The Relationships of the () Author(s): M. G. Gilbert Source: Annals of the Missouri Botanical Garden, Vol. 81, No. 2 (1994), pp. 283-288 Published by: Missouri Botanical Garden Press Stable URL: http://www.jstor.org/stable/2992098 . Accessed: 19/01/2015 03:28

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

Missouri Botanical Garden Press is collaborating with JSTOR to digitize, preserve and extend access to Annals of the Missouri Botanical Garden.

http://www.jstor.org

This content downloaded from 210.72.93.185 on Mon, 19 Jan 2015 03:28:45 AM All use subject to JSTOR Terms and Conditions THE RELATIONSHIPS OF M. G. Gilbert2 THE EUPHORBIEAE (EUPHORBIACEAE)'

ABSTRACT

The Euphorbieae and Hippomaneae, though both usually placed in subfamily , differ in many inflorescence details and can be linked unambiguously only by one cryptic character: the rod-shaped starch grains in the latex. While the Hippomaneae have inflorescences very similar to those of many other members of the Euphor- biaceae, the Euphorbieaehave an inflorescence so specialized that one seems to be forced to relate it to other members of the family through a hypothetical ancestor with a synflorescence of axillary bisexual cymes more primitive than most extant taxa possibly other than Jatropha.

A primary aim of this symposium is to review between the Old World and New World species of critically Webster's (1994) classification of the Eu- ? How do the subgenera Euphorbia and phorbiaceae. This paper is an attempt to share a Lacanthis (sensu Gilbert, 1987) relate to the rest speculation arising from misgivings over the sup- of the ? Such an analysis, in my view, could posed homogeneity of the subfamily Euphorbioi- do much to clarify these relationships and thus the deae. This is not a presentation of the results of of the tribe as a whole. It would demand prolonged and detailed research but rather spec- a much greater knowledge of the possible origins ulations arising while producing a routine Flora of the Euphorbieae than is available. In a group account of the Euphorbiaceae for the Flora of so morphologically isolated, speculation is needed. Ethiopia (31 genera, 209 species), coupled with The Euphorbieae and Hippomaneae are includ- a long-standing interest in Euphorbia (the largest ed within the subfamily Euphorbioideae along with genus within the Ethiopian flora), particularly the three other tribes (Stomatocalyceae, Pachystro- succulent species. A Flora writer should delve into mateae, and Hureae) that have usually been as- the larger scale taxonomy of those families that sociated with the Hippomaneae (Webster, 1994), must be covered, but is rarely allowed the time to primarily on the basis of their caustic milky latex carry out anything more than superficial investi- from nonarticulate laticifers, frequently glandular gations on taxa not actually included in the Flora. bracts, and often highly reduced flowers that always Thus the following ideas must not be regarded as lack petals. This juxtapositioning has not always anything more than simple-minded speculations. been the case: Mueller Argoviensis (1866) and Determining the relationships of a group as spe- Bentham (1878) placed these groups at opposite cialized as the Euphorbieae used to be of little ends in their sequences. There are indeed major concern to a Flora writer. The matter became a differences in inflorescence morphology between little more relevant with the advent of cladistics, the two groups such that I felt forced to consider where the methodology demands a working hy- the possibility that the similarities were the product pothesis of probable sister groups and character of convergence. One unusual cryptic character, polarization. Sooner or later such methods should very characteristic, rod-shaped starch grains found be applied to the Euphorbiaceae in general, and in the latex, is a good contender for a svnapo- ideally to the Euphorbieae in particular, where morphy, which suggests that they do have a com- there are inconsistencies in current generic delim- mon origin. This in turn led to an attempt to re- itation. How, for instance, would a cladistic analysis construct a possible ancestral inflorescence type relate Chamaesyce, very widely recognized as a from which the modern could have evolved distinct genus, to the various sections within Eu- most parsimoniously. phorbia subg. Agaloma and the rest of Euphor- The rod-shaped starch grains have not been bia? There are many other interesting problems reported in other members of the Euphorbiaceae, waiting to be tackled: What is the relationship although many taxa appear never to have been

I The author was employed by the Ethiopian Flora Project, funded by the Swedish Agency for Research Cooperation with Developing Countries (SAREC), at the time this paper was prepared. 2 Flora of China Project, Missouri Botanical Garden; address for correspondence: Department of Botany, Natural History Museum, Cromwell Road, London SW7 5BD, U.K. ANN. MISSOURI BOT. GARD. 81: 283-288. 1994.

This content downloaded from 210.72.93.185 on Mon, 19 Jan 2015 03:28:45 AM All use subject to JSTOR Terms and Conditions 284 Annals of the Missouri Botanical Garden

surveyed (Rudall, 1987), and their discovery else- lanthus but also various red-flowered Euphorbia where in the family would be a serious challenge species, pollinated by birds. In most members of to the taxonomic integrity of the Euphorbioideae. the Hippomaneae the styles are undivided, long The toxicity of the latex has also been mentioned and tapered, and are usually characteristically cir- as a linking character but the major compounds cinately coiled when young. There are not many involved-diterpene esters occur also in subfam- observations of pollination. The pendent inflores- ily Crotonoideae (Beutler et al., 1989). The pollen cences and long stigmas of some species suggest of the Euphorbioideae is rather uniform (Punt, that wind pollination is a possibility, but other gen- 1987), but the pollen type is apparently plesiomor- era have well-developed nectaries within the inflo- phic; thus the similarity should be treated with rescences indicating some form of animal pollina- caution. Morphological evidence is even less clear- tion, perhaps most often by small unspecialized cut. There are superficial similarities in that most insects (Bawa et al., 1985), but in at least one taxa in both groups have a sparse to nonexistent case, Mabea occidentalis Benth. (Steiner, 1983), indumentum of simple hairs but there are many predominantly by bats. Thus the significance of the exceptions in both major tribes, such as species of striking differences in the styles is not clear. An- Chamaesyce and Agaloma within Euphorbia and other possibly significant difference is the frequen- members of the Mabeinae and many species of cy of succulence within the Euphorbieae where it Stillingia within the Hippomaneae, and such ev- has clearly evolved several times independently, idence can only be regarded as essentially negative: whereas succulence appears to be almost absent in it does not disprove the possibility of a relationship. the Hippomaneae, recorded only from a few Bra- The tendency toward very reduced flowers, rare zilian species of Stillingia (Rogers, 1951). in the family as a whole, has also been invoked to The contrast in inflorescence organization be- support a common grouping. It is possible, theo- tween the Hippomaneae and the Euphorbieae seems retically, to select a reduction series for the flowers so great that one must consider whether the sirn- within the Hippomaneae from genera such as Ma- ilarities between them could be the result of con- bea, which have well-defined perianths and male vergence rather than an indication of common flowers with many stamens, through to Dalember- ancestry. If the most reasonable (parsimonious) tia, in which the male flower is reduced to a single hypothetical common ancestor has to have char- sepal and a single stamen and the female flower acters such that it would have to be placed within has only vestigial remnants of the perianth. This one of the other subfamilies, notably the Crotonoi- sequence could be regarded as continued by the deae, rather than the Euphorbioideae, the Eu- Euphorbieae, where the male flower is always re- phorbioideae would have to be regarded as a grade duced to a single naked anther and the female rather than a lade. flower to a naked ovary. There is evidence of a There seems to be little or no case for ques- female perianth in many genera within the Eu- tioning the homogeneity of the tribe Euphorbieae phorbieae, including some species of Euphorbia, as it is clearly defined by the very peculiar inflo- but the only clear indication of a male perianth, rescence the cyathium distinct from all other beyond an articulation between pedicel and fila- inflorescence types seen within the family. Lin- ment, is the vestigial male perianth seen in An- naeus and other very early authors, plus one of thostema and . There has been a the two giants of nineteenth-century Euphorbi- temptation to suggest that this sequence gives a aceae taxonomy, Baillon (1874), thought that the true guide to relationships, but this must be ques- cyathium was a hermaphroditic flower. However, tioned could the Hippomaneae really have given the alternative theory that it was an inflorescence rise directly to the Euphorbieae? was mentioned by Lamarck as early as 1788 and The major difference between the two groups is is now unquestioned, though there is still disagree- the inflorescence structure, which is discussed at ment in the exact interpretation of its organization. length below. Details of floral morphology, most The cyathium consists of a cupular receptacle with notably the very different styles, also suggest that marginal lobes usually regularly alternating with the Euphorbieae and Hippomaneae may not be glands, sometimes quite complex in structure, en- closely related. In the Euphorbieae the styles are closing a whorl of groups of male flowers inserted relatively short, usually divided, and have more or below each lobe, plus their associated bracteoles, less capitate stigmas. Studies indicate that most and a single, central female flower, often reduced species are pollinated by a variety of often unspe- to a naked ovary. There is quite a lot of variation cialized insects, the exception being a group of within this theme with regard to features such as New World taxa, most obviously the genus Pedi- the number of parts within the involucre and the

This content downloaded from 210.72.93.185 on Mon, 19 Jan 2015 03:28:45 AM All use subject to JSTOR Terms and Conditions Volume 81, Number 2 Gilbert 285 1994 Relationships of the Euphorbieae

amount of fusion between them and the occasional regard to the relative positions of male and female presence of a small perianth on the male and female flowers, and must be regarded as a special case. flowers, but no genera can be regarded as having To convert a typical Hippomaneae inflorescence inflorescences truly intermediate between a cy- into a cyathium one must postulate a considerable athium and any other inflorescence recorded within condensation of the main inflorescence axis, a rel- the family. There has been much debate on the atively straightforward and common evolutionary nature of the cyathium but the weight of evidence process. However, at the same time one must also (e.g., Schoute, 1937) seems to support the scheme postulate a complete reversal in the relative position suggested by Eichler (1878), in which each invo- of the male and female flowers or the loss of the lucral lobe/bract subtends a monochasium of male proximal female flowers plus the de novo produc- flowers. tion of a terminal female flower. The latter process The cyathium contrasts very strongly in orga- seems particularly difficult to account for unless nization with the inflorescences of the vast majority one invokes the terminal allomorphic flowers seen of other members of the family, including the other in some species of Acalypha. There would also tribes placed within subfamily Euphorbioideae by have to be a reversal of the general trend of the Webster. There is a fundamental division within reduction of the distal male inflorescences. An al- the Euphorbiaceae between the uniovulate tribes, ternative scheme would be to derive the cyathium with only one ovule per ovary locule, and the from a group of thyrses. Such a scenario is very biovulate tribes, with two ovules per locule. The similar to that proposed below and must be given biovulate tribes are distinct in so many features of consideration. The one fact against it is the cymose anatomy and morphology that they will not be arrangement of the groups of male flowers within considered in detail. It might, however, be worth the cyathium. It is difficult to imagine a racemelike making the observation that the axillary fascicles inflorescence giving rise to such a grouping. of flowers characteristic of the vast majority of A more parsimonious scenario is to postulate biovulate taxa must be considered rather special- that both the Hippomaneae and the Euphorbieae ized reduced inflorescences. Most other uniovulate evolved from a common ancestor which had a tribes, including the Hippomaneae, have inflores- thyrse-a synflorescence in which an indefinite cences that are generally regarded as cymose, but fertile axis produced a spiral series of axillary bi- many show an essentially racemelike organization sexual cymes (Fig. 1). The cymes could be either in which the cymose nature of the ultimate ele- dichasial or monochasial or, perhaps quite likely, ments is only betrayed by the fact that some of initially dichasial with monochasial ultimate the male flowers are in very reduced cymules rather branches such as is frequently seen with Jatropha. than solitary as in a true raceme. The contrast The Hippomaneae inflorescence could (and surely with the Euphorbieae is further increased by the did) evolve by a simple process of reduction with fact that, as in the vast majority of the genera of the proximal cymes being reduced to single female the Euphorbiaceae, the female flowers are proximal flowers and the distal cymes reduced to cymules and the male flowers distal. Such an arrangement of male flowers or eventually to single male flowers. within an elongated inflorescence has a structural The Euphorbieae would have to be derived by a logic: female flowers produce bulky, longer-lived more complex evolutionary process centered pri- fruits requiring the transport of nutrients; male marily on a great condensation of the main syn- flowers are usually smaller, short-lived, and only florescence axis to produce a dense head of cymes, require a transient supply of nutrients. It would be the central one, perhaps in response to the pro- functionally illogical to have to maintain an extra tected situation, becoming reduced to a single fe- length of stem at the base of an inflorescence once male flower while the surrounding whorl of cymes the male flowers had finished, especially as the became all male and their associated subtending extra length would mean that the stem would have bracts and ?stipular glands fused to form an in- to be that much stronger mechanically. By having volucre. distal male flowers, that part of the stem does not An alternative scenario is that the cyathium has need to transport as many nutrients for so long, evolved from a single cyme with a primary female does not need to be strong enough to support rel- flower and lateral male flowers such as is seen in atively heavy fruits, and can be shut down as soon Jatropha. Croizat (1938), in discussing Neoguil- as the male flowers have dispersed their pollen. lauminea, accepted such a scenario and expressed Having said this, there is an obvious exception to the belief that 4-lobed involucres were the basic such logic Acalypha-which shows a bewilder- type for the Euphorbieae, derived presumably from ing array of inflorescence types, particularly with the initial two levels of branching of a dichasium.

This content downloaded from 210.72.93.185 on Mon, 19 Jan 2015 03:28:45 AM All use subject to JSTOR Terms and Conditions 286 Annals of the Missouri Botanical Garden

This content downloaded from 210.72.93.185 on Mon, 19 Jan 2015 03:28:45 AM All use subject to JSTOR Terms and Conditions Volume 81, Number 2 Gilbert 287 1994 Relationships of the Euphorbieae

An immediate difficulty is the frequency of 5-lobed are strictly terminal and show no tendency toward involucres, which would be difficult to derive from the production of a synflorescence. Perhaps the a regular dichasial cyme but which could be ex- strictly terminal position of the inflorescence in pected as the norm if the involucre was the product such Jatropha is a factor in their having retained of the condensation of a Fibonacci spiral derived their clearly cymose organization while other groups from a thyrse. Two further bits of evidence argue with axillary cymes have evolved the thyrse (and against a development from a single cyme. First is cyathia?) now so widespread in the family. How- the situation in Neoguillauminea, where the cen- ever, the most primitive species (Dehgan & Schutz- tral female flower appears to be subtended by four man, 1994, this issue), J. curcas L., does produce separate involucres of male flowers, which indicates a synflorescence such as is postulated here as the that several inflorescences are involved. Second, common ancestral form for cyathia and for the some species in Euphorbia sect. Esula (notably E. thyrses of most other uniovulate Euphorbiaceae. cyparissioides Pax) have groups of male flowers This would seem to reinforce the view that Jatro- and glands very similar to those of the involucre pha might represent the nearest approach to an intermingled within the involucre proper in a pos- ancestral type for the family as a whole. sible spiral series. Such a situation could be ex- In conclusion, it is suggested that it is not pos- plained easily if the axis of the cyathium was a sible for the cyathium to have evolved from any contracted synflorescence, but is very much more inflorescence found within modern-day Hippoma- difficult to explain if the structure was derived from neae (and closely allied tribes). One must conjecture a single cyme. a rather primitive common ancestor belonging to Unmodified simple cymes are of rather rare neither group but rather to the Crotonoideae. If occurrence within the family and among African this is indeed the case, it might be better to extend taxa are only found in Jatropha, a genus showing the Euphorbioideae to include the Crotonoideae, a number of other apparently primitive features, as suggested in effect by Mahlberg et al. (1987). most notably the well-developed corolla and female Alternatively, if an independent relationship of each flowers with staminodes occasionally so well de- group to a very primitive member of the family be veloped (e.g., in J. tropaeolifolia Pax) that the accepted, there might be a good case for placing flowers appear to be bisexual. Another unusual the Hippomaneae and allies in a subfamily of their character of Jatropha is the diversity of laticifer own and restricting the Euphorbioideae to the Eu- types, a feature that has been interpreted as an phorbieae. indication of primitiveness (Dehgan & Craig, 1978; Rudall, 1987). The cymes of most species of Jatro- LITERATURE CITED pha are primarily dichasial, but in many species BAILLON, H. 1874. Euphorbiacees. In: Hist. P1. 2: 105- (J. spicata Pax is a good example) the ultimate 256. Hachette, Paris. branches are usually long monochasia of male flow- BAWA, K. S., S. H. BULLOCK, D. R. PERRY, R. E. COVILLE & M. H. GRAYUM. 1985. Reproductive biology of ers. This suggests that there is a natural tendency lowlandrainforest trees II. Pollinationsystems. Amer. within the organization of such cymes for reduced J. Bot. 72: 346-356. terminal elements to be monochasial, which would BENTHAM, G. 1878. Notes on Euphorbiaceae. J. Linn. support Eichler's (1878) interpretation of the cy- Soc., Bot. 17: 185-267 (pp. 201-204, 239-245). BEUTLER, J. A., A. B. ALVARADO, T. G. MCCLOUD & G. athium. M. CRAGG. 1989. Distributionof phorbol ester bio- Dehgan & Craig (1978) discussed the possibility activity in the Euphorbiaceae. Phytotherapy Re- of a Jatropha-like being ancestral to the search 3: 188-192. Crotonoideae. It seems a distinct possibility that CROIZAT, L. 1938. Notes on the Euphorbiaceae, with any common ancestor of the Hippomaneae and the a new genus of the Euphorbieae. Philipp. J. Sci. 64: 397-412. Euphorbieae might also have had Jatropha-like DEHGAN, B. & M. E. CRAIG. 1978. Types of lacticifers cymes, thus opening up the question of subfamily and crystals in Jatropha and their taxonomic impli- relationships. The cymes of most Jatropha species cations. Amer. J. Bot. 65: 345-352.

FIGURE 1. Diagrammaticrepresentation of the hypothetical evolution of, on the left, the racemelike inflorescences of the Hippomaneae (and most other Crotonoideae and Acalyphoideae) and, on the right, the cyathium from a hypothetical synflorescence of axillary bisexual cymes. For simplicity the cymes are shown as monochasia, but in practice are more likely to be primarily dichasial. On the left there is a progressive elongation of the main axis and a reduction of the axillary cymes. On the right there is a shortening of the main axis to form a dense cluster of cymes, a central, terminal cyme becoming reduced to a single female flower and surrounded by a whorl of all-male cymes, while the associated subtending leaves and stipules fuse to form an enclosing involucre.

This content downloaded from 210.72.93.185 on Mon, 19 Jan 2015 03:28:45 AM All use subject to JSTOR Terms and Conditions 288 Annals of the Missouri Botanical Garden

& B. SCHUTZMAN. 1994. Contributionstoward PUNT, W. 1987. A survey of pollen morphology in a monograph of neotropical Jatropha: Phenetic and Euphorbiaceae with special reference to Phyllan- phylogenetic analyses. Ann. Missouri Bot. Gard. 81: thus. Bot. J. Linn. Soc. 94: 127-142. 349-367. ROGERS, D. J. 1951. A revision of Stillingia in the EICHLER,A. W. 1878. Blithendiagramme construit und New World. Ann. Missouri Bot. Gard. 38: 207-259. erliutert, 2. W. Engelmann, Leipzig. RUDALL, P. J. 1987. Lacticifers in Euphorbiaceae-A GILBERT,M. G. 1987. Two new geophytic species of conspectus. Bot. J. Linn. Soc. 94: 14-163. Euphorbia with comments on the subgeneric group- SCHOUTE, J. C. 1937. On the aestivation in the cy- ings of the African members of the genus. Kew Bull. atheum of Euphorbia fulgens, with some remarks 42: 231-244. on the morphological interpretation of the cyathium LAMARCK,J. B. A. P. M. DE. 1788. Euphorbe ou Ti- in general. Recueil Trav. Bot. Neerl. 34: 168-181. thymale, Euphorbia; Caractire GeCneirique.Encycl. STEINER, K. E. 1983. Pollinationof Mabea occidentalis 5: 413. (Euphorbiaceae)in Panama. Syst. Bot. 8: 105-117. MAHLBERG,P. G., D. G. DAVIS, D. S. GALITZ& G. D. WEBSTER, G. L. 1994. Synopsis of the genera and MANNERS. 1987. Lacticifers and the classification suprageneric taxa of Euphorbiaceae. Ann. Missouri of Euphorbia: The chemotaxonomy of Euphorbia Bot. Gard. 81: 33-144. esula L. Bot. J. Linn. Soc. 94: 165-180. MUELLERARGOVIENSIS, J. 1866. Euphorbiaceae. In: A. P. de Candolle, Prodr. 15(2): 1-1286.

This content downloaded from 210.72.93.185 on Mon, 19 Jan 2015 03:28:45 AM All use subject to JSTOR Terms and Conditions