Karyological analysis of the five native Macaronesian Festuca ( Gramineae ) grasses supports a distinct diploid origin of two schizoendemic groups by

1 Dep. Biologia/CEM, Universidade da Madeira, Alto da Penteada, 9000-390 Funchal, Portugal. sequeira@uma.pt 2 Departamento de Agricultura (Botánica), Escuela Politécnica Superior de Huesca, Universidad de Zaragoza, Ctra. Cuarte km 1, E-22071 Huesca, Spain 3 Unidad de Botánica Aplicada, Instituto Canario de Investigaciones Agrarias (ICIA), c/ Retama 2, E-38400 Puerto de la Cruz, Spain Corresponding Author: sequeira@uma.pt Anales del Jardín Botánico de Madrid Vol. 66(1): 55-63 enero-junio 2009 ISSN: 0211-1322 doi: 10.3989/ajbm.2196

Different studies have demonstrated that a high percentage of the Macaronesian endemic flora has a continental Mediterranean origin (Emerson, 2002;Comes, 2004;Carine & al., 2004), and that the potential ancestors were probably located in SW Europe and NW Africa (Carine & al., 2004;Carine, 2005).The existence of large endemic plant groups has been interpreted as the result of either single continental colonization followed by successful adaptive radiation or multiple colonizations followed by in-situ speciation (reviewed in Silvertown, 2004;and Carine & al., 2004).Conversely, the presence of reduced endemic plant groups has been attributed to recent long distance dispersals followed by restricted insular speciation (Díaz-Pérez & al., 2008).However, the ages of the Macaronesian plant lineages are unknown in most cases, and their relative ancestry has been based on karyological and nuclear DNA content assessments (Löve & Löve, 1975;Suda & al., 2003) and on phylogenetic analyses (Francisco-Ortega & al., 1996;Silvertown, 2004;Carine & al., 2004).
Karyological data are essential to decipher the potential ancestry of species and populations and to support evolutionary hypotheses in angiosperms (Stebbins, 1971).Festuca shows a single chromosome base number of x = 7 but a large variety of ploidy levels ranging from diploids to duodecaploids (Ainscough & al., 1986;Dubcovsky & Martínez, 1992;Catalán & al., 2004).Chromosomal analysis conducted by Dubcovsky & Martínez (1992) found a world biogeographical distribution of Festuca ploidy levels, with almost all relict diploid taxa restricted to the Mediterranean-Eurasian region, whereas polyploid taxa were the only ones present in the southern hemisphere but also extended elsewhere.Dawe (1989), Catalán (2006) and Šmarda & al. (2008) further reported an evolutionary signal of karyotype patterns, ploidy levels and genome sizes within Festuca, with diploid races resolved basally in most Loliinae clades (Catalán, 2006).
Because chromosome data are lacking or have been barely investigated in most of the endemic Macaronesian Festuca we have conducted a chromosomal analysis of the five native species of this genus, aiming to investigate their ploidy and karyotype patterns, and their evolutionary implications.

Material and methods
Root tips obtained from planted individuals were used for the chromosomal analysis.Chromosome counts were done in mitotic cells of Festuca agustinii (5 populations from Tenerife island, 5 individuals), F. jubata (4 populations from Madeira island, 10 individuals), F. francoi (5 populations from S. Miguel, S. Jorge, Pico and Flores islands, 8 individuals), F. petraea (7 populations from St. Maria, Graciosa, S. Jorge, Pico, Faial and Flores islands, 8 individuals), and F. donax (2 populations from Madeira island, 2 individuals) (see Results and Appendix 1).Chromosome counting was carried out following the procedures of Menezes de Sequeira ( 2004) and Menezes de Sequeira & Castroviejo ( 2007) standardized for Hol- cus L. Root tips (0.5-1.5 cm) were cut and kept in distilled water at 0-4 ºC for 24-48 hours.Then they were fixed in Carnoy's solution (3:1 ethanol:acetic acid) overnight and transferred to a staining hydrochloric aceto-carmine solution (saturated) 48 hours before counting.Semi-permanent slides were prepared after a gentle squashing of the root tips in a drop of 45% acetic acid.
Chromosome observations were made using a Nikon Eclipse 90i binocular microscope and micrographs were taken using a digital camera (Applied Imaging, Progressive Scann), and treated using the program Cytovision 3.9 (Applied Imaging).Between 4 and 10 chromosome counts per plant were made, totalling 230 metaphase plates studied (50 of F. agustinii, 40 of F. jubata, 50 of F. francoi, 70 of F. petraea, and 20 of F. donax).Chromosome sizes were estimated using the biometric application AxioVision Rel.4.6 (Carl Zeiss Vision) on the captured images.The best 2-3 metaphase plates from each studied species were used to construct the respective karyograms.Chromosomes were classified according to their size and shape related to the centromere position following Stebbins (1938) and Levan & al. (1965), respectively.Karyotype asymmetry was calculated following the asymmetry index of Stebbins (1971) and the intrachromosomal asymmetry index A1 and interchromosomal asymmetry index A2 of Romero Zarco (1986).The best karyograms were used to construct the ideograms from the mean length of the long and short arms between homologous chromosomes.

Results
Our chromosome analysis indicates that all Macaronesian fine-leaved and broad-leaved native Festuca are diploid, showing 2n = 14 chromosomes (Table 1; Fig. 2), except for one individual of F. francoi that showed 2n = 28 chromosomes (Table 1).Chromosome numbers are reported for the first time for the Canarian F. agustinii and the Azorean F. francoi.Our extensive survey also confirms previous diploid counts reported for the Madeiran F. jubata (Dalgaard, 1986) and F. donax (Malik & Thomas, 1966) and for the Azorean F. petraea (Devesa & Gómez, 1988).
Karyotypes also differ between the two Festuca groups.They tend to be more symmetrical in broadleaved F. donax and more asymmetrical in the four fine-leaved taxa.F. agustinii shows an ideogram formula of 10 m + 4 sm, an asymmetry of type 2A and asymmetry indices A1 and A2 of 0.28 and 0.17, respectively (Fig. 2a).F. jubata has an ideogram formula of 10 m + 4 sm, asymmetry of type 1A and asymmetry indices A1 and A2 of 0.30 and 0.17 (Fig. 2b).F. petraea shows an ideogram of 10 m + 4 sm, asymmetry of type 2A and asymmetry indices A1 and A2 of 0.29 and 0.18 (Fig. 2c).F. francoi has an ideogram formula of 10 m + 4 sm, asymmetry of type 2A and asymmetry indices A1 and A2 of 0.32 and 0.17 (Fig. 2d), and F. donax shows an ideogram formula of 4 M + 10 m, asymmetry of Table1.Karyological data of Macaronesian Festuca.Asymmetry index of Stebbins (1975) and coefficients A1 and A2 of Romero Zarco (1986).

Taxon
No  type 1A and asymmetry indices A1 and A2 of 0.13 and 0.15 (Fig. 2e).No differences have been found in A2 indices between fine and broad-leaved groups, indicating that similar interchromosomal asymmetry relative to the respective sizes of the chromosomes exists in both cases.Contrastingly, the intrachromosomal asymmetry, which is related to the branches' length and the Stebbins' type of asymmetry, clearly separates both groups, indicating that the broad-leaved F. donax has more metacentric chromosomes than the fineleaved group.

Discussion
Our karyological analysis conducted in individuals from different islands and populations of the five native Macaronesian Festuca indicate that these endemic taxa are mostly diploid.Diploid individuals have been found in all Madeiran F. jubata populations examined, including the low altitudinal population from Curral das Freiras, which showed a genetic AFLP profile distinct from that of the remaining populations from the highest peaks (Díaz-Pérez & al., 2008).Diploidy is a common phenomenon within the Macaronesian flora (Suda & al., 2003) and a genomic trait related to ancestry within Festuca (Dawe, 1989;Dubcovsky & Martínez, 1992).Evolutionary studies have demonstrated that the more ancestral broad-leaved Loliinae lineages are totally or predominantly diploids, whereas the more advanced fine-leaved lina b c d e eages tend to show some basal and subbasal diploid elements that are usually outnumbered by the more diverse and recently evolved polyploid taxa (Catalán, 2006;Inda & al., 2008).The diploidy observed in the endemic Macaronesian Festuca agree with their respective phylogenetic placements.However, whereas the robust F. donax is nested within a relict diploid clade of paraphyletic broad-leaved F. subg.Drymanthele sect.Phaeocloa taxa, the slender diploid F. agustinii, F. jubata, F. francoi and F. petraea, together with the single continental diploid relative F. rivularis, are placed in an unresolved basal position within the otherwise recently evolved clade of high-polyploid fineleaved F. subg.Festuca sect.Aulaxyper taxa (Catalán, 2006;Inda & al., 2008).The karyotypical differences observed between F. donax and the group of slender Macaronesian Festuca also suggest a distinct origin.The usually large sized chromosomes of F. donax (7.18 ± 1.09 µm) correspond to the mean values observed within F. subg.Drymanthele sect.Phaeocloa (8.66 µm), whereas the smaller sized chromosomes of F. agustinii, F. jubata, F. francoi and F. petraea (2.7 ± 0.47 µm, 2.66 ± 0.46 µm, 2.84 ± 0.47 µm and 3.47 ± 0.61 µm, respectively) fall within the values reported for different sectional representatives of fine-leaved Festuca (3.45-3.47µm in F. sects.Eskia, Festuca and Aulaxyper; Dawe, 1989).The 2.5-fold range decrease in chromosome size from primitive broad-leaved Festuca subg.Drymanthele taxa to the more advanced fine-leaved Festuca subg.Festuca taxa reported by Dawe (1989) has been confirmed in the studied Macaronesian taxa.
Differences in genome size and karyotype evolution in the native Macaronesian fescues are therefore concordant with hypotheses recovered from molecular data by Catalán (2006) and Inda & al. (2008).These authors indicate that these two groups of schizoendemics, i.e. relict endemic diploid taxa with similar karyotype profiles (cf.Favarger & Contandriopoulos, 1961;Contandriopoulos, 1988), originated independently and probably at different times.This has been further reassessed by divergence time calculations of those lineages estimated from Bayesian relaxed-clock phylogenies of Loliinae (Inda & al., 2008).These authors dated the divergence event that led to F. donax from its common ancestor with F. drymeja c. 3.6 ± 1.6 Mya, whereas the radiation of F. agustinii, F. jubata, F. francoi and F. petraea from the ancestor of the F. rubra group took part approximately 2.5 ± 0.9 Mya.
The older origin of F. donax would probably be the result of an earlier long distance dispersal event of broad-leaved F. subg.Drymanthele founders from Europe to Madeira (Inda & al., 2008).The nemoral ecology of F. donax seems to agree with an early colonization of Madeira, possibly with other paleoendemic plants such as the Lauraceae.In fact, F. donax is one of the few grasses occurring in the deep shaded laurel forest of this volcanic island (Capelo & al., 2004), which arose ca.5.3 My ago (Gelmacher & al., 2000).By contrast, the recent origin of F. agustinii, F. jubata, F. francoi and F. petraea likely resulted from more recent colonizations of the Macaronesian archipelagos from near western Mediterranean diploid F. sect.Aulaxyper ancestors.In their molecular phylogeographical study, Díaz-Pérez & al. (2008) suggested the existence of three relatively recent long distance continental colonizations to the Canaries, Madeira and Azores archipelagos that resulted in F. agustinii, F. ju-bata, and the sister taxa F. francoi and F. petraea, respectively.The ecological preferences for open high altitudinal habitats shown by F. agustinii, F. jubata and F. francoi (Saint-Yves, 1922;Díaz-Pérez & al., 2008) correlate well with a more recent establishment of these grasses in the high mountain cliffs of these archipelagos, whereas the recent derived origin of F. petraea from F. francoi (1.1 ± 0.6 Mya; cf.Inda & al., 2008) has been interpreted as the results of ecological in-situ adaptation to seashore ecosystems (Díaz-Pérez & al., 2008).
Despite the predominant diploid state found in all native Macaronesian Festuca, one tetraploid count of 2n = 28 chromosomes was observed in one individual of the Azorean Pico island population of F. francoi.The existence of polyploid endemic Macaronesian grasses is a rather common phenomenon, as exemplarized by the cases found in the genera Dactylis (e.g.tetraploid Madeiran D. smithii Link subsp.hylopdes Parker (Parker, 1972), Holcus (e.g.tetraploid Azorean H. rigidus Hochst.(Menezes de Sequeira & Castroviejo, 2007) and Canarian Holcus mollis L. subsp.hierrensis Stierst.(Stierstorfer, 2001) (Quintanar & al., 2006).These polyploid Macaronesian endemics would correspond to neoendemics (or apo-endemics) in the sense of Favarger & Contandriopoulos (1961) and Contandriopoulos (1988), and probably derived very recently from polyploid mainland colonizers.Nonetheless, the single case of tetraploidy detected in the otherwise largely diploid Festuca francoi and the long geographical distances that separate the Azores archipelago from any other land mass precludes any hypothesis of secondary colonization.It rather suggests an in-situ origin of autotetraploid races from older and previously established diploid races in the Pico island.This potential euploidy should be confirmed, however, through a more extensive cytological survey of populations of this Azorean fescue.

Fig. 1 .
Fig. 1.Map of geographical distribution and illustrations of habits and ecological niches of the five native Macaronesian Festuca.Black and grey backgrounds indicate the distribution of the species in the islands of each archipelago.F. agustinii: western Canary islands; F. donax: Madeira island; F. jubata: Madeira and Porto Santo islands; F. petraea: Azores islands; F. francoi: all Azores islands except Graciosa.