Two New Species of Hoffmannia (Rubiaceae) from Panama, with Remarks on Circumscription of Mesoamerican Species of the Genus William C. Burger Botany Department, The Field Museum, Chicago, Illinois 60605-2496, U.S.A. Abstract. The unusual degree of variation with¬ intermediate in a number of characteristics that in Hoffmannia populations and the difficulty in distinguish two closely related species, both rep¬ clearly defining some Mesoamerican species are resented by large suites of collections, we may be discussed. Two new species, Hoffmannia divyeri confident that we have identified hybrids. If there and H. fortunensis, both endemic to the Chiriquf are many intermediates between two distinctive highlands, are described. morphological populations, we may be dealing with a cline or subspecific differentiation. In both these Having worked with Hoffmannia for the Flora instances, however, we must first describe and Costaricensis series (Burger & Taylor, 1993), it characterize two different suites of character states, seemed appropriate to expand that effort and treat delimiting our species or subspecies. The problem the genus for the Flora Mesoamericana project. in Hoffmannia is that the variation is often so great, Work with the Costa Rican material had indicated and there is so little correlation between different that most species in the genus were very variable characters, that delimiting two closely related “spe¬ morphologically, and that making specific distinc¬ cies” among a large suite of similar collections can tions in some groups was extremely difficult or ar¬ be quite arbitrary. The only effective way to tackle bitrary. Unfortunately, studying over 3000 speci¬ this kind of problem is to separate out the most mens of Hoffmannia in the region covered by the distinctive entities first, then try to tease out other Flora Mesoamericana has made these difficulties distinctive taxa, before finally dealing with a large abundantly clear. complex of similar specimens. While individuals of most plant species vary Clearly, one of the most easily distinguished spe¬ considerably, there are usually concordant suites of cies in the genus is Hoffmannia vesiculifera Stand- consistent morphological traits that allow us to ley, which has two inflated elongate vesicles along identify and distinguish species from their close the lateral sides of the usually short petiole. Hoff¬ congeners. Floral dimensions and morphology are mannia kirkbridei Dwyer also has such vesicles, but usually sufficiently constrained so that they are that species has very different petioles and leaf useful in characterizing species. Leaf dimensions form, is known from only two collections, and may may vary greatly, but leaf form and venation pat¬ prove to be a variant of //. vesiculifera. Setting aside terns are usually also helpful in identifying and dis¬ the two collections tentatively assigned to //. kirk¬ tinguishing closely related species. Many species bridei, and using the vesicles as a defining aut- of Hoffmannia, however, exhibit a wider-than-nor- apomorphy for //. vesiculifera, we can bring together mal range of variation in both floral dimensions, a number of specimens that display a wide range vesture, plant size, and leaf size and shape. This of variation in both leaf and inflorescence charac¬ wider range of variation is rarely exhibited by any teristics. The plants themselves grow from 0.1 to one plant; it becomes apparent ordy when a larger 1.5 m in height when flowering, and range from suite of collections is gathered together. central Costa Rica to central Panama. The inflores¬ We use larger suites of collections to assess pop¬ cences vary from sessile congested fascicles to pe¬ ulation variation, characterize species, determine dunculate and distally branched paniculate forms. species differences, delineate ranges, and look for Leaves range from 10 to 30 cm in length and vary local or regional differentiation. If we are able to from elliptic to oblong or obovate. Calyx lobes may effectively separate congeneric material into two be 4—20 mm long; the corolla varies from about 10 closely related species, and then find that they con¬ to 20 mm in length. These extremes are never sistently grow in different areas or slightly different found on the same individual, and they do not char¬ habitats, we are reassured that they are indeed dif¬ acterize particular habitats, localities, or differing ferent species. If we find a few collections that are elevations. Rather, the extreme individuals appear Novon 9: 13-17. 1999. 14 Novon to he nothing more than unusual variants. Some of lections makes it clear that these differences are the more distinctive variants have been the basis bridged by many intermediates at all elevations. for erecting new species, but all can be linked with While the large, broad, brilliantly colored leaves the earliest type through intermediate collections. are only found at elevations above 1200 m, there In the eastern part of this species’ range, inflores¬ is sufficient variation at all elevations to make rec¬ cences are often borne on peduncles with small ognition of more than one species dubious. leaves at their apex (modified axillary branches). In Hojfmannia subauriculata Standley is quite dis¬ the same area, there are plants with flowers densely tinctive because of its long-tapering lamina base crowded in the leaf axils, as in the type of H. morii and winged petioles, but it also exhibits consider¬ Dwyer, based on Mori & Kallunki 5915 (MO). At able variation. The winged petioles are usually first, the two forms appear to be different species, rounded and subauriculate at the stem, but not al¬ but a larger series of specimens will include some ways. The plants are often short in stature with in¬ with intermediate inflorescence characteristics. In florescences usually borne at the lower leafless short, //. vesiculifera appears to be a species easily nodes, even at rooting nodes. The species ranges identifiable by the petiolar vesicles, larger calyx from Costa Rica to Colombia. Specimens from Da¬ lobes, generally short habit, and pubescence; but it rien, Panama, are distinctive because their consis¬ is a species that varies rather dramatically in vir¬ tently oblanceolate leaves tend to dry grayish, and tually all of its morphological features. their winged petioles are not aurieulate at the base. If we think of //. vesiculifera as being typical of Also the stems are often horizontal and rooting Hojfmannia species, wide patterns of variation in along the proximal nodes in eastern Panama. The other species of the genus become easier to under¬ name 11. eliasii Dwyer would be appropriate if these stand. What gives us problems is the fact that these plants of eastern Panama were to be separated from other species usually do not possess a similarly //. subauriculata as a distinct species, but it seems unique identifying trait. Hojfmannia vesiculifera best to consider them part of a single variable spe¬ and the poorly known 11. kirkbridei Dwyer are the cies. only species of the genus to have developed petiolar An extremely variable complex is found in the pouches large enough to house small ants. A num¬ Chiriquf highlands of Panama, for which the name ber of species, such as the two new species being H. pittieri Standley has often been used. These pop¬ described below, do have a suite of distinctive traits ulations may intergrade with material referable to and limited geographic range that make them rel¬ the earlier named 11. arborescens J. Donn. Smith of atively easy to recognize, but by the time we sep¬ Costa Rica’s Talamanca mountains. I have separat¬ arate out such species we are left with wide-ranging ed these collections by annotating those with usu¬ taxa that may be catchalls for large numbers of sim¬ ally axillary branched or cymose inflorescences as ilar-looking material. H. arborescens, and those in which the flowering Hojfmannia regalis (J. I). Hooker) Hemsley was inflorescences are simpler and largely confined to based on horticultural material grown in Europe, the lower leafless nodes as H. pittieri, consistent and the name has been used for some wild collec¬ with the types. Unfortunately, there are intermedi¬ tions from southern Mexico. The geographic origin ate collections both as regards their inflorescences of the original cultivated material was not known, and in other characteristics. It may be that this but similar material of Hojfmannia is found only in complex of forms should be further divided accord¬ Chiapas and Guatemala. The large leaves are oc¬ ing to specific morphological traits, but I suspect casionally broad and brilliantly colored with shades that this is unrealistic. A closer examination of the of purple beneath and a rich dark green above, the populations in Chiriquf, and especially their cytol¬ obvious reason for its introduction as a horticultural ogy, might help unravel this problem. There is a novelty. As in the case of 11. vesiculifera, there is report of within-species polyploidy in Hojfmannia great variation in leaf size, shape, and coloring. (C. M. Taylor, pers. comm.), and this may be a fac¬ Leaf bases, petioles, and inflorescences are also tor in the variability found within the genus. highly variable. The result of this extensive within- The most commonly collected species of Hojf¬ population variation is that there is no way to clear¬ mannia in Mexico and Central America is 11. ni- ly separate those specimens that have in the past cotianifolia (M. Martens & Caleoti) L. 0. Williams. been identified as Hojfmannia lineolata J. Donn. Based on material collected in Veracruz, Mexico, Smith and 11. riparia Standley. This latter material the name suggests plants with larger leaves. Among has been distinguished by its more often elliptic Veracruz collections there are large-leaved collec¬ leaves with well-defined petioles and a lamina base tions, a greater number of smaller-leaved collec¬ that is not so long-decurrent. A large suite of col¬ tions, and some intermediates. Based on this vari- Volume 9, Number 1 Burger 15 1999 Hoffmannia from Panama ation in the region in which the earliest type was sheets and their respective parts have been anno¬ collected, I believe it is best to interpret //. nico- tated accordingly. tianifolia very broadly. However, adopting a broad¬ In central Panama there are specimens of Hoff¬ er inteq>retation can cause difficulty in dealing with mannia nicotianifolia with more ovate leaves that distinctive variants that may be common in some have been identified as H. cercidifolia Dwyer by geographical regions such as Honduras or Nicara¬ Dwyer. However, the type of H. cercidifolia (Duke gua. 15597 (2), MO) was collected in Darien and ap¬ A large percentage of Hoffmannia specimens pears to represent a different species. Unfortunate¬ from Nicaragua dry with a characteristic orange- ly, the type has only a few subsessile immature brown or olive-green coloration and many have flowers and provides insufficient information to larger obovate to oblanceolate leaves. Their small come to any firm conclusion regarding its place¬ flowers and short pedunculate inflorescences with ment. In addition, there are no similar collections thin pedicels clearly ally them with H. nicotiani- from Darien to give us a better idea of the popu¬ folia. Because they are quite distinctive and char¬ lation that this collection represents. It seems likely acteristic of so much material from Nicaragua, they that the type of H. cercidifolia may prove to be an could be recognized as a separate species, first de¬ unusual collection of an earlier named South Amer¬ ican species, but it will require better sampling of scribed by L. 0. Williams as //. oreophila. However, the Darien populations before that likelihood can with much variation and many intermediate collec¬ be verified. tions, it seems best to include H. oreophila in a At present a draft treatment of Hoffmannia for broad circumscription of //. nicotianifolia. This is the Flora Mesoamericana includes 49 species. consistent with also treating H. angustifolia Stand- Having accepted many broader species concepts, I ley of Guatemala and Honduras as a synonym of have also tended to tentatively place unusual col¬ H. nicotianifolia. After a review of many collec¬ lections within a related described species rather tions, I believe a broader, more inclusive species than erect new names. However, two small groups concept is justified, making //. nicotianifolia a spe¬ of collections from the Chiriquf highlands appear cies that ranges from northeastern Mexico to east¬ to represent two new and distinctive species. ern Panama. However, larger circumscription does not eliminate the problem of having collections that appear to be intermediate with other closely related Hoffmannia dwyeri W. Burger, sp. nov. TYPE: Panama. Chiriquf: vicinity of Boquete, Cerro species. In the case of H. nicotianifolia there are Pate de Macho, SW slope, 8°46'N, 82°25'W, collections that appear to be intermediate with H. 1800-1950 m, 19 June 1987, Croat 66403 cryptoneura Standley, //. excelsa (Kunth) K. Schu¬ (holotype, MO; isotype, F). Figure 1. mann, H. honduremis Standley, H. laxa Standley, //. oreophila, H. pittieri, H. psychotriifolia (Ben- Species habitu cum Hoffmannia valerii Standley uptime tham) Grisebach, H. regalis, and //. steyermarkii congruens, sed differt coroll is longioribus, foliis basi cu- Standley. Whether any of the “intermediate” col¬ neatis, petiolis alatis, stigmatibus latis, habitatione alti- lections are true hybrid-like products of interspe¬ tudo alta (1700—21(K) m). cific gene flow or simply variants with character¬ Herbs 0.2-0.8 m tall, leafy stems terete, 2-6 mm istics of one of the other species is difficult to thick, subglabrous to villous with reddish brown determine. hairs 0.5—2 mm long; stipules 1—3 mm long, thin, Hoffmannia psychotriifolia is quite similar to H. deciduous. Leaves opposite, subglabrous to villous, nicotianifolia but differs from the latter by the usu¬ lamina obovate to elliptic-obovate or narrowly ob¬ ally longer corolla and by the consistently sessile ovate, 7—23 X 3—9 cm, base cuneate and decurrent inflorescences. The two are sometimes collected to¬ to form a wing along the petiole, attenuate or gether. The names Hoffmannia rotundata Standley abruptly rounded and subauriculate at the base, (holotype: Purpus 7268, MO) and H. chuipensis apex acute to acuminate, 2° veins 6—12 pairs; pet¬ Standley (holotype: Purpus 7268, US) were based ioles 2—12 mm long. Inflorescences axillary, 2-6 on a collection that Standley recognized as mixed. per node, 2-5 cm long, flowers in a close cymose All the herbarium sheets seen have some elements grouping or paniculate, peduncle 10-27 mm long, of the two similar species. Having reviewed the de¬ villous to subglabrous. Flowers on pedicels 1—4 mm scriptions and examined sheets from F, US, and long, hypanthium 2-4 mm long, villous with hairs MO, it appears that H. rotundata is a synonym of 1—2 mm long, calyx lobes 2-A mm long, narrowly H. nicotianifolia in a wide sense, while H. chiapen- triangular; corolla funnelform, 8—16 mm long, white sis is a synonym of //. psychotriifolia. The type with pink tips to pinkish red, with few distal tri- 16 Novon figure 1. Hoffmannia dwyeri W. Burger. —A. Flowering stern. —B. Flower. A based on Hummel 7412 (F) and Sytsma el dl. IH.'ifr (\i<)); B from Hummel 7412 (MO). chomes, lube 4—5 mm long; anthers ca. 3.2 mm corolla, and broad stigmatic lobes are additional long, subsessile; style ca. 9 mm long, stigma ca. 3 distinguishing features. Pubescence and floral mor¬ mm long, broadly ovate. Fruits oblong, 8—9 mm phology are similar to //. valerii Standley, but that long, 5—6 mm thick, pink to red-magenta, spongy, species differs in the form of the lamina base and seeds blue. petiole, and its lower elevation habitat. The new species has only been collected around the Cerro Hoffmannia dwyeri has an unusual characteris¬ de Pate Macho and upper headwaters of Rfo Palo tic: the leal base is cuneate-decurrent and the Alto, between 1700 and 2100 m elevation. It flow¬ winged margins are often revolute near the base, ers in March-June. forming a partly enclosed space abaxially. Its short John Dwyer, who studied Hoffmannia intensively herbaceous habit, long reddish brown multicellular over more than 20 years, recognized material of this hairs (when present), short densely flowered inflo¬ species as new in two of his annotations but did rescences, prominent calyx lobes, relatively large not publish either of the names. Because Dwyer’s Volume 9, Number 1 Burger 17 1999 Hoffmannia from Panama (1969, 1980) publications and many identifications or reddish green, corolla tube 1.5—2 mm long; an¬ have been so helpful to all those working with this thers 2.5 mm long. Mature fruits not seen. very difficult genus, it seems especially fitting that The usually long petioles and very narrow leaves a distinctive species should bear his name. with clearly demarked loop-connected secondary Paratypes. PANAMA. Chiriqui: vicinity of Boquete; veins close to the lamina margins are distinctive Cerro Pate de Macho, SW slope, 1800-1950 m. 19 June features. In addition, the leaves often dry very dark 1987, Croat 66403 (MO); Rfo Palo Alto mad to Chiriqui above and grayish beneath. Slender racemose inflo¬ border with Bocas del Toro Province near peak of Cerro rescences with a single raehis drying dark, small Pate Macho, 6200 ft., 20 Nov. 1978. Hammel 5814 (MO); flowers, and restriction to the Fortuna area of the end of road past Palo Alto to Bocas, 6200 ft.. 18 Mar. 1979. Hammel 6510 (MO); Palo Alto, 4.5 mi. NE of Bo¬ Chiriqui highlands between 1000 and 1500 m el¬ quete, forest along western branch of headwaters of Rio evation are additional distinctions. This species ap¬ Palo Alto, 6300 ft., 24 May 1979. Hammel 7412 (MO); S pears to be related to Hoffmannia laxa Standley, slopes of Cerro Pate Macho along Rfo Palo Alto, 1300— with which it shares the slender glabrous inflores¬ 1800 m, 11 Nov. 1981, Knapp et al. 2083 (MO); trail to Cerro Pate Macho, above Palo Alto, 1700-2KM) m, 15 cences and small flowers. There may be lower-el¬ Mar. 1982. Knapp et al. 4249 (MO); SE slopes and summit evation collections that have a few characteristics of Cerro Pate Macho, 4 km NE of Boquete, 1700-2KM) of both species, such as von Wedel 2363 (MO). m, 26 May 1981. Systma et al. 4836 (MO). Paratypes. PANAMA. Chiriqui: vicinity of Gualca 11 mi. from Planes de Hornito, La Fortuna on road to dam HofTiiiannia fortunensis Dwyer ex W. Burger, sp. site, elev. 4(XM) ft.. 9 July 1980. Antonio 5050 (MO); N.O. nov. TYPE: Panama. Chiriqui: S.O. del cam- del Campamento Fortuna (Hornito), 1000— 12(K) m. 16 pamento de Fortuna (Hornito), sitio de presa ago. 1976, Correa et al. 2515 (MO); Fortuna dam site, subiendo hasta la finca Pittf, 1000-1200 m, forest area. 1400-1600 m, 15 Sep. 1977, Folsom el al. 5568 (MO); near site of dam, lower slopes of Cerro For¬ 14 ago. 1976, M. D. Correa, R. Dressier & C. tuna, 1150 m. 18 June 1982. Knapp & Vodicka 5584 Garibaldi 2415 (holotype, PMA not seen; iso¬ (MO). types, F, MO). Acknowledgments. David Lorence kindly pro¬ Suffrutices ad 2 m alti, caulibus teretibus, glabris. Folia vided me with information relating to Hoffmannia ad 23 cm longa, lamina anguste lanceolata, glabra, in sic- from his draft manuscript “Nomenclator of Mexican co discolor, venis lateralibus 9—15 paribus. Inflorescentiae and Central American Rubiaceae,” which was very brevi. Flores glabri, lobis calycis brevibus, corollis 4—7 helpful. Mireya Correa verified that her collection mm longis. Fructus maturi non visi. 2415, the designated holotype for H. fortunensis, Slender shrubs 0.5—2 m tall, leafy stems 2—7 mm was present at PMA. David Lorence and Charlotte thick, terete, glabrous. Leaves opposite, blades 11— Taylor provided helpful suggestions for revision of 23 cm long, 1.5—8 cm wide, linear-lanceolate to the manuscript. I am grateful to the curators of the narrowly ovate-elliptic or elliptic-oblong, eharta- following herbaria for making their collections ceous, dark above, grayish beneath, glabrous, base available for study: CAS, F, MEXU, MO, US, WIS. cuneate and decurrent, apex acute to acuminate, 2° veins 9-15 pairs, loop-connected along the margin, Literature Cited petioles 15—65 mm long. Inflorescences 1—3 per Burger, W. C. & C. M. Taylor. 1993. Flora Costaricensis. axil, at lower or leafless nodes, racemose, 3—9 cm Family #202 Rubiaceae. Fieldiana Rot. n.s. 33: 1-333. long, glabrous, drying dark, peduncles 2—4 cm, Dwyer, J. D. 1969. The genus Hoffmannia (Rubiaceae) in Panama. Ann. Missouri Bot. Card. 56: 269-286. 0.3—0.4 mm thick. Flowers glabrous, pedicels to 10 -. 1980. Rubiaceae. In R. E. Woodson & R. W. mm long, hypanthium 2—3 mm long, calyx lobes Schery (editors), Flora of Panama. Ann. Missouri Bot. 0.2—0.5 mm long, corolla 4—7 mm long, pale green Card. 67: 1-522.