1994;Keeley & Robinson 2009).Kingham (1976) created six pollen Types (I-VI) for the tribe, based on the study of 85 species (mostly African).Keeley andJones (1977, 1979) also identifi ed six basic Types (A-F) to which additional variations have been added (Robinson 1999).Currently, ten pollen Types are recognised for the tribes (Dematteis & Pire 2008).Using mainly the classifi cation of Keeley and Jones, several authors have analysed and described pollen grains of Vernonieae, mostly of Vernonia s.l.(Keeley & Jones 1977, 1979;Jones 1979Jones , 1981;;Robinson 1987aRobinson ,b,c, 1988aRobinson ,b,c, 1990;;Skvarla et al. 2005;Mendonça & Gonçalves-Esteves 2000;Mendonça et al. 2007aMendonça et al. ,b,c, 2009Mendonça et al. , 2010;;Dematteis & Pire 2008;Angulo & Dematteis 2010).
Th e pollen grains of Lychnophorinae are tricolporate, echinate, sublophate with a perforated tectum continuous between colpi (Type "A") (Robinson 1992(Robinson , 1999;;Keeley & Robinson 2009).Th e endoaperture is usually lalongate (Peçanha et al. 2008).Th e Type "A" pollen is considered a reversion from more strongly lophate ancestors in the Vernonieae, but it is nevertheless one of the most common forms in the tribe and is consistent in many groups such as the Piptocarphinae, Vernonia s.s. and Vernonanthura (Robinson 1990(Robinson , 1992(Robinson , 1999)).Robinson (1999) suggested that due to common reversion of lophate pollen to Type "A", co-occurrence in diff erent groups of the Vernonieae may have little phylogenetic signifi cance.
Th e pollen Type of Eremanthus was initially described by Stix (1960) (based on E. glomerulatus Less.and E. polycephalus (DC.)MacLeish) as Lychnophora-type, and latter called Type "A" by Keeley and Jones (1979).Kingham (1976) assigned the pollen of E. crotonoides (as Vernonia crotonoi-des DC.) to his category VI: grains spherical, tricolporate, micropunctate, subechinolophate tending to echinate.Peçanha et al. (2008) described pollen grains of Paralychnophora bicolor (DC.)MacLeish (as E. bicolor (DC.)Baker) and E. erythropappus (DC.)MacLeish (as Vanillosmopsis erythropappa (DC.)Baker).Th is study revealed some interesting variation of the exine pattern at the generic level: Paralychnophora having a sublophate exine and Eremanthus a subechinolophate one.Peçanha et al. (2008) concluded that their data do not support the inclusion of Vanillosmopsis in Eremanthus (inclusion made by MacLeish in 1987).Such conclusion however cannot be considered valid, since they actually compared pollen grains of Paralychnophora with Eremanthus subg.Vanillosmopsis MacLeish, and not the latter with E. subg.Eremanthus Sch.Bip. as intended.
Th e present study aimed to conduct a comprehensive examination of the pollen morphology in the genus Eremanthus, using LM and SEM, with almost complete species coverage, in order to evaluate the signifi cance of this structure for generic and infrageneric taxonomy and to provide additional data for the phylogenetic reconstruction.

Materials and methods
Pollen grains of 20 species of Eremanthus were examined (see Appendix).Th e samples were obtained from anthers of fl ower buds from specimen kept at the herbaria ESA, HRB, HUEFS, IBGE, MBM, SP, SPF, UB and UEC; acronyms according to Holmgren et al. (1990).
Th e terminology used for descriptions follows Punt et al. (2007), taking into consideration size, form, number of apertures and the ornamentation of the sexine.For the light microscopy (LM) study, the pollen grains were photomicrographed with a Sony Cyber-shot DSC-W7 digital camera coupled with a Zeiss Axiostar Plus microscope.For scanning electron microscopy (SEM), acetolysed and non-acetolysed pollen samples were placed on aluminium stubs covered with carbon tape and sputter-coated with a thin layer of gold palladium (c.150 Å).Th e samples were analysed using a ZEISS DS M960 microscope at Laboratório de Ultraestrutura Celular Hertha Meyer, Inst.de Biofísica (UFRJ) and a JEOL JSM-5800 microscope at Departamento de Invertebrados do Museu Nacional (UFRJ).
Pollen grains were prepared for light microscopy by the acetolysis method of Erdtman (1952), modifi ed by Melhem et al. (2003).Measurements in equatorial view (PD= polar diameter and ED=equatorial diameter) were randomly taken on 25 pollen grains per sample.Measurements of equatorial diameter from polar view (EDPV) and apocolpus were randomly taken on 10 pollen grains distributed on, at least, three diff erent slides.
Statistical procedures were carried out, calculating the arithmetic mean ( X ); the standard deviation of the mean (SD X ), the mean deviation (S X ), the coeffi cient of variation (CV%), the confi dence interval at 95% and the variation range.For the measurements of the other characteristics, such as endoapertures size, layers of exine and pollen grain diameters of comparison material, the arithmetic mean of 10 measurements were calculated randomly chosen from, at least, three diff erent slides (Salgado-Labouriau et al. 1965;Salgado-Labouriau 1973).

Results
The palynological description has been organised according to the following characteristics: size, polarity, shape, apertures and exine surface.Th e results of the measurements are summarized in Tab.1-3, Fig. 1-5.

Discussion
In the present study 20 species of Eremanthus were analysed (87% of the species), including all taxa placed by Robinson (1999) in the genus (except Paralychnophora species).Th e pollen of Eremanthus is isopolar, oblate spheroidal in most of the species, more rarely prolate spheroidal or suboblate, subtriangular amb, tricolporate and subechinolophate, i.e., corresponding to Type «A».Many of the quantitative characters have some variability, however the results show considerable overlap between species.Th erefore, it was not possible to construct a palynological key to identify the species or to defi ne some pollen subtypes.
Th e pollen shape is the only qualitative character that varies in the genus.Eremanthus argenteus MacLeish & H. Schumach.and E. veadeiroensis H. Rob. have prolate spheroidal pollen grains, the former belongs to the subg.Eremanthus and the latter was not included in Eremanthus by MacLeish (1984,1987) because of its densely pubescent stems, loose infl orescence and a large number of fl orets per capitulum (8-11); these three characters are uncommon within the genus.Phylogenetic analyses based on molecular and morphological data (Loeuille 2011) cifi c.Th e quantitative characters vary conspicuously between the three species, a variation that encompasses the subgeneric categories, E. pohlii having more similar pollen grains of the species of subg.Eremanthus than of subg.Vanillosmopsis.
Th e results of phylogenetic analyses (Loeuille 2011) have indicated that E. crotonoides, E. leucodendron and E. pabstii are not closely related to the genus Eremanthus.Again, our palynological data did not show any diff erence between these species and the rest of the genus.
The presence of meaningless differences in pollen between closely related species of the tribe Vernonieae has already been documented.For instance, in the small genus Eirmocephala H. Rob., E. brachiata (Bentham ex Oersted) H. Rob. has a Type «D» and E. megaphylla (Hieron.)H. Rob.Type «A» (Robinson 1987b) pollen.A certain degree of instability of types is also found in Distephanus      Cass.(Robinson & Kahn 1986).Th e case of Eremanthus is slightly diff erent because the only variation is found in the quantitative characters, which are usually more diffi cult to interpret.Th e pollen grains of the species here analysed are typically subechinolophate and not sublophate, i.e., not Type «A» sensu Keeley & Jones (1979) (echinate to sublophate, continuous and microperforate tectum, spines on the ridges of sublophate grains).Th erefore one might consider that the Eremanthus species are a modifi ed Type «A» (subechinolophate).Blackmore (2000) emphasizes the importance of determining at which taxonomic hierarchy level the pollen data are most useful.Th e present study clearly demonstrates that the pollen morphology is taxonomically uninformative at the specifi c level for the genus Eremanthus and probably to a certain degree at the generic level in the subtribe Lychnophorinae.Th erefore, the general idea that pollen morphology is signifi cant in delimitating at specifi c and generic levels into the entire tribe Vernonieae (Angulo & Dematteis 2010) has to be rectifi ed; the situation in some groups (especially the Lepidaploa complex) could not be used as a general rule in the tribe.
The absence of variable qualitative characters and the fact that quantitative characters show a high level of homoplasy limit greatly the use of palynological data to reconstruct the phylogeny of the subtribe Lychnophorinae.Keeley & Robinson (2009) arrived at a similar conclusion for Vernonieae.Nonetheless, because Peçanha et al. (2008) found sublophate pollen in Paralychnophora bicolor, it would be valuable to analyse the pollen grains of more species of that genus and of other genera of the subtribe Lychnophorinae to evaluate how frequent the sublophate pattern occurs and if it could be used in a systematic context.

Conclusions
Pollen Type "A" is present in all the species of the genus Eremanthus.No qualitative variation has been found in different species and the variation of quantitative characters does not correlate with the macromorphological subdivision of the genus at the generic or specifi c limits.Th ese results indicate that palynological data might not be variable enough to be relevant in the delimitation at the specifi c level and probably at the generic level in the subtribe Lychnophorinae.
X -standard deviations of the mean; IC -confi dence interval
EDPV -equatorial diameter in polar view; PAI -polar area index.