Populations used in this study were sampled from March to May 2010–2012 in various ecogeographic areas of Northern Algeria (Table 1). In each site, 5–10 bulbs were collected and cultivated in the experimental garden of Houari Boumediene University of Sciences and Technology (Algiers). Taxonomic determinations were made based on several North-Africa and Mediterranean Floras: Desfontaines (1798–1799), Battandier and Trabut (1895–1902), Maire (1958), Quézel and Santa (1962) and Davis and Stuart (1980). The specialized taxonomic and nomenclatural websites, the African Plant Database (Dobignard and Chatelain 2010–2013) and the World Check List of Selected Plants (Govaerts 2015) were also consulted.

Mitotic preparations were performed on young root-tips obtained from potted plants. The chromosome observations were performed using the standard Feulgen technique for staining tissues (Jahier et al. 1992), with little modifications. Root-tips were pretreated in 8-hydroxyquinoline (0.002%) or in 0.25 % aqueous colchicine for 5 hours at room temperature, then fixed in Carnoy fixative solution (3 : 1 (v/v) ethanol : acetic acid) at 4°C for at least 48 hours. Hydrolysis was made in 1N HCl for 7–9 min at 60°C before staining with the usual Schiff reagent. Root-tips were squashed in a drop of 45% acetic acid. The observations were made using a Carl Zeiss Axiostar-Plus microscope equipped with a Canon digital camera.

Measurements for karyotype and idiogram constructions were based on at least five well-spread metaphase plates of different individuals. The arrangement of homologous pairs was made using MICROMEASURE Software version 3.3 (Reeves 2001). Chromosomes are described according to the nomenclature of Levan et al. (1964) based on the chromosomal arm ratio (r = long arm/short arm) and the centromeric index (CI % = short arm/long arm + short arm × 100): metacentric (m), submetacentric (sm), subtelocentric (st) and telocentric (t). Karyotype asymmetry indices were estimated following the proposal of Peruzzi and Eroğlu (2013). The intrachromosomal asymmetry index is represented by the mean centromeric asymmetry M_CA = A × 100, where A is the average ratio of long arm-short arm/long arm + short arm, according to Watanabe et al. (1999). The interchromosomal asymmetry index is the coefficient of variation of chromosome length CV_CL = A₂ × 100 (Paszko 2006) where A₂ is the standard deviation of chromosome length/mean chromosome length (Romero Zarco 1986). The coefficient of variation of the centromeric index CV_CI = S_CI/X_CI × 100 is the ratio between the standard deviation S_CI and the mean centromeric index X_CI (Paszko 2006).

In order to compare the karyotypes of the studied species, a Principal Component Analysis (PCA) was performed using STATISTICA Software version 6. Analysis was based on six fundamental karyological parameters as proposed by Peruzzi and Altinordu (2014): chromosome number (2n), chromosome base number (x), total haploid chromosome length (THL), mean centromeric asymmetry M_CA, coefficient of variation of chromosome length CV_CL and coefficient of variation of the centromeric index CV_CI.

Mitotic observations showed tetraploid and octaploid cytotypes with base number x = 4. The tetraploid cytotypes 2n = 4x = 16 (Fig. 1A–B; 2A–B) was found in two populations from two contrasted biogeographical sectors. Plants from biogeographical sector of Constantine (Tiddis) grow on clayey-marly soil. Their chromosomes show a total haploid length THL = 93.05 μm with mean length per chromosome (CLR) ranging from 7.00 to 17.1 µm (Table 3). The karyotype consists of 4m + 4st + 8sm. Specimens from the biogeographical sector of Oran (Stidia) occurring on coastal sand dunes are distinguished by much larger chromosomes. The mean length per chromosome (CLR) is 10.05-20.47 µm and THL = 113.86 µm (Table 3) with a karyotype formula 4m-sat + 4st + 8sm. This karyotype is distinguished by two terminal satellites on the first largest metacentric pair (Fig. 2B). Except the occurrence of the satellites, the structure of the two karyotypes is similar regarding the centromeric asymmetry values and the coefficient of variation (Table 3).

The octoploid cytotype 2n = 8x = 32 (Fig. 1C, 2C) was found in two populations from Ouled Fayet and AinTorki of the biogeographical sector of Algiers. This cytotype is characterized by a larger THL 171.40 μm and CLR values more extensive (6.05-18.05 μm). The karyotype formula is quite similar to that of the tetraploids. One submetacentric supernumerary chromosome was occasionally observed in octoploid individuals (Fig. 1C). The centromeric asymmetry indices (M_CA) of tetraploid and octaploid cytotype are rather different while the coefficients of variation (CV_CL) are much closer.

This species is widespread in the north of Algeria. Examined populations were diploids with 2n = 18 chromosomes and a base number x = 9 (Fig. 1E). The mean length per chromosome is comprised between 1.94 µm to 10.49 µm and total length THL = 33.51 µm (Table 3). The karyotype is distinguished by two large pairs of chromosomes and seven other pairs much smaller. The first pair is telocentric; the second pair constituted by one metacentric and one submetacentric chromosome is polymorph due to structural heterozygosity (Figs 1E, 2D). All the remaining small chromosomes are metacentric. The karyotype formula is 2t + (1m + 1sm) + 14m. The values of the centromeric asymmetry (M_CA) and the coefficient of variation (CV_CL) are 73.8 and 19.97 respectively.

M. maritimum is less common. The studied population lives on the sand dunes in the steppe high plains of the Saharan border (Guelt es Stel). It is also diploid with 2n = 18 (Fig. 1F). The mean length of chromosomes is between 2.37 µm and 9.38 µm with a THL = 47.64 µm (Table 3). The karyotype is characterized by 6st-sat + 6sm-sat + 6m (Fig. 2) showing two satellites: terminal on the first subtelocentric pair, and intercalary on the fifth submetacentric pair. Compared to M. comosum, M. maritimum have a less asymmetrical karyotype reflected in a low value of its centromeric asymmetry index (M_CA).

In this species, three cytotypes were observed: pentaploid 2n = 5x = 45, hexaploid 2n = 6x = 54 and octaploid 2n = 8x = 72 (Figs 1G–I). All cytotypes were encountered in the eastern biogeographical sector of Constantine (Ain El Bey, Tiddis and Djemila). The western population of Mansourah (Tlemcen) is pentaploid (Tables 1–2). Compared to the previous species, chromosomes are markedly small with mean lengths between 1.80 µm and 5.39 µm and no significant difference among the three karyotypes (Table 3). This species is characterized by a rather symmetrical karyotype comprising only metacentric chromosomes (Figs 1–3, Table 3) with total length depending on the ploidy level. The centromeric asymmetry indices (M_CA) and the coefficients of variation are also similar.

Figure 3.

Principal Component Analysis of the eight cytotypes of Bellevalia and Muscari. A Correlation loadings of the six karyotype variables with PC1 and PC2 (abbreviations in Table 3) B Corresponding projection of the eight cytotypes: Bm Bellevalia mauritanica (Tiddis), B. mauritanica (Stidia), Mc Muscari comosum, Mm M. maritimum, Mn M. neglectum.

In order to estimate the karyological relationship among the studied taxa, a principal component analysis (PCA) was carried out on the 8 populations, each representing different species and/or cytotypes (Fig. 3). The pattern of correlation loadings of the variables (Fig. 3A) highlights the major role of PC1 and PC2. Cumulative variance explained by these two first components approaches to 85% of the total information. The formation of PC1 was due to intrachromosomal asymmetry parameter M_CA, the coefficient of variation of the centromeric index CV_CI (negative values) and to chromosome numbers 2n and x (positive values) which have a discriminant power > 0.79 (data not shown). PC2 is well described by the variables THL and CV_CL (inversely correlated) and, in least degree, once again, by the chromosomal numbers 2n and x (Fig. 3A). As expected, the projection of the taxa on the first two axes confirms the divergence between the cytotypes representing the genus Muscari from those of genus Bellevalia (Fig. 3B).

Cytotypes of Bellevalia mauritanica constitute a clearly distinct group, in which the two tetraploid cytotypes (from Stidia and Tiddis) shows close relationship. The octoploid cytotype (2n = 8x = 32) can be discreetly distinguished probably because of a higher value of the total haploid length (THL).

The karyotypes of the studied species of Muscari constitute two other clusters significantly different from each other (Fig. 3B): the first cluster is limited to the positive values of PC1 and involves all the 5x, 6x and 8x cytotypes of M. neglectum; the second cluster, showing positive values of PC2, relates to diploid karyotypes of M. comosum and M. maritimum. This distribution matches the different affiliation of the species to the two subgenera Botryanthus and Leopoldia respectively. The diploid species belonging to subgenus Leopoldia e.g. M. maritimum [= Leopoldia maritima (Desfontaines, 1798) Parlatore, 1845] and M. comosum [= L. comosa (Linnaeus, 1753) Parlatore, 1847] are well separated due to different asymmetry chromosomal indices M_CA, CV_CI and CV_CL. Within the M. neglectum group the three ploidy levels did not show any significant differentiation.

The studied populations of Bellevalia mauritanica display two ploidy levels, tetraploid (2n = 4x = 16) and octoploid (2n = 8x = 32). This species was previously known as exclusively tetraploid besides twelve other species of the genus (Brullo et al. 2009, Bareka et al. 2012).

Usually, in the genus Bellevalia, the karyotypes show satellites on either the first, the second or the third pair of chromosomes (Bothmer and Wendelbo 1981, Bareka et al. 2008, 2012). Our tetraploid B. mauritanica from Stidia shows a similar chromosome arrangement and bears one pair of satellites on the first metacentric pair.

The octoploid level is reported here in B. mauritanica for the first time The polyploidy is quite abundant in Bellevalia, 2x, 3x, 4x, 6x and 8x levels have already been reported (Musano and Maggini 1976, Özhatay et al. 1991a, Özhatay and Johnson 1996, Johnson 2003, Yaylaci et al. 2009, Bareka et al. 2012, Karabacak et al. 2014). The octoploid level is rare and it seems known in two species only, B. longistyla (Miscz.) Grossheim, 1928 (Özhatay and Johnson 1996, Johnson 2003) and B. olivieri (Baker) Wendelbo, 1985 (Bareka et al. 2012, 2015). Singular populations with 2n = 4x = 32 have been also quoted for B. glauca (Lindley) Kunth, 1843 and B. sarmatica (Pallas ex Miscz.) Woronow, 1927 (Bothmer and Wendelbo 1981).

In some octoploid cytotypes of B. mauritanica, we observed one large and metacentric supernumerary chromosome, similar to all the other homologues. It seems to be a very interesting case of aneuploidy, which has not yet been reported, to our knowledge, in genus Bellevalia (P. Bareka pers. comm.). Only B chromosomes were sometimes observed in diploids such as B. saviczii Woronow, 1927 with 2n = 8 + 1B (Gettner, 2005) and B. koyuncui Karabacak & Yildirim, 2015 with 2n = 8 + 2B (Karabacak et al. 2015). The occurrence of aneuploidy in a polyploid context, associated with vegetative reproduction, may indicate chromosomal changes in process providing evolutionary potential, as presumed for B-chromosomes (Johnson 2003, Weiss-Schneeweiss and Schneeweiss 2013, Bareka et al. 2015). The absence of structural differentiation and the total length of the octoploid complement, nearly twice that of the tetraploid (171.4 versus 93.05 µm), argue for an autopolyploidy event. Bareka et al. (2012) already concluded that autopolyploidy was the principal mechanism of polyploidization among populations occurring in Greece belonging to B. edirnensis hexaploid, B. hyacinthoides triploid and B. ciliata tetraploid.

Karyological results on M. comosum and M. maritimum agree with previous findings on the subgenus Leopoldia in which species are mostly diploids (Ruiz Rejón et al. 1985, Nersesian 2001, Jafari and Maassoumi 2011, Jafari 2012a, 2012b).

All the examined specimens of M. comosum have 2n = 2x = 18 with a markedly asymmetric karyotype consisting of 2 pairs of large chromosomes and 7 pairs of small and metacentric chromosomes. Slight variations were observed in the first pair of chromosomes, sometimes viewed as telocentric (Ruiz Rejón et al. 1981, Kostovic-Vranjes 1999) or as subtelocentric (Cuñado et al. 2000, Jafari et al. 2008). Similarly, the second pair is polymorphic with submetacentric and subtelocentric chromosomes (Ruiz Rejón et al. 1985, 1990, Cuñado et al. 2000, Kostovic-Vranjes 1999).

Concerning M. maritimum, the chromosome number 2n = 18 was previously quoted by Garbari and Di Martino (1972) for specimens with unspecified origin. It is also quoted by Troìa et al. (2014) for one Tunisian population at the Cap Bon. However, in our knowledge, the karyotype structure of M. maritimum is reported here for the first time. It would be related to that of M. gussonei (Parlatore) Todaro, 1872, an endemic species to Sicily (Garbari and Di Martino 1972, Davis and Stuart 1980). The karyotype of this species consists of 10 large and 8 small chromosomes (Ruiz Rejón et al. 1985). However, karyotype of Algerian specimens collected in the Saharan border is distinguishable in having satellites located on the 1^st and the 5^th large chromosome pair.

Muscari neglectum belongs to the subgenus Botryanthus which contrasts considerably with the precedent by the occurrence of ploidy series of 2x, 3x, 4x, 5x and 6x levels (Davis and Stuart 1980, Karlén 1984, Ruiz Rejón et al. 1985, Garbari 2003, Suárez-Santiago et al. 2007). Previous chromosomal counts for M. neglectum indicate several numbers: 2n = 18, 36, 44, 54, 55, 63 and 72 (Karlén 1984). The three ploidy levels (5x, 6x, 8x) observed in Algeria confirm the extent of polyploidy in this complex. However, no diploids or tetraploids were detected in our country. In contrast, the presence of octaploid plants is significant because the 8x level was extremely rare and only few individuals having 2n = 72 were previously quoted in a population from the northern Greece (Karlén 1984). So far, only tetraploid, pentaploid and hexaploid populations of this taxon have been observed in the western Mediterranean area, precisely in the Iberian Peninsula (Ruiz Rejón et al. 1985, Suárez-Santiago et al. 2007). This is what justifies the statement generally accepted that the diploids occur only in Greece and Turkey (Karlén 1984, Garbari 2003).