Go to species list

Eupatorium as currently delimited is a north temperate genus of about 45 species.  Traditional classifications placed almost all species of Eupatorieae that have a pappus of numerous bristles surmounting a five-ribbed cypsela into this genus, but careful work including molecular phylogenetic investigations has shown that the broader circumscription is massively para- and polyphyletic. The genus as now delimited is characterized by its herbaceous habit, opposite (or rarely whorled) leaves, flat receptacles, relatively few-flowered (3-23) heads, white, narrowly campanulate corollas which have a narrow basal tube, style which is puberulous at the base and in which the appendages are filiform, and cypselae which are glandular but usually lack any other pubescence.  

Molecular phylogenetic data show that the sister group to Eupatorium is the North American Eutrochium (Joe Pye Weeds).  Because Eupatorium + Eutrochium form a monophyletic group, it becomes a matter of judgment whether or not to recognize the two as separate genera.  There are, however, a number of fundamental differences between the two, and they are easily separable, so we have preferred to separate them at the genus level.  Among other things, no hybrids between members of Eupatorium s.s and Eutrochium have ever been reported, despite the fact that they grow in close association both in the wild and in garden settings.  In contrast, a spontaneous hybrid between the North American species Eupatorium altissimum and the Eurasian E. cannabinum has now been noted from a garden setting where the two species were growing in close proximity and confirmed to be this combination by molecular data.  The sister group to Eupatorium + Eutrochium is the Liatrinae, another North American group.  To date there is no morphological synapomorphy to support this grouping, but it makes biogeographical sense.

The species of Eupatorium can be separated into two groups based on geography, although these groups have not been recognized taxonomically.  One group occurs in North America, and the other occurs in Asia and extends into Europe where it is represented by a single species, E. cannabinum, which is the type of the genus.  Based on comparison to the geographic distribution of the outgroups, it seems most likely that the genus originated in North America with subsequent dispersal to Eurasia.  The high degree of similarity based on molecular phylogenetic data between Eurasian and North American species suggests that the dispersal event must have been relatively recent.

The North American group of species can be separated again into two subgroups.  One includes only three species that have small heads and appear to be wind pollinated and has been recognized as the “Traganthes” group ("Dog Fennels").   These species are all diploids and seed reproduction is exclusively sexual.  Hybrids are known among each pair of species.

The second North American subgroup includes a much larger number of species and has been called the “Uncasia” group.  Some are sexual and diploid; others include both sexual diploids and apomictic polyploids.  And there are also some species that are entirely apomictic and polyploid.  There is an absolute correlation between apomixis and polyploidy.  Many of the apomictic polyploids are nearly or completely pollen sterile, and this appears to limit gene flow between species and between ploidy levels. There are many reported interspecific hybrids within this group, and it is likely that any pair of sexual diploids are capable of intercrossing.

Hybridization occurs between members of the Traganthes and Uncasia groups, although it appears to have limited consequences.  The taxon Eupatorium x pinnatifidum consists of hybrids between E. capillifolium and either E. perfoliatum or E. serotinum.  It is likely that some plants labeled E. x pinnatifidum involve E. compositifolium as the Traganthes-group parent.  Plants of E. x pinnatifidum appear to be short-lived and sterile.

Two factors seem to have encouraged speciation in the Uncasia group of Eupatorium.  One is edaphic specialization, and the second is the apparent lack of gene flow via hybridization between species because the hybrids are sterile.  The edaphic specialization may have been significant in leading to highly reduced population numbers during periods of unfavorable climate and thus resulting in rapid fixation of genes. Each species of Eupatorium is characterized by distinct ITS and ETS sequences, which is in contrast to other members of Asteraceae that inhabit the same types of old field habitats and geographical region, such as Echinacea, Helianthus and Solidago.  

A number of diploid species of Eupatorium currently have very restricted ranges:  E. lancifolium is restricted to the Ozarks region; E. resinosum is known from a limited range on the Atlantic Coastal Plain; E. petaloideum is known from a small area of northern Florida and southern Georgia; E. mikanioides is endemic to Florida; the recently described E. paludicola occurs in Carolina bay habitats in a small part of the Atlantic Coastal Plain.  Eupatorium album has a somewhat wider distribution, but as narrowly defined this species occurs primarily in the southeastern portion of the U.S.

Two species of Eupatorium that are known only as sexual diploids are quite widespread in geographic distribution:  E. perfoliatum and E. serotinum.  These are two of the three species of the genus in which the heads have more than 5 flowers.  (The other is the extremely limited E. resinosum, which molecular data suggest to be closely related to E. perfoliatum).  

Several species of Eupatorium have both sexual diploid and apomictic polyploid populations.  These include E. altissimum, E. hyssopifolium, E. leucolepis, E. mohrii, E. pilosum, E. rotundifolium, and E. sessilifolium.  In each case, the range of the apomictic polyploids is much greater than that of the sexual diploids.  Molecular phylogenetic data have confirmed that, for each of these, there are autopolyploid populations.

The allopolyploid apomicts have received varying taxonomic treatment, based on how distinctive they are relative to the related diploids.  Those that are clearly separable from either parent have been recognized as distinct species.   These include E. godfreyanum (E. rotundifolium/E. sessilifolium), E. novae-angliae (E. paludicola/E. perfoliatum), E. saltuense (E. hyssopifolium/E. sessilifolium) and E. anomalum (E. mohrii/E. semiserratum, but see below for a fuller explanation). Recently hybrid apomicts involving E. petaloideum or E. album have been proposed for recognition as distinct, including E. subvenosum, E. vaseyi; E. fernaldii, and E. sullivaniae.   For some other species, the hybrids are not clearly distinct and have been recognized as infrageneric taxa.

Hybrids resembling Eupatorium album.  The distinctive, white, “petaloid” phyllaries of E. album (and of its sibling species, E. petaloideum) are both characteristic and also appear to be a dominant trait that is passed on to hybrid offspring, and several hybrid apomicts have been recognized taxonomically as varieties of E. album.  Ironically, the ones that have been named all turn out to contain a genome from E. petaloideum, rather than E. album.  These include E. subvenosum (E. album var. subvenosum = E. petaloideum/E. hyssopifolium); E. vaseyi (E. album var. vaseyi = E. petaloideum/E. sessilifolium); and E. fernaldii (E. album var. glandulosum = E. petaloideum/E. sessilifolium/E. perfoliatum).  A hybrid combination that includes a genome from E. album is E. sullivaniae, (E. album/E. lancifolium); a second hybrid combination involving E. album is E. album/E. hyssopifolium, but it is not clear that this is worthy of recognition as a distinct entity.   Note that E. saltuense has been hypothesized to be derived from E. album/E. hyssopifolium, but material that we tested proved to have E. sessilifolium rather than E. album in combination with E. hyssopifolium.

Hybrids resembling E. hyssopifolium.  The relatively narrow leaves, usually whorled nodes, and tendency for the lateral buds to initiate and then halt expansion (suppressed nodes), all characteristic of diploid and autoploid E. hyssopifolium are also displayed in its hybrid derivatives, including at least two (or more) that are apomictic and widespread.  Reflecting the combination with another species, the hybrid derivatives have broader and often more prominently toothed leaves compared to diploid and autoploid E. hyssopifolium, and this morphological syndrome has been recognized as E. hyssopifolium var. laciniatum; the name E. torreyanum is available for this taxon at the species level.  Unlike the hybrid derivatives of E. album and E. petaloideum noted above, those involving E. hyssopifolium with other species are not easily separated from one another morphologically.  Thus, material that keys to E. hyssopifolium var. laciniatum includes hybrid derivatives with E. serotinum (which fit the description of E. torreyanum) and E. semiserratum.  Some of the derivatives involving E. serotinum are recognizable because they have heads with more than 5 flowers, but others of the combination E. hyssopifolium/E. serotinum have only 5 flowers in most or perhaps all of the heads.  Specimens of the combination E. hyssopifolium/E. semiserratum tend to have somewhat narrower leaves that are less prominently toothed than those of E. hyssopifolium/E. serotinum, but others are not clearly separable (part of the problem may simply involve inadequate herbarium material, and it is difficult to collect completely typical specimens of these plants because they tend to have dropped the lower stem leaves by the time that they reach full anthesis.)  Provisionally the name E. torreyanum is considered to embrace both combinations, as well as possibly others.  Individual specimens have been analyzed that fit the general syndrome of E. torreyanum but are from the combinations of E. hyssopifolium/E. album, E. hyssopifolium/E. linearifolium, and E. hyssopifolium/E. lancifolium.  Further study is needed to assess if any of these is sufficiently widespread and adequately distinct to deserve individual taxonomic recognition.

 

Hybrids resembling Eupatorium mohrii.  Eupatorium mohrii is a species of the southeastern Coastal Plain that is distinct in its relatively short, elliptical, crenate leaves and tendency to form a thickened underground storage organ.  It occurs in both sexual diploid and apomictic polyploid populations.  The diploids were previously recognized as E. recurvans, and the polyploids were hypothesized to be derived from hybridization between the diploids and E. rotundifolium.  Molecular phylogenetic data, however, clearly show that E. recurvans/E. mohrii form a diploid/autopolyploid pair (for which the name E. mohrii has priority if they are combined) and this agrees with chemical results published much earlier.  The allopolyploid apomict E. anomalum was also ascribed to hybridization involving E. mohrii and E. rotundifolium.  Detailed analysis, however, shows that E. rotundifolium is not involved (although it does hybridize with E. mohrii, the resulting plants have a different morphology), and that the situation is a bit more complicated.  The name E. anomalum has been applied to plants from North Carolina, and plants from this region that fit the description of E. anomalum are derived by hybridization between E. mohrii and E. serotinum.  Another series of apomictic allopolyploids that occur primarily in Florida and have been called E. anomalum are actually derived from the combination E. mohrii/E. semiserratum, and these appear to fit the type of E. anomalum, which was collected in Florida.  Additional study is needed to assess whether either or both of these form independently reproducing populations or are primarily F1 hybrid derivatives.

Hybrids resembling Eupatorium rotundifolium.  The ultimate complications of hybridization affecting a species within Eupatorium can be found in E. rotundifolium.  The phenotype that is considered to be characteristic of the species involves sessile, ovate to almost rotund leaves, and appears to be passed on to almost all hybrid derivatives involving E. rotundifolium.  A further complication is that there are both sexual diploids and apomictic polyploids within E. rotundifolium, and these are often not simply separated from other hybrid derivatives.  The sexual diploid populations appear to be limited in geographic distribution to a narrow belt across northern Florida and southern Georgia. Analysis of herbarium material at TENN showed relatively few specimens that were labeled as E. rotundifolium to actually be simply that species.  A few names have been proposed, primarily as varieties of E. rotundifolium, that include some of the hybrid combinations.  Hybrids with E. perfoliatum are not infrequently encountered across a wide geographic range, and have a distinctive phenotype that has been called E. rotundifolium var. cordigerum or E. x cordigerum.  Though these have been described as being simply F1 hybrid derivatives, the plants at a minimum appear to be robust and may continue to persist in areas where one or both parental species have disappeared.  Hybrids with E. semiserratum have been called E. rotundifolium var. scabridum, and are relatively widespread.  At least two, and possibly more, hybrid derivatives involving E. rotundifolium and E. sessilifolium occur.  One morphological syndrome is that of E. godfreyanum, in which the leaves are clearly longer and narrower than in E. rotundifolium.  Another has been called E. rotundifolium var. ovatum (E. pubescens when recognized at the species level).  It is not yet clear why the same hybrid combination might produce different phenotypes, but two possibilities are that there are different dosages of the diploid genomes or that it involves a maternal effect.  There are also hybrid derivatives in the Ozark region involving E. rotundifolium and E. lancifolium.  These were noted by Sullivan (1972) but it is unclear whether they are F1 hybrids or independently occurring allopolyploid apomicts.  Sullivan (1972) also described hybrids involving E. rotundifolium and E. mikanioides.   Two misconceptions have complicated study of E. rotundifolium and its hybrids.  One was the long term misapplication of the name E. vaseyi as applying to a hybrid derivative of E. rotundifolium/E. sessilifolium (as noted above, it is actually E. petaloideum/E. sessilifolium; the name E. godfreyanum replaces it for the E. rotundifolium/E. sessilifolium combination).  The other was the misinterpretation of E. pilosum as a variety of E. rotundifolium (as E. rotundifolium var. saundersii).  Careful study shows E. pilosum to be a distinct species that includes both sexual diploid and apomictic polyploid populations, and it resembles E. rotundifolium only in having the leaf shape being similar to some of the hybrid derivatives involving E. rotundifolium (particularly those with E. semiserratum).  

In addition to apomictic allopolyploids derived from interspecific hybridization, there are also numerous hybrid combination within Eupatorium that have been documented.  These will be listed in the accompanying table (not yet available).