We designed the sampling of this study aiming to include a broad geographical (Palearctic region) and taxonomic (three subgenera and 10 sections) representation of Helianthemum (see Fig. 1). Except for one species, all the seeds came from wild plants sampled in the field. We considered karyotype features to be constant at species level, so we included seeds from one or two populations per species. In every population we harvested ripe capsules from 5 to 15 different plants which were pooled in paper bags. Then, the capsules were carefully opened in the laboratory to extract the seeds. The accessions of clean seeds were kept in a dry and cool place until study. We additionally included seeds from four species stored in seed banks (Millennium Seed Bank and Israel Plant Gene Bank) (Appendix 1). In total, we karyotyped mitotic metaphase plates obtained from about 350 seeds, representing 85 populations and 78 species and subspecies.

In this study, we followed the taxonomic adscriptions and nomenclatural recommendations for the genus Helianthemum proposed by Martín-Hernanz & al. (2021a). Notice that we present the results for H. sect. Helianthemum (s.l.) separated into two groups: (1) H. sect. ‘Helianthemum Canarian clade’, which include all the 15 species of H. sect. Helianthemum endemic to the Canary Islands (see Table 1), and (2) H. sect. ‘Helianthemum p.p.’ for the rest of the species in H. sect. Helianthemum. This is because the species from the Canary Islands conform a cohesive monophyletic lineage within H. sect. Helianthemum that rapidly diversified during the Pleistocene in the archipelago with idiosyncratic genomic, morphological, biogeographical, and ecological features (Aparicio & al. 2017; Martín-Hernanz & al. 2019; Albaladejo & al. 2021; Martín-Hernanz & al. 2021b). Helianthemum dagestanicum Rupr. has been ascribed to H. sect. Pseudomacularia Grosser on the basis of target sequencing according to Martín-Hernanz & al. (unpublished data). The voucher specimens from the populations studied have been deposited in the SEV herbarium (University of Seville) (Appendix 1).

The protocol for the germination of seeds in Helianthemum was described in detail in Aparicio & al. (2019). Briefly, all the material required for seed germination (sandpaper, Petri dishes, distilled water, filter paper, scissors, etc.) was introduced in an UV-cleaner box for about 40 minutes. Then, seeds were gently scarified by abrasion between two sheets of fine-grained sandpaper (Pérez-García & González-Benito 2006) and set for germination in Petri dishes at 20ºC. Root tips were pre-treated by immersion in 2 mM 8-Hydroxyquinoline (Tjio & Levan 1950) for 4 h at 10ºC, fixed in 1:3 glacial acetic acid and absolute ethanol for at least 2.5 h at 10ºC, stained in alcoholic-hydrochloric acid-carmine for 24-48 h (Snow 1963) and then squashed in 45% acetic acid. For every species, clear metaphase spreads were photographed in an Olympus BX41 microscope equipped with a ColorView III digital camera.

Chromosome counts and karyotype analyses were carried out from mitotic metaphase photographs. We measured between 5-14 mitotic metaphase spreads for every species and used the tools provided by the software MATO (Measurement and Analysis Tools; Altinordu & al. 2016) to compute several karyological parameters. Aiming to assess karyotype length and karyotype heterogeneity, considering both the inter and intrachromosomal components of karyotype asymmetry (Peruzzi & Eroğlu 2013), we obtained five different parameters: (1) total haploid (monoploid) length of chromosome set (THL; Altinordu & al. 2016), (2) karyotype formula (Levan & al. 1964), (3) karyotype asymmetry classification of Stebbins (Stebbins 1971), (4) interchromosomal coefficient of variation of chromosome length (CV_CL; Paszko 2006), and (5) intrachromosomal mean centromeric asymmetry (M_CA; Peruzzi & Eroğlu 2013). To obtain mean THL at species level we discarded extreme values by merging in MATO only individual measurements which yielded a mean THL value with a coefficient of variation (CV) < 10%. Satellited chromosomes can be commonly observed in metaphase plates, but due to inconsistency among them we have not considered satellites in this study.

To test for significant relationships between chromosome number and the karyotype parameters, we converted the base chromosome number into a binary response variable (n = 10 to 0 and n = 11 to 1) and ran a phylogenetic logistic regression (Ives & Garland 2010) with phylolm R package (Ho & Ane 2014). We excluded n = 5 and n = 12 because these numbers are only present in one species each (see Introduction). We accounted for the shared ancestry of chromosome numbers using the TreePL phylogenetic tree explained in the following section.

Chromosome number evolution in Helianthemum was already reconstructed by Aparicio & al. (2019) based on a phylogenetic hypothesis derived from DNA Sanger sequences (Aparicio & Albaladejo 2017). To update this analysis, we used the time-calibrated phylogeny based on GBS (genotyping-by-sequencing) data obtained by the software TreePL (Martín-Hernanz & al. 2019). This phylogenetic tree was additionally modified by the inclusion of Sanger DNA sequences of H. dagestanicum (see methodology in Martín-Hernanz & al. 2021b) and the exclusion of H. ordosicum Y.Z.Zhao, Zong Y.Zhu & R.Cao until confirmation of its polyploid (2n = 4x = 40) status. The updated analysis was also enhanced by the inclusion of all the species of H. sect. Pseudomacularia, whose chromosome numbers were unknown until this study. The time-calibrated phylogeny was pruned to keep one single tip per species. The final data set consisted of 73 species.

The updated time-calibrated phylogeny and the chromosome numbers were analysed using ChromEvol v.2.0 (Glick & Mayrose 2014; Mayrose & al. 2010) to elucidate the mode of chromosome evolution. ChromEvol determines the probability of a certain model to explain the given data (haploid chromosome numbers) along a phylogeny, based on the combination of the first two or more of the following parameters: (i) gain or (ii) loss of a single chromosome, (iii) polyploidization, (iv) half increment of the chromosome number (demi-polyploidization) and (v) increment of the base number with regard of a rate of multiplication different from a regular duplication. Furthermore, two additional parameters permit to detect linear dependency between the current haploid number and the rate of (vi) gain and (vii) loss of chromosomes. Specifically, we performed the analyses using eight models of chromosome evolution implemented in ChromEvol that combine differently these parameters for chromosome number transitions: CONST_RATE, CONST_RATE_DEMI, CONST_RATE_DEMI_EST, CONST_RATE_NO_DUPL, LINEAR_RATE, LINEAR_RATE_DEMI, LINEAR_RATE_DEMI_EST and LINEAR_RATE_NO_DUPL.

Based on our own previous reconstruction of chromosome evolution (Aparicio & al. 2019) we run the analysis fixing the root of the phylogeny at a chromosome base number of n = 10. Models were compared using Akaike information criterion (AIC and ∆AIC), which allowed us to test the alternative hypotheses of chromosome evolution. The best model was plotted on the time-calibrated phylogeny using the ChromEvol functions v. 1 by N. Cusimano (https://www.en.sysbot.bio.lmu.de/people/employees/cusimano/use_r/) in R.

We obtained new chromosome counts for 11 species and subspecies, which are illustrated in Figure 2 (see also Table 1). Therefore, the number of species of Helianthemum whose chromosome number is known increases to c. 77%. We confirmed that the predominant somatic chromosome numbers are 2n = 20 and 22, while that 2n = 10 and 2n = 24 are restricted to just one species each (H. squamatum and H. caput-felis, respectively). However, we have unexpectedly found 2n = 22 for all the species of H. sect. Pseudomacularia [H. subg. Eriocarpum (Dunal) Martín-Hernanz, Velayos, Albaladejo & Aparicio], a chromosome number never reported so far out of H. subg. Plectolobum Willk. The phylogenetic logistic regression showed that chromosome number is not correlated with the karyotype parameters analysed, except marginally and negatively with CV_CL (Table 2), i.e., the higher the number of chromosomes, the higher the interchromosomal homogeneity. As mentioned, satellited chromosomes are common (see for example Fig. 2a, b, c, h, i) but they have not been described in this study due to inconsistencies among metaphase spreads.

Our analyses showed that karyotype of Helianthemum can be considered quite symmetric (Table 1). At subgenus level, values of interchromosomal heterogeneity (CV_CL) ranged between 14.30 ± 1.94 in H. subg. Plectolobum to 17.11 ± 3.15 in H. subg. Eriocarpum, and intrachromosomal heterogeneity (M_CA) between 20.23 ± 4.05 in H. subg. Plectolobum to 24.49 ± 2.09 in H. subg. Helianthemum. At the section level, values of CV_CL and M_CA were also quite homogeneous ranging CV_CL from 12.70 ± 1.94 in H. sect. Macularia Dunal to 21.74 ± 1.79 in H. sect. Brachypetalum Dunal, and M_CA from 10.91 in the monospecific H. sect. Argyrolepis Spach to 28.07 ± 7.43 in H. sect. Lavandulaceum G.López. Moreover, CV_CL and M_CA were not correlated, and H. subg. Plectolobum showed even a more symmetric karyotype compared to the other two subgenera, particularly in mean centromeric asymmetry (Fig. 3). The asymmetry classification of Stebbins and the karyotype formula further showed regular symmetry and a predominance of metacentric (m) and submetacentric (sm) chromosomes, with some subtelocentric (st) chromosomes only present in H. teneriffae Coss. and H. kostchyanum (Table 1; Fig. 2i). We can discard chromosome structural heterozygosity, so odd numbers in karyotype formulas are the consequence of chromosome folding in photographs or subtle angle deviation in individual measurements (Yu, pers. comm.).

Nevertheless, it is very interesting to note that the total karyotype length (THL) resulted quite variable among subgenera, sections and species. THL values for H. subg. Eriocarpum, Plectolobum and Helianthemum were 22.58 µm, 38.27 µm, and 32.88 µm, respectively. Sections ranged from 18.93 µm in H. sect. Eriocarpum Dunal to 44.03 µm in the monospecific H. sect. Caput-felis G.López. The entire genus Helianthemum, the three subgenera, and H. sects. Pseudomacularia, Helianthemum (s.l.), and Macularia showed high heterogeneity in mean THL values across species, with CV > 10% (Table 1; Fig. 4). Conversely, mean THL had CV < 10% in the non-monospecific sects. Lavandulaceum, Eriocarpum, Brachypetalum and Pseudocistus Dunal. Notice that H. sect. Helianthemum p.p. and H. sect. Helianthemum Canarian clade had also CV < 10%.

The analysis of chromosome number evolution with the ancestral base chromosome number fixed at n = 10 showed that the best-fitting model for Helianthemum was CONST_RATE_NO_DUPLI with an AIC value of 66.02 (Table 3). This scenario revealed a CONSTANT_RATE background with 0.02053 gain events Myr^-1, 0.03258 loss events Myr^-1 and no polyploid events across the phylogenetic tree (Fig. 5). In agreement with previous knowledge, we detected one shift in the mode of chromosome evolution in the lineage of H. squamatum (n = 5) in which the rate of chromosome losses increased several orders of magnitude (5.34359 events Myr^-1).

Mean chromosome length in Helianthemum is about 3 µm long, albeit they range from small (1.69 µm) to medium-sized (4.91 µm) at section and subgenus levels. Remarkably, H. squamatum is both, the species with the lowest chromosome number (2n = 10) and the species with the longest chromosomes (4.91 µm), likely because its genome is the consequence of a large reorganization achieved by progressive primary chromosomal rearrangements (see Aparicio & al. 2019).

Except in one case (H. songaricum Schrenk ex Fisch. & C.A.Mey; see below) we did not find deviating chromosome numbers with respect to previous knowledge (Goldblatt & Johnson 1979; Rice & al. 2015; Aparicio & al. 2019). Also, all the new chromosome counts here provided are coincident with those of related species, such as 2n = 20 for H. confertum Dunal and H. sancti-antonii Schweinf. in H. sect. Eriocarpum, or 2n = 22 for H. frigidulum Cuatrecasas, H. marifolium subsp. andalusicum (Font Quer & Rothm.) G.López and H. origanifolium subsp. africanum M.B.Crespo, M.A.Alonso, A.Vicente & J.L.Villar in H. sect. Pseudocistus. Nevertheless, the chromosome number found for all the species of H. sect. Pseudomacularia (2n = 22) is unexpected since this number has never been reported for any species within H. subg. Eriocarpum (actually, out of H. subg. Plectolobum). Furthermore, it is remarkable that our observation of 2n = 22 for H. songaricum (see Fig. 2f) disagrees with Zhao & al. (2000) who reported 2n = 20. Overall, it seems that the whole genus Helianthemum is integrated by diploid species, since only a few tetraploid populations have been reported in the literature for some therophyte species [e.g., H. aegyptiacum (L.) Mill., H. ledifolium (L.) Mill.; Goldblatt & Johnson 1979]. It is then essential to confirm the somatic number 2n = 40 reported for H. ordosicum (Zhao & al. 2000), considering that this species is closely related to H. songaricum (in fact synonymized by Quiner & Gilbert 2007).

Karyotype symmetry has two components, one related to variation among chromosome size and the other to variation in centromere position (Peruzzi& Eroğlu 2013). Most species of angiosperms are characterized by uniform symmetric karyotypes with mostly meta or submetacentric chromosomes (Weiss-Schneeweiss & Schneeweiss 2013), and it has been classically assumed that asymmetric karyotypes derived from symmetric ones (Stebbins 1971). Nevertheless, cytogeneticists now believe that reversal situations may have occurred (Stace 2000) and that karyotype asymmetry is a transitory state rather than an evolutionary endpoint (Lysák & al. 2006). In this study we find that most chromosomes in Helianthemum are meta or submetacentric, and that indices of karyotype symmetry show consistently low values of heterogeneity and variation. Values of CV_CL and M_CA can reach 100 (or even higher), but in taxa in which variation is ostensible, the value of asymmetry indexes may span from 20-80 (see fig. 2 in Peruzzi & Eroğlu 2013). Values of CV_CL and M_CA for Helianthemum were low and scarcely variable at the subgenus and section levels (overall from 10.91-33.32). At the species level, the highest values of intrachromosomal heterogeneity are found in H. motae Sánchez-Gómez, Jiménez & Vera, H. pomeridianum Dunal and H. kostchyanum Boiss. (M_CA = 33.32, 30.68, 29.91, respectively), in which a higher proportion of submetacentric and subtelocentric chromosomes is found (see Table 1). It is interesting to note that H. subg. Plectolobum has the longest chromosomes and the lowest asymmetry (also evident in H. squamatum, at the species level), a result that supports that genomic processes involved in increasing chromosome size are also likely increasing chromosome symmetry (Levin 2002; Weiss-Schneeweiss & Schneeweiss 2013).

The total length of the karyotype is an indirect measure of the amount of DNA contained in the cell nucleus, and is thus related to the size of the genome, whose trends of variation and evolution have received increasing attention (Levin 2002; Soltis & al. 2005; Weiss-Schneeweiss & Schneeweiss 2013; Pellicer & al. 2018). In Helianthemum, the values of total haploid karyotype length between species range almost 3-fold from 16.91 μm in H. sessiliflorum (H. sect. Eriocarpum) to 47.84 μm in H. kostchyanum (H. sect. Helianthemum p.p.), disregarding chromosome number. But when THL is analysed at the section and subgenus levels remarkably interesting results appear.

THL values increase from 22.58 μm in H. subg. Eriocarpum to 33.99 μm in H. subg. Helianthemum and 38.27 μm in H. subg. Plectolobum, with high coefficients of variation (> 10%). This means that karyotype length is not a conserved cytogenetic trait neither among subgenera nor among sections (see Fig. 4). For example, in H. subg. Eriocarpum most species have low or medium THL values except those in H. sect. Lavandulaceum, and species of H. sect. Brachypetalum and H. sect. Helianthemum Canarian clade have clearly lower values of THL than the rest of H. sect. Helianthemum (i.e., H. sect. Helianthemum p.p.). In H. subg. Plectolobum the two species that integrate H. sect. Macularia have disparate THL values with H. pomeridianum having considerably lower THL values than H. lunulatum (All.) DC. (22.2 and 39.37 µm, respectively). In this case, both sister species diverged in a Pliocene vicariance now attested by an intercontinental disjunction, and their divergence also entailed an outstanding environmental niche shift (Aparicio & Albaladejo 2017; Martín-Hernanz & al. 2021b). The therophyte H. sanguineum (Lag.) Lag. ex Dunal also had a low THL value (24.09 µm). Sections Pseudomacularia and Helianthemum p.p. also showed high coefficients of variation, due to large variability among the four species of Pseudomacularia, and due to the existence of the therophyte H. aegyptiacum in H. sect. Helianthemum p.p. The THL values of the non-monospecific H. sects. Lavandulaceum, Brachypetalum, Pseudocistus were conserved among species (i.e., CV < 10%).

The pattern of chromosome number evolution in Helianthemum remains invariant even after the inclusion of some species and the entire H. sect. Pseudomacularia (with the unexpected somatic number 2n = 22): chromosome number shifts have not been a major driver in the evolution of Helianthemum, in which a constant rate of single chromosome increase or decrease is predominant. If n = 10 is the ancestral base chromosome number, Helianthemum has then evolved three independent instances of chromosome gain (see Fig. 5): (1) the ancestors of H. sect. Pseudomacularia, (2) the whole H. subg. Plectolobum and (3) the lineage of H. caput-felis. It is worth mentioning the rate of chromosome losses in the lineage of H. squamatum, which increased by several orders of magnitude compared to the whole phylogenetic tree. Although whole genome duplications have been considered a primary source of variation and evolution, dysploid changes can, indeed, be even more persistent that those achieved by polyploidy (Escudero & al. 2014).

The wealth of comparative cytogenetic information gathered in this study allows us to present a compelling picture of Helianthemum as a genus with stable chromosome numbers, whose evolution involved only three instances of slow chromosomes gain. Nevertheless, imprints of large genome reorganizations in this genus are quite evident, such as in H. squamatum, and in the high variation in karyotype (i.e., genome) size that we have found at subgenus, section and species levels. Indeed, in Helianthemum, regardless of chromosome number, karyotype size contains very relevant systematic and evolutionary information which we summarize in Figure 6. Note that when the mean THL values are averaged across sections, ‘small’ and ‘large’ karyotypes appear to be separated by a gap between 28 and 37 µm. On the one hand, small karyotypes below 28 µm are found in most sections of H. subg. Eriocarpum, all the therophyte species regardless their taxonomic position, all the species H. sect. Helianthemum Canarian clade plus H. pomeridianum. On the other hand, large karyotypes over 37 µm are found in H. sect. Helianthemum p.p. (except the therophyte H. aegyptiacum) and the whole of H. subg. Plectolobum (except H. pomeridianum). In other words, ‘small’ karyotypes are present in desert specialists, therophytes and the recently diversified species of the Canary Islands (Albaladejo & al. 2021), whose breeding system is also predominantly autogamous (Martín-Hernanz & al. 2023). On the other hand, ‘large’ karyotypes characterise mostly xenogamous chamaephyte species of Mediterranean and Eurosiberian distribution (Martín-Hernanz & al. 2021b).

In future analyses we will use the power of a high-resolution phylogenomic reconstruction based on target capture data (Martín-Hernanz & al. unpublished) and 2C values of nuclear DNA amount (Pellicer & al. 2018) to trace the direction and strength of genome size evolution and its potential relationships with shifts in extrinsic (i.e., environmental niche) and intrinsic (i.e., breeding systems and habit) characteristics in the genus Helianthemum. Cytotaxonomy has always gone hand in hand with phylogenetics for a better understanding of chromosome and species evolution (Guerra 2012).