The tribe Mutisieae (excluding Barnadesieae) traditionally comprises 84 genera and approximately 900 species in three subtribes: Gochnatiinae, Mutisiinae, and Nassauviinae. We examined whole and fractured pollen grains of 51 genera from these subtribes by scanning electron microscopy (SEM) and light microscopy (LM). Additionally, we also examined 11 genera (Adenocaulon, Berardia, Brachylaeana, Cratystylis, Dipterocome, Eriachaenium, Gymnarrhena, Hesperomannia, Hoplophyllum, Tarchonanthus, and Warionia) whose tribal positions have been controversial. We present detailed tables of pollen characters for each taxon and 13 plates of SEM photos ofrepresentative taxa. We also provide limited discussion of pollen variation in the subbribes Gochnatiinae, Mutisiinae, and Nassauviinae and the tribal and subfamilial placement of the 11 problematic genera.
The tribe Mutisieae, one of the basal lineages in the Asteraceae (Bremer, 1987, 1994; Jansen and Kim, 1996), is extremely diverse in morphology and biogeography. Classification of the tribe largely began with the system of Bentham (1873), who recognized five subtribes: Barnadesiinae, Onoseriinae, Gochnatiinae, Gerberiinae, and Nassauviinae. Bremer and Jansen (1992) elevated the Barnadesiinae to subfamilial rank as the Barnadesioideae based on both morphological and DNA data. This subfamily is defined by a number of morphological synapomorphies, including the presence of axillary spines and peculiar “barnadesioid” hairs on corollas, achenes and pappus, and a distinctive pollen morphology (Gamerro, 1985; Urtubey, 1997; Urtubey and Telleria, 1998; Zhao et al., 2000). The other three subtribes have been maintained within the heterogeneous tribe Mutisieae until recently when Panero and Funk (2002) elevated several groups to subfamilial and tribal status.
The tribe Mutisieae (excluding Barnadesieae) comprises 84 genera and approximately 900 species. Most genera of the tribe are from the New World, mainly from Central and South America, while 11 genera are distributed in Africa and Madagascar, and 12 genera in Asia. The tribe has many genera that are monotypic or that have relatively few species occurring in restricted areas that are sometimes completely isolated from their closest relatives. The tribe has very few weedy species and only a few species are cultivated.
Pollen of the Mutisieae has been included in several broad floristic and palynologic studies (Wodehouse, 1929a, b; Carlquist, 1957; Stix, 1960; Askerova, 1970; Reusser, 1971; Skvarla et al., 1966, 1977; Dimon, 1971; Parra and Marticorena, 1972; Crisci, 1974; Markgraf and D'Antonio, 1978; Wigenroth and Reusser, 1983; Moreira et al., 1981; Nair and Lawrence, 1985; Hansen, 1991a, b; Jones et al., 1995; Perveen, 1999; Rull, 2003; Lin et al., 2005), as well in treatments of individual genera. This atlas utilizes SEM (scanning electron microscopy) and LM (light microscopy) to present pollen wall patterns that will be helpful in: 1) providing a framework for understanding evolutionary history of early branches of the Asteraceae, 2) resolving the relationships among sub tribes, and 3) determining the taxonomic placement of some problematic genera.
Taxon sampling for pollen comparisons. Herbarium acronyms follow Holmgren et al. (1990).
Materials and Methods
Pollen grains of 62 genera ( 68 species) were examined, 23 genera (25 species) from Gochnatiinae, 15 genera (19 species) from the Mutisiinae, 13 genera (14 species) from Nassauviinae, and 11 genera (12 species) whose tribal positions are uncertain (Table 1). Pollen was removed from herbarium sheets and acetolyzed according to the method of Erdtman (1960). Preparation for LM and SEM was as described in our previous investigation (Zhao et al., 2000). All measurements are based upon an average of five pollen grains.
Results and Discussion
Tables 2–5 summarize various features of the pollen of the 62 genera examined and constitute an extension of those typical palynological features (size, shape, exine thickness, etc.) used in other studies of the Mutisieae. They are supplemented by SEMs of fractured surfaces through the pollen wall. Figures 1–11 illustrate SEM photos of representative species from the subtribes Gochnatiinae, Mutisiinae, and Nassauviinae of the Mutisieae. Whole pollen grains are depicted in SEM, primarily in equatorial (apertural or side) view and general shape parameters were determined from this orientation with light microscopy. However, depending upon conditions under which the measurements were taken (e.g., dry or rehydrated pollen removed from herbarium sheets, acetolyzed pollen, pollen in the vacuum of the SEM, etc.), overall shape may vary with results from other studies. Polar and partial polar views are indicated in Figures 1e (Ainsliaea), 11 (Chimantaea), 2b (Cnicothamnus), 21 (Gochnatia), 3g (Oldenburgia), 4e (Pleiotaxis), Sc (Wunderlichia), and 7e (Mutisia). With fractured grains an attempt was made to standardize the area of SEM photography in order to describe with consistency the heights of the proximal and distal columellae. This was essentially at the midpoint on the equatorial surface. When fractured walls are examined in different areas such as near aperture regions, heights become greatly skewed (see as examples Figs. 1f of Ainsliaea acerifoliaa with an explanation in table legend and Fig. 2g of Gochnatia in Roque and Silvestre-Capelato, 2001).
With the possible exception of the Vernonieae, pollen of Mutisieae has been the most intensively investigated of all Compositae tribes. Given that this paper is an atlas of Mutisieae pollen we do not discuss in the details the pollen morphology in detail. However, in concert with Tables 2–4 we provide below overall summaries of some of the more outstanding morphological characteristics of each subtribe.
Mutisiinae. See footnotes at bottom of Table 2.
Gochnatiinae. In addition to references provided earlier, pollen of the Mutisieae has been examined by Barroso and Maquire, 1973; Marticorena and Parra, 1974; Moreira et al., 1981; Robinson, 1991; Chissoe et al., 1994; Ortiz and Coutinho, 2001; Roque and Silvestre-Capelato, 2001; Sancho et al., 2005; Rodriquez et al., 2004; and Telleria and Katinas, 2005. Spines, while not prominently developed in any of the three subtribes of Mutisieae, are best expressed in the Gochnatiinae. Most taxa are considered as barely spinate (ca. 1 μm in height and most favorably viewed by light microscopy) loosely following the terminology of Erdtman (1952). The largest spines occur on pollen of Achyrothalamus (Figs. 1a, 1b; Ortiz and Couthino, 2001), Erythocephalum (Figs. 2g, 2h), Hochstetteria (Fig. 3c) and Wunderlichia (Figs. 5b, 5c; Wodehouse, 1929a). Spines in these taxa also have comparatively wide bases where they unite with their respective exine surfaces.
Columellae structure, as viewed in fractured sections by SEM, always shows at least two levels (layers) with each separated by a horizontal, often not clearly defined, internal tectum layer (Skvarla and Turner, 1966; Skvarla et al., 1977). Multilevel columellae are characteristic of all three Mutisieae subtribes, and to our knowledge, all pollen studies in the tribe, without exception, support this observation. As indicated in Table 2, in most taxa the proximal columellae are greater in height than the distal columellae immediately beneath the exine surface. The exceptions appear to be in Gochnatia argentina, Nouelia, Pertya, Pleiotaxis, Stenopadus and Stifftia where proximal and distal columellae are approximately equal. However, as indicated in Table 2 (as well as in Tables 3 and 4) it is difficult to calculate these parameters precisely. These parameters are even more difficult to calculate when the distal columellae appear to greatly ramify such as in Old enburgia (Fig. 3h) and other taxa designated with “ML.”
A columellae character not taken into consideration in this atlas format, and one possibly of significance in a more detailed SEM study, is the thickness of the columellae. Frequently, the proximal and distal columellae appear to be equal in thickness or often, the proximal columellae are thicker (wider) than the distal columellae. Sometimes, proximal columellae thickness appears to be highly exaggerated (i.e., extremely thickened) as illustrated by Nouelia (Fig. 3e), perhaps of importance beyond the scope of this report.
Yet another characteristic of potential importance in a more in depth study is that of exine surface texture. In general, in the Gochnatiinae the exine surface of at least some pollen grains appears to be more perforated (see for example Achyrothalamus, Fig. 1a and Erythrocephalum, Figs. 2d, 2f) than in Mutisiinae and Nassuviinae.
Mutisiinae. This subtribe has been studied by Erdtman, 1952; Southworth, 1966, 1983; Telleria and Katinas, 2004; and Telleria and Forcone, 2002, as well as by workers cited earlier. Pollen size is greatest in Mutisiinae, although only slightly more so than in the Gochnatiinae. With the exceptions of Eurydochus (Figs. 6 c, 6d) and Glossarion (Figs. 6i, 6j) the surface is spinose (spines less than 1 μm in height). Height of the proximal columellae is greatest in this subtribe and reaches considerable length in genera such as Chaetanthera (Figs. 5h, 5j), Gerbera (Figs. 6h), Hyaloseris (Fig. 7b), and Mutisia (Fig. 7f). Columellae thickness (width) is comparatively less variable than in Gochnatinae with the proximal columellae always thicker that the distal columellae. The most recent study of the Mutisiinae clearly indicated that infrageneric pollen morphology was uniform and that species within genera were usually similar to each other (Lin et al., 2005).
Nassauviinae. Pollen of this subtribe was studied extensively by Wodehouse, 1929b; Hernandez, 1966; Crisci, 1971a, 1971b, 1974, 1976; Crisci and Marticorena, 1978; Fagundez, 2003; Cabrera and Dieringer, 2003; and Telleria et al., 2003 (also see earlier cited references). Of the three subtribes, pollen size is clearly the smallest in Nassauviinae (Wodehouse, 1929b). Furthermore, all pollen is spinose. These characters of small size and spinose surface were key in leading Wodehouse (1929b) to distinguish this subtribe from the Gochnatiinae and Mutisiinae. In Nassauviinae there tend to be more columellae with proximal and distal segments equal than in the Gochnatinae and Mutisiinae, although, like these two subtribes the most dominant feature is that the height of proximal columellae is greater than the distal columellae. Crisci (1974) indicated several possible variations in heights of proximal and distal columellae and Cabrera and Dieringer (2003) clearly showed in various species of Acourtia differences in the heights of the proximal and distal columellae and established types on this basis. Telleria et al. (2003) indicated that exine types, also based on differences in heights of proximal and distal columellae, overlap in the Gochnatiinae and Mutisiinae. In contrast, Nassauviinae has distinct types not noted in these other subtribes. While not quantitativey measured, the overall height of columellae in this subtribe appears to be shorter than in Gochnatiinae and Mutisiinae. Further, thickness of the proximal and distal columellae is often approximately equal. The internal tectum layer in the Nassauviinae is considerably better developed than in either of the other subtribes, often distinctly demarcating proximal and distal columellae.
Nassuviinae. See footnotes at bottom of Table 2.
Taxa of Uncertain Tribal Placement
Adenocaulon. This genus has been positioned in four different tribes: Heliantheae (Bentham, 1873), Inuleae (Hoffmann, 1890), Senecioneae (Cronquist, 1955) and Mutisieae (Ornduff et al., 1967; Grau, 1980; Bittmann, 1990; Kim et al., 1998). Some features, particularly the rather deeply lobed, sometimes bilabiate corolla and the calcarate anthers, contradict a position within any of the asteroid tribes, and the testa epidermis (Grau, 1980) is consistent with a relationship to the Mutisieae. Hansen (1991a) did not include Adenocaulon in the Mutisieae because it had several features that are unknown in the group. These include: 5 or 4-lobed florets with rigid, short petals; very widened styles; dub-shaped achenes; anthers with minute tails; petals and styles indistinctly hairy; and an involucre of very few bracts. Kim et al. (1998) placed Adenocaulon in the Nassauviinae based on sequences of the chloroplast gene ndhF. All of the pollen characters described for this genus (Stix, 1960; Liens, 1969; Heusser, 1971; Skvarla et al., 1977) are consistent with its placement in the Nassauviinae (Figs. 11g,11h; Tables 4–5).
Berardia. Following Bentham (1873) and Hoffman (1890), Grau (1980) also favored a position of Berardia within the Mutisieae. Hansen (1991a) excluded the genus from the Mutisieae in his review of the tribe. Karis et al. ( 1992) placed Berardia near the genus Carlina in the Cardueae. Bremer (1994) also accepted Berardia in the Cardueae, but did not assign it to any particular subtribe. More recently a supertree of the Asteraceae clearly positioned Berardia with the Cardueae (Funk et al., 2005). Our pollen data is insufficient (Table 5; Figs. 11i, 11j) but the large spines and coarsely granulate exine surface do not favor a position within Mutisieae.
Brachylaena and Tarchonanthus. Pollen morphology has been examined by several workers including Erdtman (1952), Liens (1969), Skvarla et al. (1977) and Cilliers (1991) and is described in Table 5 and Figs. 11k, 13b, 13c. The uniform and equal lengths of the proximal and distal columellae resemble Nassauviinae of the Mutisieae. A cpDNA restriction site analysis indicated that these two genera form the distinct tribe Tarchonantheae at or near the base of the subfamily Cichorioideae (Keeley and Jansen, 1991). This tribe, along with the Cardueae and Dicomeae, is now placed in the subfamily Carduoideae (Panero and Funk, 2002; Funk et al., 2005).
Cratystylis. This genus was unassigned to a tribe in the subfamily Cichorioideae by Bremer (1994). A phylogenetic analysis based on morphological and phytochemical data also concluded that the genus represents an isolated lineage within the Cichorioideae (Anderberg et al., 1992). The unique pollen morphology (Anderberg et al., 1992; Table 5; Figs. 11L. 12a), characterized by the single but complex columellae layer (Fig. 12a) removes it from a position in the Mutisieae and more closely places it in a tribe such as Inuleae. However, a recent molecular phylogeny based on three cpDNA markers provided strong support for the placement of Cratystylis in the tribe Plucheeae of the subfamily Asteroideae (Bayer and Cross, 2003).
Taxa with uncertain tribal placement. See footnotes at bottom of Table 2.
Dipterocome. This is a monotypic genus that Bremer (1987) placed in the Mutisieae (following work of Praglowski and Grafstrom, 1980). Pollen morphology (Table 5), especially prominent spines and questionable second level of columellae (Fig. 12c), suggests that Diptercome is not related to any examined members of the Mutisieae.
Eriachaenium. This Patagonian genus was originally positioned in the primarily African tribe Calenduleae. However, no characters were provided supporting its relationships to this tribe, although the very short style branches may have contributed to this placement. Eriachaenium was transferred to the Inuleae-Adenocauliinae by Cabrera (1961). No explanation was given for this placement, although certain features such as caudate anthers and wooly pubescence of the leaves and flowers give the genus a superficial resemblance to the Inuleae. In contrast, Merxmuller et al. (1977) excluded it from the Inuleae. Hansen (1991a) noted that Eriachaenium has the mutisioid epidermal pattern of the corollas and suggested that it may belong to the subtribe Nassauviinae. Our SEM (Figs. 12d, 12e; Table 5) suggests possible placement in the Mutisiinae.
Gymnarrhena. This genus has been classified in the Inuleae because of the similarity of its habit to that of Geigeria of the same tribe (Bentham, 1873). Bremer (1994) classified the genus in the subfamily Cichorioideae but did not assign it to any tribe. Panero and Funk (2002) placed Gymnarrhena in a monotypic tribe and subfamily that is sister to a large clade including Cichorioideae, Corymboideae, and Asteroideae. Gymnarrhena pollen, as noted elsewhere (Skvarla et al., 1977), is difficult to place but it does not suggest Mutisieae. The multilevel columellae (Table 5; Fig. 12g) are found in the Mutisieae but also elsewhere (for example, Anthemideae and Cardueae) and the distinctly spinate surface (Fig. 12f) is rare, if at all present in Mutisieae.
Hesperomannia. This genus is endemic to the Hawaiian Islands, and was placed in the Gochnatiinae by Cabrera (1977). Marticorena and Parra (1975) suggested that the pollen morphology (along with Moquinia) indicated that it was isolated within the Mutisieae. Our SEMs (Figs. 12h-k) of pollen of H. arborescens and H. lydgatei, particularly, fractured sections (Figs. 12i, 12k), strongly support a position within the Vernonieae. This conclusion is based on a single columellae layer that is distally highly branched. Recent DNA studies (Kim et al., 1998) positioned the genus within Vernonieae and indicated that it was closely allied to African members of Vernonia, subsection Strobocalyx.
Hoplophyllum. This South African genus is usually classified in the Vernonieae. Bolick (1978) noted that the genus is morphologically aberrant within the Vernonieae. Bremer (1994) positioned it in the subfamily Cichorioideae but left it unassigned as to a tribe. A chloroplast DNA phylogeny based on ndhF gene sequences placed Hoplophyllum sister to Eremothamnus in the tribe Arctoteae (Karis et al., 2001). The entirely spinate pollen surface (Fig. 12L; Robinson, 1994) and single layer of columellae (Fig. 13a; Table 5) most definitely removes it from Mutisieae, Vernonieae, or Cichorioideae.
Warionia. The genus is an herb endemic to the Sahara of Africa. It has coarsely lobed leaves and large discoid capitula. Warionia was placed in the Mutisieae by Cabrera (1977), although affinities with the Cardueae have been suggested. Hansen (1991a) excluded Warionia from the Mutisieae. In the cladogram of Karis et al. (1992), the genus appeared as an independent branch between the Cardueae and the vernonioid complex. Like the latter, Warionia has vernonioid styles, but it is not readily assigned to any of the tribes in the Vernonioid complex, and it cannot be accommodated in the Cardueae (Dittrich, 1977). Bremer (1994) did not assign a tribal position to Warionia. Molecular phylogenies placed W arionia sister to Gundelia in the tribe Gundelieae, and this tribe is sister to the Cichorieae (Panero and Funk, 2002; Funk et al., 2005). Dimon (1971, Plate V, Figs. 3–5) depicted pollen of Warionia as having proximal columellae considerably greater in height than distal columellae in agreement with our SEM fractured image (Fig. 13 f) and data (Table 5). Further, the highly perforate exine surface (Figs. 12, d, 12e) resembles some Gochnatinae.
This work was supported by a National Science Foundation grant to RKJ (DEB — 9318279) and the Sidney F. and Doris Blake Centennial Professorship in Systematic Botany at the University of Texas at Austin. 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 William F. Chissoe of the Samuel Roberts Noble Electron Microscopy Laboratory, University of Oklahoma, for technical assistance with the SEM work and Greg Strout, also of University of Oklahoma, for reproduction of plates using Adobe PhotoShop 7. 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, S. A. Hall, B. B. Simpson, and T. J. Mabry, for their assistance. We also thank two anonymous reviewers for suggestions on an earlier version of the manuscript.