Phylogeny of Helianthus and related genera

Molecular phylogenetic studies have contributed significantly to our understanding of the phylogenetic relationships of Helianthus, although several problems remain to be resolved. Molecular data have resolved problems with the circumscription of Helianthus. Its sister group is Phoebanthus, a narrowly distributed genus of two species from the state of Florida. The sister group to the HelianthusPhoebanthus clade is a diverse set of taxa that occurs in Mexico and South America. These include species now classified as part of the paraphyletic Viguiera as well as such distinctive genera as Tithonia, Simsia, Pappobolus, Scalesia, Lagascea, and Alvordia. Incongruence between results based on chloroplast and nuclear-based data sets suggests that hybridization has been involved in the evolutionary history of this group. The nearest outgroup to the clade containing Helianthus is composed of members of Viguiera sect. Maculatae, which are trees and shrubs of Mexico. Other basally diverging groups in the subtribe Helianthinae to which Helianthus belongs are primarily woody members now classified in Viguiera from Mexico and nearby areas. Within Helianthus, divergence appears to be recent, based on an overall lack of divergence between species. The chloroplast-based tree suggests that Helianthus includes four phylogenetic lineages, whereas the nuclear ITS sequence data suggests that the perennial species are paraphyletic relative to the rest of the genus, with basally diverging branches consisting of species confined to the southeastern US. Because there is a lack of divergence among many groups of species, more variable markers will be required to resolve fully relationships within Helianthus.


Introduction
The familiar domesticated sunflower, Helianthus annuus, has become a popular plant well known to agronomists and also to home decorators. Less well known is the fact that it is but one of some 50 species in a genus that is native to North America. Together with its close relative, Phoebanthus, Helianthus forms a distinctive lineage within the subtribe Helianthinae in which it is classified. Subtribe Helianthinae includes a total of about 400 species and has an entirely New World distribution, ranging from eastern North America to Chile and Argentina. The subtribe appears to have its origin in Mexico, and divergence among many of its members is probably recent.
Our view of the systematics of Helianthus and its relatives has been heavily influenced by the influx of molecular phylogenetic data (Fig. 1). An initial molecular data set (Schilling and Jansen, 1989;Schilling, 1997) was based on restriction site analysis of chloroplast DNA (cpDNA), which was one of the first approaches to have widespread use in plant phylogenetic studies. The resulting tree thus represents a maternal gene tree, because the chloroplast genome is mostly inherited maternally in flowering plants and is rarely if at all subject to recombination. A data set from sequencing of the nuclear ribosomal internal transcribed spacer region (ITS) has also been obtained (Schilling and Panero, 1996;Schilling et al., 1998), which provides a second, independent gene tree. The results from cpDNA and ITS analyses are mostly but not entirely congruent (Fig. 1).

Circumscription of Helianthus
The availability of molecular data has made it possible to refine with confidence the circumscription of Helianthus. As a result, several species or groups of species can certainly be excluded from it, and two species of questionable affinities have been included. As now circumscribed, Helianthus is geographically coherent, being restricted to North America, but it is not easily diagnosed morphologically. The key feature that classically identifies Helianthus within subtribe Helianthinae is the caducous pappus of two awns, with few or no intervening scales. Helianthus is also characterized by having pales that are trilobed at the apex. Both of these features are homoplasious in subtribe Helianthinae, and are found in other groups. Furthermore, they are both absent in one species now included in Helianthus, H. porteri. A third feature that is found throughout Helianthus is the well-developed terminal style appendage, but a similar structure is also observed in a few other members of the subtribe.
The relationships of Helianthus have also been clarified through molecular phylogenetic analysis. Phoebanthus, a small genus of two species with narrowly endemic ranges in the state of Florida, is a near sister group. There is no other single genus that is particularly close to Helianthus. The sister group to the Helianthus-Phoebanthus clade is a large, diverse assemblage that includes several genera, and whose collective range spans areas from Mexico to South America.
A major adjustment in the circumscription of Helianthus has involved the transfer of the all of the South American species formerly placed in it to the genus Pappobolus.
The South American species had been placed in Helianthus because they also have a caducous pappus of two awns. Pappobolus, which is endemic to the Andes Mountain region, includes species with a wide variety of pappus types (Panero, 1992). The removal of the South American species from Helianthus had been suggested prior to molecular studies, and the molecular phylogenetic information adds ample support to this. This makes it clear, for instance, that domestication of the cultivated sunflower, Helianthus annuus, did not occur initially in South America.
Two other species that had been placed in Helianthus because of their pappus type have been returned to Viguiera. Erroneous information regarding its habit had led to the placement of one of these, V. similis, with the annual members of Helianthus (Heiser et al., 1969). This reflects in part the location of V. similis in a remote area of Baja California, Mexico, and the fact for many years the only herbarium material that was available for it was fragmentary. More recent information shows that it is clearly a shrub, like other members of V. subgen. Bahiopsis with which it is clearly allied. The relationships of the other species removed from Helianthus, V. phenax, are less well resolved. This species was placed in Helianthus under a synonym, H. ludens, and confusion regarding its affinities again reflects in part the scanty information available for it. It is an annual that is abundant locally in an area of Durango, Mexico, and apparently occurred as a waif in Texas, from where a collection was described as a Helianthus. It is still not completely clear what its exact relationships are, but it is clearly not a Helianthus, despite its caducous pappus.
Molecular data allow confirmation of the placement in Helianthus of two species whose affinities were not certain. The placement of one, H. agrestis, had been questioned because it has several distinctive traits for Helianthus, including tuberculate cypselas and self-compatibility. The other species, H. porteri, has been placed in Viguiera (or in Heliomeris by those who recognize it as distinct) by most botanists. As recently as 1985 a thorough review of morphological and chemical data done as part of a Ph.D. dissertation concluded that it was best placed in Viguiera. The molecular data, however, agree with chromosome number, crossing behavior, and biogeographic evidence in indicating that H. porteri is accurately placed in Helianthus. The inclusion of H. porteri in Helianthus creates a problem in the morphological circumscription of the genus, because it lacks two of the three apparent synapomorphies of the genus. The pales of H. porteri are not trilobed at the apex, and it entirely lacks a pappus. It does, however, have the prominent terminal style appendage that characterizes Helianthus. There are several other members of subtribe Helianthinae that exhibit a similar terminal style appendage, so there is not a single absolute synapomorphy to allow morphological diagnosis of Helianthus.

Relationships of Helianthus to other members of subtribe Helianthinae
The nearest relative to Helianthus is clearly Phoebanthus, a small genus of two species restricted to areas in the state of Florida in the southeastern U.S. In fact, the cpDNA tree would suggest inclusion of Phoebanthus within Helianthus, although the ITS tree suggests that the two are sister genera (Fig. 3).
The Helianthus-Phoebanthus clade is distinctive within subtribe Helianthinae. This clade is supported by 11 restriction site changes in the cpDNA tree and five base pair substitutions in the ITS tree. There have not yet been any morphologically synapomorphies discovered that diagnose unambiguously this clade.
Beyond Phoebanthus, there is no single genus that can be identified as being nearest to Helianthus. The sister group to the Helianthus-Phoebanthus clade is a large and diverse group that includes species now classified in Viguiera as well as the distinctive genera Tithonia, Simsia, Pappobolus, Scalesia, Lagascea, and Alvordia (Fig.1,  2). There is no morphological synapomorphy that characterizes this group, and they occur over a wide area of Mexico, Central and South America. In general, plants of this lineage are herbaceous perennials, although there are also annuals and a few woody species (notably Pappobolus and Scalesia).
An aspect of the sister clade to Helianthus is that there is significant incongruence between trees obtained from cpDNA and ITS results, respectively (Fig. 2). In some cases, the incongruence appears to be most logically explained by chloroplast transfer through past hybridization. For example, Simsia dombeyana, which is one of only two South American species of a mostly Mexican genus, is placed by cpDNA within the clade of South American species of Viguiera, whereas the ITS tree places it in a derived position within Simsia. The latter position is consistent with its morphology, because it barely differs from S. foetida (Spooner, 1990). Similarly, the cpDNA tree would suggest that Tithonia is diphyletic, with some species part of the Simsia clade and others close to Viguiera subgen. Amphilepis (Fig. 2.) In contrast the ITS tree shows Tithonia to be entirely monophyletic and sister to V. subgen. Amphilepis. Again, the result from ITS is most consistent with morphological observations. A problem with an explanation that involves hybridization between genera is that such hybrids have never been documented, either in nature or from artificial crosses. It suggests that wide hybridization should be possible, however.
The fact that the basal-most diverging clades consist of groups that are primarily or exclusively found in Mexico suggests that subtribe Helianthinae originated in this region (Fig. 1). Subsequent divergence of the subtribe led to various distinctive groups both in Mexico and also in other areas. Most of the movement out of Mexico appears to have been to Central and South America. The Helianthus-Phoebanthus clade is the only group to be present in eastern North America. The molecular phylogenetic trees are consistent in suggesting that Viguiera sect. Maculatae is the next most basal diverging group in subtribe Helianthinae to Helianthus. Viguiera sect. Maculatae is endemic to Mexico and is woody, including large shrubs and small trees.

Divergence within Helianthus
The most striking feature of molecular phylogenetic studies of Helianthus to date is that there is strikingly little variation within the genus for either ITS or cpDNA (Table  2). In addition, there is incongruence between results from the two molecular markers (Fig. 3). The lack of divergence in particular has limited the conclusions that can be drawn regarding relationships within Helianthus. This observation was extended by our recent survey of variation for the trnT-trnF spacer/intron region. Comparison of H. porteri, H. angustifolius, and H. giganteus revealed only 2 base pair changes and 6 insertion/deletion changes from a total of 1,417 based that were sequenced for this region of the cpDNA genome (Table 2). By contrast, studies of this gene region in other plant families have revealed extensive variation, sometimes even within a single species.
The incongruence between results from cpDNA and ITS data sets for relationships within Helianthus is not readily explainable, although it may reflect past hybridization between different lineages. The cpDNA results would suggest a relatively straightforward division of the genus into four main lineages (Fig. 3), with Phoebanthus as part of the perennial group. The ITS data suggest a more complex scenario in which the perennial species are paraphyletic to the rest of the genus, based on placement of H. heterophyllus and H. carnosus as basally branching clades. In the ITS tree, Phoebanthus is placed as the sister group to Helianthus, rather than as part of the perennials.
Neither the tree from cpDNA nor the one based on ITS is in accord with the current infrageneric classification of Helianthus (Schilling and Heiser, 1981). Section Helianthus, which contains the bulk of the annual species including H. annuus, is shown to be monophyletic in both trees. Both trees also place the peculiar annual species, H. agrestis and H. porteri, as distinctive lineages. In neither tree is H. sect. Ciliares shown to be monophyletic. The three species of H. ser. Ciliares are a monophyletic ingroup to the perennial species in both trees. For the most part, however, there are few differences among the perennial species from either data set.
The lack of variation leaves unresolved the relationships of the polyploid species of Helianthus, all of which are perennial. It has not been possible even to identify the maternal genome of any of the polyploids. The only exception to this involves the sterile triploid, H. x multiflorus, in which it has been possible to show that the cpDNA genome is from a polyploid species and that it contains rDNA genomes from both H. annuus and a perennial species. Thus, further work is needed to identify molecular markers that show the appropriate, high level of variation that will be needed to resolve relationships of the polyploid species of Helianthus.