LIMONIUM MARISOLII L. LLORENS (PLUMBAGINACEAE) REVISITED*

SÁEZ, L., Á.C. CARVALHO & J.A. ROSSELLÓ (1998). Limonium marisolii L. Llorens (Plumbaginaceae) revisited. Anales Jard. Bot. Madrid 56(1): 33-41. A new triploid agamic species, L. leonardi-llorensii, is described from coastal populations of South-West Mallorca. The new species is closely related, on morphological grounds, to the endemic L. marisolii L. Llorens, from which it could be distinguished by several morphological characters (leaves without papillae, longer calyx, more flowers per spikelet) and by a different chromosome number (2n = 26, L. leonardi-llorensii; 2n = 27, L. marisolii). The relationships between both species are discussed, and it is suggested that both taxa might share some common evolutionary history despite their divergent chromosome number. The origin of L. marisolii from L. leonardi-llorensii through chromosome fission or by a hybridization event is considered.

Limonium marisolii L. Llorens is a triploid taxon endemic to the Baleario islands (LLORENS, 1986a;ERBEN, 1993), belonging in section Limonium subsect.Dissitiflorae Boiss.This species has a narrow discontinuous distribution along southern Mallorca, where it grows on the sandy slopes and on calcareous rock crevices near the seashore.The eastern populations (from which the plant was described) have more individuáis than the western ones, which are very rare and contain scanty specimens as a consequence of habitat disturbances caused by touristic development.Limonium marisolii has been assigned to the L. gibertii complex (SÁEZ & ROSSELLÓ, 1996) on the basis of overall morphology, but this placement is currently tested by other cytological and molecular evidence (SÁEZ & al., unpublished data).In the course of this revisión, attention was paid to the morphological differences encountered between the western and eastern populations of L. marisolii, which were supported by different chromosome numbers and cuticular micromorphology.Evidence is presented here that supports taxonomic recognition for the western populations which are hereby described as L. leonardi-llorensii.

MATERIAL AND METHODS
Seeds, living plants and herbarium specimens were collected from all known populations, including the type locality of L. marisolii.Additional specimens were borrowed from the following herbaria: BCC, G.MAandMAF.

Breeding system and pollen fertility.
Flowers were removed from herbarium specimens and the stigma and pollen grains were stained according to the ALEXANDER (1980) technique.

Phytodermology.
Dried leaves were rehydrated, decolorized and stained with Bismarck Brown, using standard techniques.Thirty stomatal guard cells from both leaf surfaces were measured for each accession.
Karyology.Seeds were germinated in Petri dishes on moistened filter paper.Root-tips were pretreated for about 4h with 0,2% colchicine, fixed in ethanohglacial acetic acid (3 : 1) at 4 °C for 24h, hydrolysed in HC1 IN for 3 minutes at 60 °C, and stained with acetic orcein overnight.Root tip squashes were made in 45% acetic acid.Photographs of metaphase plates were taken at a final magnification of x 2500.In order to make the idiograms comparable, the length of the short and long arms of the chromosomes in each taxon was expressed in relative valúes (chromosome set = 100 %).The description of the chromosome sets follow the nomenclature of LEVAN & al. (1964).
Derivatio nominis: Named after L. Llorens, who was the first to find a population of the new species and included it under L. marisolii.

KARYOLOGY
Metaphase plates from individuáis belonging to eastern populations consistently yield a chromosome count of 2n = 27 (fig.3a), with a chromosome formula of 4M + 15m + 8sm.One aneusomatic cell with an aneuploid karyotype (2n = 26) was observed in one individual (# 4) from one of the two examined accessions.In contrast, the two western populations had a 2n = 26 complement (fig.3b), with a chromosome formula of 7M + lOm + 8sm + lst.Also, one cell with a deviating chromosome number (2n = 27) was encountered in a 2« = 26 individual (#5).These chromosome numbers are in accordance with a polyploid (triploid) level for   overall morphology, which has favoured the confusión of both taxa under a single entity.However, conspicuous differences in morphology (table 3), cuticular ornamentation pattern, stomatal guard cells length and chromosome number support the view that the western and eastern allopatric populations of L. marisolii deserve taxonomic recognition at the specific level.The divergent chromosome number and the presence, in one taxon, of a long metacentric marker chromosome, strongly support the idea that both taxa are not so closely related as their macromorphology could suggest.In fact, according to ERBEN'S (1978,1979) hypothesis about the origin of polyploid taxa in the genus, L. marisolii and L. leonardillorensii should not share a common evolutionary pathway.If Erben's hypothesis is correct, then L. leonardi-llorensii, or its ancestor, originated through a cross between a reduced gamete of a diploid taxa of basic chromosome number x = 8 with an unreduced gamete of a species having 2n = 18 chromosomes (thus with basic basic number x = 9).According to the same hypothesis, L. marisolii should have arisen in a similar way, by a combination of gametes of plants of different ploidy level belonging solely to taxa with x = 9.Erben's explanation of the formation of the polyploid agamic taxa assumes that the long metacentric marker chromosomes are homologous and, therefore, their presence in the triploid and tetraploid taxa comes from diploid ancestors belonging to x = 8 chromosome lineages.However, no 1.4-1.9x1.5-1.9 3.9-4.4x2.9-3.1 (0.7)0.8-0.9(1) 8-16 1-3 3.8-4.40.3-0.6 x 0.6-0.9 Reaching or depassing teeth basis  Karyological rearrangements and chromosomal losses in unstabilized zygotes could account for the low chromosome number found in L. marisolii.Genomic in situ hybridization (GISH) is currently underway, to further explore both compelling hypotheses about the genomic relationships and evolution between L. leonardi-llorensii and L. marisolii.

TABLE 2 PHYTODERMOLOGICAL
FEATURES OF LIMONIUM LEONARDI-LLORENSII AND L. MARISOLII DISCUSSIONLimonium marisolii and the new proposed species, L. leonardi-llorensii, share a similar Carvalho, unpublished data).This high allelic similarity would not be present in two taxa of such presumed different evolutionary pathways, as inferred from ERBEN'S (1978) data, at least if one considers the modérate level of isozyme differentiation found within and between the taxa of the L. gibertii complex (Carvalho, unpublished data).On the other hand, RFLPs of noncoding chloroplast DNA (trnC-trnD) show that L. leonardi-llorensii and L. marisolii have different haplotypes, a noteworthy feature if both taxa have an ancestordescendent relationship.These conflicting molecular data could be reconciled assuming an hybrid origin of L. marisolii through a cross between L. leonardi-llorensii and another still unidentified taxa, which should be the ovule donor (chloroplast DNA transmission is assumed to be maternal in Limonium;HARRIS & INGRAM,1991).