The subfamily Barnadesioideae of the Asteraceae consists of nine genera and approximately 90 species. Both molecular and morphological phylogenies indicate that this subfamily is sister to the rest of the family. We have used scanning electron microscopy (SEM) and transmission electron microscopy (TEM) to study pollen of 41 species from all genera of the Barnadesioideae. Three general pollen types are described in the subfamily: Barnadesia-type (Barnadesia, Huarpea), Chuquiraga-type (Chuquiraga, Doniophyton, Duseniella, Fulcaldea) and Dasyphyllum-type (Dasyphyllum and Schlechtendalia). A fourth type, Amaldoa-type, consisting solely of Arnaldoa, is intermediate between the Chuquiraga- and Dasyphyllum-types. These types parallel and confirm findings from previous studies. Psilolophate grains are found only in the Barnadesia-type. Pollen with a cavity (cavea) between pollen wall units in each of the three interapertural regions is present in Barnadesia (Barnadesia-type), Dasyphyllum (Dasyphyllum-type) and Arnaldoa (Arnaldoa-type). The Chuquiraga-type does not have cavate pollen. Intercolpar concavities occur only in the Dasyphyllum- and Arnaldoa-types. In the latter, intercolpar regions are accompanied by pairs of indentations flanking the colpi. The presence of intercolpar concavities in Dasyphyllum and Schlechtendalia, often cited as a synapomorphy for the Barnadesioideae and Calyceraceae, has apparently evolved independently within the sub-family. Chuquiraga pollen exhibits the least derived palynological features in the subfamily. Palynological characters, when placed in the context of current phylogenies for the Barnadesioideae, suggest additional phylogenetic analyses are needed to re-evaluate intergeneric relationships within the subfamily.
Recent phylogenetic studies of the Asteraceae have resolved many systematic issues at higher taxonomic levels in the family (Jansen and Palmer 1987; Jansen et al. 1991; Jansen and Kim 1996; Kim and Jansen 1995; Kim et al. 1992; Karis et al. 1992; Bremer and Jansen 1992; Bremer 1987, 1994). The subfamily Barnadesioideae is now widely regarded as sister to the remaining members of the Asteraceae. Although there have been many efforts to resolve the higher level phylogeny of the family, relationships between the Barnadesioideae and other proposed basal lineages of Asteraceae remain unresolved. In particular, relationships among genera within the par-aphyletic tribe Mutisieae (Cichorioideae) and the subfamily Barnadesioideae are still problematic.
Past classifications of the tribe Mutisieae have included members of the subfamily Barnadesioideae. Bentham (1873) and later Cabrera (1965, 1977) recognized the subfamily as one of the four or five subtribes of the Mutisieae (Barnadesiinae, Gerberiinae, Gochnatiinae, Mutisiinae, and Nassauviinae). Cladistic analyses of morphological and molecular data prompted Bremer and Jansen (1992) to elevate the subtribe Barnadesiinae to subfamilial rank as the Barnadesioideae. The subfamily is defined by a number of morphological synapomorphies (e.g., axillary spines, “barnadesioid” trichomes present on corollas, achenes, and pappus) and absence of a large chloroplast DNA inversion which is present in all other members of the Asteraceae (Jansen and Palmer 1987). Bremer (1994) recognized one tribe within the Barnadesioideae (Barnadesieae) consisting of nine genera (Arnaldoa, Barnadesia, Chuquiraga, Dasyphyllum, Doniophyton, Duseniella, Fulcaldea, Huarpea, and Schlechtendalia) and approximately 90 species.
Extensive pollen studies of the Mutisieae (including the subtribe Barnadesiinae) using light microscopy have been carried out on selected genera by Wodehouse (1928, 1929a, b), Carlquist (1957), Stix (1960), and Parra and Marticorena (1972). Wodehouse (1928, 1929a, b) observed .the lophate grains of Barnadesia and stated that “this genus bears little or no relationship to the Mutisieae but that its affinities are closer to the Vernonieae”. Wodehouse (1929b) also concluded that Mutisieae were polyphyletic. Carlquist (1957) surveyed pollen morphology of enigmatic taxa of the Mutisieae from the Guyana Highlands. Stix (1960) found six types of internal exine patterns in the six genera she studied: Dicoma, Erythrocephalum, Mutisia, Ameghinoa, Oxyphyllum, and Trixis. Parra and Marticorena (1972) recognized eight exine patterns among Chilean genera of Mutisieae (Chuquiraga, Dasyphyllum, Chaetanthera, Mutisia, Gochnatia, Proustia, Trixis, and Leucheria-types).
Later studies utilizing electron microscopy by Skvarla et al. (1977) and Hansen (1991a, b) showed that the most diverse assemblage of pollen grains within the Asteraceae were found in the tribe Mutisieae. These authors noted that Barnadesia and related genera currently placed in the subfamily Barnadesioideae exhibited ultrastructural exine features that were strikingly similar to those found in the Calyceraceae. In particular, they noted a strong resemblance of the wall ultrastructure of Dasyphyllum to Nastanthus (Calyceraceae).
Urtubey (1997) used SEM to distinguish two symmetry patterns in Barnadesia pollen: radiosymmetric (B. corymbosa, B. glomerata, B. odorata, B. parviflora, B. spinosa) and radioasymmetric (B. aculeata, B. arborea, B. blackeana, B. carypophylla, B. dombeyana, B. horrida, B. jelskii, B. lehmannii, B. macbrideana, B. macrocephala, B. polycantha, B. pycnophylla, B. reticulata). Most recently, Urtubey and Telleria (1998) examined pollen morphology of 59 species of Barnadesioideae with light microscopy (LM) and SEM. They recognized three pollen types, constructed a key for their identification, and discussed their phylogenetic significance.
In the current study we survey pollen grain morphology of 41 species of all nine genera of the subfamily Barnadesioideae, including representatives from all subgenera and sections of the three larger genera Barnadesia, Chuquiraga, and Dasyphyllum. Also included for purposes of discussion and comparison is Gamocarpha alpina pollen of the Calyceraceae. Our study extends previous contemporary SEM palynological investigations of this subfamily (Urtubey 1997; Urtubey and Telleria 1998) by incorporating data from freeze fracture SEM and TEM. We use these data in combination with previously published data to examine the similarity between intercolpar concavities and exine types in the subfamily Barnadesioideae and the Calyceraceae. Additionally, we examined the distribution of pollen data on two recently published morphological phylogenies of the Barnadesioideae (Bremer 1994; Stuessy et al. 1996).
Materials and Methods
Pollen Sampling. Pollen was obtained from herbarium sheets (Table 1) for 41 species representing all nine genera of the Barnadesioideae and one genus of the Calyceraceae (Gamocarpha). There is strong evidence based on molecular (Jansen and Kim 1996), morphological (DeVore 1994; DeVore and Stuessy 1995; Pesecreta et al. 1994; Carlquist and DeVore 1998) and phytochemical (Bohm and Stuessy 1995; Bohm et al. 1995) data that the Calyceraceae is sister to the Asteraceae. Furthermore, Gamocarpha possesses pollen features typical for basal members of the family (M. DeVore, Z. Zhao, J. Skvarla, and R. Jansen, unpublished).
Taxon sampling for pollen comparisons. Herbarium acronyms follow Holmgren et al. (1990). Taxonomic circumscriptions and nomenclature follow recent treatments of Arnoldoa (Stuessy and Sagastegui, 1993), Bamadesia (Urtubey, 1999), Chuquiraga (Ezcurra, 1985), Dasyphyllum (Cabrera, 1959), Doniophyton (Katinas and Stuessy, 1997) and Huarpea (Cabrera, 1951).
Light Microscopy. Acetolyzed pollen grains were stained and mounted on glass slides according to Nair (1970) and then examined with transmitted light using a Leitz Wetzlar microscope. The slides are housed in the reference collection at the University of Texas at Austin (TEX).
SEM. Pollen was placed in tapered test tubes, acetolyzed according to the method of Erdtman (1960), screened with fine wire mesh to remove undigested coarse plant fragments (Skvarla 1966), and placed on sucrose pads to remove finer particles (Chissoe and Skvarla 1974).
For whole grain pollen mounts, dehydration was accomplished using 5 min washes in graded ethanol (EtOH) solutions, three 5 min washes in absolute EtOH, and two 5 min treatments in 100% hexame-thyldisilazane (Chissoe et al. 1994). After dehydration, all samples received a sputter-coating with gold for 4–5 min. Finally, pollen grains were examined with a JEOL JSM 880 scanning electron microscope.
Freeze-sectioned pollen grains for structural study were prepared following the method described by Vezey et al. (1994). A drop of Optimal Cutting Temperature (OCT) compound (Tissue-Tek embedding medium) was placed on an IEC Microtome-Cryostat specimen mount (precooled to - 20°C). Within seconds the OCT drop was frozen and a drop of concentrated pollen/water was immediately placed on the frozen OCT drop. The resulting pollen/ice mixture was sectioned with a razor blade into 8–15 juim sections and placed on a precooled boat-shaped specimen mount. Mounts were then warmed to room temperature and placed in a desiccator for several hours to evaporate thoroughly the melted ice. Sections were then sputter-coated with gold and examined with a JEOL JSM 880 scanning electron microscope.
Results and Discussion
Pollen Types in the Barnadesioideae. The distribution of pollen characters for the 41 examined taxa is summarized in Table 2 and in representative photomicrographs (Figs. 1,2). Accordingly, four pollen types could be distinguished. They are outlined in Table 3 and discussed below. It is evident from these tables that the most useful characters for determining pollen types are the presence or absence of lophate surfaces, intercolpar concavities, and microspines.
Barnadesia-type. Psilolophate pollen of Barnadesia exemplifies the Barnadesia-type and is characterized by having many lacunae (Fig. 1I). Most species have irregular (radioasymmetrical) lacunae, however, radiosymmetrical lacunae are found in three taxa: B. odorata, B. parviflora, and B. spinosa. Urtubey (1997) previously reported that Barnadesia can be divided into two groups based on this character. All examined radiosymmetrical pollen grains have 32 lacunae even though their patterns may not be the same. For example, B. parviflora has an equatorial lacuna with four neighboring lacunae, whereas B. spinosa has an equatorial lacuna with six neighboring lacunae. Pollen of Huarpea is smaller than most Barnadesia pollen (Fig. 1B, Table 2), although B. aculeata has the smallest grains in this group.
Dasyphyllum-type. This type is distinctive in the possession of intercolpar concavities between aperture furrows (Figs. 1D, 1H). The feature was first recognized by Wodehouse (1928) and he termed such concavities as intercolpar (Fig. 1H). The intercolpar regions of some species of Dasyphyllum are strongly concave while others are only slightly concave or even somewhat flattened, or perhaps absent as noted by Urtubey and Telleria (1998) for D. donianum, D. infundibulare, D. reticulum and D. velutinum. Dasyphyllum is divided into two subgenera (Cabrera 1959), Archidasyphyllum (with two species) and Dasyphyllum (with 34 species) with two sections (sect. Dasyphyllum (= Microcephala) and sect. Macrocephala). The only species of subg. Archidasyphyllum examined, D. excelsum, has strongly concave intercolpar concavities. In sect. Microcephala (with 24 species) four examined species have strongly concave intercolpar regions and in sect. Macrocephala (with 10 species) pollen of the two examined species have more or less flat intercolpar regions. This suggests concavities may be useful for distinguishing the two sections. Urtubey and Telleria (1998) indicate that intercolpar depressions also characterize Archidasphyllum, most species of sect. Microcephala and only 3 of 11 species of sect. Macrocephala. Urtubey and Telleria (1998) examined 39 species of Dasyphyllum and established two major types and 3 subtypes (of Type 2) based on the strength and weakness of intercolpar depressions. Pollen of Schlechtendalia is a modified dasyphylloid type (Figs. 1D, 2I, 2L). Like Dasyphyllum, the pollen grains have a large equatorial depression in each intercolpar region. Also present in Schlechtendalia are two small depressions on each side of the farrows. Schlechtendalia does not have a cavus above the foot layer as is always noted in Dasyphyllum. This confirms similar observations by Skvarla et al. (1977) and Urtubey and Telleria (1998). In a comprehensive examination of Dasyphyllum (39 species examined) they noted an absence of a cavus only in D. velutinum.
Pollen types in subfamily Barnadesioideae.
Arnaldoa-type. Pollen grains of Arnaldoa (Figs. 1A, 1K, 2A) appear intermediate between Chuquiraga- and Dasyphyllum- types. They resemble the former in lacking distinctive intercolpar concavities and the latter in having them, although modified, occurring as four depressions surrounding each aperture (i.e., paraporal depressions) which are symmetrical to the long axis of the pollen grain (Fig. 1A). Urtubey and Telleria (1998) noted that these depressions can vary from weak to very strong. A cavus is also present in this pollen, as is present in Dasyphyllum.
Chuquiraga-type. The prolate-subspheroidal pollen grains of Chuquiraga (Fig. 1G) are the smallest of all Barnadesioideae species and the pollen is very similar to each other. Fulcadea pollen is somewhat different from the other chuquiragoid members in having wider germinal farows and larger microspines (Fig. 2G, Table 2).
In summary, these four pollen types, based on sculptural features of the pollen wall, essentially agree with the types established by Urtubey and Telleria (1998) who used presence or absence of intercolpar depressions and lophate pollen in Barnadesia to recognize three pollen types. Their “Type I” corresponds to our Barnadesia-type, “Type II” to our Dasyphyllum- and Arnaldoa-types, and “Type III” to our Chuquiraga- type.
Exine Structure. Urtubey and Telleria (1998) suggested that exine layering was an important phylogenetic character for cladistic analyses of the Barnadesioideae. Using LM they recognized one-, two- and three-layered exines based on observations of bleached pollen grains. Our study, which includes freeze fractured SEM and limited TEM, provides an excellent opportunity to extend such observations. Micrographs from fractured (SEM) and sectioned pollen (TEM) indicate that exine structure is not clearly defined in the Barnadesioideae. Indeed, these micrographs reveal somewhat diffuse structural patterns. We recognize three structural patterns.
Single-layer exines—Consist of thick columellae extending through the exine. They are present in lophate grains (i.e., Barnadesia-type, Fig. 2K). As indicated elsewhere (Skvarla et al. 1977), they are also present in intercolpar portions of Dasyphyllum and Schlechtendalia exines (while other portions are more complex). A single layer may also be present in Arnaldoa but freeze-fracture SEM is inconclusive. Similar doubt of Arnaldoa exine structure was expressed by Urtubey and Telleria (1998).
Double-layer exines—Consist of a basal layer composed of thick columellae and an upper layer of highly branched or divided columellae. Exine columellae are rod-like (as is the case with all Barnadesioideae taxa) with the rods close to the pollen surface palisade-like and the rods close to the exine base partly broken up to form minute globular bodies (Fig. 1L). Double-layer exines occur in Chuquiraga, Duseniella (Fig. 2N), Fulcaldea (Fig. 2G) and Doniophyton (see Fig. 23E of Skvarla et al. 1977).
Triple-layer exines—Schlechtendalia (Fig. 2L) and Dasyphyllum (Fig. 2D), in addition to possessing single-layers (see above) these taxa also have complex layering. The basal layer is about 0.5 juum wide and has a highly reduced reticulate layer (some portions of the basal layer are totally devoid of reticulate exine). The medial layer consists of both fine and thickened columellae and the upper layer consists only of short and thick columellae. Similar morphology was observed by TEM in these taxa elsewhere (see Figs. 23B, C of Skvarla et al. 1977). Triple-layered exines were observed by Urtubey and Telleria only in Schlechtendalia.
The three types of exine layers discussed above are in close agreement with those proposed by Urtubey and Telleria (1998) and any disparities have already been noted. Additionally, these data allow general comparisons with pollen of Calyceraceae (to be discussed later).
There may be a biomechanical factor that favors a single-layered exine in lophate pollen and in the intercolpar concavity regions of grains such as Schlechtendalia and Dasyphyllum, which otherwise, are triplelayered. One may assume that triple-layered exines would be more widely distributed within the family if they better fit the functional role of lophate pollen and/or grains with intercolpar concavities. The ultrastructure of Fulcaldea provides a possible clue to the evolution of pollen within the subfamily. Fulcaldea shows a nearly single-layer exine structure. This suggests that the reticulate layer, which is distinct from the columellae, became more developed in derived members of the Barnadesioideae.
Urtubey and Telleria (1998) also suggested that the five exirie sculpture patterns (i.e., granulate sparsely microechinate, microechinate, scabarate microechinate, smooth, and spinulate) in the Bamadesioideae were of phylogenetic significance. We closely examined the ultrastructure of these sculpture patterns to better ascertain their phylogenetic utility. All sculpture patterns, except for smooth exine, are derived from the conjunction of columellae (Fig. 2L-N). The columellae extend up into the apex of the spinulate tip. The various sculpture patterns differ in the relative length of the central columellae and in the degree of concavity among adjacent spinules. Among the five patterns recognized by Urtubey and Telleria (1998), the scabarate microechinate and granulate sparsely microechinate are not different based on our high magnification freeze-fracture SEM and TEM data. The surface of the scabarate microechinate pattern is flattened and Urtubey and Telleria's (1998) example of this pattern is Doniophyton. All of our micrographs of Doniophyton show the triangle spinule sculpture pattern (Fig. 1C, 2E). The granulate sparsely microechinate pattern has continuous nodules and Urtubey and Telleria (1998) use Schlechtendalia as their example of this exine type. Our data (Figs. 1D, 2L) clearly demonstrate that Schlechtendalia has the triangle spinule sculpture pattern. The only difference between the microechinate and spinulate patterns is the size of the spinules.
Exine structure of the Calyceraceae—Pollen ultrastructural studies of Nastanthus (Skvarla et al. 1977) and Gamocarpha (Fig. 2J) indicate that the Calyceraceae have a double exine layer. The basal exine layer consists of thick columellae and the upper exine consists of highly branched or divided columellae (Fig. 2J). This is not homologous with the double exines found within the Barnadesioideae for three reasons: 1) thick columellae extend through the entire exine in Barnadesioideae (Fig. 2L); 2) a reticulate, spongy network extends throughout the exine (Fig. 2M) or is only found in the basal layer in Barnadesidoideae (Fig. 2L); and 3) the reticulate layer and thick columellae are often found together.
Phylogenetic Implications in the Barnadesiodeae. It is extremely difficult to determine homologies, if any, between the subfamily Barnadesioideae and other members of the Asteraceae. This is particularly true for palynological data because the pollen ultrastructure for the subfamily is so distinctive. In particular, the pollen exine columellae layers of all genera are rodlike and more or less divided into fine divisions that appear globular in section (Fig. 2A-I). This is the only palynological characteristic that supports the monophyly of the Barnadesioideae.
Cabrera (1959, 1977) suggested that Dasyphyllum is basal and that it gave rise to Chuquiraga, Barnadesia, Schlechtendalia, and Fulcaldea. Doniophyton is believed to have originated from an ancestral line including Chuquiraga, while Huarpea is derived from Barnadesia.
Bremer (1994) provided the first phylogenetic analysis of the Barnadesioideae based on 22 morphological characters (Fig. 3). His cladogram identified two major clades, one including Schlechtendalia, Doniophyton and Duseniella, and the second including the remaining six genera. Mapping pollen morphology on Bremer's tree (Fig. 3) indicates that many of these characters change multiple times. For example, the intercolpar areas change from convex (Chuquiraga) to concave (Dasyphyllum), then to convex (Fulcaldea), then to concave (Barnadesia) again. This character is lost twice and gained twice on the Bremer phylogeny, which seems very unlikely.
Stuessy et al. (1996) proposed an alternative phylogeny based on 19 morphological characters (Fig. 4). Their cladogram suggested Schlechtendalia to be the basal genus. Pollen characters also change very frequently in this phylogeny. For example, the intercolpar concavities change from many (Schlechtendalia) to none (Fulcaldea) to many (Huarpea).
Phylogenetic Distribution of Intercolpar Concavities. Skvarla et al. (1977) first noted that the Asteraceae and Calyceraceae have intercolpar concavities, and this is one of the major reasons why they suggested a close relationship between these families. Hansen (1992) suggested that the intercolpar concavities of the Calyceraceae and Barnadesioideae may represent a synapomorphy. Our pollen investigations suggest that in both the Barnade-sioideae and Calyceraceae the intercolpar concavities are derived from convex pollen. The only members of the Barnadesiodeae with intercolpar concavities are Dasyphyllum and Schlechtendalia. Dasyphyllum does not occur in a basal position in phylogenies produced by either Bremer (1994, Fig. 3) or Stuessy et al. (1996, Fig. 4) but Schlechtendalia is basal in the Stuessy et al. cladogram. Thus, it is likely that intercolpar concavities have evolved independently within the Barnadesioideae and Calyceraceae.
Phylogenetic relationships within the Barnadesioideae are still unresolved. Palynological characters, when placed within the context of Bremer's (1994, Fig. 3) and Stuessy et al.'s. (1996, Fig. 4) phylogenies, would have had to undergone some difficult character state transformations. However, of the two morphological phylogenies, pollen morphology is more concordant with Stuessy et al.'s hypothesis of relationships (Fig. 4). Palynological data illustrate that further analyses of the Barnadesioideae are needed and incorporating pollen characters can enhance future comparisons.
The present study indicates that based on pollen characters alone, there are three lineages within subfamily Barnadesioideae, each with a distinctive pollen type. Members of the subfamily exhibiting Chuquiraga-type pollen are most likely basal within Barnadesioideae. In contrast, taxa with Barnadesia-type pollen probably represent the most derived lineage within the subfamily. Most importantly, those taxa within the family with intercolpar concavities are apparently not basal. This suggests that intercolpar concavities are derived independently within both the Calyceraceae and Barnadesioideae and should not be viewed as a synapomorphy uniting the Asteraceae and Calyceraceae.
This work was supported by a National Science Foundation grant to RKJ (DEB-9318279). We are grateful to the staff of the Plant Resources Center for use of their facility and to the following herbaria for permission to remove pollen from specimens: K, OS, TEX, UC, and US. We thank Tod F. Stuessy for providing material of the Barnadesioideae. We also thank William F. Chissoe of the Samuel Roberts Noble Electron Microscopy Laboratory, University of Oklahoma, for technical assistance with the SEM and TEM work. This paper represents a portion of the Ph. D. dissertation of ZZ presented to the Graduate School of The University of Texas. He thanks his committee members, B. L. Turner, A. A. Hall, B. B. Simpson, and T. J. Mabry, for their assistance. We also thank H. Robinson and an anonymous reviewer for helpful suggestions on an earlier version of this manuscript.