Theor Appl Genet (1992) 85:331-340 ~4?~:IN 2~~ ~ '~ ,~I ~+~N~ ~>~ 9 Springer-Verlag 1992 Phylogeny of Brassica and allied genera based on variation in chloroplast and mitochondrial DNA patterns: molecular and taxonomic classifications are incongruous A.K. Pradhan 1, S. Prakash 2, A. Mukhopadhyay 1, and D. Pental 1 1 Tata Energy Research Institute, 90 Jor Bagh, New Delhi 110 003, India 2 Biotechnology Centre, Indian Agricultural Research Institute, New Delhi 110 012, India Received April 14, 1992; Accepted May 20, 1992 Communicated by R. Hagemann Summary. Chloroplast D N A (cpDNA) variability of 60 taxa of the genus Brassica and allied genera comprising 50 species was studied. RFLPs for seven enzymes were generated and F values were estimated from five frequently cutting enzymes. Phenetic clusterings indicated a clear division of Brassica coenospecies into two distinct lineages referred to as the Brassica and Sinapis lineages. Two unexplored genera, Diplotaxis and Erucastrum, also exhibited two lineages in addition to the genera Brassica and Sinapis. This finding is inconsistent with the existing taxonomic classification based on morphology. Mitochondrial D N A (mtDNA) variability studied from EcoRI R F L P patterns, by hybridizing total D N A with four cosmid clones containing non-overlapping m t D N A fragments, did not show any congruence with c p D N A variation patterns. However, at the cytodeme level, the patterns of genetic divergence suggested by the c p D N A data could be correlated with m t D N A variation. In the Brassica lineage, Diplotaxis viminea was identified as the female parent of the allotetraploid D. rnuralis. The chloroplast DNAs of Erucastrum strigosum and Er. abyssinicum were found to be very closely related. In the Sinapis lineage, Brassica maurorurn was found to be the diploid progenitor of autotetraploid B. cossoneana. B. amplexicaulis showed a very different cpDNA pattern from other members of the subtribe. Brassica adpressa was closest to Erucastrurn laevigaturn and could be the diploid progenitor of autotetraploid Er. laevigatum. Based on the close similarity of the c p D N A pattern of Diplotaxis siifolia with that of D. assurgens, we have proposed the retention of this species in the genus Diplotaxis. The taxonomic positions of some other species have also been discussed. Correspondence to." A.K. Pradhan Key words: Brassica coenospecies - Subtribe Brassicinae - Chloroplast D N A - Mitochondrial D N A - Phylogeny Introduction The subtribe Brassicinae of the family Brassicaceae includes a considerable number of wild and crop species. Based on morphological observations, Schulz (1936) recognized 11 genera and 90 species in this subtribe. GomezCampo (1980), however, recognized only nine genera. The relationship of species belonging to the Brassicinae has also been extensively studied cytogenetically. Harberd (1972, 1976) took into consideration the chromosome number, chromosome pairing, and the extent of fertility in the hybrids and established 46 cytodemes which included the following genera - Brassica, Diplo- taxis, Erucastrum, Raphanus, Hirschfeldia, Trachystoma, Sinapis, Enarthrocarpus, Sinapidendron, Eruca and Hutera. He referred to these as Brassica coenospecies. The limits of this grouping did not strictly correspond with that of the subtribe Brassicinae as proposed by Schulz. Harberd removed Reboudia and transferred two genera, Raphanus and Enarthrocarpus, from the subtribe Raphaninae to the Brassicinae. Takahata and Hinata (1983) recognized a total of 53 cytodemes in the Brassica coenospecies. These studies helped in defining the relationships of different disputed taxa within a species but the relationship between the demes remained obscure. Extensive studies have been conducted on meiotic chromosome pairing of interspecific and intergeneric crosses between species of Brassica and related genera (Mizushima 1968; Harberd and McArthur 1980). However, due to wide morphological variations within genera, the evolutionary relationships based on morphological data 332 (Takahata and Hinata 1986) were not always congruent with the cytological data. Species relationship can be studied at different levels morphological, cytogenetical (chromosome structure and extent of pairing), biochemical (isozyme and protein electrophoresis patterns) and molecular (restriction patterns and R F L P analysis of organelle and nuclear genomes). The small, relatively constant, size and conserved nature of the chloroplast (cp) genome makes it an ideal molecule for phylogenetic studies of different plant species (Palmer 1985). In brassicas polymorphism in c p D N A (Erickson et al. 1983; Palmer et al. 1983; Yanagino et al. 1987; Warwick and Black 1991) and nuclear D N A (Song et al. 1988a, b, 1990; Hosaka et al. 1990) have been used to study interspecific relationships. However, most of these studies were confined to a limited number of species. The only exception to-date is the study of Warwick and Black (1991) who established the molecular systematics of 33 taxa (28 species) of the subtribe Brassicinae on the basis of c p D N A polymorphism as detected by the hybridization of c p D N A probes of B. juncea with total cellular D N A . However, the phylogenetic relationship of some of the genera closely related to Brassica, namely Erucastrum and Diplotaxis, were not examined in detail. The present study was undertaken with the objective of establishing the phylogenetic relationship of 60 taxa of the genus Brassica and allied genera comprising 43 diploid and seven tetraploid species based on variability in c p D N A and mitochondrial (mt) D N A restriction profiles. This study, therefore, extends the work of Warwick and Black (1991) by including a number of species belonging to Diplotaxis and Erucastrum and compares the phenetic classification derived from c p D N A and m t D N A analyses. An attempt has been made to relate the present findings to the available data on R F L P analysis of nuclear D N A , and the existing cytogenetic and taxonomic classifications. Materials and methods The different species used in the present investigation are listed in Table 1. Healthy, well expanded leaves from 30-45 day-old plants were collected and stored at - 7 0 ~ until used. Chloroplast DNA analysis Chloroplast DNA was isolated from 10 g of leaf material following Kemble (1987). DNA was digested with BamHI, EcoRI, ClaI, HindlII, HaelII, XhoI and PstI and electrophoresed on either 0.6% or 0.8% agarose gels in TBE (89 mM Tris, 2.5 mM EDTA and 89 mM Boric acid). Gels were photographed on a UV transilluminator. Mitochondrial DNA analysis Total DNA was extracted using the method described by Dellaporta et al. (1983). DNAs were digested with EcoRI, elec- trophoresed and transferred to nylon membranes (Hybond N, Amersham). Filters were sequentially hybridized with four cosmid clones containing non-overlapping mitochondrial DNA fragments from a mtDNA library of alloplasmic male-sterile Brassica campestris containing B. oxyrrhina cytoplasm (a detailed map of these clones will be published elsewhere). These clones, hereafter named as pCos42, 61,88 and 131, have mtDNA inserts of 40.0, 32.7, 32.4 and 34.8 kb, respectively. Probes were labelled with a2p dCTP using a multiprime labelling kit (Amersham). Hybridization was carried out overnight at 42 ~ in the presence of 50 % formamide (Perbal 1988). Hybridization filters were washed under stringent conditions (2 x 15 min at room temperature and 2 x 15 rain at 65~ in 0.2 x SSC, 0.1% SDS) and exposed to Kodak X-OMAT AR films. For re-probing, membranes were stripped of radioactive probe by washing, for 30 rain each at 42 ~ in denaturing (0.4 M NaOH) and neutralizing (0.2 M Tris, pH 8.0; 0.1 x SSC; 0.5% SDS) solutions. Data analysis The F value (an estimate of similarity coefficient, F = 2Nxy/ Nx + Ny) was computed according to the method of Nei and Li (1979) where Nx was the number of bands detected in one cytoplasmic type, Ny the number of bands in the other cytoplasmic type, and Nxy the number of bands held in common between the two types. These parameters were computed from photographs of the same size as the original gel in the case of cpDNA and from the bands detected in autoradiograms in the case of mtDNA. F values were computed for each enzyme separately or else pooled for more than one enzyme. Phenograms were constructed according to F values by a computer programme following the unweighted pair-group procedure (Gentzbittel and Nicolas 1989). In case of cpDNA, phenograms were constructed either on the basis of F value estimates from fragment patterns of five enzymes (EcoRI, BamHI, ClaI, HindIII and HaeIII) or three enzymes (BamHI, ClaI and HindIII). Bands of up to 0.5 kb were taken into consideration in all enzymes with the exception of ClaI where bands up to 1 kb were considered. Results and discussion Analysis of chloroplast DNA Chloroplast DNAs, isolated from 60 taxa comprising 50 species belonging to Brassica and allied genera, were restricted with seven different endonucleases. As an example, the patterns of BamHI-generated fragments are shown in Fig. 1. It was found that the use of frequently cutting enzymes was more informative than rarely cutting enzymes for distinguishing very closely related species. Hence, the phenetic clusterings in the present study were based on the R F L P patterns of the frequently cutting enzymes (EcoRI, BamHI, ClaI, HindlII and HaelII). Phylogenetic trees, constructed on the basis of length and restriction site mutations of c p D N A , show a close congruence with a phenetic classification based on genetic distance or simple matching similarity coefficient (Dally and Second 1990; Warwick and Black 1991). This rationale was followed in the present study to establish the phylogenetic relationship among Brassica coenospecies by a phenetic classification based on c p D N A variation. Species relationship, as determined by phenetic cluster- 333 Table 1. List of the species of Brassica and allied genera included in this study Table 1 (continued) Taxa a Taxa ~ Brassica adpressa Boiss. B. fruticulosa Cyr. B. maurorum Durieu B. spinescens Pomel B. nigra (L) Koch cv. IC257 ~ B. oleracea L ~ var. alboglabra (chinese-kale) var. botrytis (cauliflower) var. capitata (cabbage) var. italica (Broccoli) B. oxyrrhina Coss. B. barrelieri (L) Janka B. tournefortii Gouan B. campestris L ~ ssp. oleifera var. brown sarson var. yellow sarson var. toria ssp. japonica ssp. chinensis ssp. parachinensis ssp. narinosa B. gravinae Ten. B. repanda (Willd.) DC B. amplexicaulis (Desf.) Pomel B. cossoneana (Boiss. & Reut.) Maire B. carinata Braun cv. IC218 ~ B. juncea (L) Czern. cv. Pusa bold ~ B. napus L ~ cv. Regent cv. BO-15 Diplotaxis erucoides (L) DC D. siettiana Maire D. virgata (Cav) DC D. catholica (L) DC D. assurgens (Del.) Gren D. tenuisiliqua Del. D. siifolia G. Kunze D. viminea (L) DC D. tenuifolia (L) DC D. cretacea Kotov D. pitardiana Matte D. harra (Forsk.) Boiss. D. muralis (L) DC Eruca sativa (Mill.) Thell. Eruca vesicaria (L) Cav. Erucastrum varium Durieu Er. virgatum Presl. Er. leucanthum Coss. & Durieu Er. strigosum (Thunb.) O.E. Schulz Er. cardaminoides (Webb ex Christ) O.E. Schulz Er. laevigatum (L) O.E. Schulz Er. abyssinicum (Rich.) O.E. Schulz Sinapis aucheri (Boiss.) O.E. Schulz S. arvensis L S. pubescens L. Code no. Gametic b chromosome number Bad Bf Bm Bs 7 8 8 8 B 8 C1 C2 C3 C4 Bo Bb Bt 9 9 9 9 9 10 10 A1 A2 A3 A4 A5 A6 A7 Bg Br Ba Bc 10 10 10 10 10 10 I0 10 10 11 16 BC 17 AB 18 ACI AC2 De Ds Dv Dc Da Dt Dsi Dvm Dte Dcr Dp Dh Dm Es Ev Erv Eri Ere Ers Erc 19 19 7 8 9 9 9 9 10 t0 11 11 11 13 21 11 11 7 7 8 8 9 Erl Era Sa Sar Sp 14 16 7 9 9 S. alba L. S. flexuosa Poir. Raphanus raphanistrum L. Raphanus sativus L. Hutera leptocarpa Gonzalez-Albo Rhynchosinapis pseuderucastrum (Brot.) Gz. Campo Moricandia arvensis (L) DC Erucaria ollivieri Maire no. Gametic b chromosome number Sal Sf Rr Rs H1 Rp 12 12 9 9 12 12 Ma Ero 14 8 Code a Seeds from Prof. K. Hinata, Tohoku University, Sendal, Japan and Prof. C. Gomez-Campo, Universidad Politechnica, Madrid, Spain b Chromosome numbers are from Gomez-Compo and Hinata (1980) ~ Collection from our germplasm ing using three enzymes (BamHI, CIaI and HindIII), is shown in Fig. 2. Similarly, five enzymes (EcoRI, HaeIII, BamHI, ClaI and HindIII) were used to construct the phenogram for establishing the relationship between the species within a genus (Fig. 3 A - D ) . Chloroplast D N A analysis shows two distinct evolutionary lineages Phenetic classification of the c p D N A patterns of different species of Brassica and allied genera (Fig. 2) clearly established two evolutionary pathways which we refer to as the Brassica and Sinapis lineages and which correspond to the rapa/oleracea and nigra lineages of Warwick and Black (1991). Our results, based on phenetic clustering, confirm the earlier work of Warwick and Black (1991) and extend it in a significant way by showing that two distinct lineages are also present in Diplotaxis and Erucastrum (Fig. 3 B, D). In addition, these molecular analyses (Warwick and Black 1991; this study) show that the generic classifications of Schulz (1936) and GomezCampo (1980) do not reflect the natural relationship of species and genera within the subtribe Brassicinae. Comparison o f phenetic classification based on cpDNA with cytodeme classification and chromosome pairing data The cytodeme classification of Harberd (1972, 1976) and Takahata and H i n a t a (1983) is primarily based on crossability using high fertility of the F 1 as an important criterion for establishing cytodemes. A high level of congruence was observed between the recognized cytodemes and the c p D N A data. In order to establish the relationships between species belonging to different cytodemes, 334 chromosome pairing data were considered. However, the pairing data were not congruent with the phenetic classification based on cpDNA, since the extent of chromosome pairing between species across the lineage could be as high as, or even more than, that between species within a lineage. For example, hybrids of Diplotaxis erucoides (Brassica lineage) x Brassica nigra (Sinapis lineage) form up to seven bivalents (Quiros et al. 1988) while D. erucoides x B. oleracea (Brassica lineage) form only up to four bivalents (Mizushima 1972). Similarly, Diplotaxis tenuifolia belonging to the Brassica lineage shows very little homology with B. oleracea ( 0 - 2 bivalents) but forms 0 - 6 bivalents with Hirschfeldia incana and 0 - 4 bivalents with Hutera (Harberd and McArthur 1980). Significant chromosome homology has also been shown between many other species belonging to the two distinct lineages, e.g., B. campestris x B. nigra (up to eight bivalents, Prakash 1973), Sinapis arvensisx B. campestris/ oleracea (up to five bivalents, Mizushima 1950) and Hutera x B. oleracea (up to seven bivalents, Harberd and McArthur 1980). The extent of pairing, therefore, does not reflect the extent of divergence shown by c p D N A analysis. Comparability of pairing between species within a lineage and across lineages could be due to frequent hybridization between species since a large number of them are sympatric in distribution. Phenetic classification based on cpDNA in comparison with R F L P analysis of nuclear DNA Song et al. (1988 a, b, 1990) have studied the taxonomy of Brassica species based on RFLP analysis of nuclear DNA. Using the RFLP data they were able to establish a relationship between sub-species of B. oleracea and B. campestris and between the closely related B. nigra and Sinapis arvensis (Song et al. 1988b, 1990). However, the cluster analysis of distantly related species is incongruent with the chloroplast data. While cpDNA analysis shows D. erucoides to be closely related to oleraeea/campestris the nuclear RFLP shows D. erucoides to be very distant. Thus the nuclear RFLP data is only useful for establishing the relationships of closely related species and may give aberrant results when applied to more diverse species. Phenetic classification based on mtDNA analysis For constructing a phenogram based on m t D N A diversity, RFLP patterns were generated by hybridizing EcoRI- Fig. 1. BamHI restriction fragment patterns of cpDNAs from different species of the genus Brassica and its allied genera. The restriction pattern of Er. virgatum is not included. The code numbers of the species are given in Table 1. M, 2 DNA marker restricted with HindlII 335 0.6- j 0.7. 1 I 0.8 tI 0.9- 1.0- I BRASSICA LINEAGE ~~ - ~ - ~.~ SINAPIS I I ~ ~-~ ~. LINEAGE I Fig. 2. Phenogram showing relationship among species of the genus Brassica and its related genera based on chloroplast DNA variability. F value estimates from the combined data of three enzymes (ClaI, BamHI and HindIII) were used to construct the UPGMA phenogram. B. barrelieri and Er. virgatum are not included in the construction of the phenogram. B, Brassica; D, Diplotaxis; E, Eruca; Er, Erucastrurn; Eru, Erucaria; H, Hutera; M, Moricandia; R, Raphanus and S, Sinapis. BO-15, a synthetic B. napus with B. campestris cytoplasm; bs, brown sarson; nar, narinosa and chi, ehinensis; " includes Rhynchosinapis pseuderucastrum; b includes alboglabra, botrytis, eapitata and italica; c includes yellow sarson, toria, japonica and parachinensis 0.6- DIPLOTAXIS BRASSICA SINAPIS ERUCASTRUM C D I 0.7 I "= " 0.8 0.9 iii A B ~ Fig. 3. Phenogram showing cpDNA diversity patterns based on F value estimates using five enzymes (EcoRI, BamHI, HindIII, ClaI and HaeIII) in the genera Brassica (A), Diplotaxis (B) and Sinapis (C) and using four enzymes (ClaI, HindIII, EcoRI and HaeIII) in the genus Erucastrum (D). bs, brown sarson; nat, narinosa; ys, yellow sarson; jap, japonica; par, parachinensis and chi, chinensis digested total D N A with four heterologous probes of m i t o c h o n d r i a l origin. A n example o f the hybridization pattern with three different probes is shown in Fig. 4. The p h e n o g r a m based on the R F L P patterns is presented in Fig. 5. In general, the diversity pattern of m t D N A did not show any congruence with the c p D N A diversity pattern (Figs. 2, 5). However, at the cytodeme level the d a t a are congruent. F o r example, D. tenuifolia, D. cretacea and D. pitardiana, belonging to same cytodeme ( H a r b e r d 1976), also show a close relationship in b o t h cp and 336 Fig. 4, Autoradiographs of EcoRI-digested total DNA probed with the heterologous mtDNA probes pCos42 (a), pCos61 (b) and pCos 131 (e) showing variation among some of the Brassica species. The code number of the species are given in Table 1. Faint bands were detected by longer exposure of the autoradiograms 0.6 [ 0.7 L _1 0.8, u. -'1i--i 0,9" 1,0 ~-g ~~ z~ ~.~ ~ -2~ .~% ~ ~~ . ~ ~@ ~.~ ~ ~ ~-~ ~ g ~ . ~ % b , ~N ~ ~a~-~ .k, ~ ~ .~ ~.~ ~. " ~'~'~ ~ i~ ~ ~ ~ ~~,~;~ ~ ~~ ~~ ;~~' ~,~ ~ ~'~ ~ ~..~ ~% ~_ ~,:~ ~.~ ~-~ ~-- 7,- z. ~ i Fig. 5. Phenogram showing mtDNA diversity pattern in the genus Brassica and its allied genera based on F value estimates form EcoRI RFLP patterns detected from total DNA hybridized to four cosmid clones (pCos42, 61,88 and 131) containing non-overlapping mtDNA fragments, bs, brown sarson; " includes R. pseuderucastrum; b includes alboglabra, botrytis, capitata, italica; c includes yellow sarson, toria, japonica, chinensis, parachinensis and narinosa m t D N A patterns (Figs. 2, 3 B and 5). In plants, m t D N A is k n o w n to be extremely dynamic and, while its gene sequences are highly conserved, their location changes very rapidly due to intragenomic recombination. Therefore, phenetic classification based on m t D N A R F L P pattern cannot index the divergence in a critical manner. Terachi and Tsunewaki (1986), in their w o r k with Triticum and Aegilops, pointed out that the m t D N A restriction patterns are useful in clarifying phylogenetic relationship between different accessions of a species, or even between species that have very similar chloroplast genome. On the other hand, mitochondrial genome 337 variability is so extensive among species with different chloroplast genomes that phylogenetic relationships cannot be inferred from mtDNA. It can be concluded that the chromosome pairing, nuclear RFLP and mtRFLP data can be used only to ascertain relationships amongst closely related types. These data are difficult to use for developing a natural relationship among all the species in an ascending order of divergence. Phylogenetic analysis Phylogeny of Brassica. Cytogenetic evidence has shown that the three basic diploid species, B. nigra, B. oleracea and B. eampestris, evolved in ascending order and are secondary balanced polyploids from a common archetype having the chromosome constitution x = 6 (Robbelen 1960; Prakash and Hinata 1980). However, our findings clearly discount the monophyletic origin of the diploid species and agree with the proposals of Song et al. (1988a) and Warwick and Black (1991) that the genus is of biphyletic origin (Fig. 3A), where B. eampestris/oleracea had a common progenitor and B. nigra evolved through another pathway. The suggestion that B. campestris and B. oleracea evolved from the same ancestor is supported by several lines of evidence: close similarities in their cytoplasms (Palmer et al. 1983; Erickson et al. 1983; Yanagino et al. 1987; Warwick and Black 1991; the present study), high chromosome pairing in F 1 hybrids of B. campestris x B. oleraeea (up to nine bivalents, Olsson 1960; Namai 1971), serological similarities in seed proteins (Vaughan et al. 1966), and nuclear DNA studies (Song et al. 1988a, b, 1990; Hosaka et al. 1990; McGrath and Quiros 1991). Diplotaxis erueoides appears to be the closest ancestor involved in the origin of B. oleracea and B. eampestris, as earlier proposed by Warwick and Black (1991). High homologies of repeat sequences between D. erucoides and B. campestris/oleracea (96 and 94% homology to B. eampestris and B. oIeracea, respectively, Harbinder and Lakshmikumaran 1990) substantiate the view that D. erucoides is related to the ancestor of B. campestris/B, oleraeea. Our cp and mtDNA data indicate that B. nigra is very close to Sinapis arvensis (Figs. 2, 5). This close relationship between S. arvensis and B. nigra had been suggested earlier from cpDNA data (Yanagino et al. 1987; Warwick and Black 1991) and nuclear RFLP data (Song et al. 1988a). Warwick and Black (1991) proposed that the genus Sinapis could be redefined to include three species of Brassica, viz., B. nigra, B. frutieulosa and B. tournefortii. Our cpDNA analysis shows that some of the Diplotaxis and Erucastrum species are more closely related to the Sinapis complex than this complex is to B. fruticulosa and B. tournefortii (Fig. 2). Therefore, only the transfer of B. nigra to the genus Sinapis can be justified. Three species, B. fruticulosa, B. maurorum and B. spinescens, all with n = 8, have been grouped into one cytodeme by Harberd (1972, 1976) and Takahata and Hinata (1983). The cpDNA and mtDNA analyses in the present study are congruent with this classification. B. cossoneana (2n--32) was proposed as an autotetraploid of B. fi'uticulosa (Harberd 1972). However, in view of its close similarities in cp and mtDNA with B. maurorum, it would be more reasonable to assume that it originated from B. maurorum. Song et al. (1990) held the view that B. tournefortii evolved from a B. campestris/oleracea ancestor with introgression from B. nigra on the basis of similarity in cpDNA RFLP pattern (only one probe and one enzyme was used) with B. campestris/oleracea and high homology in nuclear DNA with B. nigra. Our results contradict this view and agree with Warwick and Black (1991) that B. tournefortii belongs to the Sinapis lineage. The taxonomic position of B. arnplexicaulis has long been controversial. It was placed in the section Melanosinapis alongwith B. dimorpha and B. nigra by Schulz (1936). Takahata and Hinata (1986), however, did not find any justification to group it with B. nigra. Harberd (1976) considered it to be an ancient relict since it showed very strong reproductive isolation and proved to be negative in reaction both as male and female parent. The phenetic clustering in the present study puts B. amplexicaulis at the base of the Sinapis lineage and most distant from the other taxa (Figs. 2, 3 A). Takahata and Hinata (1983) provisionally included B. gravinae into the B. repanda cytodeme. Our cpDNA data confirms the closeness of these two species and place them in the Brassica lineage (Figs. 2, 3 A). The HindIII restriction pattern of B. barrelieri was not available. The phenogram constructed on the basis of four enzymes (EcoRI, BamHI, ClaI and HaeIII) indicated that B. barrelieri is close to B. oxyrrhina and belongs to the Brassica lineage, as has been reported by Warwick and Black (1991). Phylogeny of Diplotaxis. The cpDNA data presented in this study clearly divides the different species of Diplotaxis into two distinct lineages suggesting a biphyletic origin for this genus (Fig. 3 B). A high level of congruence was observed between recognized cytodemes (Harberd 1976; Takahata and Hinata 1983) and clusters defined by cpDNA variation. D. muraIis (n = 21) is a natural allotetraploid between D. tenuifolia ( n = l l ) and D. viminea (n= 10, Harberd 1972). Close similarities of cp and mtDNA between D. viminea and D. muralis points to D. viminea as the cytoplasmic donor in the formation of D. muralis and also indicates that this allopolyploid is of relatively recent origin. The status of D. siifolia in the genus Diplotaxis has been a point of contention. Gomez-Campo and Tortosa 338 (1974) and Takahata and Hinata (1986) assigned it to Brassica, based on similarities in cotyledon shape. The present study showed that D. siifolia has a cpDNA very similar to that of D. assurgens and both are known to have sub-biseriate seeds. These facts indicate that D. siifolia might have evolved from D. assurgens so that it should remain within the genus Diplotaxis. Phylogeny of Erucastrum. Seven species of genus Erucastrum, comprising five diploids and two tetraploids, were found to be classified into two distinct groups (Fig. 3 D). One group with gametic chromosome numbers of 8 and 16 showed association with the Brassica lineage whereas the others, with n = 7 , 9 and 14, were associated with the Sinapis lineage. The status of Brassica adpressa (syn. Hirschfeldia incana, n = 7) is controversial; it was treated as B. adpressa by Mizushima (1968) whereas Harberd (1972), Gomez-Campo (1980) and Takahata and Hinata (1983) regarded it as separate from Brassica. The cpDNA pattern of this species is very close to those of Er. laevigatum and Er. virgatum. We suggest that it should be placed under the genus Erucastrum and disagree with the proposal of Warwick and Black (1991) for its inclusion in the genus Sinapis. Er. laevigatum (n = 14) has been reported to be an autotetraploid of Er. virgatum (n = 7) because the hybrid behaves as a typical trivalentforming triploid (Harberd 1976). It also forms seven bivalents and a high frequency of trivalents in the hybrid with B. adpressa (Harberd and McArthur 1980). Our cpDNA analysis indicates that Er. laevigatum is closer to B. adpressa than to Er. virgatum (Fig. 3 D) although all three species form a close cluster with a F value of 0.95. On the other hand the mtDNA pattern shows that Er. laevigatum is close to Er. virgatum. Further work is required to identify the exact diploid progenitor of Er. laevigatum. Er. abyssinicum (n=16) is close to Er. strigosum (n = 8) in cpDNA. Er. abyssinicum has been proposed to be an autotetraploid of an unknown species (Harberd 1976). It is probable that Er. strigosum, or else a closely related species, is the diploid parent of Er. abyssinicum. At the intrageneric clustering level Er. cardaminoides was found to be more closely associated with Er. varium than with any other Erucastrum species (Fig. 3 D). However, at the coenospecies level the former showed a closer association with Diplotaxis assurgens and D. tenuisiliqua than with Er. varium (Fig. 2). A similar situation was observed in terms of mtDNA classification (Fig. 5). This probably indicates a common origin of these taxa. Phylogeny of other related genera. We believe that the transfer of Raphanus into the subtribe Brassicinae is justified. Our cpDNA analysis is in agreement with the earlier report of Warwick and Black (1991) that Raphanus is close to S. aucheri. Eruea sativa and E. vesi- caria, belonging to same cytodeme (Harberd 1972), are very close in their cpDNA pattern, and are clustered with the Diplotaxis species tenuifolia, cretacea and pitardiana (Fig. 2). Both the Eruca and the Diplotaxis species have two rows of seeds in each fruit loculus. Therefore, Eruca spp. are closely related to Diplotaxis spp. of the Brassica lineage. Song et al. (1990) proposed a common origin of Eruca and Raphanus, based on nuclear RFLP data. However, the results of cpDNA analysis (present study) do not agree with this. Gomez-Campo (1977) studied the clinal variation in the Hutera-Rhynchosinapis complex and concluded that both should be placed in the genus Hutera. A high fertility of F 1 hybrids between Rynchosinapis and Hutera has been reported by Harberd (1972) and Vesperinas (1991). Our observations on cp and mtDNA also reveal that both genera have exactly the same type of cytoplasm and hence it is justified to merge them. Moricandia arvensis belongs to the subtribe Moricandinae. Its cpDNA restriction pattern differs considerably from other species of the subtribe Brassicinae and is aligned at the base of the Brassica lineage. Close genetic affinities are reported between Moricandia arvensis and the diploid Brassica species as revealed by the high degree of chromosome pairing in intergeneric hybrids [a maximum of six bivalents in hybrids with B. campestris and B. oleracea and five bivalents with B. nigra (Takahata 1990; Takahata and Takeda 1990)]. Although there is some evidence favouring the inclusion of Moricandia under the subtribe Brassicinae, further studies are required to confirm its taxonomic position. Similarly, the taxonomic classifications of Erucaria and Reboudia remain controversial (Schulz 1936; Harberd 1976). The cpDNA analyses of Reboudia (Warwick and Black 1991) and Erucaria (present study) place these near the base of the Sinapis lineage. However, the lack of any other supporting evidence makes it difficult to resolve this problem. Conclusion We have shown that there are two divergent lineages in the subtribe Brassicinae, which we have termed the Brassica and Sinapis lineages. This 'two-lineage concept' is reflected even at the genus level. Being highly conserved, the cpDNA analysis can be applied to resolve the taxonomic relationships of a large number and a wider range of taxa, whereas mtDNA, nuclear DNA RFLP, and cytogenetic approaches remain useful only for delineating the relationship of closely related taxa and species. We find that the phenetic classification based on cpDNA analysis is, by and large, incongruent with the taxonomic classification. Due to geographical distribution and breeding behaviour, the morphological characters used to distinguish different genera might have appeared independently in the two lineages, or free genetic exchange 339 might have occurred between the various species. To seek further confirmation for the presence o f two phylogenetic lineages in the Brassicinae we propose the two following lines o f work. (1) Use of a large n u m b e r o f speciesspecific probes o f nuclear origin. O u r preliminary w o r k with a B. nigra-specific p r o b e (probe pBNBH35, G u p t a et al. 1992) that does not hybridize with B. eampestris, and a B. campestris probe ( L a k s h m i k u m a r a n and R a n a d e 1990) that does not hybridize with B. nigra, has shown that, under high stringency conditions, the B. nigra probe hybridizes p r e d o m i n a n t l y with species o f the Sinapis lineage, and the B. campestris probe with species of the Brassica lineage. However, under low stringency there are some overlaps. Therefore, a number o f speciesspecific probes will have to be used on allopatric populations o f species belonging to the two lineages so that material with frequent nuclear gene exchange can be avoided. (2) Analysis o f the nuclear r D N A non-transcribed spacer and the study o f the sequence divergence o f some conserved nuclear genes. I f the presence o f two distinct lineages can also be shown at the nuclear D N A level then a m a j o r revision o f the t a x o n o m y o f the species of the subtribe Brassicinae will be fully justified. Acknowledgements. Technical assistance was provided by Mr. B. S. Yadav. The research was supported by a CEC grant number ECII-0193-IND(BA). References Dally AM, Second G (1990) Chloroplast DNA diversity in wild and cultivated species of rice (Genus Oryza, section Oryza). Cladistic-mutation and genetic distance analysis. Theor Appl Genet 80:209-222 Dellaporta SL, Wood J, Hicks JB (1983) A plant molecular DNA mini preparation. Version II. P1 Mol Biol Rep 1:19-21 Erickson LR, Straus NA, Beversdorf WD (1983) Restriction patterns reveal origins of chloroplast genomes in Brassica amphidiploids. Theor Appl Genet 65:201-206 Genztbittel L, Nicolas P (1989) A BASIC program to construct evolutionary trees from restriction endonuclease data. J Hered 80:254 Gomez-Campo C (1977) Clinal variation and evolution in the Hutera-Rhynchosinapis complex of the Sierra Morena (south central Spain). Bot J Linn Soc 75:119-140 Gomez-Campo C (1980) Morphology and morphotaxonomy of the Tribe Brassiceae. In: Tsunoda S, Hinata K, Gomez-Campo C (eds) Brassica crops and wild allies. Japanese Scientific Societies Press, Tokyo, pp 3-31 Gomez-Campo C, Hinata K (1980) A check list of chromosome numbers in the Tribe Brassiceae. In: Tsunoda S, Hinata K, Gomez-Campo C (eds) Brassiea crops and wild allies. Japanese Scientific Societies Press, Tokyo, pp 51-63 Gomez-Campo C, Tortosa ME (1974) The taxonomic and evolutionary significance of some juvenile characters in the Brassiceae. Bot J Linn Soc 69:105-124 Gupta V, Lakshmisita G, Shaila MS, Jagannathan V, Lakshmikumaran MS (1992) Characterization of species-specific repeated DNA sequences fi'om Brassica nigra. Theor Appl Genet (in press) Harberd DJ (1972) A contribution to cyto-taxonomy of Brassica (Cruciferae) and its allies. Bot J Linn Soc 65:1-23 Harberd DJ (1976) Cytotaxonomic studies of Brassiea-related genera. In: Vaughan JG, MacLeod A J, Jones BMG (eds) The biology and chemistry of the Cruciferae. Academic Press, London, pp 47-68 Harberd DJ, McArthur ED (1980) Meiotic analysis of some species and genus hybrids in the Brassiceae. In: Tsunoda S, Hinata K, Gomez-Campo C (eds) Brassica crops and wild allies. Japanese Scientific Societies Press, Tokyo, pp 65-87 Harbinder S, Lakshmikumaran M (1990) A repetitive sequence from Diplotaxis erucoides is highly homologous to that of Brassiea campestris and B. oleraeea. P1 Mol Biol 15:155-156 Hosaka K, Kianian SF, McGrath JM, Quirts CF (1990) Development and chromosomal localization of genome-specific DNA markers of Brassiea and the evolution of amphidiploids and n = 9 diploid species. Genome 33:131-142 Kemble RJ (1987) A rapid, single leaf, nucleic-acid assay for determining the cytoplasmic organelle complement of rapeseed and related Brassiea species. Theor Appl Genet 73:364370 Lakshmikumaran M, Ranade SA (1990) Isolation and characterization of a highly repetitive DNA of Brassica campestris. P1 Mot Biol 14:447-448 McGrath JM, Quirts CF (1991) Inheritance of isozyme and RFLP markers in Brassica campestris and comparison with B. oleracea. Theor Appl Genet 82:668-673 Mizushima U (1950) Karyogenetie studies of species and genus hybrids in the tribe Brassiceae of Cruciferae. Tohoku J Agr Res 1:1 14 Mizushima U (1968) Phylogenetic studies on some wild Brassica species. Tohoku J Agr Res 19:83-99 Mizushima U (i 972) Evolution of species in Brassiceae and their breeding. Kagaku to Seibutsu 10:78-85 Namai H (1971) Studies on the breeding of oil rape (Brassica napus var oleifera) by means of interspecific crosses between B. campestris ssp. oleifera and B. oleraeea. I. Interspecific crosses with the application of grafting method or the treatment of sugar solution. Jpn J Breed 21:40-48 Nei M, Li WH (1979) Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc Natl Acad Sci USA 76:5269-5273 Olsson G (1960) Species crosses within the genus Brassica II. Artificial Brassica napus L. Hereditas 46: 351:396 Palmer JD (1985) Chloroplast DNA and molecular phylogeny. BioEssays 2:263-267 Palmer JD, Shields CR, Cohen DB, Orton TJ (1983) Chloroplast DNA evolution and the origin of amphiploid Brassiea species. Theor Appl Genet 65:181 189 Perbal B (1988) A practical guide to molecular cloning, 2nd edn. John Wiley and Sons, Chichester, UK Prakash S (1973) Non-homologous meiotic pairing in the A and B genomes of Brassica: its breeding significance in the production of variable amphidiploids. Genet Res 21:133-137 Praksah S, Hinata K (1980) Taxonomy, cytogenetics and origin of crop Brassica, a review. Opera Bot 55:1-57 Quirts CF, Ochoa O, Douches DS (1988) Exploring the role of x = 7 species in Brassiea evolution: hybridization with B. nigra and B. oleraeea. J Hered 79:351-358 Robbelen G (1960) Beitr/ige zur Analyse des Brassiea Genomes. Chromosoma 11:205-228 Schulz OE (1936) Cruciferae. In: Engler A, Prantl K (eds) Die natiirlichen Pflanzenfamilien, 2nd edn. Vol 17. Engelmann, Leipzig, 227-658 Song KM, Osborn TC, Williams PH (1988 a) Brassica taxonomy based on nuclear restriction fragment length polymor- 340 phisms (RFLPs). 1. Genome evolution of diploid and amphidiploid species. Theor Appl Genet 75:784-794 Song KM, Osborn TC, Williams PH (1988 b) Brassiea taxonomy based on nuclear restriction fragment length polymorphisms (RFLPs). 2. Preliminary analysis of subspecies within B. rapa (syn. campestris) and B. oleracea. Theor Appl Genet 76:593-600 Song KM, Osborn TC, Williams PH (1990) Brassica taxonomy based on nuclear restriction fragment length polymorphisms (RFLPs). 3. Genome relationships in Brassiea and related genera and the origin of B. oleracea and B. rapa (syn. campestris). Theor Appl Genet 79:497-506 Takahata Y (1990) Production of intergeneric hybrids between a C3-C 4 intermediate species Moricandia arvensis and a C 3 species Brassiea oleracea through ovary culture. Euphytica 46:259-264 Takahata Y, Hinata K (1983) Studies on cytodemes in subtribe Brassicinae (Cruciferae). Tohoku J Agric Res 33:111 - 124 Takahata Y, Hinata K (1986) A consideration of the species relationships in subtribe Brassicinae (Cruciferae) in view of cluster analysis of morphological characters. Plant Sp Biol 1:79-88 Takahata Y, Takeda T (1990) Intergeneric (intersubtribe) hybridization between Morieandia arvensis and Brassica A and B genome species by ovary culture. Theor Appl Genet 80:38 -42 Terachi T, Tsunewaki K (1986) The molecular basis of genetic diversity among cytoplasm of Tritieum and Aegilops. 5. Mitochondrial genome diversity among Aegilops species having identical chloroplast genomes. Theor Appl Genet 73:175181 Vaughan JG, Waite A, Boulter D, Waiters S (1966) Comparative studies of the seed proteins of Brassiea eampestris, B. oleraeea and B. nigra. J Exp Bot 17:332-343 Vesperinas ES (1991) Experimental hybrids in the complex Hutera-Rhynehosinapis (Cruciferae) of Sierra Morena (Spain). Eucarpia Cruciferae Newslett 14/15:8-9 Warwick SI, Black LD (1991) Molecular systematics of Brassiea and allied genera (subtribe Brassicinae, Brassiceae) - chloroplast genome and cytodeme congruence. Theor Appl Genet 82:81-92 Yanagino T, Takahata Y, Hinata K (1987) Chloroplast DNA variations among diploid species in Brassica and allied genera. Jpn J Genet 62:119-125