1. Introduction to the Trilliaceae

  2. History of the Taxonomy

  3. Relationships within Trilliaceae

  4. Recent Molecular and Morphological Studies

  5. Previous studies


Introduction
Trillium L., Paris L., Kinugasa Tatew. and Sutô, and Daiswa Ref. are genera of plants found in north temperate forests which make up the family Trilliaceae as defined by Takhtajan in 1987. Confusion has arisen about the classification of the group: Are the traditional generic groupings correct, or would these species be better considered to be a single genus? If the traditional genera are correct, what are the relationships within this group? In addition, even though the plants are morphologically distinct and quite similar systematists disagree as to whether the plants comprise a separate family or should be included within a larger family such as the Liliaceae in the broad sense.

back to top

History of the Taxonomy
The genera that are always placed in the Trilliaceae are Trillium and Paris. Morphologically, these plants consist of a stem topped by a whorl of leaves with a single actinomorphic flower.

Table 1 contains the type species and generic synonymy for the genera included in this family. Takhtajan splits Paris into Paris, Kinugasa Tatew. & Sutô, and Daiswa Raf. Other authors, notably Hara, et al. (1978) recognize both Trillium and Trillidium Kunth. In Species Plantarum published in 1753, Linnaeus named four of the species included in this family: Trillium sessile, T. cernuum, T. erectum, and Paris quadrifolia shown at the top of this page or in Figure 1. He separated the genera based on number of locules and anthers. Other names that have been used for the genera in this family and the rationale for being named are given in Table 2.

The composition and placement of the Trilliaceae has varied since its recognition by Lindley in 1846. It has been afforded tribal status within various families, reincorporated back into the Liliaceae, or elevated to familial status within different orders within the monocots. Takhtajan, alone, placed the family in four different orders (1969, 1980, 1987, 1997). Although Paris and Trillium are always paired when segregated within or from the Liliaceae, two other genera, Medeola L. and Scoliopus J.Torr., are frequently included in the family. Sometimes both taxa are included (Dalla Torre and Harms (1908) and Watson and Dallwitz (1991)), others include only Medeola (Engler, 1888; Melchoir, 1964), and still others include only Scoliopus (Dahlgren, et al., 1985; Brummit, 1992). Clintonia was included by Bentham and Hooker (1883). The additional taxa included in the family are shown in Figure 2. An overview of the placement and composition of this family is given in Table 3.

Recent studies such as those using the rbcL gene (Chase, 1994; Chase, et al., 1995a; Kato, et al., 1995b), chloroplast DNA (cpDNA) restriction-site analysis (Davis, 1994, 1995; Kato, et al., 1995a), and combined morphological-restriction site analysis (Stevenson, 1994; Chase, et al., 1995b) are reinforcing the concept that these taxa are not a part of the Liliaceae (sensu stricto) and should in fact be an independent family. It should be understood that the taxa included in the Trilliaceae are not removed from the Liliaceae alone. The Liliaceae sensu lato is a heterogeneous assemblage of genera united only by the fact that they possess 6 perianth segments, usually 6 stamens, and a 3-carpellate ovary (Smith, 1977). Within the broadly defined Liliaceae there is extensive variation for many features. For example, leaves vary from simple, undifferentiated, and strap-like leaves to having leaves with a distinct petiole and lamina; leaves are alternate, opposite, or whorled; leaves have parallel or palmate venation; the infloresence is a raceme or solitary; flowers are perfect or unisexual; floral symmetry is actinomorphic or zygomorphic; the perianth is composed of tepals or distinct sepals and petals; anthers are basifixed or dorsifixed; the ovary is 3-locular with axilary placentation or unilocular and parietal. It is when the Liliaceae is more narrowly defined as a more homogeneous family that the Trilliaceae, as well as many other well-defined families such as Agavaceae, Amaryllidaceae, Asparagaceae, Melanthiaceae, and Smilacaceae are separated out. The molecular evidence indicates that a narrower definition of Liliaceae is probably more consistent with phylogeny.

back to top

Relationships within Trilliaceae
The limited results from the rbcL studies (Figure 3), suggested that Paris and Trillium may not represent monophyletic genera; in fact there is no clear and consistent set of morphological characters that separates any of the groups within the family. Traditionally, Trillium has been segregated from Paris based on floral merosity. In Trillium, leaves and floral parts occur in 3s or multiples of 3; however, 4- and 5-merous Trillium monstrosities exist. In Paris, merosity ranges from multiples of 4 to multiples of 11, but trimerous Paris exist. In fact, T. taiwanense was recently listed by Huang, et al. (1989) in the synonymy for P.fargesii var. brevipetala. Other than Paris japonica, which has been placed in every genus except Daiswa, this is the first occurrence of cross-generic synonymy. Indeed, the genus Trillidium underscores the difficulty that sometimes occurs in separating these genera: Trillidium, when named (Kunth, 1850), contained what is now known as P. japonica and T. govanianum. Trillium govanianum possesses characters normally seen in only Paris such as filiform petals and ellipsoid, monosulcate pollen (Zomlefer, 1996), but it shares having trimerous flowers with Trillium. Paris japonica appears to possess chromosomes from both Trillium and Paris (Tatewaki and Sutô, 1935; Haga, 1937). Characters that it shares with Trillium include having a large showy inflorescence even though it has the filiform petals characteristic of Paris, but characters it shares with Paris include merosity as well as narrow leaves. See Figure 4 for photographs of Paris japonica and Trillium govanianum as well as T. grandiflorum and T.stamineum two other morphologically distinct Trillium.

Relationships within Paris have been studied by Takhtajan (1983) and Li (1984b, 1986). Takhtajan separates Paris from Kinugasa and Daiswa on the basis of a slender, creeping rhizome, round ovary, and fruit being an indehiscent, black berry. Daiswa is characterized by a thick rhizome, an ovary with a transverse rim, a fleshy loculicidal capsule, and seeds with a scarlet sarcotesta. Kinugasa shares several features with Paris such as slender stigmatic branches, indehiscent berry, and seeds without a sarcotesta; it also shares features with Daiswa such as a thick rhizome, and angular ovary. Li prefers to retain the single genus Paris, but separates Takhtajan's genera into similar subgenera and sections on the basis of placentation and number of locules, presence/absence of an aril, fruit type, rhizome morphology, and basal merosity (e.g., parts 2x-6x the base number). Even though some of these characters may adequately separate between Daiswa and Paris, they do not adequately differentiate between Paris sensu lato and Trillium (i.e., none of the characters occur solely within Paris).

Relationships among the sessile-flowered Trillium were studied by Freeman (1969a, 1975). He separated the sessile-flowered Trillium, subgenus Phyllantherum, into informal groupings based on morphology. The "Trillium recurvatum group" possess basally recurved sepals, incurved anthers, and slender elongated rhizomes. Prolonged anther connectives, introrse or extrorse dehiscence, angled ovaries, linear stigmas, and thick compact rhizomes distinguish the "Trillium sessile group." The "Trillium maculatum group" is separated on the basis of lateral anther dehiscence, no stigma prolongation, and subulate stigmas. These groupings were formalized by Murray (1983, 1984) as Section Recurvata, Section Sessilia, and Section Maculata. Most sessile-flowered Trillium have spotted leaves, but the trait occurs sparingly in pedicellate Trillium as well as Paris.

Although there have been numerous studies of various aspects of the pedicellate-flowered Trillium (see Section 2.3), and there are several studies that draw some preliminary phylogenetic conclusions; there has not been a single unifying study comparable to the ones that Li has performed on Paris and Freeman has done with the sessile-flowered Trillium. The last monographic study of the pedicellate-flowered Trillium was done by Gleason in 1906.

back to top

Recent Molecular and Morphological Studies
Recent studies show that the Trilliaceae are a cohesive group. In studies that include most of the Trillium taxa, the monophyly of the sessile-flowered sub-genus of Trillium is evident. However the pedicellate Trillium and Paris are polyphyletic. These studies have included general analyses of the monocotyledons as well as studies specifically on the family, and have used sequences from the rbcL gene (Chase, 1994; Chase, et al, 1995a; Kato, et al., 1995b), cpDNA restriction site data (Davis, 1995; Kato et al., 1995a), as well as morphological characters (Goldblatt, 1995; Kawano and Kato, 1995).

Results of rbcL studies show that even though the Trilliaceae sensu stricto segregates out into a distinct, cohesive unit (Figure 3b) obviously separate from other plants examined, not only do the taxa examined not separate out into genera, they do not separate out into recognizable morphological groups. The earliest study (Chase, 1994) used several taxa within the Trilliaceae sensu lato (Trillium kamtschaticum, T. sessile, T. erectum, Paris polyphylla, P. tetraphylla, Kinugasa, Medeola, and Scoliopus) as part of a study of the monocotyledons. Mark Chase, who performed the molecular analysis, said, "Medeola is some distance from Trillium and is closer to Lilium, ... From our molecular work, it seems pretty clear that there is only one genus in the family. ... It is possible that with our limited sampling that we are not getting the 'right' tree, but it is worth taking a broad view to determine if the genera are good." Figure 3 shows excerpts from cladograms that were produced in Chase study. The Kato study (Kato, et al., 1995b) used a subset of the taxa used in the Chase study (monocots versus Trilliaceae), except that T. erectum was omitted. Not surprisingly, the results of the Kato study corroborated those produced by Chase.

The restriction-site analyses performed thus far have shed no light on the composition of Trilliaceae or generic relationships within the family. Because the earliest restriction-site analysis (Davis, 1995) used only Trillium grandiflorum from the Trilliaceae, it only indicates that Veratrum and Amianthum were the nearest neighbors to Trillium as was indicated in the rbcL study. Kato, et al. (1995a) also performed a restriction site analysis, using 24 species of Trillium, with Paris tetraphylla as the outgroup to the Trillium taxa. The monophyly of the sessile-flowered Trillium is supported as is the monophyly of the T.erectum complex.

In a morphological study, Kawano and Kato (1995) showed the monophyly of i subgenus Phyllantherum (the sessile-flowered species of Trillium), but because the sessile-flowered variety of T. pusillum fell out with T. pusillum var. pusillum, the sessile-flowered plants are not monophyletic. Even though the pedicellate-flowered Trillium are polyphyletic, three species groups showed up in their analysis. Their analysis indicated that the T. erectum group is monophyletic, the T. grandiflorum-T. ovatum group is monophyletic, but the rest of the pedicellate Trillium do not form a distinct clade. Stevenson and Loconte (1995) also performed a morphological analysis of the monocots that included Trilliaceae only as a family rather than as a distinct species. In this analysis, Trilliaceae was basal to the rest of the monocots and was sister to the Stemonales rather than the Melanthales.

Much of the current molecular work places the Trilliaceae basal to the Liliales confirming its place as a primitive monocot. This study does not address the ordinal placement of the family, but addresses the placement of the taxa within the family. Are the traditional generic groupings correct, or is there only one genus as the molecular work to date seems to indicate? Because no clear picture of the composition of the Trilliaceae has emerged from these studies, the family is circumscribed as containing Trillium and Paris sensu lato for the present.

back to top

Previous studies
Though many authors have studied various aspects of Trilliaceae either through taxonomic groupings or a character across several different members of the family, no one has yet conducted a thorough morphological study of the family. Over the years, there have been several generic and sub-generic monographs of various portions of the family including Paris (Franchet, 1888; Li, 1984b, 1986), pedicellate Trillium, (Gleason, 1906), North American Trillium (Gates, 1917), southern Appalachian pedicellate Trillium (Barksdale, 1938), east Asian Trillium (Samejima and Samejima, 1962), sessile-flowered Trillium (Freeman, 1969a; Murrell, 1967), North American pedicellate Trillium (Ihara and Ihara, 1978, 1982), Daiswa (Takhtajan, 1983), and Trillium (Samejima and Samejima 1987; Case and Case, 1997).

Trillium has 5 large chromosomes which have been referred to as the Drosophila of the plant world (Warmke, 1937), and have been the subject of extensive studies of structure and function. Many of these studies were simply DNA or chromosomal studies rather than studies of Trillium chromosomes in specific. Paris, even though similar chromosomally, has not been as extensively studied until recently (Miyamoto, et al., 1990, 1992). The heterochromatin forms specific banding patterns when treated with cold and then stained that have been found to be of taxonomic significance. Various aspects of chromosome morphology and banding patterns have been studied by Bailey (1951, 1954a, 1954b, 1958), Haga (1934), Haga and Watanabe (1966) and Kurabayashi (1952, 1957). Chromosome morphology has also been used to identify the putative parents of the Japanese hybrid species of Trillium (Haga, et al., 1974). In addition, Warmke (1937) could usually distinguish between four Pacific Coast trilliums. Trillium rugelii could be separated cytologically from other species in Rutherford Co., North Carolina by Serota (1969). Other studies elucidated differences between the H-segments of Japanese and American species (Darlington and Shaw, 1959).

Berg studied the embryology of Scoliopus, Paris, Trillium, and Medeola (1962a) and seed dispersal and morphology of Trillium (1958). The use of pollen morphology as taxonomically significant has been extensively studied by Patrick (1982), Takahashi (1982, 1983, 1984, and 1989), and Wei (1995).

One of the reasons that the Trillium life-cycle has been studied so extensively is because the process from seed dispersal to flowering takes from four to seven or more years (personal observation; Patrick 1973). These studies have included such aspects of the life of the plant as growth stages, pollination and breeding systems, fruit and seed dispersal, and life span (Ihara, 1973; Ohara and Kawano 1986a, 1986b; Kawano, et al., 1986, 1992; Ohara and Utech 1986, 1988; Ohara, 1989). Even though this information is important biologically, it appears to convey little or no information of phylogenetic importance.

back to top

Back to main Trilliaceae pages

restored: 28 May 2004 sfarmer@goldsword.com