The onion platycaule (now in Alliaceae) from Soldier Creek, Modoc County. Photograph by J. Vale; its use courtesy of the Jepson , UC.

CONSIDER THE LILIES by Dean G. Kelch

Consider the lilies of the field, , it wasn’t a member of the are a host of other “imposters” (see ho w they gro w; they to il no t, true lily . Table 1). neither do they spin. This particular response reflects These common names refer to Matthew VI: 28. a general tendency toward vague- that aren’t necessarily closely ness on the part of botanists. The related to each other. Not all of here is some controversy as term lily has been used in a very them are placed in the lily , to just what wildflowers are loose fashion to describe a broad s.l. ( lato, in the broad Treferred to in the biblical array of plants. Consider the great sense). The most you can say is that passage above. Some have suggested diversity falling under the rubric they are all monocots. However, if that it is the autumn daffodil, lily. Besides the true lilies (Lilium) you count the water lily (Nymphae),a lute.a Most scholars are that include the tiger lily, the Ori- you can’t even say that! agreed, however, that whatever the ental lilies, and the Asian lilies, there The monocots (short for mono-

VOLUME 30:2, APRIL 2002 FREMONTIA 23 of most monocots are TABLE 1. PLANTS THAT HAVE THE three-parted, an ancient character in flowering plants that is otherwise COMMON NAME OF LILY THAT ARE NOT found in only a few ancient lineages TRUE LILIES ( LILIUM ) of non-monocots. In many of the “lilies” s.l., the three and three (Hemero calli)s are similar in appearance and Toadlily (Tricyrtu)s are referred to collectively as . Corn lily () The vast majority of monocots are Desert lily (Hesperocall)is herbaceous. The few woody groups Trout lily () (such as palms and ) have odd Lily-of-the-valley (Co nvallaria) wood that is unlike the dense, often Lily turf () ringed wood of non-monocot Rain lily (Zephyranthe)s and shrubs. Voodoo lily (Amorphophallu)s As you can see from the above Sego lily (Calochortu)s characteristics, most monocots are lily (Zantedeschi)a relatively simpler in structure than Lily-of-the-Nile () other flowering plants. Therefore, Scarborough lily (Vallo ta) the various natural groups of mono- Plantain lily () cots are less different from each Ginger lily (Hedychium) other than many groups of non- Glory lily () monocots. For example, a strap- Checker lily () shaped does not lend itself to as Fortnight lily (Dietes) much shape variation as a net-veined Foxtail lily (Eremurus) leaf (think of the great leaf Spear-lily (Doryanthe)s in such non-monocot families as the Palm lily (Curculigo) carrot family, Umbelliferae, and the rose family, Rosaceae). This explains ) are one of the largest monophyletic groups of flowering plants. The remainder of the flow- ering plants are generally referred to as the dicots (), but this latter group is not monophyl- etic, as the monocots seem to have arisen from within the “dicots” early in the history of flowering evolution. The monocots include such diverse groups as , palms, gingers, grasses, irises, orchids, and, of course, lilies s.l. The members of these groups all have a single leaf (), while most flower- ing plants have two seed (think of a bean or alfalfa sprout). The cotyledon often isn’t around to look at for very long, but there are other characteristics that are common in monocots. Most monocots have linear or strap- shaped leaves with parallel veins. These leaves generally lack a well- defined (stalk), and the base Sand lily, montanum (now in Anthericaceae), from Lassen County. forms a sheath around the stem. Photograph by B. Ornduff; its use courtesy of the Jepson Herbarium, UC.

24 FREMONTIA VOLUME 30:2, APRIL 2002 TABLE 2. GENERA INCLUDED IN THE LILIACEAE SENSU LATO IN THE JEPSON MANUAL AND THEIR ASSIGNMENTS TO LILIACEAE SEGREGATE FAMILIES DISCUSSED IN THE TEXT

Assignment of genera to families is based on the references provided. Agavaceae Lilium (lily) Sc o lio pu(fos etid adder’s tongue) (camas) Strepto pus(t wisted-stalk) Chlo ro galum( soap plant, amole) Hastingsia Hesperocalli(sd esert lily) Veratrum (corn lily, false ) Hesperoyucc(aour lord’s candle) Xero phyllum(b ear-grass, ( Spanish bayonet) Indian basket-grass) Alliaceae (d eath camas) Allium (onion, garlic) Ipheio n(st ar flower) ( bog asphodel) (fa lse garlic) Anthericaceae sensu stricto Smilax (green briar) Leucocrinum(s and lily) Themidaceae Andro stephium Alo e Bloomeria( goldenstar) Aspho delu(sa sphodel) Bro diaea Convallariaceae (b lue dicks, lily) ( false lily-of-the-valley) Muilla (b eargrass) ( Ithuriel’s spear, pretty face) Smilac ina (false Solomon’s seal) Hyacinthaceae ( bog asphodel) (grape ) (possibly within the Melanthiaceae) Liliaceae (wakerobin, trillium) Calochortu(sm ariposa lily, globe lily) Uvulariaceae Erythronium(fa wn lily) (’s cup; potential placement) Fritillaria (fritillary) ( twin bells) why the lily family (Liliaceae) has way. In the late 19th century, it approach was not useful in discern- long been a “catch-all” group that was common to recognize all “lil- ing the natural groups within the included most monocots with some- ies” bearing flowers with inferior Liliaceae s.l. Luckily, the work of what showy, radially symmetrical ovaries as the family many botanists resulted in the accu- flowers and no obvious specializa- (). While propo- mulation of a lot of information tions. (This contrasts with the very nents of this approach could boast about the microscopic and chemical specialized flowers of orchid family of its simplicity, it resulted in such characters of monocots. A group members, for example, which have diverse plants as daffodils and led by the Danish botanist Rolf the fused male and female flower being lumped together, while Dahlgren decided to synthesize all parts forming a column.) yuccas (close relatives of agaves) of this information and revise the The lily problem was no secret were left in the Liliaceae. The tra- classification of monocots. This to botanists. There never was a ditional Liliaceae s.l. was like a gi- work was published in 1985 as The doubt that the lily family contained gantic house of cards: remove one Families o f Mo no co tyledo(Dnsahlgren, groups of quite disparate plants. The piece and the whole structure was et al. 1985). In this book, plants problem consisted in how to break liable to come crashing down. formerly in the Liliaceae s.l. were up the family in a practical, natural Obviously, the single character placed in 40 different families in

VOLUME 30:2, APRIL 2002 FREMONTIA 25 Beargrass, (now in Convallariaceae), from the Kingston Mountains. Photograph by C.S. Webber; its use courtesy of the Jepson Herbarium, UC. three different orders! Botanists and cluding the members of the mono- system of Dahlgren et al., such as horticulturists are, by and large, cotyledons. (For more information the listing of cultivated taxa in Kelch rather conservative, and so it has on this approach, see the sidebars (2000), will have to be modified in taken a while for these changes in on pages 4–7 and page 15.) the future. to be accepted. The preliminary published re- While it may seem as if our en- During the 1990s, acceptance sults of this research confirm many tire classification of plants is sliding of the Dahlgren classification has of the conclusions reached in The into the abyss, things aren’t as bad as been hastened by the explosion of Families o f Monoco tyledoThens. re- they appear. The molecular data, by the academic discipline molecular sults also show that the story may and large, has confirmed much of . By comparing the DNA be more complicated than we had our understanding of plant relation- sequences for a particular or hoped. This should come as no ships. Many traditional plant groups genetic marker for a large number surprise, as increased knowledge seem to be monophyletic. Of the of , biologists found large leads us to a more sophisticated un- seriously unnatural groupings, per- numbers of new characters to use in derstanding of things. This is how haps the families of monocots rep- elucidating the evolutionary rela- science advances. Nevertheless, resent the most extreme case. How- tionships of living organisms, in- publications that closely follow the ever, there were earlier indications

26 FREMONTIA VOLUME 30:2, APRIL 2002 that the traditional taxonomy of The inclusion in the agave Hyacinthaceae and close to Hosta monocots was seriously flawed from group, based on evidence from com- based on serological data. the work of Dahlgren et al. Other parison of sequences of the chloro- Placing Hastingsia, , groupings that have proven unnatu- plast gene rb cL, of some and Camassia in the Agavaceae ren- ral include the figwort family genera that Dahlgren had placed ders that family difficult to identify (Scrophulariaceae s.l.; see the article in the Hyacinthaceae (Camassia, based on macroscopic characters. It by Olmstead on page 13), and the Hastingsia , and Chlo ro galum) was is possible that further sampling will dogwood family (Cornaceae). something that no one had pre- identify two related lineages: one a Liliaceae s.l. will be broken dicted (Chase et al. 1995). These desert-adapted Agavaceae and an- down into a number of smaller fami- plants are very similar to such Old other the -adapted Hostaceae lies (see Table 2 on page 25). To World Hyacinthaceae taxa as squill (this name replaces the illegitimate prepare you for this change, I have () and grape hyacinth (Musc ari). Funkiaceae). If, as seems likely, these included the following listing of the However, based on the rb cL analy- taxa are all hopelessly related, we larger families, with information on sis, Camassia, Hastingsia, and may have to place them in one big, their defining characteristics. I have Chlo ro galum are not closely related dysfunctional family. Until we de- also included a discussion of why to the Old World taxa. Some previ- velop a field lens powerful enough certain genera are placed in par- ous evidence had provided clues. to count , or invent a ticular families. Some of these con- For example, the genus Camassia is pocket DNA sequencer, this group clusions are pretty firm, but others quite distinct from Old World may be hard to define based on are tentative at this point in time. Agavaceae. In The Families of Bear-grass, tenax (now in the Melanthiaceae), from Glacier National Park, Da, hlgren, Clifford, . Photograph by C. Webber; its use courtesy of the Jepson Herbarium, UC. and Yeo recognized that this family was not a natural one if it included such old world taxa as dragon (Drac aena), Australian grass tree (Xantho rrho e),a and flax (). As the earth be- came drier in the late Tertiary Period, drought-adapted, fibrous- leaved, giant herbs evolved several times from smaller, herbaceous plants native to moister, shadier areas. Therefore, the superficial similarity of these plant groups in different areas of the globe is the result of convergent evolution and not shared ancestry. In the case of the Agavaceae, it seems likely that the desert-adapted plants like agaves evolved from a woodland herb like Hosta via some intermediate plant resembling the desert lily (Hesperocall)i sor tuberose (Polianthe)s. This hypothesis re- ceives support from the chromo- somes of Hosta, which are similar in size and number to those of Agave and Yucca. Also, the flowering spike of Hosta is very similar to those of such plants as the desert lily and tuberose; these latter genera are in- terpreted as including some of the less specialized members of the tra- ditional Agavaceae.

VOLUME 30:2, APRIL 2002 FREMONTIA 27 field characters. However, all in- smell. In , our only native surprisingly, these were once placed cluded species have a of genus is Allium itself. All Allium in the Agavaceae. However, a look , often undulate leaves. The species have the onion odor, as does at their small, starry, cream-colored flowers are borne on a or the introduced Ipheio n uniflo rum. flowers reveals their true relation- panicle, with along its length Nothoscordum inodor,u imntroduced ship lies with the forest herbs in- and subtending the flowers. The from , is also a mem- cluded in the lily-of-the-valley fam- petals are nearly free, generally be- ber of the Alliaceae, but lacks the ily (Convallariaceae). Further DNA ing joined at the base. typical odor. The Jepson Manuare-l sampling of genera indicates that The monotypic genus Hespero- ports this latter species as a noxious the Nolinaceae probably evolved yucca contains the variable H. weed in California. from within Convallariaceae, and whipplei. This genus rarely has been The other lineage comprises should be included within this fam- recognized as distinct from Yucca. Western and Southwestern North ily. Therefore, in California, the It can be easily distinguished from American plants. It includes such Convallariaceae includes two groups Yucca s.s. (sensu strict,o in the narrow genera as , blue dicks with similar flowers, but different sense) in having ro- (Dichelostemma), and Ithuriel’s spear ecology; one consists of a group of settes. The technical, generic dif- (Triteleia). These have been sepa- woodland herbs, such as false lily- ference is that the style is slender rated in current classifications as of-the-valley (Maianthemum) and with a capitate , while the the Themidaceae. false Solomon’s seal (Smilacina), style of Yucca is stout, with a six- Several characters can be used while another consists of large, lobed stigma. Although this may to distinguish Alliaceae from woody, desert plants (Nolina). seem a relatively minor distinction, Themidaceae. While Alliaceae have Liliaceae/Uvulariaceae. So, the style morphology is extremely a pair of bracts that encloses the what is left of the actual Lily Fam- important in the plant’s relation- flower buds, Themidaceae have sev- ily? In The Families of Monocotyle- ship with its obligate , eral bracts that do not enclose the dons the Liliaceae s.s. is sadly deci- the yucca . Interestingly, young flowers. Alliaceae have a true mated. In California, it includes the Tegeticula maculat, athe re- (composed of swollen leaf true lilies (Lilium spp.), the fritillar- sponsible for of H. bases) with a membranous coat, but ies (Fritillaria spp.), and the trout whipplei, is the sister to all other the in Themidaceae lilies ( pp.). Molecular true yucca moths (Pellmyr et al. is a (composed of stem tis- evidence indicates that more gen- 1996). According to evidence from sue). Most, if not all, Themidaceae era are related and could be included chloroplast restriction site analysis, lack the onion odor. in this family. Calochortu, splaced by Hesperoyuc cias more closely related Convallariaceae. Another Dahlgren et al. in its own mono- to Hesperalo ethan it is to Yucca s.s. group of desert-dwelling, large, fi- typic family in The Families of Mono- (Bogler and Simpson 1995). brous-leaved plants recognized by cotyledonsis, closely related to the Alliaceae/Themidaceae . The Dahlgren are the Nolinaceae, a Liliaceae s.s., despite its having well- onion family, as circumscribed in group including the bear-grasses differentiated sepals and petals The Families of Monocotyledoisn verys, (Nolina), the (Dasylirion), and (most members of the Liliacaeae easy to define. It includes all plants the pony-tail palm (). Not s.s. have tepals). with scapose flower spikes, a supe- Also, the genera Strepto pusan d rior , and flowers borne in Tofieldia glutinosa ssp. occidentalis, bog Sc o lio pu, splaced by Dahlgren in the . Once again, the evidence asphodel (now in the Tofieldiaceae), from Uvulariaceae, belong here. Some from the gene rb cL has indicated Gold Beach, . Photograph by C.S. might advocate including these Webber; its use courtesy of the Jepson that there are at least two unrelated Herbarium, UC. -like taxa in a separate fam- lineages in this family in North ily, the Tricyrtidaceae, but I would America (Fay and Chase 1996). rather see them added to the lily The true Alliaceae contains the family. This family is difficult to few Old World genera, as well as characterize. It has showy flowers, the circumboreal onions (Allium), often with spotted tepals. The in- and the South American taxa. This florescence is usually leafy, but can group seems closely related to the be scapose, in which case it is single- amaryllis family (Amaryllidaceae), flowered (as in Sc o lio pusan d some a group that differs in having flow- Tulipa). are not black-coated. ers with an inferior ovary. World- The of this group of related wide, most of the genera of Alliaceae taxa are quite variable, as they can have the familiar onion or garlic be dry or fleshy.

28 FREMONTIA VOLUME 30:2, APRIL 2002 Disporum smithii, twin bells (now in the Uvulariaceae), from Eureka, California. Photograph by C.S. Webber; its use courtesy of the Jepson Herbarium, UC.

Melanthiaceae/Narthecia- evolved from a lineage within the P.F. Yeo. 1985. The Families of Mono- ceae/Tofieldiaceae . The Melan- Melanthiaceae. This result is well- cotyledonsS. pringer-Verlag. Berlin, thiaceae once included all three of supported by the initial molecular . these families, but it turns out that data, but it is so odd based on mor- Fay, M.F. and M.W. Chase. 1996. Res- many of the characters are prob- phology that I prefer to wait for urrection of Themidaceae for the ably primitive characters in the further evidence and confirming Brodiaea , and recircum- monocots. These include the three- macroscopic characters before scription of Alliaceae, Amarylli- daceae, and . Taxo n lobed or three-parted , the lack merging the two families. 45:441–451. of a black coating on the seeds, and Hickman, J.C. 1993. The Jepson the presence of calcium oxalate crys- Manual: Higher Plants o f Califo r- tals in the cells. There are three LITERATURE CITED nia. University of California Press. distantly related groups in this larger Berkeley, CA. family. All three groups are repre- Bogler, D.J. and B.B. Simpson. 1995. Kelch, D.G. 2000. What happened to sented in California. Both Tofield- A chloroplast study of the Agava- the lily family? Pacific iaceae and Nartheciaceae have ceae. Sytematic Bo tan y20:191–205. 61:76-79. equitant (V-shaped in cross ) Chase, M.W., M.R. Duvall, H.G. Pellmyr, O., J.N. Thompson, J. leaves. Tofieldiaceae have tailed Hills, J.G. Conran, A.V. Cox, L.E. Brown, and R.G. Harrison. 1996. seeds, while the of Narthe- Eguiarte, J. Hartwell, M.F. Fay, L.R. Evolution of pollination and mutu- Caddick, K.M. Cameron, and S. ciaeae have specialized air spaces. alism in the yucca moth lineage. Hoot. 1995. Molecular phylo- Melanthiaceae have various types Americ an Naturalis t148:827–847. genetics of . In Mo no c o tyle- of leaves, but they are not equitant. dons: Systematics and Evolutioedsn, . They also lack the specialized seed P.J. Rudall, P.J. Cribb, D.F. Cutler, Dean Kelch, University and Jepson Her- and characters of the other two and C.J. Humphries. Whitstable baria, 1001 Valley Sciences Building families. Litho Printers Ltd. Whitstable, En- #2465, University o f Califo rnia, Berke- One surprise from rb cL analy- gland. ley, CA 94720-2465. dkelch@sscl. ses is that the Trilliaceae may have Dahlgren, R.M.T., H.T. Clifford, and berkeley.edu

VOLUME 30:2, APRIL 2002 FREMONTIA 29