Evolution. 44(2), 1990, pp. 390-402

EVOLUTIONARY SIGNIFICANCE OF THE LOSS OF THE CHLOROPLAST-DNA INVERTED REPEAT IN THE LEGUMINOSAE SUBFAMILY PAPILIONOIDEAE

MAlT LAVIN,' JEFF J. DOYLE, L. H. Bailey Hortorium, Cornell University, Ithaca, NY 14853

AND

JEFFREY D. PALMER2 Department ofBiology, University ofMichigan, Ann Arbor, MI48109

Abstract.-The distribution of a rare chloroplast-DNA structural mutation, the loss of a large inverted repeat, has been determined for 95 representing 77 genera and 25 ofthe 31 tribes in the legume subfamily Papilionoideae. This mutation, which is regarded as a derived feature of singular origin within the subfamily, marks a group comprising six temperate tribes, the , Hedysareae, Carmichaelieae, Vicieae, Cicereae, and Trifolieae, an assemblage traditionally con­ sidered to be monophyletic. This mutation also occurs in the chloroplast genome of , a member ofthe tropical tribe whose other members so far surveyed lack the mutation. These new DNA data, together with traditional evidence, support the hypothesis that Wisteria is an unspecialized member of a lineage that gave rise to the temperate tribes marked by the chlo­ roplast-DNA mutation; the probable paraphylesis of Millettieae is revealed. Two other tribes, Loteae and Coronilleae (traditionally regarded as a derived element of the aforesaid temperate tribes) do not possess this chloroplast-DNA structural mutation and, therefore, presumably rep­ resent a distinct temperate lineage. This hypothesis is supported by additional evidence from pollen, inflorescence, and root-nodule morphology that suggests that the Loteae and Coronilleae share a more recent ancestry with tropical tribes such as and Millettieae than with other temperate tribes.

Received January 13, 1989. Accepted November 17, 1989 The Leguminosae (or ), with ap­ The higher-level systematics ofthe Papil­ proximately 18,000 species and 650 genera, ionoideae, particularly tribal circumscrip­ is the third largest , after the tions and interrelationships, has been the Compositae and Orchidaceae. Its taxonom­ subject of much recent study (e.g., Polhill ic relationships with otherangiosperm plant and Raven, 1981; Stirton, 1987). Evidence families are obscure and, consequently, the presented in these investigations of chro­ family is the sole member of the order Fa­ mosome numbers, seed chemistry, anato­ bales (Cronquist, 1981). It is divided into my, pollen morphology, and phytogeogra­ three subfamilies, Caesalpinioideae, Mimo­ phy has provided a greater insight into the soideae, and Papilionoideae (Polhill and evolutionary patterns in Papilionoideae and Raven, 1981); of these, Papilionoideae is has resulted in a restructuring and rear­ the largest (with approximately 12,000 rangement ofmany tribes originally circum­ species and 440 genera) and economically scribed by Bentham (1865) in his taxonom­ the most important. Peanuts, beans, peas, ic system ofPapilionoideae. However, many sesban, guar, tragacanth, rosewood, indigo, characters used at the higher systematic and black locust are examples of econom­ levels, such as the degree ofstaminal fusion ically important papilionoids which pro­ or legume dehiscence, show a high degree vide food, fiber, oils, gums, hardwoods, dyes, ofhomoplasy, and evaluation ofsuch char­ and ornamentals. Species ofPapilionoideae acters affects not only tribal relationships, predominate in nearly every vegetation type but sometimes tribal circumscriptions as throughout the world, from tropical rain well. As stated by Polhill (198la), much still forests to deserts, and nearly all species root­ remains ill-defined in the subfamily. nodulate and fix atmospheric nitrogen. Recent work by Palmer and coworkers (summarized in Palmer et al. [1987] and Palmer et al. [1988b]) has shown that the , Present address: Department ofBiology, Montana State University, Bozeman, MT 59717. chloroplast genome ofdiverse Leguminosae 2 Present address: Department of Biology, Indiana possess several major structural mutations, University, Bloomington, IN 47405. including large inversions and a major dele- 390 LOSS OF cpDNA INVERTED REPEAT 391

168238 II 125 kb II .1 12 3 4 sp.

12 3 214 II III II II • II II 150 kb 88e L8e region \rm 2 168238 region 238168 rm 2 1 r.Jl§ 19 iW:!A Coronilla varia

FIG. 1. Linearized representation ofthe chloroplastgenome ofCoronilla and Carmichaelia showing locations, relative to the inverted repeats, ofgenes discussed in the text and the four probe fragments (numbers 1, 2, 3, and 4). The heavy lines connecting rpl2 with 23S indicate the region of the inverted repeated sequences. SSC = small single copy; LSC = large single copy. tion; such mutations have been shown in sess the taxonomic distribution of the loss other plant groups to be phylogenetically of the inverted repeat in the Leguminosae informative (Jansen and Palmer, 1987; and its evolutionary significance. Palmer et al., 1988a). Ofparticular interest here is the loss ofa large (ca. 25-kb) inverted MATERIALS AND METHODS repeated sequence containinga duplicate set Ninety-five species of Papilionoideae, ofribosomal-RNA genes (Fig. 1). This loss representing 77 genera and 25 ofthe 31 tribes is remarkable in that this repeat is an evo­ in this subfamily, have been surveyed for lutionarily conserved feature of land-plant the loss of the inverted repeat (Table 1; chloroplast DNA and is known to be absent voucher specimen information is available otherwise only from certain conifers (Strauss upon request from M. Lavin). In addition, et al., 1988). Structurally conserved features 12 species from the subfamilies Caesalpini­ of the chloroplast-DNA inverted repeat, oideae and Mimosoideae were tested. such as its size and position (dividing the Techniques. -A fraction enriched in genome into small and large single-copy re­ chloroplast-DNA sequences was obtained gions of approximately 20 kb and 80 kb, from most species by isolating a chloroplast respectively), location relative to such pellet in a high-salt buffer (Bookjans et al., flanking genes as psbA, and ribosomal-RNA­ 1984) at a ratio of approximately 30 g of gene transcriptional orientation (toward the leafmaterial to 500 ml ofbuffer. DNA was small single-copy region), suggest that the then isolated from the chloroplast pellet by inverted repeat was present in the common adding an equal volume of the CTAB iso­ ancestor of land (Palmer, 1985). lation buffer of Doyle and Doyle (1987) to Palmer et al. (1987), from a limited sam­ the pellet and incubating at 60°C for one­ pling oflegumes, hypothesized that the loss halfhour. Subsequent chloroform/isoamyl­ of the chloroplast-DNA inverted repeat in alcohol extraction, centrifugation, and al­ Leguminosae occurred but once during the cohol precipitation followed Doyle and evolution of the subfamily Papilionoideae. Doyle (1987). This method was especially The survey presented here attempts to as- useful for obtaining DNA from species with 392 M. LAVIN ET AL. resinous or mucilaginous extracts should often detect the same two fragments. (e.g., those in the tribes Aeschynomeneae When the inverted repeat is absent, probes and Amorpheae). When leaftissue was lim­ 1 and 2 have only a single region of ho­ ited, total DNA was extracted by the CTAB mology on the chromosome and should de­ method of Doyle and Doyle (1987). tect only a single fragment. Only ifa restric­ DNAs were subjected to restriction-en­ tion site is located within its region of donuclease digestions and electrophoresis homology will such a probe detect more than in 0.7% agarose gels and transferred to MSI one fragment in a genome lacking the in­ (Fisher) or Zetabind (Cuno) nylon mem­ verted repeat, in which case the adjacent branes. Membrane-bound DNA fragments probe will detect only one of these frag­ were hybridized serially with each of the ments. clones described below that had been la­ Cloned restriction fragments in the sec­ beled with 32p by nick-translation. Electro­ ond set contain sequences flanking that seg­ phoresis, transfer, hybridization, and au­ ment of the inverted repeat that is deleted toradiography were performed as described in legumes (Palmer et al., 1987). These frag­ in Doyle and Beachy (1985), except that ments were used in this survey to test for nylon filters were used in place of nitrocel­ the absence of the inverted repeat (Fig. 1: lulose and Dextran sulfate was omitted from probes 3 and 4). Probe 3 (a 3.l-kb BamH the hybridization solution. I-Pst I fragment cloned in pUC 8 from Vig­ DNA of all species listed in Table 1 was na radiata; Palmer et al., 1987) is at the digested with at least two of the following margin ofthe small single-copy (SSC)region restriction enzymes: Hind III, Pst I, Pvu II, and probe 4 (a 1.2-kb Pst I-EcoR I fragment Sac I, Sal I, Xba I, and Xho 1.Two or more cloned into pUC 8 and containing the 3' single digests of a given DNA of a single end of psbA from Pisum sativum; Oishi et species were run side-by-side to maximize al., 1984) is at the margin ofthe LSC region reproducibility of the hybridization assays flanking the end ofthe inverted repeat seg­ that were used to diagnose the presence or ment. Each therefore has only a single re­ absence ofthe repeat. Additionally, a single gion of homology on a chloroplast chro­ digest ofthe DNA ofa species was run side­ mosome. These regions are normally by-side with other species in the same tribe separated by approximately 25 kb but are in order to assess more readily whether the known to be adjacent in species with chlo­ repeat was consistently present or absent in roplast DNA lacking an inverted repeat. Us­ a given tribe. ing a restriction enzyme with many recog­ Strategy. - Two sets of small cloned nition sites (e.g., Hind III or Xba I), these probes were used to diagnose the presence two probes should generally detect the same or absence ofthe inverted repeat. The first filter-bound fragment ofa DNA lacking the set included cloned restriction fragments inverted repeat but should detect two dif­ from Nicotiana tabacum (Fig. 1: probes 1 ferent fragments of a genome with an in­ and 2) that were used to test for the presence verted repeat. of the chloroplast-DNA inverted repeat. We emphasize that it is the structural con­ These two probes (coordinates 88778-87745 servation ofthe inverted repeated sequence for probe 1 [a 1.0-kb BamH I-Pst Ifragment in the chloroplast genome of land plants cloned into pTZ19R] and 90181-88778 for that allows us to diagnose readily its pres­ probe 2 [a l.4-kb BamH I fragment cloned ence or absence by the simple hybridization into pTZ19R]; Shin ozaki et al., 1986) are assays described above. The use of probes located near the large single-copy (LSC) homologous with the end points of the in­ margin ofthe inverted repeats and therefore verted repeat is possible because it is pre­ normally have two regions ofhomology on cisely these end points that are evolution­ the chloroplast chromosome. When chlo­ arily conserved (Palmer, 1985). For roplast DNA is digested with a restriction example, the LSC region contains psbA at enzyme having few recognition sites (e.g., one end and rps19 at the other end in di­ Pst I or Pvu II), probes 1 and 2 will usually verse species ofmonocots and dicots (Palm­ detect two filter-bound fragments ifthe in­ er, 1985). In addition, rpl2 is located at the verted repeat is present and, furthermore, LSC margin ofeach repeat in nearly all an- LOSS OF cpDNA INVERTED REPEAT 393 giosperms (Fig. 1). Consequently, the use of to lack the inverted repeat (e.g., Cicer, Wis­ probes containing such sequences should teria, Medicago, Trifolium, and Pisum; provide reliable results for diverse groups Palmer et al., 1987). of angiosperms, let alone members of Le­ Probes 3 and 4 (located in the single-copy guminosae. regions immediately flanking the inverted repeat that is putatively lost in the Papilio­ REsULTS noideae; Palmer et al., 1987) provided a test A total of 107 legume species have been for the absence ofthe inverted repeat. These examined for the presence/absence of the two probes always detected the same filter­ large inverted repeat in the chloroplast ge­ bound DNA fragment in species ofthe tribes nome. The results of the survey are sum­ Galegeae, Carmichaelieae, Hedysareae, Ci­ marized in Table 1, and a complete list of cereae, Trifolieae, and Vicieae, as well as in fragments detected by probes 1, 2, 3, and 4 Wisteria (Fig. 3A, B). They hybridized to for each enzyme and species is available different fragments of the DNA digests of from M. Lavin upon request. In all 12 species species from other tribes (Fig. 3C, D). of Caesalpinioideae and Mimosoideae sur­ The fact that probes 3 and 4 hybridized veyed, the inverted repeat was present. Of to the same DNA fragment in individual the 95 species of Papilionoideae so far ex­ DNA digests of Wisteria and members of amined, 70 species have the chloroplast­ the six tribes listed above suggests that, of DNA inverted repeat. Only Wisteria ofthe the four possible inverted repeat "segment tribe Millettieae and all 24 species exam­ orientations" (Palmer et al., 1987), the one ined in the tribes Galegeae, Carmichaelieae, flanked by psbA and the sequence homol­ Hedysareae, Vicieae, Cicereae, and Trifoli­ ogous with the 3.l-kb mung-bean fragment eae had a fragment phenotype consistent (i.e., probe 3) is the one that was lost. Probes with the absence ofthe inverted repeat. The 3 and 4 would only rarely hybridize to the presence or absence of the repeat was con­ same fragment if the inverted repeat that sistent within each and, except for the was lost had other flanking-sequence ori­ tribe Millettieae, consistent within each tribe entations. (Table 1). This hybridization assay using probes ho­ Probes 1 and 2 (containing the LSC mar­ mologous with the end points of the in­ gin of the inverted repeat) provided a test verted repeated sequences has an advantage for the presence ofthe inverted repeat. They in allowing a simultaneous test for the pres­ both hybridized to two filter-bound frag­ ence (probes 1 and 2) and absence (probes ments in DNA digests of most species, in­ 3 and 4) of the repeat. These two indepen­ cluding members of three genera (, dent assays always gave the same answers Lupinus, and Phaseolus) known to possess in the species ofLeguminosae we surveyed. a chloroplast chromosome with an inverted Furthermore, identical results were nearly repeat (Palmer et al., 1987). Moreover, these always obtained using multiple enzymes for two adjacent probes nearly always detected a given DNA. Species with chloroplast DNA the same two fragments, thus ruling out the that we interpret as lacking the inverted re­ possibility that the two fragments detected peat would occasionally show hybridization with one probe were caused by a restriction of probes 3 and 4 to different fragments in site within the area ofprobe homology. We one ofthe several enzyme digests (presum­ consider this to be strong evidence for the ably because of a restriction site between presence of the chloroplast-DNA inverted the regions ofhomology ofthese two probes). repeat (Fig. 2A, B). This fragment pheno­ Likewise, species with chloroplast DNA that type was not observed from the DNA di­ we interpret as having the inverted repeat gests of species from the tribes Galegeae, would show, in one of several enzyme di­ Carmichaelieae, Hedysareae, Cicereae, Tri­ gests, hybridization of probes 3 and 4 to folieae, and Vicieae or from Wisteria; rath­ seemingly the same fragment (presumably er, the same single band was detected by two different fragments with equal mobili­ these adjacent probes in each DNA (Fig. 2C, ty). D). Among these species were several with Additional confirmation ofthe presence/ a chloroplast chromosome already known absence of the inverted repeat is found in 394 M. LAVIN ET AL.

TABLE 1. Species of Leguminosae surveyed for the loss of the chloroplast-DNA inverted repeat. Asterisks indicate absence ofthe inverted repeat. Vigna radiata and Viciafaba are reported here from Palmer et aI. (1987) and Koller and Delius (1980), respectively. Voucher specimen information and details offragment sizes detected by probes for each species are available upon request from the first author.

Subfamily Tribe Species Papilionideae Abreae Abrus precatorius Aeschynomeneae Aeschynomene indica A. pfundii A. rudis A. virginica Arachis hypogea Diphysa robinioides Amorpheae Amorpha fruticosa Dalea candida D. purpurea Eysenhardtia texana Psorothamnus kingii Bossiaeeae Goodia latifolia Brongniartieae Brongniartia pacifica Carmichaelieae Carmichaelia sp.* Cicereae Cicersp.* Coronilleae Coronilla varia Crotalarieae Crotalaria mucronata Dalbergieae Pterocarpus indica Tipuana tipu Desmodieae Desmodium grandiflorum Lespedeza sp. Galegeae aboriginum* A. adsurgens* A. alpinus" A. cicer* A. pectinatus" Caragana arborescens" formosus" Glycyrrhiza glabra* deflexa" O. campestris" O. multiceps" O. splendens" salsula* Genisteae Genista canariensis Laburnum anagyroides Lupinus polyphyllus Hedysareae Hedysarum alpinum" Onobrychis sativa" tetragonoloba miniata Loteae Lotus australis L. corniculatus L. strigosus Millettieae Dalbergiella nyasae elliptica blackii eriocarinalis dura piscipula cinerea T. heckmanniana T. rhodesica T. villosa LOSS OF cpDNA INVERTED REPEAT 395

TABLE I. Continued.

Subfamily Tribe Species

Wisteria floribunda" Xeroderris stuhlmanii Mirbelieae Chorizema chordata Phaseoleae Cajanus cajan Canavalia maritima Centrosema virginiana Erythrina herbacea Glycine max rubicunda Mucuna sp. Phaseolus vulgaris Sphenostylis sp. Vigna radiata Robinieae heterantha C. hypoleuca Daubentonia tripettii Glircidia sepium Glottidium vesicarium cubense tesota pseudoacacia Sesbania sesban constrictum Sabinea carinalis Sophoreae Bolusanthus speciosus Calpurnea aurea Myrospermum sousanum Ormosia columbiana Sophora affinis S. secundiflora Swartzieae Swartzia sp, Thermopsideae Baptista leucophaea Thermopsis divaricarpa Trifolieae Medicago sativa* Melilotus alba* Trifolium pratense* Vicieae Lathyrus sp.* Lens esculenta" Pisum sativum" Viciafaba* Mimosoideae Acacia sp, Calliandra emarginata Desmanthus subulatus D. virgatus Dichrostachys cinerea Enterolobium cyc!ocarpum Leucaena sp, Neptunia pubescens Caesalpinioideae Bauhinia sp, Brownea sp, Caesalpinia platyloba Gleditsia triacanthos published reports on chloroplast-DNA ri­ cia; Koller and Delius, 1980). Confirming bosomal-gene arrangements for Cicereae evidence is also found in chloroplast-DNA and Phaseoleae tCicer, Glycine, and Vigna; restriction maps of Genisteae (Lupinus), Chu and Tewari, 1982) and for Vicieae (Vi- Millettieae (Wisteria), Phaseoleae (Glycine, 396 M. LAYIN ET AL.

Kb 25-

13-

probe 1 2A probe 2 28 23­ .... 16- ... --.... -_ _--

probe 1 2C probe 2 2D

FIG. 2. Autoradiographs of filter-bound DNA fragments of A, B) Coronilla varia and C, D) Carmichae/ia sp. hybridized with probes I and 2. The lanes represent, from left to right, restriction-enzyme digestions ofSac I, Sac I-Sal I, Sal I, Sal I-Pst I, Pst I, Pst I-Pvu II, Pvu II, Pvu II-Xho I, Xho I, Sac I-Pst I, Sac I-Pvu II, Sac I-Xho I.

Phaseolus, and Vigna), Trifolieae (Medi­ only comprehensive scheme to date. The cago and Trifolium), and Vicieae (Lathyrus, taxonomic distribution of the inverted-re­ Lens, Pisum, and Vida) (Palmer et al., 1987, peat mutation occurs almost exclusively 1988b) and, more recently in chloroplast­ within one of Polhill's lineages which in­ DNA restriction maps of Carmichaelieae cludes temperate herbaceous tribes that ac­ (Carmichaelia), Coronilleae (Coronilla), cumulate canavanine (Galegeae, Carmi­ Galegeae (Astragalus and Clianthus), Hed­ chaelieae, Hedysareae, Vicieae, Cicereae, ysareae (Onobrychis), Loteae (Lotus), Mil­ and Trifolieae). It is among this group of lettieae tPiscidia and Tephrosia), and Ro­ legumes that this new chloroplast-DNA binieae tCoursetia, , Hebestigma, character shows promise in testing existing Sesbania, and Sphinctospermum) (Lavin taxonomic hypotheses. and Doyle, unpubL). Our interpretations We emphasize that the evolutionary po­ from the experiments reported here using larity of this mutation is absolutely unam­ small probes from the ends ofthe inverted biguous; all species examined from each of repeat and single-copy regions are in com­ more than 60 angiosperm families (Palmer, plete agreement with these data. 1985, unpubL), as well as Caesalpinioideae and Mimosoideae (Table 1), have a chlo­ DISCUSSION roplast chromosome with an inverted re­ Polhill's (1981 a p. 199) hypothesis of peat. Loss of the inverted repeat has oc­ tribal relationships in Papilionoideae is the curred in six tribes (Galegeae, Hedysareae, LOSS OF cpDNA INVERTED REPEAT 397

Kb 1 2 3 4

23-

7.4-

38, 1 2 3 4 21- 6.8-.

3.5-

probe 3 !3C probe 4

FIG. 3. Autoradiographs of filter-bound DNA fragments of A, B) Cicer sp. and C, D) Canavalia maritima, hybridized with probes 3 and 4. Lanes 1-4 represent, respectively, restriction-enzyme digestions of Hind III, Pvu II, Sac I, and Xba I.

Carmichaelieae, Vicieae, Cicereae, and Tri­ Hedysareae, Trifolieae, Cicereae, and Vi­ folieae) and also in the genus Wisteria of cieae) should be of little surprise to legume the tribe Millettieae. Given the general con­ systematists. These tribes share a predom­ gruence of the distribution of this chloro­ inantly herbaceous growth habit, centers of plast-DNA structural mutation with the diversity in temperate regions of the Old classification of Papilionoideae, as well as World, and marker characters (such as the consistency of its presence/absence epulvinate leaves and stipules adnate to the within a genus, we substantiate the hypoth­ petiole) (Polhill, 1981a). These chloroplast­ esis of Palmer et al. (1987) that the loss of DNA data support Polhill's (l98ld) place­ the chloroplast-DNA inverted repeat oc­ ment ofGlycyrrhiza within this group, from curred only once among legumes and that which it has been excluded because of its it marks a monophyletic subgroup ofPapili­ pulvinate leaves with stipules free from the onoideae. petiole. The recognition of this monophyletic Other findings revealed by these chloro­ group of tribes (Galegeae, Carmichaelieae, plast-DNA data are completely unexpected 398 M. LAYIN ET AL. and provide new insights into the evolution anomalous in temperate groups. For ex­ of the Papilionoideae. The inconsistencies ample, the tribe Carmichaelieae is entirely between these chloroplast-DNA data and woody (the extinct Streblorrhiza ofthis tribe the classification of the Papilionoideae are is described as having an overall appearance the absence ofthe repeat in the genus Wis­ ofgenera in Millettieae, as well as "Wiste­ teria, and its presence in the tribes Loteae ria-like" flowers in Polhill [198lb]), as are and Coronilleae. It is the taxonomic distri­ Caragana and members of other genera in bution ofthis inverted repeat structural mu­ the tribe Galegeae. The tribes Vicieae and tation that, for the first time, clearly suggests Cicereae are characterized by seedlings with a relationship oftemperate tribes with trop­ hypogeal germination, and Glycyrrhiza of ical ones and yields new insight into the the tribe Galegeae has pulvinate leaves. evolutionary change of certain morpholog­ Thus, it is possible that further study will ical features, such as the pseudoraceme. show that Wisteria can be accommodated Wisteria. - The genus Wisteria possesses among the temperate tribes. At the least, a woody liana habit and seedlings with hy­ Wisteria should be regarded as an unspe­ pogeal germination, two characteristics of cialized member of the woody papilionoid woody Papilionoideae inhabiting tropical lineage that gave rise to the temperate, pre­ forests with a closed canopy. It is for this dominantly herbaceous tribes marked by the reason and others (e.g., pulvinate leaves and loss of the inverted repeat. As such, Gee­ stipellate leaflets) that Wisteria is placed in sink's (1984) suggestion that Millettieae is the tropical woody tribe Millettieae (Gee­ paraphyletic is supported by this finding. sink, 1984). However, Wisteria is atypical The possibility of a parallel loss of the amonggenera ofMillettieae not only in hav­ inverted repeat in Wisteria is diminished ing this unusual chloroplast-DNA structur­ because probes 3 and 4, those that flank the al mutation, but also in having a completely inverted repeat, detect the same fragment temperate distribution and a base chro­ in all enzyme digestions. These same results mosome number ofx = 8. These latter two are observed in the hybridization assays features are characteristic of tribes marked performed on species of Galegeae, Carmi­ by the chloroplast-DNA structural muta­ chaelieae, Hedysareae, Vicieae, Trifolieae, tion. The eight other genera of Millettieae and Cicereae, and this suggests that Wisteria surveyed (Table 1) have a chloroplast chro­ lost the same inverted repeat, flanked by mosome with an inverted repeat, a base the same sequences, as did these other tem­ chromosome number ofx = 11 and 12, and perate tribes (see Palmer et al. [1987] for a a tropical distribution (these latter two fea­ discussion of the inverted repeat flanking­ tures are known from nearly all genera of sequence orientations). Millettieae). Wisteria also is anomalous in Many genera presently placed in Millet­ Millettieae in lacking a pseudoracemose in­ tieae, Galegeae, and other related tribes were florescence, a trait distinctive ofmost other once included in a large heterogeneous tribe Millettieae and closely related tribes (Tuck­ Galegeae (sensu Bentham, 1865). However, er, 1987). subsequent anatomical (Dormer, 1946) and However, the actual disposition of Wis­ chromosomal (Turner and Fearing, 1959; teria remains problematic. A cladistic anal­ Goldblatt, 1981) evidence revealed a dis­ ysis ofadvanced woody Papilionoideae (i.e., tinction between the temperate represen­ those that accumulate nonprotein amino tatives of this tribe (with a predominantly acids; Lavin, unpubl.) suggests that Wisteria herbaceous growth habit and a base chro­ did not attain features typical ofMillettieae mosome number ofx = 8) and those more (woody habit, pulvinate leaves, hypogeal tropical in distribution (with a predomi­ seedling germination, reduced hypanthium, nantly woody habit and a base chromosome and intrastaminal nectary) secondarily from number of x = 11). The loss of the chlo­ herbaceous ancestors lacking these traits; roplast-DNA inverted repeat further distin­ therefore, it does not seem to be readily ac­ guishes these temperate genera from puta­ commodated into other temperate tribes. tive close relatives in tropical regions. Although such features weigh heavily in Other members of the Millettieae yet to placing Wisteria in Millettieae, they are not be examined may also have this chioro- LOSS OF cpDNA INVERTED REPEAT 399 plast-DNA structural mutation. For ex­ a very reduced to absent foot layer, reduced ample, Millettia japonica is distinctive in columellae, and a thick exine. This mor­ having a base chromosome number ofx = phology is commonly found in genera ofthe 8 and a temperate distribution, and this Phaseoleae and Millettieae (Polhill, pers. species has been postulated to be closely comm.; see Ferguson and Skvarla, 1981), as related to Wisteria (Geesink, 1984). The ge­ well as in Coronilleae and Loteae, but it is nus Sarcodum is also considered by Geesink rarely found elsewhere in the family. (1984) to be closely related to Wisteria and, Root-nodule morphology, which Corby therefore, Sarcodum may have a chloro­ (1981) suggests is taxonomically useful at plast chromosome lacking an inverted re­ the tribal level, provides additional evi­ peat. Finally, the New World species of dence that the Loteae and Coronilleae are Wisteria have been generically segregated related to tropical tribes. Loteae and Coro­ from the Old World species (Stritch, 1984), nilleae possess the determinate nodule, as but it is unknown whether the mutation is do Phaseoleae, while other temperate tribes consistently present in both groups. Al­ surveyed that have the chloroplast-DNA though DNAs ofMillettiajaponica, species structural mutation (i.e., Galegeae, Hedy­ ofSarcodum, and New World Wisteria were sareae, Vicieae, Cicereae, and Trifolieae) not available for this study, the final as­ have an indeterminate root nodule (Sprent, sessment of the distribution of the loss of 1981). Corby (1981) suggests that the in­ the chloroplast-DNA inverted repeat in determinate nodule (his Astragaloid type) is Millettieae will be significant in revealing the ancestral type and that determinate root the closest tropical relatives ofthe temper­ nodules are more specialized. That this de­ ate, predominantly herbaceous tribes. rived trait is shared by Loteae, Coronilleae, Loteae and Coronilleae. - The tribes Lo­ and tropical tribes such as Phaseoleae is teae and Coronilleae traditionally have been strong evidence of relationship. Sprent regarded as derived elements of the tem­ (1981) expressed difficulty in speculating on perate tribes here shown to be characterized the possible advantages ofdeterminate nod­ by the chloroplast-DNA mutation. Evi­ ules (as found in the Loteae and Coronil­ dence used to group these two tribes with leae) in the temperate line of evolution in other temperate tribes comes largely from which the indeterminate nodule was pre­ their herbaceous growth habit and associ­ dominant (as in the Galegeae and closely ated characters (e.g., a closed vascular sys­ related tribes). A new hypothesis formulat­ tem and stipules adnate to the petiole; Pol­ ed from these chloroplast-DNA data sug­ hill, 1981a) and a center of diversity in gests that the Loteae and Coronilleae rep­ temperate regions ofthe Old World. Loteae resent an independent temperate lineage that and Coronilleae together share many pre­ inherited the determinate type ofroot nod­ sumably derived features, such as special­ ule from tropical ancestors similarto Phase­ ized pollen (Polhill, 1981c; Ferguson and oleae. Skvarla, 1981), seedlings with exceptionally Inflorescence morphology may provide long and narrow cotyledons, and a sup­ additional evidence for the relationship of pressed plumule (Duke and Polhill, 1981), the Loteae and Coronilleae to tribes such as and a base chromosome number of x = 7. Phaseoleae and Millettieae. The umbellate The number of postulated specializations inflorescence found in nearly all species of found in these two tribes has been used as Loteae and Coronilleae is readily distinct evidence for their derived condition within from that found in supposedly related tem­ the temperate lineage. perate tribes; multiple flowers emanate from However, some of these specializations a single node and reach anthesis synchro­ may not be ofunique derivation but, rather, nously. This condition is very similar, and the result of recent common ancestry with perhaps homologous, to the pseudorace­ tropical tribes such as Phaseoleae and Mil­ mose condition found in the tribes Abreae, lettieae. In particular, characters ofthe pol­ Desmodieae, Millettieae, Phaseoleae, and len are suggestive of this relationship. Lo­ . Tucker (1987) suggests that these teae and Coronilleae have a specialized five predominantly tropical tribes represent pollen-wall morphology, which consists of a monophyletic group marked by this inflo- 400 M. LAYIN ET AL.

rescence type. Perhaps Loteae and Coronil­ loss ofthe inverted repeat was a single mu­ leae can be added to this group iftheir um­ tational event in Leguminosae and not of bellate inflorescences are found to be parallel origin in a family with inherently homologous with the pseudoraceme, as variable chloroplast chromosomes. characterized by Tucker (1987). From its taxonomic distribution, this Finally, two other characters are in need mutational event ofthe chloroplast genome of close study, as they may also reveal the occurred later than much of the evolution relationships of Loteae and Coronilleae. and diversification ofthe Papilionoideae. It Ingham (1981) points out that benzofurans occurred after the origin ofcanavanine syn­ are known in Papilionoideae only from a thesis (an important taxonomic marker in few genera in the Phaseoleae and Loteae. Papilionoideae that occurs in all the tribes Additionally, staminal filaments that are with the chloroplast-DNA mutation) but distally expanded have been used to char­ predates the evolution ofthe predominantly acterize the Loteae and Coronilleae; how­ herbaceous growth habit (the mutation is ever, this attribute is also found in some found not only in the woody liana Wisteria, genera of the Phaseoleae. but also in the woody tribe Carmichaelieae and in woody Galegeae such as Caragana). Conclusion Because no tropical representatives have The deletion ofthe chloroplast-DNA in­ been found to have lost the inverted repeat, verted repeat provides a means of testing the mutation very likely originated in a pa­ the traditional evidence used in reconstruct­ pilionoid group already established in a pa­ ing the phylogeny of Papilionoideae. The leotemperate habitat. Ifwe extrapolate from discovery of this mutation has been re­ our sample, this group of temperate tribes markable in that, with the exception of marked by a chloroplast chromosome lack­ Tucker's (1987) work on inflorescence mor­ ing the inverted repeat comprises ca. 3,800 phology, little new information has become species and 45 genera and includes the larg­ available in the last few years bearing on est genus of flowering plants, Astragalus, relationships at the tribal level in Papilio­ with approximately 2,000 species. Many noideae. The recognition of a large mono­ cultivated plants are also found in this group, phyletic group comprising the tribes Gale­ such as licorice root, peas, lentils, chickpeas, geae, Carmichaelieae, Hedysareae, Vicieae, alfalfa, and clover, as well as economically Cicereae, and Trifolieae should be of little important weeds such as the . surprise to botanists familiar with the Whatever the effect ofthis mutation on the subfamily. It will be ofmuch more interest, structural stability of the chloroplast ge­ however, to reassess the relationships ofthe nome, it apparently does not have a dele­ tribes Loteae and Coronilleae, as well as to terious effect on the ability of this lineage find the specific genera in the Millettieae to evolve numerous species that predomi­ that are the closest tropical relatives of the nate in many vegetation types. temperate tribes marked by this chloro­ These chloroplast-DNA data suggest a plast-DNA structural mutation. new taxonomic hypothesis for a large group Palmer et al. (l988b) have speculated as oftemperate legume tribes. Theyreveal spe­ to why the chloroplast DNA of Legumi­ cific characters in need ofinvestigation that nosae is so much more variable than that may prove useful in reformulating the cur­ ofmost other land plants. Particularly, they rent classification of temperate Papilionoi­ asked whether the chloroplast DNA of le­ deae. Not only should taxonomic relation­ gumes is such that it is predisposed to dele­ ships be better resolved, but our knowledge tions and inversions or whether the loss of of morphological evolution in this group .the inverted repeat has a destabilizing effect should increase (e.g., by possibly revealing on the chloroplast chromosome. The evi­ unrecognized modifications of the pseudo­ dence gathered in the present study address­ , such as that postulated for the Lo­ es only the issue of the origin of this chlo­ teae and Coronilleae). The distribution of roplast-DNA mutation. Given that its the loss of the chloroplast-DNA inverted distribution is generally congruent with the repeat and other structural mutations cur­ existing classification, it is likely that the rently under study should provide many new LOSS OF cpDNA INVERTED REPEAT 401 insights into the evolution of the Legumi­ H. Raven (eds.), Advances in Legume Systematics, nosae. Part 2. Royal Botanic Gardens, Kew, U.K GEESINK, R. 1984. Scala Millettiearum. Leiden Bo­ tanical Series, Vol. 8. Brell/Leiden Univ. Press, Lei­ ACKNOWLEDGMENTS den, Neth. We thank the following individuals and GOLDBLATT, P. 1981. Cytology and the phylogeny of institutions for leafor seed material used in Leguminosae, pp. 427-464. In R. M. Polhill and P. H. Raven (eds.), Advances in Legume System­ this study: A. Alvarez, A. Delgado, H. Hed­ atics, Part 2. Royal Botanic Gardens, Kew, U.K rick, C. Hughes, G. Lewis, M. Luckow, F. INGHAM, J. L. 1981. Phytoalexin induction and its Gregoire, P. Manos, G. Nesom, B. Ozbim, taxonomic significance in the Leguminosae V. Phelps, A. M. Powell, R. Riggins, M. (subfamily Papilionoideae), pp. 599-626. In R. M. Sousa, J. Vukelich, N. Weeden, A. Whitte­ Polhill and P. H. Raven (eds.), Advances in Legume Systematics, Part 2. Royal Botanic Gardens, Kew, more, Adelaide Botanical Garden, Hong U.K. Kong Botanical Garden, Kew Royal Botan­ JANSEN,R.K,ANDJ.D.PALMER. 1987. Achloroplast ical Gardens, and Zimbabwe National Bo­ DNA inversion marks an ancient evolutionary split tanic Gardens. We thank A. Bruneau and in the sunflower family (Asteraceae). Proc. Nat. Acad. Sci. USA 84:5818-5822. J. Doyle for technical assistance and R. C. KOLLER, B., AND H. DELIUS. 1980. Vicia faba chlo­ Bameby and R. M. Polhill for helpful com­ roplast DNA has only one set of ribosomal RNA ments. Financial support for this study was genes as shown by partial denaturation mapping provided by grants from the National Sci­ and R-Ioop analysis. Molec. Gen. Genet. 178:261­ ence Foundation (BSR-8700164 to M.L.; 269. OISHI, K. K, D. R. SHAPIRO, AND KK TEWARI. 1984. BSR-85 16630 and BSR-8805630 to J.J.D.; Sequence organization of a pea chloroplast DNA BSR-8717600 to J.D.P.) and from the gene coding for a 34,500-dalton protein. Molec. United States Department of Agriculture Cell. BioI. 4:2556-2563. (Hatch NYC-l 87405 to J.J.D.). PALMER, J. D. 1985. Comparative organization of chloroplast genomes. Ann. Rev. Genet. 19:325-354. PALMER, J. D., R. K JANSEN, H. J. MICHAELS, M. W. LITERATURE CITED CHASE, AND J. R. MANHART. 1988a. Chloroplast BENTHAM, G. 1865. Papilionaceae, pp. 465-562. In DNA variation and plant phylogeny. Ann. Missouri G. Bentham and J. D. Hooker (eds.), Genera Plan­ Bot. Gard. 75:1180-1206. tarum, Vol. I. Reeve & Co., London, U.K PALMER, J. D., B. OSORIO, J. ALDRICH, AND W. F. BOOKJANS, G., B. M. STUMMANN, AND K. W. THOMPSON. 1987. Chloroplast DNA evolution HENNINGSEN. 1984. Preparation of chloroplast among legumes: Loss ofa large inverted repeat oc­ DNA from pea plastids isolated in a medium of curred priorto other sequence rearrangements. Curro high ionic strength. Anal. Biochem. 141:244-247. Genet. 11:275-286. Cau, N. M., AND K. K TEWARI. 1982. Arrangement PALMER, J. D., B. OSORIO, AND W. F. THOMPSON. of the ribosomal RNA genes in chloroplast DNA 1988b. Evolutionary significance of inversions in of Leguminosae. Molec. Gen. Genet. 186:23-32. legume chloroplast DNAs. Curro Genet 14:65-74. CORBY, H. D. L. 1981. The systematic value of le­ POLHILL, R. M. 1981a. Papilionoideae, pp. 191-208. guminous root nodules, pp. 657-670. In R. M. Pol­ In R. M. Polhill and P. H. Raven (eds.), Advances hill and P. H. Raven (eds.), Advances in Legume in Legume Systematics, Part 1. Royal Botanic Gar­ Systematics, Part 2. Royal Botanic Gardens, Kew, dens, Kew, U.K. U.K. --. 1981b. Carmichaelieae, pp. 364-366. In R. CRONQUIST, A. 1981. An Integrated System ofClas­ M. Polhill and P. H. Raven (eds.), Advances in sification ofFlowering Plants. ColumbiaUniv. Press, Legume Systematics, Part I. Royal Botanic Gar­ N.Y. dens, Kew, U.K. DORMER, K. J. 1946. Vegetative morphology as a --. 1981c. Loteaeand Coronilleae, pp. 371-375. guide to the classification ofPapilionatae. New Phy­ In R. M. Polhill and P. H. Raven (eds.), Advances tol. 45:145-161. in Legume Systematics, Part 1. Royal Botanic Gar­ DOYLE, J. J., AND R. N. BEACHY. 1985. Ribosomal dens, Kew, U.K. gene variation in soybean (Glycine) and its rela­ --. 1981d. Galegeae,pp. 357-363. In R. M. Pol­ tives. Theoret. Appl. Genet. 70:369-376. hill and P. H. Raven (eds.), Advances in Legume DOYLE, J. J., AND J. L. DOYLE. 1987. A rapid DNA Systematics, Part 1. Royal Botanic Gardens, Kew, isolation procedure for small quantities offresh U.K. tissue. Phytochem. Bull. 19:11-15. POLHILL, R. M., AND P. H. RAVEN (eds.). 1981. Ad­ DUKE, J. A., AND R. M. POLHILL. 1981. Seedlings of vances in Legume Systematics, Part I. Royal Bo­ the Leguminosae, pp. 941-950. In R. M. Polhill tanic Gardens, Kew, U.K and P. H. Raven (eds.), Advances in Legume Sys­ SHINOZAKI, K., M. OHME, M. TANAKA, T. WAKASUGI, tematics, Part 2. Royal Botanic Gardens, Kew, U.K. N. HAYASHIDA, T. MATSUBAYASHI, N. ZAiTA, J. FERGUSON, 1. K., AND J. J. SKVARLA. 1981. The pollen CHUNWONGSE, J. OBOKATA, K. YAMA­ morphology of the subfamily Papilionoideae (Le­ GUCHI-SHINOZAKI, C. OHTO, K TORAZAWA, B. Y. guminosae), pp. 859-898. In R. M. Polhill and P. MENG, M. SUGITA, H. DENO, T. KAMOGASHIRA, K 402 M. LAYIN ET AL.

YAMADA, J. KUSUDA, F. TAKAIWA, A. KATO, N. STIRTON, C. (ed.). 1987. Advances in Legume Sys­ TOHDOH, H. SHIMADA, AND M. SUGIURA. 1986. tematics, Part 3. Royal Botanic Gardens, Kew, U.K. The complete nucleotide sequence of the tobacco STRITCH, L. R. 1984. Nomenclatural contributions chloroplast genome: Its gene organization and to a revision of the genus Wisteria. Phytologia 56: expression. EMBO J. 5:2043-2049. 183-184. SPRENT, J. I. 1981. Functional evolution in some TUCKER, S. C. 1987. Pseudoracemes in papilionoid papilionoid root nodules, pp. 671-676. In R. M. legumes: Their nature, development, and variation. Polhill and P. H. Raven (eds.), Advances in Legume Bot. J. Linn. Soc. 95:181-206. Systematics, Part 2. Royal Botanic Gardens, Kew, TURNER, B. L., AND O. S. FEARING. 1959. Chromo­ U.K. some numbers in the Leguminosae II. African STRAUSS, S. H., J. D. PALMER, G. T. HOWE, AND A. H. species including phyletic interpretations. Amer. J. DOERKSEN. 1988. Chloroplast genomes oftwo co­ Bot. 46:49-57. nifers lack a large inverted repeat and are exten­ sively rearranged. Proc. Nat. Acad. Sci. USA 85: Corresponding Editor: R. L. Honeycutt 3898-3902.