CLASSIFICATION OF THE GradyL. Webster2 EUPHORBIACEAE' ABSTRACT The familyEuphorbiaceae appears to be monophyletic,despite proposals for segregate families. The Euphorbiaceae displaya greatvariety of growthforms, including at least 17 "models" of Hallk. Anatomicalcharacters particularly usefulfor classificationinclude wood structure,laticifer type, trichomes, and stomata.Inflorescences are basically dichasial,and pseudanthiahave evolved several times. Pollen nuclear numberand exine structureprovide useful criteriafor characterizing genera, tribes, and subfamilies.Structure of the seed coat is characteristicfor the family and does not provideevidence for a polyphyleticorigin of the family.Pollination is prevailinglyentomophilous, and seed dispersalby ants is importantin manytaxa. Geographicdistribution patterns of generashow a concentrationof primitivetaxa in Africaand Madagascar,although in subfamilyCrotonoideae there is evidenceof a neotropicalcenter. Disjunctionsbetween Africa and South Americaare common.Bentham's hypothesis of an Old World originof the familyappears well supported.The basic distributionpatterns appear to reflectearly (Cretaceousand Paleogene) dispersalacross land or narrowwater barriers and spectacularbut rathertrivial instances of long-distancedispersal in the late Tertiaryand Pleistocene;Tertiary high-latitude dispersals via the Beringland bridgeappear to have been relativelyinsignificant. The Euphorbiaceae, although one of the largest uninformative about relationships, and the later dicot familiesand conspicuous throughoutthe trop- ones of Bentham (1880) and Pax (1890), who ics, have been relatively neglected by systematists accepted the general framework of Mueller with in the 20th century. While other families such as relativelyminor exceptions. Later revisions by Pax Compositae, Leguminosae, and Solanaceae have & Hoffmann (1931) and the treatment by Huru- been the subjects of various symposia, the very sawa (1954) continued to reflectthe original ideas firstinternational conference with a major focus of Mueller, despite considerable changes. Although on Euphorbiaceae was held at Kew in 1986 (Jury my recent classification (Webster, 1975) appears et al., 1987). The Kew symposium, on the Eu- very differentfrom that of Mueller at firstglance, phorbiales, was heavily biochemical in orientation I was struck by how many of the great Swiss and focused to a considerable extent on relation- systematist's ideas still survive. Even though he ships between the Euphorbiaceae and other fami- went astray in his major subdivisions (based on lies. The present symposium in St. Louis is the first cotyledon shape) of the "Stenolobeae" and "Pla- in which the classification of the family and its tylobeae," Mueller arranged the genera of Eu- constituentinfrafamilial taxa is the major focus of phorbiaceae into subtribes that, in an impressively attention. large number of instances, reflectphylogenetic af- The historyof classificationsystems for the Eu- finityas presently understood. In reviewing our phorbiaceae at the subfamilial and tribal level has effortsat this symposium to improve the classifi- been reviewed in the Kew symposium on Euphor- cation of Euphorbiaceae, it seems evident to me biales (Webster, 1987). In my opinion, the two that it is Mueller's foundation we are standing on. major milestones in this history were the classifi- In my classificationof 1975, the 300 genera of cations of Adrien Jussieu (1824), who identified Euphorbiaceae were grouped into 52 tribes in five the major series of genera that (after much later subfamilies, with several of the tribes divided into revision) correspond roughlyto currentsubfamilies, subtribes. The classification presented at this con- and Jean Mueller (1866), who provided the first ference shows only a limited number of changes, detailed classificationof the familyinto subfamilies, even though it is quite probable that futureresearch tribes, and subtribes. The originalityof Mueller is will show that substantial modifications will have particularly strikingin comparison with the earlier to be made. As has oftenbeen remarked, the linear systemof Baillon (1858), which was disappointingly arrangement of taxa in a classification is an im- I I wishto thankRobert Rhode forhis assistancein preparationof the manuscriptand illustrations.Outline maps wereprovided through the courtesyof Daniel Axelrod. 2Section of Plant Biology,University of California,Davis, California95616, U.S.A. ANN. MISSOURI BOT. GARD. 81: 3-32. 1994. This content downloaded from 169.237.8.36 on Mon, 29 Dec 2014 19:08:16 PM All use subject to JSTOR Terms and Conditions 4 Annals of the MissouriBotanical Garden perfect way of indicating phylogenetic relation- Ixonanthaceae are accepted as possible "out- ships. Cladistic analysis of the familyis badly need- groups," as speculated earlier (Webster, 1967), ed in order to provide a clear idea of the evolution then it is feasible to compile a list of characters in of characters and to test the implicit phylogenetic the Euphorbiaceae for which plesiomorphic (prim- hypotheses that are lodged in the classification itive) and apomorphic (advanced) states are defined scheme. Unfortunately,the informationavailable (Table 1). for the Euphorbiaceae is imperfectand this, along An important contributionto identifyingsignif- with the size of the group, makes it difficultto icant synapomorphiesin subfamilyPhyllanthoideae proceed. In my discussion of the familyat the Berlin has been made by Levin (1986c) in his cladistic congress in 1987 (unpublished), I provided some analysis of the subfamilybased on foliarcharacters. informallyconstructed "family trees" that reflect- Pollen studies on the subfamily, mostly still un- ed evolution of characters and used these as the published (Punt, 1987) or in this symposium (Levin basis for a schematic model of the evolutionary & Simpson, 1994a, b; Nowicke, 1994), show great and biogeographic history of the subfamilies and promise of adding important informationfor con- tribes. The present essay is intended to set forth structingphyletic models. The really major lacuna this model in greater detail, in the hope that it will in our knowledge, for this and other subfamilies, suggest topics fordiscussion and criticism.In doing is in biochemical data. Although Hegnauer (1962- so, I emphasize that, while our classificationsystem 1973) and Gibbs (1974) have compiled informative is a descendant of the system of Jean Mueller in resumes of reports fromthe biochemical literature, 1866, my scenario for the evolution and bioge- this informationis frustratingto the systematist, ography of the Euphorbiaceae is indebted to the because it demonstrates a great biochemical di- perspicacious essay of Bentham (1878), who from versity in the Euphorbiaceae but at the same time extremelyimperfect data deduced a model that has reveals a meager and erratic level of sampling. As proved to be almost clairvoyant in its anticipation a consequence, biochemical data are still of limited of later work on the biogeography of major angio- usefulness for meaningfulcomparisons at the tribal sperm taxa. and subfamilial levels. In my 1987 review of the classification of the family, the Euphorbiaceae were implicitlytreated RELATIONSHIPS BETWEEN THE SUBFAMILIES as a monophyletic group by the inclusion of such Since the perceptive discussion of Bentham proposed segregate families as Bischofiaceae and (1878), it has been accepted by later workers (e.g., Hymenocardiaceae. However, for those interested Pax, 1924), that the subfamilyPhyllanthoideae is in relationships within and outside the Euphorbi- the primitive group from which other subfamilies aceae, it should be noted that this delimitation of are derived. Within the Phyllanthoideae, genera the familyis not accepted by all of my colleagues. withinthe tribe Wielandieae, such as Heywoodia Hymenocardiaceae are accepted as a family by and Savia, appear to represent relicts of the orig- Leonard & Mosango (1985) and by Radcliffe-Smith inal euphorbiaceous complex that arose probably (1987). Furthermore, Radcliffe-Smith accepts in the late Cretaceous. These primitiveEuphorbi- Pandaceae and retains three genera (Antidesma, aceae are dioecious shrubs or small trees of seasonal Bischofia, and Uapaca) only provisionally within forest habitats in tropical latitudes. They have, as the Euphorbiaceae. Jensen (1994) suggests that expected, relativelyunspecialized flowerswith well- on the basis of serological data there appear to be developed petals, floraldisk, and a pistillode in the two main groups of Euphorbiaceae: Phyllanthoi- staminate flower. The pistillate flower usually has deae + Oldfieldioideae; and Acalyphoideae + Cro- a 3-locular ovary, with two anatropous ovules be- tonoideae + Euphorbioideae. This binary arrange- neath an obturator. The fruitis a 3-celled capsule, ment of "biovulate" and "uniovulate" taxa parallels and the seeds have a dry testa, copious endosperm, the arrangement proposed by Mahlberg (1987) on and a large embryo with a radicle much shorter the basis of laticifer morphology. On a phenetic than the cotyledons. As far as the evidence goes, basis, this basic dichotomy appears intuitivelyrea- the chromosome base number is n = 13 in these sonable, but it remains to be seen whether it will "archaic" taxa of subfamily Phyllanthoideae. An- be substantiated by critical phylogenetic studies. atomically, the relict Phyllanthoideae have such Meeuse (1990) went much furtherand proposed unspecialized features as vessel elements with