Karyological notes onCentaurea sect.Acrocentron (Asteraceae)

Plant Systematics and Evolution

PL Syst. Evol. 179 1-18 (1992)

© Springer-Verlag1992 Printed in Austria

Karyological notes on Centaurea sect. Acrocentron

(Asteraceae) NI]RIA GARCIA JACAS and ALFONSO SUSANNA DE LA SERNA

Received January 2, 1991; in revised version April 18, 1991

Key words: Angiosperms, Asteraceae, Centaurea, Acrocentron.- Karyology, biogeography. Abstract: The karyology of Centaurea sect. Acrocentron is surveyed. 19 chromosome counts on 8 species are reported; those on C. acaulis, C. crocata, C. galianoi, C. pubescens, and C. malinvaldiana are new. The basic chromosome numbers of the section are x = 11 and x = 10. Karyological arguments have been used to show that evolution was from x = 11 to x = 10. This is supported by biogeographical data. Two main centres of diversification of sect. Acrocentron were studied from that point of view: the East and the Southwest Mediterranean region.

Chromosome counts and several karyotype analyses of sect. Acrocentron species have allowed us to resolve previous doubts about the basic number of this section of Cenlaurea L. Acrocentron is one of the karyologically best known sections, as is obvious from studies of GUINOCHET & FOISSAC (1962), GARDOU (1969, 1975), FERNANDEZ MORALES ~¢ GARDOU (1975), FERNANDEZ CASAS • FERNANDEZ MORALES (1979), VALDI~S BERMEJO ~; AGUDO MATA (1984), WAGENITZ ~¢ GAMAL-ELDIN (1985), FERNANDEZ CASAS & SUSANNA (1986), and numerous other short notes (cf. WAGENITZ & GAMAL-ELDIN 1985). Generally speaking, our delimitation of sect. Acrocentron coincides with that of WAGENITZ (1975) and WAGENITZ & GAMAL-ELOIN (1985). We agree with these authors as to the difficulty of establishing an intrasectional classification. According to our experience, the division into four sections proposed by TZVELEV (cf. WAGENITZ & GAMAL-ELDIN 1985) proves to be inapplicable to the western species of sect. Acrocentron. To give an idea of the problem: Centaurea x polymorpha is a very frequent hybrid of C. ornata and C. cephalariifolia WlLLK., both from sect. Acrocentron s.1. However, for DOSTAL (1976), who applies TzveLev's classification, C. polyrnorpha ought to belong to the eastern sect. Orientales. We also believe that a study of sect. Acrocentron must include sect. Chamaecyanus WILLI(. for palynological reasons put forward by WAGENn'Z (1955); this is also supported by the ease of hybridization between the two groups (FERNANDEZ CASAS & SUSANNA 1986: 58; 100, Table 29). We agree with GARDOU (1972) in connecting the two sections, but not in subordinating Charnaecyanus to sect. Acrocentron with

2

N. GARCIA JACAS & A. SUSANNA DE LA SERNA:

subsectional status. Although the type of pollen and the ease of hybridization indicate a very close relationship, Chamaecyanus has sufficient peculiarities to deserve sectional status. As for sect. Rizanthae with another type of pollen, it is clear that it cannot be included in sect. Acrocentron, contrary to the ideas of GARDOU (1972). On the other hand, we do include the species of the so-called sect. Borjae in sect. Acrocentron: Its basic n u m b e r x = 11 is frequent in sect. Acrocentron too, and the lack o f pappus is a reduction not u n c o m m o n in Centaurea s.1., and thus of little taxonomic relevance according to DITTRICH (1968). Finally, we would like to indicate that the last attempts to establish guidelines for the karyosystematics of sect. Acrocentron in its entirety are relatively out of date and were based on very few species only. N o w we have far more chromosome counts and a better understanding of the basic numbers involved. Material and methods

Mitosis. These observations were made on somatic metaphases using the squash technique. Radicular meristems of wild plants, cultivated in pots and the results of the germination of seeds, collected from wild populations, were used. Caulinary meristems from the vegetative shoots of wild plants were also used. The most satisfactory pre-treatment was with 8-hydroxiquinoline (8-quinolinol) at 4 °C for 8 h. The material was fixed with Carnoy for 24 h at low temperatures. Before staining, the material was hydrolized with HC1 5 N. It was dyed with acetic orcein (with Feulgen for the caulinary meristems) and mounted in acetic acid at 45%. In one case only, the material was dyed with alcoholic chlorhydric carmine, according to SNow's technique (1963). Meiosis. The observations were made on mother cells of the pollen grains, using the squash technique. Floral buds of wild plants were used, fixed with Carnoy at room temperature, hydrolized with HC1, stained with Feulgen, and mounted with acetic carmine. The photographs were taken with a Zeiss Standard microscope. Both the permanent preparations and the negatives are deposited in the laboratory of the Botanical Institute of Barcelona. Vouchers are preserved in the herbarium of the Botanical Institute of Barcelona (Be). Results and comments

The results are summarized in Table 1. Subsequently we shall c o m m e n t briefly on the results for each species; in some of them the chromosome formula is indicated according to LEVAN & al. (1964). Centaurea acaulis L. 2 n = 60. As far as we know, this is the first time that the chromosome n u m b e r o f this species has been studied. It is hexaploid, the third to be found in sect. Acrocentron. Its ecological affinities are those that are usually considered as typical of a young polyploid. It is a colonizing species inhabiting open areas, especially banks and ditches that have been dug up, with the help of its considerable capacity for vegetative reproduction. Centaurea clementei Boiss. 2 n = 20. There was a disparity of results for this species. HUMPHRIES • al. (1978: 393,402) published a count of 2 n = 18 on African material. For their part, FERNANDEZ CASAS & FERNANDEZ MORALES (1979: 116) took it for granted that C. clementei had 2 n--20, perhaps supported by a count that FERNANDEZ MORALES included in his doctoral thesis, a work which we have been unable to consult. To resolve the doubt, we have studied both African and Iberian materials and

Karyological notes on Centaurea

3

Table 1. Results of the chromosome counts carried out in Centaurea sect. Acrocentron Species

Chromosome no.

Population

C. acaulis DESF.

2n=60

Algeria: 24 km from Aflou to E1 Bayadh, GARCIA JACAS & SUSANNA 1269.

C. clementei BoIss.

2n=20

Morocco, T&ouan: Azla near T6touan, 700 m s.m., GARCIA JACAS & SUSANNA 1328. Spain, C6rdoba: Mt Cuevas Altas near Las Cuevas de San Marcos, 900m s.m., GARCIA JACAS & SUSANNA 1338.

C. crocata FRANCO

2n=40

Portugal: 5.5kin from Monchique to Saboia, GARCIA JACAS & SUSANNA 1329.

C. gabrielis-blancae FDEZ. CASAS

2n=20

Spain, Granada: 4 km S. from Alhama de Granada to Jfitar, 900 m s.m., GARCIA JACAS & SUSANNA 1222. Spain, Granada: 11.6 km from Granada to Sierra Nevada, c. 1 000m s.m., GARCIA JACAS & SUSANNA 1225.

C. galianoi FDEZ. CASAS • SUSANNA

2n=20

Spain, Huelva: Sierra de Aracena, from Puerto de la Cruz to La Nava, 550 m s.m., GARCIA JACAS, JULIA, J. M. MONTSERRAT 1929, SUSANNA & VENY.

C. granatensis BoIss.

n=10, 2n=20

Spain, Almeria: near Maria, Sierra de Maria, 1 600 m s.m., GARCIA JACAS c~LSUSANNA 1240. Spain, Almeria: near V61ez Blanco, Mt Maim6n, 1 300 m s.m., GARCIA JACAS & SUSANNA 1241 Spain, Almeria: near Maria, Sierra de Maria, 1 600 m s.m., GARCIA JACAS & SUSANNA 1237. Spain, Granada: 25.6 km from Granada to Sierra Nevada, c. 1 800m s.m., GARCIA JACAS & SUSANNA 1224. Spain, Granada: Sierra Nevada, near San Jer6nimo, c. 1 600m s.m., GARCIA JACAS & SUSANNA 1226. Spain, Granada: 26.9 km from Granada to Sierra Nevada, c. 1 850m s.m., GARCIA JACAS & SUSANNA 1228. Spain, Granada: 26.9 km from Granada to Sierra Nevada, c. 1 850m s.m., GARCIA JACAS & SUSANNA 1229. Spain, Granada: 26.9 km from Granada to Sierra Nevada, c. 1 850m s.m., GARCIA JACAS & SUSANNA 1230. Spain, Granada: near Casa Forestal de la Cortichuela in Mt Trevenque, 1 500 m s.m., GARCIA JACAS & SUSANNA 1231. Spain, Granada: near Casa Forestal de la Cortichuela in Mt Trevenque, 1 500m s.m., GARCIA JACAS & SUSANNA 1232.

2n=20

C. malinvaldiana ]~ATTAND.

n=10, 2n=20

Algeria, Mecheria: Djebel Antar, 5 km N. of Mecheria, near tower of TV, 1 400 m s.m., GARCIA JACAS & SUSANNA 1270.

C. pubescens WILLD.

n=10

Morocco, A1 Hoceima: Djebel Azrou Akechar near Tizi Ouzli, 1 400m s.m., GARCIA JACAS & SUSANNA 1255.

4

N. GARCIAJACAS t~LA. SUSANNADE LA SERNA;

in both cases the result has been 2 n = 20. It is clear that we can discount the existence of the basic number x = 9 in sect. Acrocentron, as the other count of 2 n = 18 for C. atropurpurea also turned out to be wrong; cf. GU~NOCHEa"& FOISSAC (1962: 380) and GARDOU (1975: 539, Table 1). Centaurea crocata FRANCO. 2 n = 40. According to our information, this is the first time the chromosomic number of this species has been studied. As can be found in the greater part of the polyploids in sect. Acrocentron, it is a colonizing species typical of banks, ditches, and other new habitats. Centaurea gabrielis-blancae FERNANDEZ CASAS. 2 n = 20 (Fig. 1 a: pop. S-1222; b: pop. 1225). 3M+4m+2sm+ 1st (S-1222); 5 M + 2 m + 3 sm (S- 1225). The chromosome number of this species is already known, as in principle all the counts of C. ornata var. microcephala WILLK. have to be referred to it (FERNANDEZ MORALES & GARDOU 1975, VALDI~S BERMEJO & AGUDO MATA 1985). As for C. gabrielis-blancae, its count was published by FERNANDEZ CASAS& GA~ARRA (1986 a: 6) with the same result as ours. The chromosome formula varies slightly according to the population studied; the difference can be attributed to the long time of pre-treatment employed. Centaurea galianoi FERNANDEZ CASAS & SUSANNA. 2 n = 20 (Fig. 1 c). 2 M + 5 m + l s m + 2 s m Sat. According to our information, this is the first time that the chromosome number of this species has been studied. However, taking into account the distribution of C. ornata, it is very likely that the count published by UBERA (1981: 238, no. 216) as "C. ornata WIILLD. subsp, ornata", from Huelva, should be attributed to C. galianoi: according to our Figures, C. ornata s. str. does not reach the province of Huelva, where it is substituted by C. galianoi. The result of this count, n = 10, coincides with our own. Centaurea granatensis Boiss. n = 10 (Fig. 2a, b: pop. S-1240; Fig. 2c, d: pop. S-1241), 2 n = 2 0 (Fig. ld: pop. S-1228; Fig. 1 e: pop. S-1232; Fig. 1, f: pop. S1237). 3 M + 4 m + 2 sm + 1 sm S"t (S- 1228); 3 M + 1 m + 2 m sat + 3 sm + 1 sm sat (S- 1232); 2 M + 3 m + 1 m sat + 2 sm + 2 sm ~at (S- 1237). Our counts coincide with those published by FERNANDEZ & MORALES & GARDOU (1975), FERNANDEZ CASAS (1977), LEAL & al. (1980), and VALD~S BERMEJO & AGUDO MATA (1984). The latter authors also discovered a triploid cytotype in the Sierra del Segura (Albacete). In the same way as in C. gabrielis-blancae, slight differences can be seen in the chromosome formulas between the populations studied (Fig. 1 d - f ) , which can, in our opinion, be attributed to the technique used; by applying a long pre-treatment, the chromosomes are very condensed and the separation of the chromosome types M or m - according to LEVAN & al. (1964) -- is very difficult. As far as meiosis is concerned (Fig. 2 a - d ) , we have not observed the irregularities perceived by FERNANDEZ MORALES & GARDOU (1975). They assume that these irregularities are due to hybridization between populations of C. granatensis already partially isolated genetically. This is possible, but we think that the irreg-

Karyological notes on Centaurea

5

l lllllii.'-"

i l l l l i . - m i l l in a i l - i

1oi

IIIIiI

i i

in;"

; | : , 1_,: . - : . . ,_.,

]

e l

| l

n ,,. n n l U l

.':, -'

=

-' |-''-'

?

= . m n u n

. .-'

Fig. 1. Somatic metaphases and haploid karyograms of different Centaurea spp. a, b C. gabrielis-blancae, a pop. S-1222; b pop. S-t225. c C. galianoi, d - f C. granatensis; d pop. S-1228; e pop. S-1232; fpopo S-1237. The karyograms are based on one plate

6

N. GARCIA JACAS & A. SUSANNADE LA SERNA:

Fig. 2. Meiosis of Centaurea spp. a Diakinesis of C. granatensis, pop. S-1240; b anaphase I of the same population, c, d Diakinesis of C. granatensis, pop. S-1241. e Diakinesis of C. malinvaldiana; f anaphase I of the same species, g, h Diakinesis of C. incana

Karyological notes on Centaurea ularities can be explained more easily: The ditches in the Andalusian region where C. granatensis grows are full of another species of the group, C. gabrielis-blancae (= C. ornata var. microcephala), which also has 20 chromosomes. Following the road from Granada to the Sierra Nevada, in Andalusia, it is easy to see both species growing almost together; populations are found that show various signs of introgression and the hybrid between the two species has already been seen in the Sierra del Segura by PAJARON (1988). It is very probable that the materials studied by FERNANDEZ MORALES & GARDOU (1975) belong in a lesser or greater extent to one of these hybrid populations, which would explain the irregularities in the meiosis. Centaurea mah'nvaldiana BATTAND. n = 10 (Fig. 2 e - f), 2 n = 20. According to our information, this is the first time that the chromosome number of this species has been studied. The meiosis is regular. This result does not resolve doubts about the affinities of the species, subordinated to C. granatensis by some authors; it supports the idea that the diploids with the basic number x = 10 in sect. Acrocentron generally have a more limited ecological behaviour and are found in more stable habitats than related polyploids; C. malinvaldiana is a rock plant, typical of the rocky limestone areas of the Saharian Atlas range. Centaurea pubescens WILLO. n = 10 (Fig. 2 g, h). This is a new count, according to our information: however, GARDOU (1975: 540) published a count for "Centaurea incana DESF. subsp, ornata (W~LLD.) MAIRE var. hookeriana (BALL) MAIRE subvar. cuprea MA~RE", with the result of 2 n = 20; a count that can perhaps be considered to be included in C. pubescens sensu lato. The meiosis is regular. As far as the ecological behaviour of this species is concerned, the population studied sometimes behaved as rock plants, but a l s o - more f r e q u e n t l y - a s ruderal.

Discussion Our results allow us to discard the presence of the basic number x = 9, which was only supported in the count of C. clementei that we have already mentioned. The basic number's remain x = 11 (confirmed in some SW. Mediterranean species and in many E. Mediterranean species, as can be seen in Tables 2 and 3) and x = 10. The next question is which one of them is ancestral in the section and which one derived. It is worth to say that the difference of basic number does not seem to involve drastic differences in the karyotype of the species. At least two hybrids are known between taxa of different basic n u m b e r s - a c t u a l l y one of them is fertile (RUNEMARK1969: 168, FERNANDEZCASAS~¢ SUSANNA 1986: 47, FERNANDEZCASAS GAMARRA 1986a: 6). FERNANDEZ CASAS & FERNANDEZ MORALES (1979) and FERNANDEZ CASAS ~1; SUSANNA (1986) have argued that the general direction of the chromosome number change has been from x = 11 to x = 10. WAGENITZ & GAMAL-ELDIN (1985: 125) left the change of direction open. We shall now look at the-karyological and biogeographical arguments in favour of our idea that the basic number x = 11 is ancestral. Karyological arguments. According to FERNANDEZ CASAS ¢~¢FERNANDEZ MORALES (1979: 121) and SILJAK-YAKOVLEV(1986) there is a tendency in Centaurea towards a reduction in the basic number and an increase in the asymmetry of the karyotype. This trend has been known for some time in other genera of the same

N . GARCIA JACAS & A. SUSANNA DE LA SERNA: Karyological notes on Centaurea

8

Table 2. Chromosome numbers of the E. Mediterranean species of Centaurea sect. Acrocentron (! indicates a count that we consider to be doubtful) Species

Count

Reference

C. achaia Boiss. & HELDR. C. aetholica PHITOS & GEORGIADIS C. atropurpurea WALDST. & KIT.

2n=22 2n=22 2 n = 20! 2n=18! 2n=22

C. corinthiaca Bolss. & HELDR. C. euboica RECH. f. C. grbavacensis (ROHLENA) STOJ & ACHT. C. psilacantha BoIss. & HELDR.

2n=22 2n=22 2n=22

WAGENITZ (~; GAMAL=ELDIN(1985) WAGENITZ & GAMAL-ELDIN(1985) GUINOCHET & FOISSAC(1962) GARDOU (1975) WAGENITZ & GAMAL-ELDIN(1985), ROUTSI & GEORGIADIS(1988) WAGENn'Z & GAMAL-ELDIN(1985) WAGENITZ & GAMAL-ELDIN(1985) WAGENITZ & GAMAL-ELDIN(1985)

2 n = 22 2 n = 20I 2n=22

WAGENITZ ~¢ GAMAL=ELDIN(1985) ROUTSI & GEORGIADIS(1988) WAGENITZ & GAMAL-ELDIN(1985)

C. acicularis SM.

2n=20

C. ebenoides HELDR, C. graeca GRISEB.

2 n = 20 2n=20

WAGENITZ & GAMAL=ELDIN(1985), ROUTSI & GEORGIADIS(1988) ROUTSI & GEORGIADIS(1988) WAGENITZ & GAMAL-ELDIN(1985), ROUTSI & GEORGIADIS(1988) KUZMANOV(1985) WAGENITZ (~ GAMAL=ELDIN(1985) ROUTSI & G~ORGIADIS(1988) WAGENITZ & GAMAL-ELDIN(1985), ROUTSI & GEORGIADm(1988) WAGENITZ & GAMAL-ELDIN(1985), ROUTSI & GEORGIADTS(1988)

Species with basic number x = 11

C. rechingeri PHITOS S p e d e s with basic number x = 10

C. C. C. C.

immanuelis-loewii DEGEN laconica Bolss. macedonica Boiss. raphanina SM.

2 n = 20 2n=20 2n=20 2n=20

C. redempta HELDR.

2n=20

C. salonitana VIs.

2n=20 2n=40 2n=20 2 n = 100 2n=l10 2n=20 2n=40

C. scabiosa L. C. spruneri Bolss. & HELDR. C. urvillei DC.

WAGENITZ ~¢ GAMAL-ELDIN(1985) WAGENITZ • GAMAL=ELDIN(1985) WAGENITZ (~ GAMAL=ELDIN(1985) WAGENITZ (% GAMAL-ELDIN(1985)

family (in Crepis and Leontodon by STEBBINS 1950: 449, 456, table 89, and 1971: 93 - 96). The mechanism by which the reduction in Centaurea has taken place also has been proposed: FERNANDEZ CASAS & FERNANDEZ MORALES (1979) suggested a series of reciprocal translocations followed by the elimination o f a chromosome without important genes; SILJAK-YAKOVLEV(1986) proposed Robertsonian translocations. In both cases, we should expect the production of a pair of longer

Table 3. Chromosome numbers of the SW. Mediterranean species of Centaurea sect. Acrocentron (the sign ! indicates a count that we consider to be doubtful; the sign ? a count of doubtful allocation Species

Count

References

Species with basic number x = 11

C. borjae VALDI~SB. & RIVAS GODAY n=33 2n=66 C. carolipauana FDEZ. CASAS & SUSANNA C. lainzii FDEZ. CASAS

2n=22

C. ultreiae SILVA PANDO

n=11?

2n=33

VALD]~S BERMEJO~1; RIVAS OODAY (1978), VALDI~SBERMEJO 8¢ AGUDO MATA (1984) FERNANDEZ CASAS & SUSANNA (1982b)

FERNANDEZ CASAS & FERNANDEZ MORALES (1979); VALDt~S BERMEJO& AGUDO MATA (1984) estimated

Species with basic number x = 10

C. collina L.

2n=60 2n=18! 2n=20 2n=60

C. crocata FRANCO C. gabrielis-blancae FDEZ. CASAS

2n=40 2n=20

C. galianoi FDEZ. CASAS & SUSANNA

n = 10 2n=20

C. granatensis BoIss.

2n=20

C. acaulis L. C. clementei BoIss.

2n=30 2n=20 2n=40

GARCIA JACAS ~¢ SUSANNA(this paper) HUMPHRIES ~¢ al. (1978) GARCIA JACAS & SUSANNA(this paper) GUINOCHET 8L FOISSAC (1962), SEIDENBINDER~7. VERLAQUE (1985) GARCIA JACAS & SUSANNA(this paper) QUEIROS (1973), FERNANDEZ MORALES & GARDOU (1975), VALDt;S BERMEJO & ACt:DO MATA (1984), FERNANDEZ CASAS & SUSANNA(1986 a) UBERA (1981) FERNANDEZ & QUEIR6S (1971), GARCIA JACAS & SUSANNA (this paper) FERNANDEZ MORALES & GARDOU (1975), FERNANDEZ CASAS (1977), LEAL & al. (1980), VALDtS BERMEJO & AGUDO MATA (1984), GARCIA JACAS & SUSANNA(this paper) VALD~S BERMEJO& AOUDO MATA(1984) QARDOU(1975), FERNANDEZCASAS& SUSANNA(1982a) FERNANDEZ CASAS d~; GAMARRA(1986 a)

C. josiae HUMBERT C. legionis-septimae FDEZ. CASAS 8¢ SUSANNA C. malinvaldiana BATTAND.

n=10

C. nana DESF. C. ornata WILLD.

2n=20 2n=40 2n=40

GARCIA JACAS • SUSANNA (this paper) FERNANDEZ CASAS & SUSANNA (1982a) FERNANDEZ 8,5 QUEIROS (1971), FERNANDEZ MORALES 8¢ GARDOU (1975), VALDI~S BERMEJO 8¢ AGUDO MATA

C. polymorpha LAG. C. prolongi Botss.

2 n=40 2n=20

C. pubescens WILLD. C. saxicola LAG.

n=10 2 n = 20? 2 n = 60

GARDOU (1972) FERNANDEZ CASAS & FERNANDEZ MORALES (1979), VALDI~S BERMEJO d~; AGUDO MATA (1984) GARCIA JACAS & SUSANNA (this paper) GARDOU (1975)

C. scabiosa L. s.1.

n = 20

(1984)

2n=20 2n=40

GARDOU (1972), FERNANDEZ MORALES & GARDOU (1975), VALD~S BERMEJO& AGUDO MATA (1984) UBERA (1981) GUINOCHET & FOISSAC (1962) KUMMER (1977), VALD~S BERMEJO & ACUDO MATA (1984), FERNANDEZCASAS & GAMARRA(1986 b)

10

N. GARCIAJACAS • A. SUSANNADE LA SERNA:

metacentric chromosomes, as actually reported in the karyotypes of species with x--- 10 by FERNANDEZ CASAS ~¢ FERNANDEZ MORALES (1979), FERNANDEZ CASAS SUSANNA (1982 a, 1986), and SlLJAK-YAKOVLEV (1986). That the change has taken place in the opposite direction (from x = 10 to x = 11) seems to us more difficult to maintain, if we follow STEBBINS (1950: 451, 458). In fact, it has been proposed only once, in the case of C. lainzii FERNANDEZ CASAS (which has 2 n = 3 x = 33) by VALDt~S BERMEJO & AGUDO MAYA (1984: 134). These authors doubt the reliability of the count x = 11 in sect. Acrocentron, as does GARDOU (1975: 539). Therefore they suggest that C. lainzii is not a triploid, but an aberrant form of C. prolongi, a species with 2 n-- 20. Nevertheless, FERNANDEZ CASAS & FERNANDEZ MORALES (1979: 118, Fig. 14) demonstrate that the chromosomes of C. lainzii can be gathered into three series of 11 chromosomes; on the other hand, C. lainzii is morphologically closer to C. carolipauana with x = 11 than to C. prolongi. Therefore it appears more reasonable to interpret 2 n = 33 as triploid based on x = 11. In the subtribe in general, the basic chromosome number is correlated with the pollen type: the genera or sections with a "primitive" pollen t y p e - p o l l e n of the types Crupina, Serratula, and Centaurea ("Centaurium") of WAGENITZ (1955) -- generally have higher basic numbers (x = 15 and x = 13 above all) than the sections with a more advanced type of pollen (the C. jacea and C. scabiosa types) that usually have x = l l , 10, 9, and 8. This correlation is supported by other correlations between the basic number and morphological characters. Higher basic chromosome numbers correspond to a more archaic floral anatomy: presence of staminodes in the marginal florets (WAGENITZ 1955); marginal florets not specialized (TONIAN1980); and the same p r i m i t i v e - not specialized-type of stigma that can be found in the related subtribe Cardueae-Carduinae. Other correlations more obscure but in the same line were suggested by GUINOCHET & FOISSAC (1962) and TONIAN (1980). In order to compare the frequency of the two basic numbers in the E. and the SW. Mediterranean region, we have collected the known counts in sect. Acrocentron in Table 2, corresponding to the E., and Table 3 corresponding to the SW. Mediterranean region. In the E. Mediterranean region, the proportion between the two basic numbers is fairly balanced: there are 8 species with x = 11, and 12 species with x = 10 (Table 2). In the SW. Mediterranean region - which should be considered as being a more recent speciation centre as we shall try to demonstrate l a t e r - t h e r e are only 4 species with x = 11, as opposed to 17 with x = 10 (Table 3). There is a great difference between the two studied regions in the proportion of polyploid species. In the E. Mediterranean region (Table 2), polyploidy is unknown among species with x = 11. This is probably due to the fact that these species are strongly allopatric, with restricted areas (Fig. 3) and without any possibility of hybridization, and generation of allopolyploids (compare STEBBINS 1950: 316, GRANT 1981:305 -- 306). A m o n g species with the basic number x = 10, with comparatively larger areas, a few polyploids are known; see PinTos 1970). In the SW. Mediterranean region, all species with the basic number x-- 11 in sect. Acrocentron (Table 3) and sect. Chamaecyanus are polyploids, with one exception (it is probable that the origin of the triploid C. lainzii could lie in the genesis

Karyological notes on Centaurea h,

, !

7

11

(,

,,,

..::::: ..:.:.:-

(

iiiiii

.~'

•' •" -"

,.,[,,-

( C.

grbavacensis

, ....

\

~

1

0 /

C. C.

rechingeri

aetolica [ C.

psilacantha>

aca

C.

~

~

C.

rechingeri

achaia

D

' ~

5 ~

%

~0 ~

C.

atropurp .... ~J

100 km

Fig. 3. Distribution areas of the species of Centaurea sect. Acrocentron with basic chromosome number x = 11 in the E. Mediterranean region

of a polyploid from an ancestral species with 2 n = 22; see LEwis 1980: 136, GRANT 1981: 309). But here the archaic character of these species is stressed by the type of its polyploidy: the polyploids with x = 11 are all palaeopolyploids without exception (FAVARGER1975: 5, EHRENDORFER 1980: 58); as will be argued later all of them have a relictic and therefore more archaic distribution (Fig. 5) and all of them are hexaploids. A noticeable difference in the proportion of polyploids between one region and another also exists in the species with x = 10: in the W. Mediterranean, 47% are polyploids; in E. Mediterranean, this p r o p o r t i o n is only 25% (but polyploids with x = 10 are always neopolyploids in both regions!). We can conclude that in the SW. Mediterranean, polyploid contributed more to diversification of sect. Acrocentron than in the E. Mediterranean. Biogeographic arguments. The change from x = 11 to x = 10 is also supported by the geographical distribution of these basic numbers. The hypothesis we shall start f r o m is that the genus Centaurea is of eastern origin. F r o m the revisions of BENTHAM (1873), HOFFMANN (1894) and more recently DrrTRICH (1977), it iS clear

12

N. QARCIA JACAS • A. SUSANNADE LA SERNA:

that the subtribe Cardueae- Centaureinae is more Asiatic than European in origin: of the 27 genera that DI3:TRICH (1977:1011) accepts in his revision of the subtribe, 14 are exclusive to Central or southern Asia and almost all of the rest are at least Asiatic Mediterranean, an only one genus-Stephanochilus Coss. & DuR. from N o r t h A f r i c a - i s restricted to the western Mediterranean. As far as sect. Acrocentron is concerned, the origin of the section if mostly oriental, too: of the 98 taxa recognized by WAGENITZ (1975) and DOSTAI. (1976), 55 have an E. Mediterranean distribution area; 3 are c o m m o n to the Italic Peninsula and the Balkans, 3 taxa are endemic to the Central Mediterranean region (Italic Peninsula and Sicily) and 21 are endemic of the SW. Mediterranean region. The remaining 16 taxa are dispersed in Central Europe, but seven of them are mostly oriental in its distribution and only one (C. scabiosa) is to be considered panEuropean. Moreover, WAGENIa'Z (1975) states that many species of sect. Acrocentron from the Flora of Turkey are Irano-Turanian elements, but he did not find any W. Mediterranean element. It seems clear that the expansion of Acrocentron has followed an east-to-west route and hence the E. Mediterranean centre of speciation must be considered more archaic. We re-examined the geographical distribution of the basic numbers x = 10 and 11 in the two main centres of endemization of sect. Acrocentron: the E. and the SW. Mediterranean regions. E. M e d i t e r r a n e a n r e g i o n . Fig. 3 shows the species with basic number x = 11 and Fig. 4 with x-- 10. The distribution areas are presented according to HAY~K (1928 - 1931), WAGENITZ (1975), and above all WAGENITZ & GAMAL-ELDIN (1985). As we do not have sufficiently detailed maps of the region, it was impossible to locate all the places exactly; unfortunately herbarium labels are often prepared without taking into account foreign investigators who do not have easy access to local geographic information. We have not been able to show the distribution area of C. macedonica because it was impossible to find the places mentioned by WAGENITZ & GAMAL-ELDIN (1985) on our maps; neither is the Bulgarian distribution area of C. immanuelis-loewii shown. However, given the aims of the figures, we think they are accurate enough. SW. M e d i t e r r a n e a n r e g i o n . Fig. 5 shows the distribution areas of species with x = 11 and Fig. 6 with x = 10. We have been able to be more precise with these figures. The distribution areas of C. borjae, C. carolipauana, C. clementei, C. crocata, C. granatensis, C. lainzii, C. prolongi, and C. ultreiae are as exact as the graphic representation used allows: they have been drawn according to the materials and the bibliography of the majority of European herbaria (see GARCIA JACAS & SUSANNA 1989, 1990, for the distribution areas of C. elementei and C. prolongi). The distribution areas of the remaining species are partly based on the usual floras, partly on the evidence of the BC, COI, and LISU herbaria, and partly on the personal observations of the authors. However, we want simply to give an approximate idea of the corresponding distribution areas. One thing must be made clear concerning the distribution area of C. ornata s.1. Over the last few years, the concept of the WILLDENOW species has changed drastically; nowadays it is more correct to talk of a complex of five different species: C. ornata s. str., C. saxicola, C. legionis-septimae, C. gabrielis-blaneae, and C. gal-

Karyological notes on Centaurea q ' . v."' . . v

.'.'.."

."i.

. . . . . . .

...'.'.'."

.'.v

13

. ' . . v . " ." . ' . ' . " .",~ c, .'." . v . " . ' . v

t. . . . . . . . . . . . . . . . .

".a'.

.'.'.'.'.'.'.'.'.'.'.'.'.'.

..........-.~.......

-....

-..

~ . . c.

.'.'2.

:._.....-....

." . ' . ' . . ' . ' . ' . " ." . v . . ' . ' .

scabxosa

.....

." ." . v

/

. ' . " ." . ' . ' . ' . " . ," . ' ," . ' . ' "

".'.,-,-2.\ .......

-2C-.,:'-:,:~.-..

~-

~

-,--,

. . . . . ..'

./

.:.:.:.:.:.;::.:+:.:+:.:+:.:..

~i.~,:~.:,-.:.:.:.:.:,:.:+:,:.:o:•:.:,

," /

°.'.'. .........

%',"

C,

immanuelis-Ioew~i

Western limit C. urvillei

i

~i;;~,~. !!iiiiiiiii?i?iii?i?i?i?i!i!i o ~

0

of

,/•

:C

%~

i

I

~ : c-'x,~', ("'x i

Q~

":~i!!~L... '::'"" ~,* :~iii~

".',

C.

L.

lactucifolia

' •

J 100

km

Fig. 4. Distribution areas of the species of Centaurea sect. Acrocentron with basic chromosome number x = 10 in the E. Mediterranean region ianoi. The last three have been described very recently and their distribution areas are not known well enough; for this reason they have not been separated on the figure. However, in relation to C. ornata of Central Spain, they all occupy quite large and peripheral areas in the Iberian Peninsula: C. galianoi is situated in the southwestern region (mountain ranges of Huelva and the South of Portugal); C. gabrielis-blancae occupies the eastern zone of the distribution area of the group (eastern Andalusia, the Levante, and Aragdn), while C. legionis-septimae replaces C. ornata in the northern region of the Peninsula• The only species of the complex that is very localized is C: saxicola, which grows in the extreme SE. of Spain (the coastal regions of Alicante and Murcia). Finally, so as not to make them almost illegible, we have not shown the distribution areas of the species of sect. Charnaecyanus on the figures of the SW. Mediterranean region• We shall discuss them later on. Comparison of the figures shows that: (1) The distribution areas of the species with x = 11 in the E. Mediterranean region (Fig. 3) are generally smaller, sometimes disjunctive and comparatively more

14

N. GARCIA JACAS & A. SUSANNADE LA SERNA:

t ~

~..~.

°

(-

C. carolipauana i t r

I00 km I

Fig. 5. Distribution areas of the species of Centaurea sect. Acrocentron with basic chromosome number x = 11 in the SW. Mediterranean region (? = estimated distribution)

relictic than those with x = 10 (Fig. 4). This is not a hard and fast rule, as there are exceptions. However, although there are some species with x = 10 with very small areas, none of the species with x = 11 has a really wide distribution. Though smaller areas are not a priori archaic, we must indicate that the species with x = 11 in E. Mediterranean are very frequently part of the chasmophytic flora or, if not, they are part of the mountain flora. The relictic character of the chasmophytes in E. Mediterranean has been noted by RUNEMARK (1969:111) and SNO~ERUP (1971 166). Further, EHRENI)ORFER (1971 : 212) points out that alpine places, cliffs and mesic forests are habitats in which relic populations have been preferably preserved. In the rest of the Mediterranean, we can see that none of the species of sect. Acrocentron with x = 11 has escaped this trend; the most extensively distributed species all have the basic number x-- 10: C. acaulis, C. eollina, the C. ornata and C. scabiosa complexes (Fig. 6), C. nana (not illustrated, but widely distributed from the Middle Atlas to Algeria), and the C. salonitana and C. urvillei complexes (Fig. 4). It is worth emphasizing that they are all polyploids or, if not, polyploid races are known amongst them. This may be interpreted as proof for the greater ecological plasticity and colonizing capacity that is usually associated with recent polyploidy (cf. STEBBINS 1950: 348--349, 1971: 183; FAVARGER 1975: 5; EHRENDORFER 1980:

Karyological notes on Centaurea

15

. .....:"/5D..~:."i.:i::.,L. " • " . '

•

• - :~.t¢ ::

" '.."/!!"..'D'~'~.? '~"i~-ikT';'~-:!~'~";:'?: