Pollen morphology of the tribe Hemimerideae: possible evidence of ancestral pollen types and parallel evolution in the basalmost clade of Scrophulariaceae s.str.

Abstract: Pollen morphology of nine species belonging to six genera of tribe Hemimerideae (representing the basalmost clade of Scrophulariaceae s.str.) was studied and illustrated with light microscopy (LM) and scanning electron microscopy (SEM). Pollen grains in Hemimerideae are 3- or 6–8-colpate or 3- or 5–8-colporate, prolate, spheroidal or oblate-spheroidal in shape; they are mainly medium-sized or occasionally small. The outline in polar view is 3-lobed, rounded-3-lobed or 5–8-lobed; the outline in equatorial view is elliptic or orbicular. Exine sculpture is rough, rough-foveolate, foveolate, microreticulate, rugulate, rugulate-foveolate or rugulate-microreticulate. Four major pollen types are recognized, based on original data: 3-colpate (Alonsoa), 6–8-colpate (Diclis, Hemimeris), 3-colporate (Colpias) and 5–8-colporate (Diascia, Nemesia). Within two of these pollen types, two subtypes can be further distinguished based on pollen size, exine sculpture, details of the colpi and endoapertures. Based on pollen morphological data presented here and in our previous studies combined with published molecular phylogenetic data and molecular clock estimates, we conclude that (1) the ancestral pollen type in Scrophulariaceae was 3-colporate with a rather “primitive” exine sculpture; (2) major trends and pathways of further morphological evolution of pollen among the all lineages of Scrophulariaceae were established at the early stages of diversification of the family, about the time of divergence of its main lineages; and (3) the signatures of parallel evolution of the main morphological pollen characters, combined with progressive diversification of exine sculpture patterns, are apparent in all major lineages of the family. Citation: Mosyakin S. L. & Tsymbalyuk Z. M. 2017: Pollen morphology of the tribe Hemimerideae: possible evidence of ancestral pollen types and parallel evolution in the basalmost clade of Scrophulariaceae s.str. — Willdenowia 47: 15–27. doi: https://doi.org/10.3372/wi.47.47102 Version of record first published online on 13 February 2017 ahead of inclusion in April 2017 issue.


Introduction Current opinions on taxonomy and phylogeny of
Hemimerideae and some other early-branching clades of Scrophulariaceae s.str.
The tribe was established in the new series of Ed wards's Botanical Register continued by Lindley (1835), who explicitly credited the authorship to Bentham: "For this valuable contribution we are indebted to George Bentham, Esq. the learned author of the Genera and Species Labiatarum." Initially Bentham included in that 16 Mosyakin & Tsymbalyuk: Pollen morphology of Hemimerideae tribe the following four genera: Angelonia Bonpl. (now placed in Plantaginaceae s.l. or Veronicaceae s.str.: see Olmstead & al. 2001;Tank & al. 2006;Olmstead & al. 2012), Hemimeris (incl. Diascia), Nemesia and "Phyla canthus Nees et Mart." (a typographical error corrected in Bentham 1835: Thylacantha Nees & Mart., now considered a synonym of Angelonia). Further discussion followed in Bentham's Scrophularineae indicae (Bentham 1835) and an article specifically on this tribe (Bentham 1836), in which he recognized six genera (with addition of Diascia and Diclis). Discussion continued later on in his monographic treatments of Scrophulariaceae (Bentham 1846(Bentham , 1876, in which his vision of the tribe was somewhat changed, but not considerably. Baillon (1888) proposed the name Alonsoeae and included in this tribe the genera Alonsoa, Angelonia, Colpias, Diascia, Diclis, Hemi meris and Nemesia. Since Hemimeris is the type of Hemimerideae, the tribal name coined by Baillon was nomenclaturally superfluous when published; however, it is not illegitimate because it is based on the stem of a legitimate generic name, Alonsoa (Art. 52.3 of ICN: McNeill & al. 2012). Consequently, the name Alonsoeae can be used if the tribe is accepted in a different circumscription, excluding Hemimeris. Barringer (1993) described Alonsoeae as a new tribe (attributed to Barringer) including only Alonsoa and Schistanthe; in fact, he re-circumscribed this tribe, and thus its authorship should be still attributed to Baillon (see Reveal 2012). Steiner (1996), after considering morphological and karyological data, proposed to include Colpias in tribe Bowkerieae Barringer (now in Stilbaceae), which contradicts the recent molecular data, and included Alonsoa in Hemimerideae. Kornhall (2004) supported the placement of Colpias in Hemimerideae, based on molecular data and morphology of the oil-secreting trichomes, and placed Cha radrophila Marloth in Stilbaceae. Fischer (2004) placed Alonsoa and Charadrophila in Alon soeae, while leaving Basistemon Turcz., Colpias, Dias cia, Diclis, Hemimeris and Nemesia in Hemimerideae (based mainly on morphological evidence). Takhtajan (1987Takhtajan ( , 1997Takhtajan ( , 2009 accepted Hemimerideae as a tribe of Scrophulariaceae in all recent versions of his system, but in somewhat differing circumscriptions. According to the latest version (Takhtajan 2009), the tribe contains Basistemon (now placed in Plantaginaceae s.l.: see Oxelman & al. 2005;Tank & al. 2006;Olmstead 2012), Colpias, Diascia, Diclis, Hemimeris and Nemesia. However, Alonsoa and the monotypic genus Charadrophila were placed by Takhtajan (2009) in a separate tribe, Alonsoeae (attributed by Takhtajan to Barringer). Molecular data (Olmstead & al. 2001) did not support the segregation of Alonsoa in a tribe separate from Hemimerideae. Doweld (2001) erected the separate family Hemimeridaceae; however, as far as we know, nobody followed that taxonomic concept. Reveal (2012) placed Hemimerideae in the newly established subfamily Hemimeridoideae and subdivided the tribe into two subtribes, Alonsoinae and Hemimeridinae.
All recent molecular phylogenetic studies (Olmstead & al. 2001;Kornhall 2004;Oxelman & al. 2005;Tank & al. 2006;Schäferhoff & al. 2010, etc.) have revealed the clade of Hemimerideae (in the circumscription discussed above; see Olmstead 2012) as the earliest-branching (basal) lineage of Scrophulariaceae s.str., which is sister to the large clade containing all other representatives of the re-circumscribed family. Molecular clock estimates of the initial evolutionary radiation of Scrophulariaceae s.str. and the divergence time of its basal groups remain controversial (Bremer & al. 2004;Wikström & al. 2001; see an overview and new assessments in Datson & al. 2008). The Hemimerideae clade, in any case, represents an ancient lineage within the family (Scrophulariaceae s.str.), the most recent common ancestor (MRCA) of which existed not later than 42 -47.5 million years ago (the Eocene or earlier epochs) or most probably even earlier (Datson & al. 2008).

Importance of Hemimerideae for understanding pollen evolution, and objectives of the present study
Representatives of Hemimerideae are also interesting from the viewpoint of their life forms and chromosome evolution (Steiner 1996;Datson & al. 2008), pollination syndromes (Kampny 1995;Renner & Schaefer 2010) and some other aspects (Steiner 2006(Steiner , 2009. In particular, some species of that group are pollinated by specialized oil-collecting bees (Buchmann 1987;Steiner 1990;Steiner & Whitehead 1990, 1991, 2002Renner & Schaefer 2010) and (in Hemi meris) possess a very peculiar form of heterostyly (Pauw 2005). Members of this tribe are also biogeographically important and morphologically diverse, especially in the Cape flora of S Africa (Manning & Goldblatt 2012; Snijman 2013).
However, published data on pollen morphology of representatives of Hemimerideae are very scarce. Erdtman (1952), using light microscopy only, studied pollen grains of Alonsoa acutifolia Ruiz & Pav., Diclis petiolaris Benth., D. reptans Benth. and Nemesia affinis Benth. Roux (1986) provided scanning electron photomicrographs of pollen grains of Alonsoa peduncularis (Kunze) Wettst. (Schistanthe peduncularis Kunze) without descriptions and discussion. Because of that, the palynomorphological data provided by Roux (1986) cannot be used in our comparative analysis below. We are not aware of any other specialized studies of pollen morphology of the tribe.
The basalmost (earliest-branching) phylogenetic position of Hemimerideae within Scrophulariaceae makes this group very interesting and promising for revealing the patterns and trends of morphological pollen evolution in the family, especially in comparison with pollen morphology data obtained for other early-branching lineages of Scro phulariaceae as summarized in Tsymbal yuk & Mosyakin (2013) and Mosyakin & Tsymbalyuk (2015a, 2015b. Willdenowia 47 -2017 Because of that, our main objectives for the present publication were to study pollen morphology patterns of selected species of Hemimerideae representing all genera of the tribe (at least those now definitely placed here), to compare the obtained data with the data available on other taxa of Scrophulariaceae and to try to reconstruct the main trends and some general patterns of pollen evolution in the early-branching lineages of the family.

Material and methods
Pollen from nine species, belonging to all six genera of Hemimerideae, was sampled in the herbarium of the Missouri Botanical Garden, St. Louis, Missouri, U.S.A. (MO). Pollen from one species of Nemesia was sampled in the herbarium of the M. G. Kholodny Institute of Botany of the National Academy of Sciences of Ukraine, Kyiv (Kiev), Ukraine (KW; herbarium codes according to Thiers 2016+). The investigated specimens are cited according to the label information, sometimes with changes in the order of data, for consistency.
Pollen morphology was studied using light microscopy (LM) and scanning electron microscopy (SEM), following the techniques described in our previous articles (Mosyakin & Tsymbalyuk 2015a, 2015b and briefly reported here. For LM (Biolar, × 700), the pollen was acetolysed and mounted on slides following Erdtman (1952). For size determinations, 20 measurements were taken along the polar (P) and equatorial (E) axes for each species.
For SEM (JSM-6060LA), pollen grains were treated with 96 % ethanol, then sputter-coated with gold at the Center of Electron Microscopy of the M. G. Kholodny Institute of Botany. The obtained micrographs were minimally edited with Adobe Photoshop 6.0 to enhance the images, with no alteration of the images themselves.
Terminology used in our descriptions of pollen grains mainly follows the glossaries by Punt & al. (1994) and Tokarev (2002), with some minor adjustments.

Pollen types and subtypes
Pollen grains in the studied taxa can be subdivided into four basic types, based on their aperture types; two types can be additionally subdivided into two subtypes each. The subtypes are distinguished mainly according to the pollen size, exine sculpture and details of colpi and endoapertures.

Descriptions of pollen grains
Genera and species within genera are listed alphabetically. Main pollen morphology characters are additionally summarized in Table 1.

Comparative pollen morphology of genera of Hemi merideae
Pollen grains in Alonsoa, in particular in A. unilabi ata, are 3-colpate with a rugulate-microreticulate exine sculpture. They have the widest colpi among pollen grains of the studied species of the tribe, and a distinctly granulate colpus membrane (a unique feature in Hemi merideae).
Pollen grains of Diclis ovata, the only species of Diclis studied by us, are small-sized, 6-colpate, with foveolate, microreticulate exine sculpture. Erdtman (1952) described pollen grains of D. petiolaris as 4-or 5-colpate and those of D. reptans as 6-or 7-colpate, small in both species. Judging from these data, species of the genus differ in the number of colpi, or at least demonstrate a considerable variation of that character. Additional palynomorphological studies in this genus are needed.
The two studied species of Hemimeris are similar in their type of apertures (6-or 7-colpate in H. racemosa Willdenowia 47 -2017 and 6 -8-colpate in H. sabulosa), pollen size and exine sculpture; they differ, however, by their colpi structure: in pollen grains of H. racemosa colpi are wider (2 -2.7 μm) with acute ends, whereas in H. sabulosa colpi are narrower (0.7 -1.6 μm) with acute and rounded edges. Further studies are needed to reveal any other distinctive characters of species of the genus.
In general, pollen grains of Diclis and Hemimeris are similar in their aperture types, but they differ somewhat in their size, thickness of exine, distinct versus indistinct columellae of the infratectum, and characters of their exine sculpture.
Pollen grains in Colpias are 3-colporate, with rough and foveolate exine sculpture, clearly differing by these characters from those of other genera of the tribe. The two studied species of Diascia are similar in having the 6 -8-colporate (6-or 8-colporate in D. capsularis and 6-or 7-colporate in D. elongata) aperture type, rugulate-foveolate and rugulate-microreticulate exine sculpture and rather wide colpi; they differ, however, by some elements of their colpi structure: in D. capsularis colpi ends are rounded, whereas in D. elongata colpi ends are acute and pointed. It should be noted that the two species that we sampled represent the two currently recognized sections within the genus, D. sect. Diascia (D. elongata) and D. sect. Racemosae (D. capsularis) (Hilliard & Burtt 1984). However, the pollen diversity within the genus could be much greater than revealed in the present study.
Pollen grains of the studied species of Nemesia (N. cheiranthus and N. strumosa) are distinguished by their size, wide colpi, clarity and structure of endoapertures, and exine sculpture. Pollen grains of Nemesia show similarities to those of Diascia by their aperture types, pollen grain size, and sculpture of aperture membranes, but differ in the structure of apertures and exine sculpture characters (see Table 1).

Comparison of palynomorphological and molecular phylogenetic evidence
Judging from molecular phylogenetic data (Oxelman & al. 2005), Diascia and Nemesia are included in one clade, which is weakly supported as sister to all other members of Hemimerideae. The close links of these two genera are also confirmed by palynomorphological characters: the same pollen type IV, characterized by 5 -8-colporate pollen, and similar exine sculpture patterns (rugulate, rugulate-foveolate, rugulate-microreticulate and/or microreticulate).
In turn, Colpias may be sister to the clade that includes the genera Alonsoa + (Diclis + Hemimeris) (Oxelman & al. 2005). Our data are consistent with this sug-gestion and demonstrate that 3-colporate pollen grains of Colpias are more similar to (but also distinguishable from) 3-colpate pollen of Alonsoa, but easily distinguished from 4 -8-colpate pollen observed in Diclis and Hemimeris.
According to molecular phylogenetic data (Oxelman & al. 2005), Diclis and Hemimeris are sister members of the same terminal clade of the tribe. Palynomorphological data also suggest a close affinity of Diclis and Hemi meris because they share 4 -8-colpate pollen. Alonsoa is probably sister to the Diclis + Hemimeris clade; however, it differs in having 3-colpate pollen, at least in the two species studied here and by Erdtman (1952). In general, the clade that includes Alonsoa, Diclis and Hemimeris is characterized by colpate pollen, as opposed to colporate  Main trends of pollen evolution in early-branching Scrophulariaceae Furness & Rudall (2004) indicated that there is a general trend, both in angiosperms in general and in their main clades, toward an increase in pollen aperture number, suggesting that aperture number is under strong selection pressure. A similar pattern is observed in Scrophu lariaceae and it is best manifested in taxa with advanced pollination syndromes (specialized entomophily), which demonstrate a trend toward developing more apertures than less specialized taxa. However, morphological patterns in early-branching clades do not always indicate ancestral character states (see Crisp & Cook 2005) and ancestral character state recognition often needs a very careful approach (Cunningham 1999;Cunningham & al. 1998). Nevertheless, when recurrent patterns are observed in various early-branching clades of a particular group (or a phylogenetic clade), one can assume that these patterns may indicate possible ancestral character states. The recent attempts at reconstruction of the early evolution of pollen grains of angiosperms (Wortley & al. 2015) confirm these predictions. Thus, reconstructions of ancient pollen patterns in Scrophulariaceae are in agreement with those ideas. Our data demonstrate that pollen grains of the studied species of Hemimerideae are characterized by colpate and colporate aperture types. We can assume that pollen with the 3-colporate type of apertures and rough and foveolate exine sculpture (similar to the characters observed in Colpias) was ancestral in Hemimerideae and probably in Scrophulariaceae in general. Further evolutionary trends manifested in the tribe are the increase in the number of colpi leading to the formation of 5 -8-colporate pollen, such as revealed in Diascia and Nemesia. Pollen grains of representatives of these genera have rugulate, rugulate-foveolate, rugulate-microreticulate and microreticulate exine sculpture patterns (the last two with indistinct muri). This can be seen as an evolutionary bridge to more advanced and more typical rugulate-microreticulate and microreticulate sculpture with well-developed muri, as observed in Alonsoa and many members of crown clades of Scrophulariaceae. Further reduction of endoapertures (ora) leads to development of 4 -8-colpate pollen grains (apertures are represented by colpi only), like those in Diclis and Hemimeris. Further evolutionary changes through reduction in the number of colpi may result in formation of the 3-colpate type, as that seen in Alonsoa. However, the type of pollen observed in Alonsoa may also develop from the 3-colporate type by reduction of ora.

Conclusions
The palynomorphological data obtained for members of Hemimerideae agree with the results of molecular phylogenetic studies. In particular, close relationships of the two sister groups, Diascia + Nemesia and Dic lis + Hemi meris, are supported by our palynomorphological studies.
As we have already demonstrated (Mosyakin & Tsymbalyuk 2015a, 2015b and in the present article), in all main basal clades of Scrophulariaceae we see a recurrent pollen morphological pattern, namely representatives of early-branching lineages within these clades (Colpias in Hemimerideae, Androya in the Leu cophylleae / Myoporeae / Androya clade, Buddleja L. in Buddlejeae and Freylinia Colla in Teedieae) usually have 3-colporate pollen with rather "primitive" (smooth, rough and/or foveolate) exine sculpture. In the core Scrophulariaceae clades (Scrophularieae and Limosel leae) we also often observe 3-colporate pollen, but usually with more advanced sculpture.
That peculiar pattern, in our opinion, supports our hypothesis of the ancestral pollen type in Scrophulari aceae (Mosyakin & Tsymbalyuk 2015a, 2015b. Judging from pollen morphological evidence compared with molecular phylogenetic data and molecular clock estimates mentioned above, we may conclude that (1) the ancestral pollen type within the Scrophulariaceae was 3-colporate with a rather "primitive" exine sculpture; (2) main trends and pathways of further morphological evolution of pollen in all lineages of Scrophulariaceae had already formed at the early stages of diversification of the family, about the time of divergence of its main lineages; (3) because of that we observe now in all main lineages of the family the signatures of parallel evolution of the major morphological pollen characters, combined with progressive diversification of exine sculpture patterns.
Head Curator of the National Herbarium of Ukraine (KW; herbarium of the M. G. Kholodny Institute of Botany, National Academy of Sciences of Ukraine), for their cooperation and assistance in extracting pollen samples from herbarium specimens. The kind help and cooperation of Dmytro O. Klymchuk, Head of the Center of Electron Microscopy (M. G. Kholodny Institute of Botany, National Academy of Sciences of Ukraine), are greatly appreciated. Thanks are due to Victoria McMichael and Mary Stiffler (Peter H. Raven Library, Missouri Botanical Garden) for providing copies of some publications not available from other sources. The authors are grateful to two anonymous reviewers for their detailed reviews of our manuscript, valuable comments and suggestions. The editorial comments of Nicholas Turland are also greatly appreciated.