Pacific Science, vol. 54, no. 2: 159-167 © 2000 by University of Hawai'i Press. All rights reserved Phylogenetic and Biogeographic Aspects of Sophora Sect. Edwardsia (Papilionaceae) 1 R. C. PENA,2 L. ITURRIAGA,2 G. MONTENEGRO,z AND B. K. CASSELS 3 ABSTRACT: Sophora comprises 45-50 species of worldwide distribution, but no general proposal as to the evolution of this group has been put forth. We used cladistic relationships of the quinolizidine alkaloids (matrine, sparteine, methylcytisine, anagyrine, and sophoranol) with morphological and palynolog­ ical characters to suggest a hypothesis of evolutionary and biogeographic rela­ tionships. The mainland Chilean species of Sophora appear to have been de­ rived from' ancestors phylogenetically near the extant Argentinean species S. linearifolia and S. rhynchocarpa and the psammophyte S. tomentosa, growing at tropical coastal sites around the world. The Boreotropic hypothesis of Lavin and Luckow is incorporated in our model as the most parsimonious explana­ tion of the evolution of the species of Edwardsia. Sophora is a taxonomic group that meets the following criteria: a center of diversity in North America, an early Tertiary record in North America, and a pantropical distribution. Styph­ nolobium and Sophora (including Calia) are representatives of Sophora s.l. in the United States, suggesting a migration of the latter from the Northern Hemi­ sphere to South America. Consistent with the Boreotropic hypothesis, a pri­ mary diversification center in South America and subsequent migration to the Indian Ocean and New Zealand, the Juan Fernandez Archipelago, Easter Island, and possibly the Hawaiian Islands is the simplest explanation for the evolution of the Edwardsia species. VALID SPECIES OF Sophora following revisions The genus Sophora is a taxonomic group of Allan (1961), Yakovlev (1967), Green of pantropical distribution but with its center (1970), Godley (1979), Isely (1981), Tsoong of diversity and an early Tertiary record in and Ma (1981), and Heenan (1998) are listed North America. Sousa and Rudd (1993) dis­ in Table 1. Although Tsoong and Ma (1981) tinguished Sophora as perennials, shrubs, or merged Calia and Styphnolobium, it is better trees without bracteoles and with a dehiscent to consider them as two distinctive taxa legume, and Calia and Styphnolobium as taking into account first the cytogenetic trees with flowers with bracteoles and non­ differences, chromosome number, and karyo­ dehiscent fruit. Styphnolobium differs from types, and second the distinctive germination Calia in having leaves with stipels and le­ mode of Styphnolobium and Sophora sect. gume with fleshy meso- and endocarp. Calia (Palomino et al. 1993). Styphnolobium affine (Torr. and A. Gray) Walp grows well in deciduous forests and grasslands. Styphnolobium japonicum is a deciduous tree endeniic to'China.- Sophora 1 This study was made possible by the financial sup­ port of the Fondo Nacional de Ciencia y Tecnologia macrocarpa is a characteristic endemic shrub (FONDECYT) Chile, Grant No. 1980967 to G.M. and from Chile that grows in open places or NIH Grant 2 UOI TW 00316-06 to B. Timmermann. under the forest canopy. Sophora prostrata, a Manuscript accepted I August 1999. divaricate suckering shrub from New Zea­ 2Ecology Department, Pontificia Universidad Cato­ lica de Chile, Casilla 114-D, Santiago, Chile. land, grows well in grassland and rocky 3Faculty of Sciences, Universidad de Chile, P.O. Box places. In contrast, the majority ofthe species 653, Santiago, Chile. of Edwardsia prefer lowland or lower mon- 159 160 PACIFIC SCIENCE, Volume 54, April 2000 TABLE 1 NOMENCLATURE OF STUDIED SPECIES, INCLUDING SECTIONS SPECIES NAME SECTION Sophora secundiflora Ortega ex DC. Calia Sophora chrysophylla (Salisb.) Seem. Edwardsia Sophora denudata Bory Edwardsia Sophorajernandeziana (phil.) Skottb. Edwardsia Sophora howinsula (Oliv.) Green Edwardsia Sophora longicarinata G. Simpson & J. S. Thompson Edwardsia Sophora macnabiana (Grah.) Skottb. Edwardsia Sophora macrocarpa J. E. Sm. Edwardsia Sophora masajuerana (phil.) Skottb. Edwardsia Sophora microphylla Ait. Edwardsia Sophora prostrata Buchan. Edwardsia Sophora tetraptera J. Mill. Edwardsia Sophora toromiro L. Edwardsia Sophora rhynchocarpa Griseb. Sophora Sophora linearifolia Griseb. Sophora Sophora tomentosa L. Sophora Styphnolobiumjaponicum (L.) Schott Styphnolobium tane areas along rivers and forest margins or that S. macrocarpa is derived from S. micro­ outskirts. Sophora longicarinata (a recently phylla of New Zealand on the basis of the reinstated species close to S. microphylla) is a similarity of the polypeptide and seed protein shrub or tree of stressed habitats; it is known compositions of both taxa. It is surprising from the Takaka area of the northern South that those authors also recognized the strong Island of New Zealand, restricted to marble similarity between these taxa and Pisum sati­ or limestone sites (Heenan 1998). Sophora vum in this regard, which seems to indicate denudata, S. chrysophylla, and Sophora from that their method does not even discriminate Juan Fernandez are found in uplands-So at the suprageneric level. Similarly, Mark­ Jernandeziana appears on hills; S. denudata, a ham and Godley (1972) could not segregate small tree from La Reunion, grows over the Chilean, Gough Island, or Chatham Is­ Acacia heterophylla Willd. and is frequently land taxa from New Zealand material from associated with the ericaceous Philippia mon­ both North and South Islands using their tana Klotsch.; similarly S. chrysophylla is the phenolic constituent patterns. Additional cir­ major tree in the treeline ecosystem on both cumstantial evidence cited by Murray (1986) Mauna Kea (Hawai'i) and Haleakala (Maui) considered the hydrochoric character of S. and is the successional replacer of Metrosi­ microphylla seeds. Ramirez and Romero deros polymorpha Gaud. (Mueller-Dombois (1978) also observed the potential for hydro­ 1987). choric seed dispersal in S. macrocarpa. There The first hypothesis put forth to explain is no difficulty, however, in considering S. the origin of Edwardsia suggests that their macrocarpa as the founding species of the disttibution center lies in the-West Pacific section as a- result of its possible relationship with subsequent migration and differentia­ to American species in the section Sophora tion from west to east, reaching South and because of its potential for hydrochoric America, Hawai'i, Juan Fernandez and seed dispersal. Two species from Argentina Easter Islands, Gough Island in the Atlantic with a large, unwinged legume and dark Ocean, and La Reunion in the Indian Ocean brown seeds (cf. Burkart 1952), S. linearifolia (Sykes and Godley 1968). Recently, Murray and S. rhynchocarpa, restricted to Cordoba (1986) and Murray and Porter (1980) stated and San Luis Provinces, and to Salta Prov- , - . .....'" , .....'" • "..,.-..-" .~, ... , ,I, ... ,~... ,.,.,r .or '....~_........ ,,_ Phylogenetic and Biogeographic Aspects of Sophora Sect. Edwardsia-"PENA ET AL. 161 ince, respectively, are closest to Chilean S. B.P. and Masafuera between 1.0 and 2.4 mil­ macrocarpa. In the same vein, Skottsberg lion yr RP. (Stuessy et al. 1984). Lava from (1956) noted that the Chilean insular taxa, S. the Poike crater of Easter Island is 2.5 mil­ jernandeziana, S. masajuerana, and S. toro­ lion yr old (Zizka 1988). Thus, an American miro, are related and might be united in S. origin for this section becomes an attractive tetraptera. In the Juan Fernandez Islands alternative contrasting with the Gondwanic there is a high degree of endemism, and the origin suggested for the Edwardsia species, as principal source of its flowering plants is the may deduced from Sykes and Godley's continental flora of South America, repre­ (1968) and Murray's (1986) reasoning (M. senting about 47% (Hoffmann and Marti­ Kalin Arroyo, pers. comm.). corena 1987). We have incorporated Lavin and The link between sections Sophora and Luckow's hypothesis in our model, because Edwardsia is S. macrocarpa. In Edwardsia, Sophora is a taxonomic group that meets the change in flower morphology can be re­ the following three criteria: a center of lated to the presence of a new pollination diversity in North America, an early Tertiary vector. Calia, Sophora, and Styphnolobium fossil record in North America, and a have an erect flag and are melittophylic, pantropical distribution (Lavin and Luckow but Edwardsia has an extended one and is 1993). Styphnolobium, Sophora, and Calia visited by birds. Such a change has been are representatives of Sophora s.1. in the demonstrated in Harpalyce (Galegeae­ United States, suggesting migration of the Brongniartieae), whose resupinate flowers are latter from the Northern Hemisphere to South inadequate to ensure successful pollination America. Another two cases demonstrated in by bees (Arroyo 1976). Drepanididae visit the Papilionaceae are Robinia (Lavin and Sophora chrysophylla in Hawai'i and the tro­ Luckow 1993) and the Psoraleeae (Grimes chilid Patagona gigas visits S. macrocarpa in 1990). The fossil record of Styphnolobium is Chile (Arroyo 1981); other hummingbirds present in Middle Eocene deposits in North pollinate S. tetraptera and S. microphylla in America (Sousa and Rudd 1993). New Zealand. It is interesting that the nectar Both palynological and chemical evidence of the ornithophilous S. microphylla kills might corroborate Or weaken the hypothesis bees, but is apparently not harmful to birds of a closer relationship between the Sophora (Arroyo