Aliso: A Journal of Systematic and Floristic Botany Aliso: A Journal of Systematic and Floristic Botany Systematics and Relationships of Fallugia Systematics and Relationships of Fallugia

The paper presents a systematic monograph of Fallugia (Rosoideae, Rosaceae) consisting of one variable species, Fallugia paradoxa. Morphological. cytological and molecular data clearly support its relationship with Geum s.l. rather than Purshia-Cowania. with which it is often allied. The species was named twice independently in 1825 from the specimens and drawings, respectively. assembled by Sesse and Mocifio for their proposed Flora Mexicana. The paper discusses the nomenclatural history, morphological variation, and the polygarno-dioecious mode of reproduction of the species. No infraspecific taxa are recognized.


INTRODUCTION
Fallug ia p aradoxa (D. Don) Endl., the Apache plume, is a co mmon shr ub in de sert s of the southwe stern United States an d northern Me xico . The specie s has long been co nside red rel ated to Cowania stans bu riana Torr., the cliffros e now placed in Purshia as Purshia stansburiana (Torr.) Henrickson (Henrickson 1986). In mo st phylogenetic treatments de aling with the relation ship of rosaceous ge nera (e .g., Bentham and Hooker 1862;Focke 1894 ;Rydber g 1913;Schulz-Menz 1964) Fallugia is placed with Cowania near or between Dryas and Geum (se nsu lato) within the tribe Dryoid eae of the subfa mily Rosoideae . Likew ise, in most so uthwes tern floras usin g a ph ylogenetic arrange me nt of ge nera, Fallug ia lies between Cowania and Geum (Abrams 1944 ;Kearney a nd Peebl es 1951;Mun z 1959). Howe ve r, rece ntly publ ished cy to log ica l dat a (Mc Arthur et al. 1983 ) and mol ecular data (Morgan et al. 1994) give strong evidence for Fallugia being closely rel ated to Geum, but not Cowania and Cercocarp us.
Whil e cy to log ica l and molecular data have provided strong eviden ce of the relationship of Fallugia, it will be shown that the same concl usion can be obta ined from ev e n a cursor y ex amination o f morphological and anato mica l character istics.

MAT ERIA L S AND METHOD S
Thi s study is based pr imaril y on e mpirical e vid ence derived from herbarium and field ob servations of Fallugi a, Purshia (Co wania), Geum, and from the literature . Material o f Fallugi a wa s borrowed from or observed at ARIZ, ASU , A-GH, CSLA, JEPS, MO , NY, RSA-POM, TEX-LL, UC , UNM, and US and observed in the fi eld in eastern California, Arizona, ce n-tral New Me xico , and throughout the Chihuahu an Dese rt, parti cul arly in ce ntral Coahuila. SEM was done with a Jeol JSM T200 Scanning Ele ctron Microscope at CSLA from dried materi als .

Comparative Structure of Fallugia, Pur shia (Cow ania) and Geum
Fallugia is often sy mpatric with Purshia stansburiana, and the two share the foll owing characteri stics. Both are shrubs in the southwes tern de se rts, with fibrous bark that se pa ra tes into many thin , ci nna monbrown layer s. Both have a basic ves riture of unicellular hairs along with pinnately divided leaves that are green above, and stro ng ly ves titured ben eath with revo lute margins that co nverge tow ard s thick ened ce ntral vein s. In both ta xa, the leav es are crowded on long and short shoots, and as in many Rosaceae, the blad es a bsc ise above the per sistent conj oined leaf bases and stipules. Both produce large flow ers. The hypanthi a bear 5 large, imbricate sep al s, 5 orbicular petals, many stamen s with yellow, longi scidal anther s borne on s lender filaments, and multipl e ovaries w ith terminal , non -articulating sty les that form accrescent, plumose awn s on the mature I-seeded ac he nes. Th e seeds have basal radicles and no endosperm . In both, the fruit s are wind d isper sed .
Fallugia ( Fig . 1-1 9), how e ver, has a large number of distinctive traits. In sa ndy ar royo s, Fallugia produ ces rhizome s and tend s to form colonies (Fig . 4 ,5) . Vegetative por tions have a distinctive orange-r ust-co lored lepidote und erstory ves titure (Fig. 24) and seedling lea ves have scatt er ed uni seri are, multi cellular, stipitate glands (Fig. 26 ;Kyle et al. J 986). Its mature leav es are egl andular, thin a nd pinn ati fid, or so me times bipinnatifid (Fig. 27 ,28 ). The flowers are produced at the tips of elongated stems, either solitary or in irregular corymbs (Fig. 6 , 17). The ri m of the hemispherical hypanthium has conspicuous leaf-like bracts that alternate with the sepa ls to form an epicalyx (Fig. 29,30). The hypanthia are den sely hirsute within. The sepals have one or more linear, leaf-like terminal or subterminal appendage s (Fig . 29, 30). The petals are white; the ovaries number 50-120, borne in a spiral pattern on a raised conica l receptacle each on a distinct, hirsute stalk, and each ovary has 2 superposed amphitropous basal ovules, one borne above the other (Fig. 15). The fruit wall s are 2-veined, with thin lateral walls, and the stigmatic surface extends about 0.2-0.3 mm down the style (Fig. 16). Furthermore, FalLugia is polygarno-dioecious, with so me plants producing pistillate flowers (with redu ced, s terile stamens; Fig. 32, 34 ) and other plants have larger staminate flowers with fertile stamens with larger anthers on longer filaments (Fig. 31, 33 ), and pistil s that do not develop, except that in some plants the terminal flowers of a stem may be perfect and produ ce fruit.
In contrast, Pursh ia (C owania) has been shown to have a deep root sys te m. grows in rocky habitats, and is not rhizomatou s (Henrickson in prep. ). It has multiseriate sti pirate glands on stems and hypanthia and se ssile gl ands im bedded in the leaf surfaces. The flowers are borne on the lateral short-shoots all along the upper ste ms. The rim of the obconic hypanthia lack s an epicalyx ; the petals are cream-yellow in color, the ovaries are far fewer (4-10) and whorled at the base of the hypanthium , and the stigmatic surfaces extend 2-3 mm down one side of the styles. The ovules are solitary, but, as in Fa llugia. amphitropous with a basal micropyle . The mature fruit walls are thick and strongly 10-12-veined and all flowers are perfect and complete. Furthermore, species of Purshia (Cowania) , like Cer cocarpus, form a symbiotic relationship with the Actinomycete Frankia, which results in nodul ation and nitrogen fixation in the host roots (Nelson 1983, see Schwintzer andTjepkema 1990). Such nodulation and association has not been reported in Fallugia.
Many of the shared characters appear associated with adaptation to desert habitats, i.e., the woody growth habit, the dissected leaves with revolute margins, the long-and short-shoot development. Probably the most conspicuous difference lies in the epicalyx of leaf-like bracts alternating with the sepals on the rim of the hypanthium in Fallugia (Fig. 29, 30) . This feature occurs elsewhere in the subfamily Rosoideae, in tribe Potentilleae (i.e., Potentilia. Fragaria, Ivesia, Horkelia, Sibbaldia, etc.), but in that tribe, the styles are lateral on the ovary and are deciduous at maturity. An epicaJyx also occurs in largely herbaceous Geum and allies (Geum, Waldsteinia. Colura etc.), in which the styles are, as in Fallugia, terminal (not lateral) (Fig. 16).
Geurn s.l. is highly diverse, consisting of up to 12 subgenera (Gajewski 1957(Gajewski , 1959; several of these are treated as separate genera by Rydberg 1913, Yuzepchuk 1941, and others. Some of these subgroups have plumose styles as in Fallugia and are wind dispersed (i.e., in subgenera, sections or genera Sieversia, Neosieversia, Oreogeum, and Erythrocoma) and others (subgenus or section Geum) the style is articulated with the terminal portion deciduous and the tip of the basal portion hooked (for animal dispersal) . In Waldsteinia and Colura, in contrast, the styles are deciduous at the base and the achenes are papillate-hirtellous. Gajewski (1957Gajewski ( , 1959 considers their fruits to be ant dispersed . Those taxa with long plumose styles also have short stigmatic areas as in Fallugia. Basic observations show that Fallugia and Geum s.l . have identical fruit-wall structure with thick dorsal and ventral traces with thin, inconspicuously vascularized lateral walls, whereas Purshia (including Cowania) and Cercocarpus have thicker fruit walls with many (10-12) thickened veins. However, Fallugia has two ovules, while all the Geum s.l. observed had but one ovule as does Cowania and its cohorts.
Fallugia with its epicalyx of bracts on the hypanthial rim, 2-veined achenes, small stigmatic surfaces, and high number of spirally arranged ovules on an expanded cylindrical, hirsute receptacle, shares many more characteristics with the largely herbaceous-suffruticose Geum s.l. than with Purshia (Cowania) and cohorts. Also, many of the Geum group have creeping rootstocks and become colonial (Yuzepchuk J941) as does Fallugia. Gajewski (1959) considered that the most primitive members of the Geum group had long plumose styles adapted to wind dispersal and that these plants migrated southward from high latitudes into the high mountains of Europe and North America . He considers that it is perhaps from this stock that Fallugia arose and adapted to the arid habitats of western North America, perhaps entering what is often called the Madro-Tertiary Geoflora (Axelrod 1958). Nevertheless, a number of characteristics appear to be confined to Fallugia: the distinctive orange-rust lepidote-stellate vestiture, two ovules per ovary, and its polygamo-dioecious mode of reproduction.
Studies of Rosaceae pollen also favor relationships of Fallugia with Geum and Waldsteinia. Hebda and Chinnappa (1994) note that there is distinct scuJpturing variation in the Rosoideae and that the above noted three genera (along with Coluria and Orthurus-segregates of Geum s.l.) share a distinct striate microperforate sculpturing pattern.
Interestingly, hybrids have been reported between Purshia stansburiana (n = 9) and Fallugia (n = 14) (Blauer et al. 1975). Baker et al. (1984) have shown that one such hybrid was just an aberrant individual of Purshia (as Cowania) stansburiana (2n = 18) with stamens developing into pistils and petals developing into sepal-like structures. In my studies of Purshia, I have found other collections of Purshia (Cowania) stansburiana with similar aberrant conditions . No hybridization between Purshia (Cowania) and Fallugia has ever been documented.
Based on the characteristics shared by Fallugia and Geum s./., one might be tempted to place Fallugia within Geum s.l. However, while molecular data obtained from the chloroplast rbcL gene by Morgan et al. (1994) placed Fallugia closest to Waldsteinia and Geum s.l., both taxa of Waldsteinia and Geum sampled shared a duplication of 19 base pairs near the 3' end that distinguished them from Fallugia. This implies that Fallugia is a sister group to both Geum and Waldsteinia and was not derived from either (Morgan et al. 1994). Eriksson et al. (1998) found a similar pattern in their ITS sequence data separating Geum and Waldsteinia from Fallugia and indicate that the Fallugia-Geum-Waldsteinia clade, with the inclusion of Rubus, is the sister group to all the rest of the genera of Rosoideae.
The initial rbcL molecular data on the Rosaceae by Morgan et al. (1994) indicate that the traditional Ro- soideae, long recognized on the basis of achene-type fruit and x = 7, 8 or 9, is not a monophyletic group. In their phylogenetic analysis, they found that members of three tribes of the traditional Rosoideae, the Kerriae (Ne viusia, Rhodotypos), Adenostomateae (Adenostoma) and part of the traditional Dryadeae (C ercocarpus, Purshia), all n = 9(-8), were allied elsewhere in the family, the latter two groups with follicular-fruited groups. Within the Dryadeae, Cercocarpus. Purshia, [and Dryas. (D . Morgan, pers. comm.)] are allied with the follicle-fruited Lyonothamnus (n = 27), while Fallugia, Waldsteinia and Geum (n = 7) were retained as a sister group to the remainder of the Rosoideae along with the x = 7(-8) Alchemilla, Poteniilla, Fra ga ria. Rosa, Agrimonia, Rubus, and Filipendula. Morgan et a1. (1994) also note that their rearrangement is supported by the distribution of various chemicals (sorbitol, cyanogenic glycosides, ellagic acid, flavones) as well as the distribution of rusts and nitrogenfixing root nodules. Clearl y achene-type fruits have been deri ved independently from follicles more than once.
Of interest, D. Don (1825) originally described our taxon as a Sieversia, with which it agrees in almost all characteristics except that Fallugia is a larger shrub with imbricate (not valvate) sepals, paired ovules, and pinnately divided (not truly pinnate) leaves.

Taxonomic History
The species we now know as Fallugia paradoxa, was named as new to science twice in 1825 . It was initi ally published by David Don, who was the librarian-curator of the Lambert Herbarium in England from 1820 to 1836. Lambert had acquired sets of the Se sse and Mocifio Mexican collections through Pavon beginning in 1817 and it was from these collections that D. Don observed and published Sieversia paradoxa D. Don . in 1825(Miller 1970). In the same year, Seringe (1825) also published the taxon w ith in Geum in his Rosaceae treatment for the senior de Candolle's Prodromus. His description was based on Sesse and Mocifio drawings of their Mexican collections and the name attributed to de Candolie.
The story of the Sesse and Mocifio expeditions, their collections and drawings of the collections has been detailed in a series of publications by McVaugh (1977McVaugh ( , 1980McVaugh ( , 1982McVaugh ( , 1987McVaugh ( , 1990McVaugh ( , 1998McVaugh ( , 2000 and a recent CD-Rom containing images of original illustra-tions. Sesse and Mocifio, together or se parately, collected throughout central , southern and western Mexico, coastal Alta California, southern Alaska, with additional expeditions into Central America and the West Indies, from 1787 to 1803. Their goal wa s the preparation and publication of a great Flora Mexicana. To provide illustrations to the same, a series of about 1800 paintings were prepared by artists who accompanied the expeditions. Sesse and Mocifio returned to Spain in 1803, and for various political reasons the flora was never published. The specimens arrived in Spain in 1804 and duplicates were later distributed by Pavon. Sesse died in 1808 , and during the French occupation of Spain, Mocifio was exiled in 1812; he took about 1300 of the illustrations to A. P. de Candolle in Montpellier, France. When conditions improved in Spain, Mocifio requested the return of the illustrations from de Candolle, who at this time was in Geneva, and de Candolle employed mo re than 100 artists in 1817 to make copies of about 1000 of the illustrations in a period of ten days. Not all drawings were copied; some of Mocifio's illustrations were duplicates and these were given to de Candolle, while others were considered, by de Candolle, too common to copy. Mocifio returned to Spain with the original illustrations, eventually became ill and died in 1820. The original illustrations fell into unknown private hands and were not seen again by botanists until 1979-they now reside at Hunt Center for Botanical Documentation (Me-Vaugh 1982 ) where they form the " Tom er Collection." According to McVaugh (1980, p. 102) about 279 names were published in de Candolle 's Systema and the Prodromus from these copies of the original illustrations, including a Geum that is referable to Fallugia (Seringe 1825). Sesse and Mocifio's floras, Flora Mexicana (Sesse and Mocifio 1894) and Plantae Novae Hispaniae (Sesse and Mocifio 1887 to 1891) were eventually published in Mexico in installments from 1887-1897 in the journal La Naturaleza, but by that time most of the species had been described and published elsewhere; their floras included a few species of Geum, but none relate to what is now known as Fallugia. The Lambert herbarium, from which D. Don described his Sieversia paradoxa, was sold in auction in 1842 after Lambert's death (Miller 1970 (Raphael 1970;Gage and Stearn 1988 )  Prodromus in mid-November of the same yea r (Stafleu and Cowan 1976) , and the treatment included, as Geum? cercocarp otde s, a new species based on Sesse and Mocifio illustrations. The taxon wa s attributed to de Candolle (" D C adnot. in   the copied plate at Geneva and that de Candolle was very involved with the ultimate manuscript generated for the Prodromus. The copies of the original illustrations, that would have been available to Seringe, are still at Geneva (McVaugh 1980, p. 106) and some, including that of Geum? cercocarpotdes, were published with Sesse and Mocifio's Flora Mexicana (1894). Some 50 years later the younger Alphonse de Candolie (1874) published ten sets of tracings of the copies of 279 taxa that were described as new taxa in the de Candolle 's Prodromus or Systema as Calques des Dessins [tracings of the copies]. Tracing No. 297 is of Geum? cerco carpoides, which I have seen on microfiche.
Few specimens from the northern deserts regions are included in the Sesse and Mocifio collections, as their expeditions did not extend into the area (Me-Vaugh 1977). As location data did not accompany the collections or illustrations, the source of the two type specimens has not been determined, although one can presume they came from what was then Mexico as the species occurs only in what was then Mexico. Me-Vaugh (1977, p. 181)   Cocke re ll c ircumscribed his F. m ic rantha as havin g s ma ller flowers . sma ll s ta me ns. sma ll (sterile) anthers. the ca rpe ls protruding a bove the anthers. brac ts no t div ide d o r tooth ed and the o ute r se pals wi th 1(-2) a ppe ndag es -bas ically as havin g male s teri le (i .e .. pist illate) flow ers. No Cocke re ll spec ime ns were ob se rved in any her barium that co uld se r ve as a typ e of thi s taxon. Cockere ll did note that he o bser ved spec ime ns in Woo ton's her bariu m and it is fro m these co llec tio ns tha t the lect ot yp e is here designated . Cock ere ll's charac terizatio n of F. 1'. var. a cumina tu as mal e fe rtile (stamin ate ) is nor in ag ree me nt w ith a ll sy nty pes as o nly the lec tot ype at NMC and iso lectoryp es at GH a nd NY ac tua lly have stami nate-flowered ste ms present [in eac h ca se mix ed with ma le-s te rile (pis tilla te) stems]-therefore Cock e re ll' s c haracte riza tion of Wooton 's va r. acumina to as male -fe rti le is not acce pted .
The species is widely distributed from the Eastern Mojave Desert in California to southern Nevada, southern Utah, southern Colorado, through central and eastern Arizona, through most of New Mexico, southern Oklahoma, Trans-Pecos and central Texas and south in Mexico to central Coahuila, northern Durango, Zacatecas, Chihuahua and northern Baja California del Norte at (1500-)2500 to 7000(-9500) ft elevation (Fig. 37). The species extends from the Mojave Desert region with its winter rains, through the uplands of Arizona and New Mex ico with its winter and summer rain s and into the Chihuahuan Desert region with summer rains.
Throughout much of its range the species typicall y occurs in sandy to rocky drainage s, but in highlands of Arizona, New Mexico and Texas it also occurs on higher rocky areas often in juniper. oak , pinyon or pine gra sslands. Common associates in the Moj a ve Desert, where it occurs from 4000-6500 ft elevation, include :

DISCUSSION
Fallugia paradoxa exhibits considerable variation, much of it attributable to water resources. Well watered plants typically produce well developed, slender, long-shoot stems 30-50 cm long with well spaced nodes and long lateral branches, each terminating in one or more flowers. In contrast. plants growing in drier conditions are more strongly branched with much shorter long-shoot stems, shorter internodes, a tighter branching pattern and a gnarled appearance. Overall there is considerable variation in the thickness of young stems, and the amount of lateral branching in both stems and inflorescences.
The plants vary in their ability to form sucker shoots that results in the formation of distinct colonial clusters of plants. In California, most plants form distinct colonies when growing in sandy arroyos, with new plants clearly forming from woody rhizomes (Fig. 4-5) . In the Chihuahuan Desert, Fallugia often occurs on upland sites where it is not colonial, as well as sandy drainages where it mayor may not be colonial. In Arizona and New Mexico, I found there to be considerable variation in the development of offshoots from a plant, but colonial plants were more common in sandy drainages. Due to the variation in this feature , it can not be recognized taxonomically.
Leaf size and lobing also are highly variable and again appear to reflect moisture available to the plant.
In most specimens long-shoot leaves are larger than short-shoot leaves. The long-shoot leaves typically are 10-25 mm long, divided into 5-7, slender to broad, sometimes secondarily lobed divisions (Fig. 27). These leaves appear to be produced during periods of strong terminal growth when adequate water is available and associated hormone production is high. Leaves of the short-shoot spurs. in contrast. are often smal ler, 5-8 mm long. mostly divided into 3. sometimes 5 divisions, with some leaves being undivided (Fig. 28). Long-shoot leaves were also observed to be 3-lobed or undivided in some specimens. As conditions dry. the larger long-shoot leaves often absci se as do the older short-shoot leaves allowing the plant to be in balance with its water resources . This pattern is found throughout the range of the species. showing no geographical consistency.
Plants also vary considerably in their total vestiture. In some plants leaves are sparsely hirtellous with only short erect hairs or with a few scattered longer appressed hairs; other plants are more villous-sericeous with a moderate to dense covering of both short and long, straight and curved slender hairs (Fig. 25). This longer vestiture is usually also present on stems and flowers . The vestiture is variably glabrescent and falls from overwintering leaves . Vestiture is variable throughout the range of the species and exhibits no geographical patterns .
Variation also occurs in the bracts and sepals that border the hypanthium both as to size of the structures, the density of their vestiture, and the amount of lobing in the bracts and sepals (Fig. 29-30) . The bracts, that alternate with the sepals on the margin of the hypanthiurn , are leaf-like in structure. They may be short or long, unlobed or variously lobed in the distal half or even sometimes divided to the base so that two individual bracts appear to occur between adjacent sepals. The sepals are imbricate. strongly vestitured on the exposed outer surface. with the innermost sepals having broad, thin lateral unvestitured margins. The sepals typically are broadly ovate in shape and are rounded below a slender terminal or usually subterminal green, leafy tip. When subterminal, the tips may be separated to 1.1 mm from the actual margins of the innermost sepals. The outermost sepals sometimes have three, separate, slender tips (Fig. 30). On occasional plants the outermost, or all sepals, are not distinctly rounded below the tip, rather the sepals are distinctly acuminate with convex margins below the acuminate tips. This was one of the characteristics used by Wooton (1898) to distinguish his variety acuminata. Therefore, analysis was undertaken to determine if there were any sepal or bract characteristics that could be used to distinguished geographical subunits within the species. Wooton (1898), in describing his var. acuminata, noted that plants from southern New Mexico tended to have acuminate outer sepals that contrasted with the threetoothed sepals from Texas and Mexico. While specimens from Texas and adjacent Mexico do tend to have three teeth on the outer sepals, the characteristic is not consistent on all flowers of a particular plant and similar three-toothed sepals are found in plants throughout the range of the species. A brief analysis of the frequency of multi-toothed sepals using available dried herbarium specimens for each state revealed the following percentages of specimens with at least some 3toothed sepals (n = number of collections diagnosed): Mexico: (Chihuahua and Coahuila) 91.7% (n = 26); Texas: 95.1 % (n = 62); New Mexico 23.6% (n = 72); Arizona 42.3% (n = 52); and California 40.6% (n = 32). Likewise there was no consistency in the epicalyx bract lobing, with both simple and terminal lobed and often completely divided bracts occurring on the same plant and often even on the same calyx of a flower. A separate analysis of the frequency of specimens with lobed versus unlobed or divided bracts was also conducted from available dried herbarium specimens. The percentages of specimens (n) that showed at least some lobed or divided bracts were: Mexico (Chihuahua and Coahuila) 86.4% (n = 22); Texas 51.8% (n = 56); New Mexico 35.9% (n = 64); Arizona 50% (n = 56); and California 38.7% (n = 31). While these data show a west-east trend, it was considered that the sepal and bract characteristics cannot be used to recognize geographical subunits within the species.
The plants also show considerable variation in flower size throughout the season and male flowers typically have larger petals than pistillate flowers (Fig. 31 ,32,35). Flowers produced during the active growing season are often large, with petals 10-21 mm long and 8-17 mm wide. Nevertheless, the same plants may continue to produce occasional flowers during the dry season that have smaller petals only 6-9 mm long and 6-8 mm wide. In any population petals of male-fertile flowers typically are larger than those of pistillate plants. In one brief study in the Clark Mountains in California petals on male-fertile plants averaged 16.2 X 12.4 mm in size, whereas those of the pistillate flowers were 10.8 X 8.4 mm in size. Similar differences were noted throughout the range of the species. Throughout its range the species is polygamo-dioecious with some plants bearing male-sterile, pistillate flowers that bear abundant fruit and have sterile anthers 0 .3-0.4 mm long on reduced filaments. Other plants in the same populations are male fertile or staminate, producing anthers 0.7-1.2 mm long on longer filaments. Some of these plants produce no fruit and can be considered true staminate plants, while other pollen-producing plants have scattered fruit-producing flowers, with some plants producing few such flowers and others many fruit-producing flowers . Usually these hermaphroditic flowers are terminal on a stem and inflorescence. Jn all pollen-producing flowers , anthers mature well before the pistils and the pistils develop in a broad central column within the flower. Richards (1986Richards ( , 1997 considers this as a type of dicliny, specifically polygamo-dioecious dicliny, where there are potentially five types of plants: plants with only pistillate flowers ; plants with staminate flowers; plants with all hermaphroditic flowers; plants with both hermaphroditic and staminate flowers; and plants with both hermaphroditic and pistillate flowers. Not all of these five types of flowers will occur in a species. Fallugia has plants with three of these flower types: pistillate, staminate and both staminate and hermaphroditic. A series of field tallies of pistillate plants versus strictly staminate plants (i .e. , with no fruit developing) versus staminate-hermaphroditic plants (staminate, but some flowers developing fruit) was conducted throughout the range of the species (Table I) .
As can be seen from the data, the relative frequencies of staminate plants and staminate-hermaphroditic plants is not consistent across the range of the species. Plants sampled in California have many more true staminate plants than the other populations sampled and those from Mexico showed the fewest true staminate plants.
In pistillate plants, all ovaries mature and develop their plume-like elongate styles. The styles apparently will elongate with or without pollination. At Rancho Santa Ana Botanic Garden in Claremont, California, the display gardens have only pistillate plants; the staminate plants have been removed as they do not show c haracteristic " Apac he plum es." In these isolated pistill ate plants that presumabl y are not poll inated, all ovaries still form e longated plumose styles. However, the plumose awn s do not greatl y e longate and the fruit are e mpty -they lack mature seeds. Mill er and Ven able (2000) discuss the ori gin of dioecy associated with pol yploidy, whi ch may pertain to this genus.