Home at last: the enigmatic genera Eriachaenium and Adenocaulon (Compositae, Mutisioideae, Mutisieae, Adenocaulinae)

Abstract The genera Eriachaenium and Adenocaulon (Compositae) have distinct but complex histories and both have been placed in a number of tribes across the family. For the first time the two genera are included in a molecular study and the results show that they are best placed in the tribe Mutisieae s.s. and are the only genera in the re-instated subtribe Adenocaulinae. When described, this subtribe contained only Adenocaulon and was found in the Inuleae. The study also confirms one of the conclusions of a recent morphological study that Eriachaenium and Adenocaulon are sister taxa. Past difficulties in tribal assignment are attributed to the distinct and unusual morphology of each genus. Both genera and the subtribe are described and a key to separate the genera is provided.


Introduction
Eriachaenium Sch. Bip. and Adenocaulon Hook.  are perennial herbs that were left unplaced in the most recent genus level classifi cation of the Compositae family (Hind 2007). More recent phylogenies, based on molecular data, have divided the Compositae into 42-43 tribes; about half of them are small (Panero and Funk 2008;Funk et al. 2009). Within these family level phylogenies there are four main areas (as well as a number of very small subfamilies), beginning with the crown group: the highly nested subfamily Asteroideae (asters, ragworts, sunfl owers: monophyletic) which contains about 65% of the species in the family; the subfamily Cichorioideae s.l. (gazanias, dandelions, ironweeds: monophyletic); the subfamily Carduoideae (thistles and African Mutisieae: monophyletic), and fi nally there is a paraphyletic basal area of the phylogeny that contains most of what used to be the tribe Mutisieae s.l. (Gerbera daisies) whose former members are now in 15 tribes (Funk et al. 2009). Th roughout its history Eriachaenium has been placed in three tribes in the subfamily Asteroideae and one in the basal grade; Adenocaulon has previously resided in six tribes in the Asteroideae and two in the basal grade.
Unlike many taxa that are diffi cult to place, Eriachaenium is rarely discussed or debated. Perhaps it's remote location (endemic to Patagonia; Fig. 3) or the fact that it is monospecifi c (Eriachaenium magellanicum Sch. Bip) and relatively rare, has fostered this lack of attention. Th is small herb has an unusual compressed underground stem and staminodes in the marginal fl orets (Figs 1, 2). Th e species was described by Schultz Bipontinus (1855) who placed it near Osteospermum, an African genus in an almost exclusively African tribe (Asteroideae: Calenduleae). Although today this placement is diffi cult to understand it was probably based on the long corolla lobes and short anther bases. Bentham (1873) and Hoff man (1894) did not disagree with this placement and there it remained until Cabrera (1961) in his key to the Argentine genera of Asteraceae moved it to the Inuleae (Asteroideae) without comment, but in the subtribe Adenocaulinae along with Adenocaulon. Gray described Adenocaulinae (1873) but included only the type. Eriachaenium was left in the Inuleae by Muñoz Pizarro (1966) as well as Moore (1983). Robinson and Brettell (1973) moved the genus to the Mutisieae s.l., based mainly on pollen characters and Bremer (1994) put the genus in the Mutisieae subtribe Nassauviinae (now tribe Nassauvieae). Eriachaenium was not included in recent molecular phylogenies such as Funk et al. (2005), Panero and Funk (2008), Funk et al. (2009, 2014. Adenocaulon has fi ve species that grow in temperate forests (Fig. 4A) in four widely disjunct areas (Fig. 3): Northwest USA and adjacent Canada (1 species), East Asia (2), Mesoamerica (1), and Patagonia (1) (Bittman 1990a, b). In contrast to Eriachaenium, Adenocaulon has received quite a bit of attention possibly because it is more widespread and most of its taxa are found in areas with numerous botanists. Th ese various studies have moved the genus from tribe to tribe; in fact, over the years it has been placed in eight tribes. Along with a few other genera, Adenocaulon was unplaced to tribe by Bremer (1994). Th e various placements are as follows: (1) Eupatorieae: Edgeworth (1851); (2) Heliantheae s.l. (in tribe Millerieae): Bentham and Hooker (1873), followed by Gardner (1977); (3) Inuleae: Gray (1873, as the separate subtribe Adenocaulinae), followed by Hoff mann (1894), Britton and Brown (1943), Cabrera (1961Cabrera ( , 1971, Muñoz Pizarro (1966), Merxmüller (1977), and Moore (1983); (4) Adenocauleae: Rydberg (1917), monogeneric; (5) Senecioneae: Cronquist (1955), followed by Wagenitz (1964), and Morton (1978); (6) Anthemideae: Stix (1960), followed by Leins (1968), and Skvarla et al. (1977); (7) Mutisieae: Stebbins as quoted in Ornduff et al. (1967), followed by Nordenstam (1977), Bittman (1990a, Bremer (1994), Jansen and Kim (1996), Panero and Funk (2008); Katinas et al. (2008) and (8) Cardueae: Maximova (1999). Cabrera (1961) and Katinas (2000) were the only ones to consider the two genera together. Cabrera (1961) published a key to the genera of Asteraceae of Argentina and, probably following the classifi cation of Gray (1873) who placed Adenocaulon in the subtribe Adenocaulinae of the tribe Inuleae. Cabrera (1961) accepted that placement and also included Eriachaenium in the subtribe. Katinas (2000) conducted a cladistic analysis using 38 morphological characters and 52 genera from across the family to investigate whether or not Adenocaulon and Eriachaenium were closely related to one another and to determine a tribal assignment for the two genera. Her study showed that the two genera were closely related based on sharing four characters: 1) involucral bracts in 1-2 series, 2) length/width ratio of anthers was 2.5-4.5, i.e., very small; this ratio was only found in members of Anthemideae, 3) the lack of a pappus (characters that are found in other parts of the phylogeny), and 4) a re-occurrence of the plesiomorphic short bifi d style. As far as placement in the family, Katinas results showed that 'fl oret dimorphism' and 'pseudobilabiate fl orets shared with Anthemideae, Adenocaulon and Eriachaenium that are female or neuter and the tubular and/or pseudobilabiate fl orets' place them above the Mutisieae and Cardueae. Also they share two synapomorphies, 'fl oret dimorphism' and 'marginal fl orets female or neuter', that group them with the Liabeae, Arctotideae, and Asteroideae. Th e placement of the two genera was determined to be (in todays classifi cation) in the Cichorioideae (s.s.) above the Lactuceae/Vernonieae clade and the sister group of the Liabeae. Th is was yet another new position for both of these genera. Adenocaulon was included in the chloroplast DNA phylogenies of Panero and Funk (2008) and Funk et al. (2009) and placed in the Mutisieae s.s. but the generic representation from the Mutisieae s.s. was not extensive and Eriachaenium was not included.
In order to ascertain the best placement for Eriachaenium and Adenocaulon and to test the proposed sister group relationship between the two genera we used a molecular approach that included species of both taxa and a wide sampling of outgroups. In 2009, the members of a collecting expedition to Patagonia located populations of Eriachaenium and Adenocaulon (A. chilense Less.) and this fresh material combined with two additional herbarium specimens (Adenocaulon: A. bicolor Hook., and A. lyratum) has allowed us to fully discuss these two genera and to provide an estimate on where they should be placed in the phylogeny of the family.

Materials and methods
We sequenced the nuclear ITS and the plastid molecular markers trnL-trnF and rpl32 of two diff erent populations of Eriachaenium magellanicum, one each of Adenocaulon chilense, A. lyratum and A. bicolor, the sequences are deposited in GenBank and the numbers will be included in Pasini et al. (MS submitted). Th ese data were shared with the authors of a separate study that encompassed a broad selection of taxa from the Mutisioideae and related tribes (Pasini et al. MS submitted) in order to determine the proper placement for these problematic genera. Th e molecular dataset contained species of the three tribes of the subfamily Mutisioideae (Mutisieae-8 genera, 21 species, Onoserideae-four genera, four species, and Nassauvieae-13 genera, 25 species) and four species of the subfamily Barnadesioideae. Th ese data contain new sequences generated for the Pasini et al. (MS, submitted) as well as sequences from GenBank generated for several publications (Katinas et al. 2008;Simpson et al. 2009;Baird et al. 2010;Funk et al. 2014).
Details of the sampling strategy, DNA extraction, amplifi cation, and sequencing methods and data analysis discussion are included in the Pasini et al. paper (MS, submitted). Here we show a part of the fi nal cladogram that highlights the placement of Eriachaenium and Adenocaulon (Fig. 6).

Results and discussion
Phylogeny Th e results of all three datasets, the matK, the combined ITS-trnL-F, and the total combined show Eriachaenium and Adenocaulon forming a clade nested in the Mutisieae s.s. (Fig. 6). Th e support values of the clade formed by these two genera are high (Bayesian inference of 1) but the relationships among the three species of Adenocaulon sampled in the analysis and Eriachaenium magellanicum are not clear. Here we show a section of the phylogeny (Fig. 6) for the purpose of facilitating the discussion in this paper. For the complete phylogeny based on the combined ITS, trnL-F, matK, and trnL-rpl32 markers see Pasini et al. (MS, submitted).

Morphology
Most of the confusion in the placement of Eriachaenium and Adenocaulon is caused by a lack of understanding of character evolution within the family complicated by the fact that the characters that were most often used to defi ne the Mutisieae s.l. are often missing or modifi ed in both genera. Now that Eriachaenium and Adenocaulon form a clade nested in the tribe Mutisieae s.s. (Mutisioideae) we can re-examine the morphology of the two genera and how their characters fi t with those of the Mutisieae s.s.
Prior to the advent of molecular data, the Mutisieae s.l. were considered to be highly derived because some were humming bird pollinated and many had some form of colorful and/or dimorphic corollas (especially bilabiate or pseudo-bilabiate), long tails on the anthers, short bifi d styles often with a rounded apex, and psilate or microechinate pollen. Later characters such as anthemoid (ecaveate) pollen (Ornduff et al. 1967, Robinson andBrettell 1973), testa epidermis type (Grau 1980); chromosome number (n = 23; Ornduff et al. 1967), and "thickened apical appendage on anthers" and "obtuse-rounded style hairs" (Bremer 1994), were added to the list. Th ese were thought to be derived characters because they were uncommon in the family and restricted, for the most part, to South American taxa, and because it was commonly believed by many taxonomists that studied the Compositae that the Heliantheae s.l. represented the ancestral morphology. Since the groundbreaking work of Jansen and his co-authors (Jansen et al. 1987(Jansen et al. , 1990(Jansen et al. , 1991(Jansen et al. , 1996 and subsequent contributions by Panero and Funk (2008) and Funk et al. (2009) we now know that the Mutisieae s.l. are actually a number of independent lineages strung out along the basal area of the phylogeny; some are even part of the thistle subfamily (Carduoideae) or independent lineages (Pertyeae; Pertyoideae).
Within the latest classifi cation the tribe Mutisieae s.s. falls into the subfamily Mutisioideae with two additional tribes: Onoserideae, Nassauvieae. Th is more restricted version of the tribe is defi ned by the presence of many of the same characters mentioned above because many of the taxa that lack those characters are now placed elsewhere.
However, as Katinas et al. (2008) correctly pointed out, some of these characters are found in other tribes. Eriachaenium and Adenocaulon have some of these characters (dimorphic corollas; short bifi d styles with a rounded apex; microechinate, ecaveate pollen) but lack others (colorful corollas, long tails on anthers). While Adenocaulon has bilabiate corollas, Eriachaenium does not, however it does have a variable number of corolla lobes (4 or 5) so technically they both have dimorphic corollas.

Pollen and chromosome numbers
Th e pollen grains of Eriachaenium and Adenocaulon (Fig. 6) are very similar to one another diff ering only in the size of the grain and the thickness of the exine and both genera have some grains with a compact aspect and Anthemoid pattern (see also Skvarla et al. 1977;Bittmann 1990a). A study by Zhao et al. (2006) has the only SEM images that show the exine structure that we know of for both genera and they appear to be of the standard pollen type for the Mutisioideae. Adenocaulon and Eriachaenium have pollen of medium size (P × E = 26-32 × 26-30 μm in Adenocaulon, and 30-36 × 24-30 μm in Eriachaenium), are tricolporate, with the colpi long with thin margins and a microgranulate membrane and a diff use mesoaperture (Katinas et al. 2008). Both genera have a Mutisia type exine (Tellería et al. 2003). Overall the pollen grains of both genera are similar to that of Artemisia verlotiorum (Anthemideae) a condition that exemplifi es one of the major problems with trying to identify unique morphological characters to defi ne groups within the family (Hansen 1991  "We are not convinced that Adenocaulon belongs in Senecioneae where it has been placed by various workers (Ornduff et al. 1963), but the count of n = 23 for the very local and distinctive Central American A. lyratum is a report for the fourth … member of the genus to be examined. Each species has consistently had n = 23.… Stebbins (personal communication) has suggested that Adenocaulon shows affi nities with Mutisieae on the basis of a common possession of distinctive pollen characters. Th e bilabiate tendencies of marginal corollas, the shape and pubescence of leaves, and the chromosome number of Adenocaulon further suggest relationships to Mutisieae…." Description. Herbs perennial, dwarf or scapiform with cylindrical or planate rhizomes, stems simple, erect or prostrate to ascending, glabrous or with stipitateglandular hairs. Leaves glabrous to subglabrous above, tomentose beneath; basal leaves alternate or rosulate to sub-rosulate; sessile or petiolate to pseudopetiolate; blades oblanceolate, elliptic, ovate, obovate, to deltoid, margin entire to lyrate, pinnately or palmately veined, glabrous to subglabrous above, tomentose beneath. Infl orescences terminal or axillar, monocephalous or laxly racemose to corymbose, pedunculate; heads heterogamous, disciform; receptacle epaleate; involucre uniseriate. Florets dimorphic; marginal fl orets female, with or without staminodes, corolla sub-bilabiate (3+1 corolla lips), tubular-funnelform, shortly to deeply 4-to 5-lobed, rarely bilabiate; central fl orets bisexual or male with a rudimentary ovary, corolla tubular-funnelform, deeply 5-lobed; anther apical appendages rounded to acute at the apex, basally constricted and demarcated from the thecae, basally auriculate with tails very short, smooth to slightly papillose, fi lament with anther collar; style shortly bifi d, branches dorsally papillose. Achenes truncate at the apex, densely pubescent, shaggy (long, fi liform, uniseriate hairs) or glandulose (glandular multiseriate capitate hairs), dimorphic, marginal cypselae conspicuously bigger than the central ones; pappus absent. Pollen spheroidal to prolate, tricolporate, exine Mutisia type, microechinate.
Habitat and distribution: Genus of fi ve species with a disjunct distribution in Patagonia, Mesoamerica, northern United States and southern Canada, and temperate southeastern Asia (Fig. 3). Inhabits moist forests in the shade of Pinus spp., Quercus spp. and Nothofagus spp. (Bittmann 1990a, b, and fi eld observations). Details of the fl ower morphology, including the diff erences between the male and female fl owers, can be found in Ayers (1900).
Species list: fi ve species falling into three morphological groups that are biogeographically distinct (according to Blake 1934

Conclusion
Perhaps the best conclusion is to review synapomorphies for the Eriachaenium + Adenocaulon clade. With the phylogeny available we can examine the characters that group the two genera. Th at does not mean that none of the other species in the family or even in the Mutisioideae have these characters, it means that, when examined in the light of the phylogeny they are deemed synapomorphies for the Eriachaenium + Adenocaulon clade.  (Figs 2H, 5G). 5. It is interesting to note that in these two genera the fl orets' dimorphism is not conspicuous, while in all the other genera of the tribe it is. In fact, within the Mutisieae s.s. there is an impressive variety of colors of the marginal fl orets which easily distinguishes them from the central fl orets. Th erefore the character of "inconspicuously dimorphic fl orets" found in the Eriachaenium + Adenocaulon clade and not found in the Mutisieae s.s., can also be considered as a synapomorphy. 6. Th e presence of tubulose 4-lobed corollas in the marginal fl orets in both genera indicates a strong affi nity between the two genera because while tetramerous central fl orets are common in Compositae, such corollas rarely occur as marginal ones. 7. Even though Eriachaenium and Adenocaulon have a Mutisia exine type of pollen, the grains are small and spheroid with a thin exine, whereas those of Mutisioideae (excluding Nassauvieae) are usually large and elliptic with a thick exine. Th is type of pollen grain is unique in the Mutisieae and it approaches the Artemisia exine type (Anthemideae). 8. Both genera lack a pappus (Figs 2D-H, 5D-H). Th e absence of a pappus is widespread in other tribes of Compositae (e.g., Heliantheae s.s.) but it is very rare in Mutisioideae (only found in Adenocaulinae and Cephalopappus and Panphalea of the tribe Nassauvieae).
9. Both genera have their achenes covered with unusual pubescence: Eriachaenium has multicellular, fl agellate, fi liform hairs that are confi ned to the achene (Fig. 2D, F-H, K) and Adenocaulon has multiseriate, capitate, glandular pubescence (the glands are dark purple) and this pubescence is found on other parts of the infl orescence (Figs 4D, E, 5B, D, E, H, K). 10. Eriachaenium and Adenocaulon both grow in habitats that are unusual for the family: Eriachaenium practically buries itself in the sandy mud (Fig. 1B, E) and when we found it, it was a few feet above the water line of some, but certainly not all, lakes/ponds in the mountains of Patagonia (Fig. 1A); Adenocaulon inhabits the fl oor of relatively moist forests ( Fig. 4A-C). Both of these are rather extreme limits of the habitat for the family. Perhaps this movement into these habitats has triggered their unusual morphology.
Another potential synapomorphy is the bullate leaves found in both genera. But, more data need to be gathered to be sure of its distribution in the Mutisioideae.
We can also list some characters that we now think are plesiomorphic for the Eriachaenium + Adenocaulon clade in that they are shared with other parts of the basal grade: central corollas deeply lobed; style shortly bifi d with an apex that is rounded or slightly acute; style apex shortly papillose; pollen with exine psilate or microechinate and pollen of the Anthemoid pattern. Th e pollen grains in Eriachaenium and Adenocaulon share features with many taxa in the Mutisioideae and with Anthemideae. However, at this point, the occurrence of the "anthemoid" pollen in the Anthemideae is considered to be independent of its occurrence in the Mutisioideae.
In future studies we hope to expand these lists as well as determine the point on the cladogram where the plesiomorphic characters are actually apomorphic.
Finally, we thank the editors of the amazing online databases that we use daily and often forget to mention: International Plant Names Index, JSTOR Global Plants, Index Herbariorum, and Index to chromosome numbers in Asteraceae (IPNI 2012, JSTOR-GP continuously updated, Th iers continuously updated, Watanabe 2015).