Review of chromosome cytology in Moraea (Iridaceae: Irideae): what chromosomes reveal about the evolution of the genus

A review of the chromosome cytology of the African and Eurasian geophytic genus Moraea Mill. (currently 214 spp.); including 51 new counts, many for taxa poorly known cytologically or not counted before, that shows that 167 species, representing 78% of the total, have been counted from one or more populations. The inferred ancestral base number is x = 10. Polyploidy is relatively rare; available counts indicate that both Eurasian species are tetraploid, but that, among the sub-Saharan species, just nine species (less than 5%) are exclusively polyploid and an additional 15 (7%) have diploid and polyploid populations. Chromosome rearrangement leading to reduced base numbers has occurred in subg. Polyanthes (x = 10), in which four sections have a base number of x = 6. Three subgenera, Grandifl orae, Homeria and Vieusseuxia, also have x = 6, but have different karyotypes. Several species and one subspecies are dysploid, all but one with haploid numbers lower than in related species, and are neodysploids. Except for M. virgata subsp. karooica, dysploidy is interpreted as descending. Fourteen species have diploid and polyploid populations, notably M. crispa (subg. Polyanthes) and M. cookii (subg. Homeria), in which the distribution of populations with 2n = 12, 24 and 36 is correlated with geography. Seven species have euploid and dysploid populations at the diploid level and M. inclinata has populations with 2n = 12 and 22. Differences in chromosome number within species are not normally refl ected in external morphology. Compared to most other genera of Iridaceae in sub-Saharan Africa, chromosome number and karyotype are unusually variable so that sampling of multiple populations of species is required to establish these characters. Although many species remain to be examined cytologically, those uncounted are mostly in the species-rich subg. Grandifl orae and subg. Vieusseuxia, both of which exhibit little variation in chromosome number and karyotype. Chromosome rearrangement and polyploidy appear to have been important in the early evolution of the genus as well as in recent speciation.


INTRODUCTION
The cytology of the Afro-Mediterranean geophytic genus Moraea Mill.(currently 214 spp.) of Iridaceae, subfamily Iridoideae, has been the subject of extensive investigation because of the unusually wide variation in chromosome number and karyotype that closely correlate with morphological variation.Moraea also exhibits one of the most extensive dysploid reduction series in fl owering plants.The correlation between karyology and morphology led Goldblatt (1976a) to propose an infrageneric classifi cation using cytology as a major factor in circumscribing subgenera and sections.The circumscription of Moraea was subsequently expanded to include the related genera Barnardiella Goldblatt, Galaxia Thunb., Gynandriris Parl., Hexaglottis Vent., Homeria Vent.and Roggeveldia Goldblatt.These genera were reduced to synonymy in Moraea on the basis of inferred morphological, and in some instances cytological, synapomorphies with species groups within Moraea (Goldblatt 1998).A molecular DNA-based phylogenetic study of Moraea subsequently confi rmed Goldblatt's conclusion that Moraea is paraphyletic without the inclusion of these genera, thus endorsing its expanded circumscription (Goldblatt et al. 2002).That study and a second, more extensive one (Schnitzler et al. 2011) also showed that Goldblatt's (1976a) infrageneric classifi cation of Moraea needed extensive revision.Although most of the more distinctive species clusters are monophyletic, subg.Moraea included disparate elements and required major revision.In addition, some species or species clusters did not fall in the subgenera or sections to which they had been assigned.Our new classifi cation based on phylogenetic principles and integrating molecular and morphological data recognizes 11 subgenera (Goldblatt et al. 2013) and provides a framework for interpreting cytological evolution in the genus.Here we provide a review of chromosome cytology in Moraea following this new classifi cation.We include 51 new chromosome reports, several for species known until now from a single count and fi ve for species previously unknown cytologically.Chromosome numbers have now been reported for 167 of the 214 species of Moraea currently recognized, representing 78% of the genus, a signifi cant albeit incomplete survey.

MATERIALS AND METHODS
Data on chromosome numbers in Moraea were extracted from Goldblatt's (1971) review of the cytology of southern African Iridaceae and from indexes to plant chromosome numbers that included chromosome numbers published after that time (Moore 1974(Moore , 1977;;Goldblatt 1981cGoldblatt , 1984cGoldblatt , 1985aGoldblatt , 1987a;;Goldblatt & Johnson 1990, 1994, 1996, 2000, 2003, 2006, 2010).Counts are tabulated following the revised, phylogenetic infrageneric classifi cation (Goldblatt et al. 2013), with species names corrected to refl ect current nomenclature and taxonomy.Original sources were consulted for all counts.
We interpret patterns of change in chromosome number and karyotype in light of the plastid DNA-based Bothalia 43,1 (2013) phylogeny, also published in Bothalia 43(1), which includes a phylogenetic tree that may be consulted as companion to this review (Goldblatt et al. 2013).We then relate the patterns to established hypotheses for interpreting these phenomena by Stebbins (1950Stebbins ( , 1971)), Jones (1970) and Raven (1975).Briefl y, polyploid sequences always proceed from lower to higher numbers by doubling.Stepwise changes, rather than doubling of base numbers are, we believe, largely descending in a process involving translocation of chromosome material to a second chromosome, accompanied by loss of a centromere, some heterochromatin and those genes associated with cell division.The process is here and throughout the text referred to as dysploidy.Descending dysploid reduction sometimes results in translocation of a long chromosome arm of an acrocentric chromosome to the short arm of another and loss of the centromere portion of the donor chromosome, resulting in a large metacentric chromosome and a lower base number, a process often called chromosome fusion or Robertsonian translocation.Translocation of the long chromosome arm of a short acrocentric chromosome to the distal end of the long arm of a second chromosome and loss of the centromere of the donor chromosome yields a longer, acrocentric chromosome.

Original counts
For original counts, root tips of newly sprouted corms were harvested in mid-morning and prepared according to a root tip squash method described by Goldblatt & Takei (1993).Vouchers are housed at the Missouri Botanical Garden Herbarium (MO) and Compton Herbarium (NBG).Counts are based on samples of three to four individuals unless otherwise stated and, following widespread practice in plant cytology, are assumed to represent entire populations.

Chromosome number
The most parsimonious interpretation of the available data shows ancestral basic chromosome number in Moraea is x = 10.This was fi rst established by outgroup comparison, where x = 10 is the base number (and only chromosome base number) for the related genera Bobartia L., Dietes Salisb.ex Klatt and Ferraria Burm.ex Mill.(Goldblatt 1971(Goldblatt , 1981c;;Goldblatt & Takei 1997).Subsequent DNA sequence-based phylogenetic analyses (Goldblatt et al. 2002;Schnitzler et al. 2011;Goldblatt et al. 2013) are congruent with this hypothesis-species with base numbers of x = 10 are consistently retrieved as ancestral to species with other base numbers.
A base of x = 10 characterizes the monospecifi c subg.Plumarieae, fi ve (of eight) species counted in subg.Visciramosae, and seven (of eight) species counted in subg.Umbellatae (Table 1).The three subgenera are collectively sister to the remaining species of Moraea [see phylogenetic tree in Goldblatt et al. (2013)].Counts for all but two species are diploid, 2n = 20, but the two counts for M. margaretae (subg.Umbellatae) are tetraploid, 2n = 40.The single count for M. linderi is 2n = 30, thus triploid, but only two individuals of this extremely rare species were available for study and it seems likely that the species is normally diploid.Subgenera Acaules, Monocephalae and Moraea also have x = 10.Two species of subg.Moraea are dysploid: M. indecora has 2n = 16 and, as circumscribed at present, M. papilionacea has southern populations with 2n = 18 and a northern Cedarberg population has 2n = 16 (Goldblatt 1971;1976a).
In subg.Acaules, Moraea tricolor has dysploid populations in the west of its range (three populations documented with 2n = 18), but two pink-fl owered populations from the Caledon District (the eastern part of its range) have 2n = 20.In M. ciliata two of ten counts, both from the Caledon District, are tetraploid, 2n = 40; and one population from Glenlyon, Nieuwoudtville, includes both euploid and dysploid individuals (Goldblatt 1976a).Based on available counts, these six subgenera together have 26 species exclusively diploid, one exclusively tetraploid, one heteroploid, two dysploid at the diploid level, and one (M.tricolor) with dysploid and euploid populations.
The specialized, acaulescent subg.Galaxia, with all but two of 17 species counted, has x = 9.This is the only base number in ser.Galaxia (Table 1) and also in the monospecifi c ser. Unguiculatae (sister to ser.Eurystigma and Galaxia).Ser.Eurystigma, in contrast, exhibits an extended dysploid series, n = 8, 7, and 6.M. versicolor of ser.Eurystigma has most populations with n = 8 but populations from the northern Cape Peninsula have n = 7. M. galaxia (ser.Galaxia) is heteroploid, with most populations 2n = 36, thus tetraploid, but one population in the eastern part of its range is diploid, 2n = 18, and another from the northern end of its range has 2n = 54, thus hexaploid.Three counts for M. stagnalis are also tetraploid, 2n = 36.Counts for all members of ser.Eurystigma are at the diploid level except M. citrina, in which three populations counted are diploid with 2n = 16, one is tetraploid and one dysploid with 2n = 14 (Goldblatt 1979a).
Subg.Polyanthes, now including several species previously assigned to subg.Moraea (Goldblatt 1976a) has x = 10 in sects.Deserticola and Serpentinae (Table 1).The four remaining sections all have the derived base of x = 6.In the M. fugax complex (sect.Deserticola), M. gracilenta and M. macrocarpa have exclusively 2n = 20 (fi ve populations counted) but races of M. fugax have n = 10, 9, 8, 7, 6 and 5. Closely allied M. fi licaulis (also treated as a subspecies of M. fugax), has diploid numbers of 2n = 12 and 10.As noted elsewhere (Goldblatt 1986b), this dysploid series in M. fugax is remarkable for a single species, albeit a variable one with distinct morphological races.Outgroup comparison compels us to interpret the series as descending.Signifi cantly, populations of M. fugax with 2n = 10 have about the same genome size as species with 2n = 20 based on measurements of chromosome length (Goldblatt et al. 1986b;Goldblatt & Takei 1997).This independently supports our inference that the pattern in M. fugax represents descending dysploidy.An early count of 2n = 28 in M. fugax by Sakai (1952) must be seen as an error: most likely his study material was misidentifi ed.
The remaining four sections of subg.Polyanthes (Table 1) have x = 6.Sect.Polyanthes has exclusively  (Goldblatt 1987b as Hexaglottis).Among the diploid species of the section, the widespread M. virgata subsp.virgata has 2n = 12 over most of its range but 2n = 10 in the immediate southwestern Cape populations, these evidently derived from the ancestral x = 6 for the section.Subsp.karooica of M. virgata is evidently also dysploid but has 2n = 14 in two populations counted (of three known at present).Two very small chromosomes account for the extra pair.
The pattern in sect.Gynandriris is more complex (Table 1).The two Mediterranean-Middle Eastern species, M. sisyrinchium and M. mediterranea, are tetraploid, 2n = 24, for the many populations counted (at least 19 records for M. sisyrinchium) (Goldblatt 1980b), and the southern African G. simulans is evidently also polyploid, based on two populations sampled, one 2n = 24 and the other 2n = 36.Counts for fi ve of the remaining six southern African species, all from the winter-rainfall region of the subcontinent, are exclusively diploid, 2n = 12, but M. contorta remains uncounted.
Largest section of subg.Polyanthes, sect.Pseudospicatae with 13 of 18 species counted, has nine species exclusively diploid, all 2n = 12, and two, M. elliottii and M. inclinata, with diploid and tetraploid populations (Table 1).The latter has 2n = 22 in the one polyploid population so far recorded, thus hypopolyploid.The widespread M. stricta, the range of which extends from Stutterheim in Eastern Cape, South Africa to Ethiopia, has southern populations tetraploid and hexaploid, 2n = 24, 48, and a population in Malawi, pentaploid, 2n = 60.Lastly, the comparatively widespread M. crispa has diploid, tetraploid and hexaploid populations, all based on n = 6.The six Great Karoo and Roggeveld populations sampled are diploid, four Cedarberg populations are tetraploid, and the single Cold Bokkeveld population, the southernmost sampled, is hexaploid, thus 2n = 36.The only count for M. monticola is tetraploid, thus 2n = 24, in one of its two known populations.
The remaining subgenera, Grandifl ora, Homeria and Vieusseuxia all have x = 6 (Table 1).Although the molecular phylogeny (see phylogenetic tree in Goldblatt et al. 2013) retrieves them as possibly sharing a common ancestor (with weak statistical support), each has a different karyotype, suggesting that x = 6 may have been achieved independently in each.The majority of species are diploid.For subg.Grandifl ora, cytologically least well known of the subgenera, 11 of 28 species have been counted: all are diploid, 2n = 12, and have an almost identical karyotype (Goldblatt 1971;Goldblatt & Takei 1997).In subg.Vieusseuxia, 28 of 42 species are known from at least one count.Of these 25 are exclusively diploid and just three, M. tricuspidata, M. tulbaghensis and M. villosa, have diploid and tetraploid populations.No species is exclusively polyploid.
The cytology of subg.Homeria is particularly interesting for its diversity: 21 of the 35 species (all counted) are exclusively diploid with 2n = 12.Dysploid reduction at the diploid level has been documented in Moraea pallida, which has its easternmost populations diploid, 2n = 12, and the western populations 2n = 8.The latter are complex heterozygotes forming chromosome rings at meiosis (Goldblatt 1980c).In western South Africa, M. demissa has 2n = 10, 9 and 8; and the only number so far known for M. fl avescens (two populations sampled) is 2n = 9.These last two species, like unrelated M. pallida, are also complex heterozygotes, forming various patterns of chromosome rings at meiosis (Goldblatt 1986c).Both are autogamous and produce full capsules of normal-looking viable seeds.Seven species, M. bifi da, M. britteniae, M. bulbillifera, M. cookii, M. miniata, M. ochroleuca and M. pyrophila, have diploid and tetraploid populations (just one each of the several populations counted for M. bifi da and M. miniata).The particularly widespread M. cookii has diploid, tetraploid and hexaploid populations and a pattern with a clear geographical component, with diploids occurring in the Western Karoo, tetraploids in the mountains of the Great Karoo, and hexaploids in the interior mountains of Western Cape.M. fl accida has both tetraploid and hexaploid populations and several counts for H. collina and the only one for M. marlothii are tetraploid, 2n = 24.Thus on available data, only three (8%) of the 35 species of subg.Homeria counted, are exclusively polyploid.

Genome size
Chromosomes of all Moraea species, and most other members of Iridoideae, are relatively large compared with those of other subfamilies of Iridaceae (Goldblatt 1971;Goldblatt & Takei 1997), directly refl ecting larger genome sizes.Total DNA per cell in most diploid species is in the range 15-27 pg (Goldblatt et al. 1984c) using the correction factor determined by Goldblatt & Takei (1997).Dietes and Ferraria, the other African members of tribe Irideae (to which Moraea belongs), have a similar genome size and comparably large chromosomes.In Moraea, chromosomes of subg.Galaxia are the smallest in the genus (Goldblatt 1971), evidently a derived condition: the one diploid species sampled, M. fugacissima, has just 4.8 pg per cell, somewhat less than half that in other subgenera.In subg.Vieusseuxia, species of sect.Villosae have particularly large chromosomes and a genome size of ± 27 pg in the diploid M. calcicola and ± 54 pg in polyploid samples of M. tulbaghensis and M. villosa.The single species of sect.Vieusseuxia sampled, M. unguiculata, has a genome size of ± 19.7 pg, thus consistent with most other subgenera.Genome size in subg.Grandifl orae is evidently the largest in the genus, as determined by chromosome size alone.Comparing chromosome volume (not length), Goldblatt (1971) showed that M. spathulata (2n = 12) (subg.Grandifl orae) has about twice as much chromosome material as M. ramosissima (subg.Moraea ) and Dietes, both 2n = 20.By extension, we infer a genome size in subg.Grandifl orae of about twice that for diploid species of other subgenera, excluding Vieusseuxia sect.Villosae.

Karyotype morphology
Although not clear from early illustrations made using sections of root tips (e.g.Goldblatt 1971), the x = 10 karyotypes appear moderately bimodal under root tip squash methods (Goldblatt 1976a;Goldblatt & Takei 1997).The chromosome complements consist of three long acrocentric chromosome pairs and seven medium to short ± acrocentric pairs about half the length of the long pairs, e.g. in M. elsiae (subg.Visciramosae), M. anomala (subg.Monocephalae) and even in M. herrei (Goldblatt 1976c as Barnardiella spiralis) and M. namibensis (subg.Polyanthes).Karyotypes in Ferraria, sister to Moraea, are comparable, as are those for Dietes (Goldblatt 1971(Goldblatt , 1981a;;De Vos 1979), sister to Ferraria + Moraea (Goldblatt et al. 2002;Schnitzler et al. 2011).Most species have small satellites on the distal end of one of the longest chromosome pairs but satellite position is variable and satellites are located in different positions in several species.
Each of the three subgenera in which x = 6 is ancestral are characterized by somewhat different karyotypes, thus consistent with a hypothesis that the reduced base number evolved independently in each.In subg.Grandifl orae karyotypes are strongly conserved and are fairly symmetric, with all chromosomes acro-to subtelocentric.Size differences are minimal in all species counted (Goldblatt 1971) but the longest and fourth longest pairs are consistently acrocentric.Satellites are located on the short arm of a long telocentric chromosome pair (third longest pair of the complement) in most species counted.Karyotypes in subg.Homeria are also conserved and symmetric in all species with x = 6.The complement comprises six nearly equal acrocentric pairs, with a small satellite located on the distal end of the short arm of one of the two longest pairs (Goldblatt 1971(Goldblatt , 1980a)).In the three species with lower numbers, 2n = 10, 9 and 8, karyotypes are asymmetric and include one or more pairs of very long metacentric chromosomes ± twice the length of the acrocentrics.
Subg.Vieusseuxia exhibits some variation in its karyotypes, but in general the largest chromosome pair is metacentric or almost so and there are two additional, smaller metacentric or submetacentric pairs.The position of satellites is especially variable, but small satellites are located on the short arm of a long, acrocentric chromosome pair in the majority of species examined cytologically (Moraea algoensis, M. amabilis, M. barnardii, M. caeca, M. calcicola, M. cuspidata, M. loubseri, M. marionae, M. tricuspidata, M. unguiculata, M. worcesterensis).A notable exception is M. tripetala, in which a large satellite is present on the short arm of a nearly telocentric chromosome pair (Goldblatt & Manning 2012).This feature readily separates it from close allies, which have different karyotypes, most of them typical of the subgenus.
Other notably different karyotypes in subg.Vieusseuxia include very large satellites on the short arm of an acrocentric chromosome pair in Moraea insolens and M. lurida, a closely allied species pair, and small satellites in the longest, metacentric pair in M. bellendenii, M. mutabilis, M. tulbaghensis and M. villosa.Particularly unusual for the subgenus, the karyotype of the taxonomically isolated M. thomasiae has fi ve acrocentric chromosome pairs, one submetacentric pair, and unusually large satellites on the short arm of a long acrocentric pair, a feature reminiscent of the karyotype of M. tripetala.M. fergusoniae, the only species of the subgenus with multiple leaves (and referred in the past to subg.Moraea) has a karyotype comparable to that of M. thomasiae with large satellites on a long acrocentric chromosome pair, but the smallest chromosome pair is metacentric (Goldblatt 1971, 1976a andunpublished).Evidently chromosome rearrangement, often involving the position of the satellite, has been important in the evolution of the subgenus, and is sometimes a useful indicator of phylogenetic relationship among species.Karyotypes are, as far as known, consistent within a species but most (with the exception of M. fergusoniae, M. tripetala and M. amabilis) are known from only one or two counts, rendering this conclusion open to verifi cation.
In subg.Polyanthes, karyotypes of sect.Polyanthes and Pseudospicatae, with x = 6, are comparable and differ mainly in the position of the satellites.The longest and one other pair are submetacentric and the remaining pairs are acrocentric.The karyotype of M. polystachya documented by Goldblatt (1976a) has a small satellite at the distal end of the long arm of the longest acrocentric pair, the same karyotype as in M. venenata (published as M. polystachya and not illustrated).Other species examined have small satellites on a small acrocentric pair.Satellite position is evidently variable, even with the same species (e.g. in M. crispa and M. polyanthos) and evidently without taxonomic signifi cance (Goldblatt 1980a(Goldblatt , 1986c)).One of two populations of the Tanzanian M. callista examined is structurally heterozygous with one long and one short metacentric chromosome (Goldblatt & Takei 1997), the signifi cance of which remains to be established.
In sect.Hexaglottis the ancestral karyotype consists of six nearly equal acrocentric pairs, with a small satellite present on the distal end of the short arm of one of the two longest pairs (Goldblatt 1987b).Moraea lewisiae subsp.secunda has a distinctive karyotype for the group, with the longest chromosome pair metacentric and with a small satellite located at the distal end of the shortest chromosome pair.As discussed in more detail below, in M. virgata (x = 6), southwestern Cape populations of subsp.virgata have 2n = 10 and M. virgata subsp.karooica has 2n = 14, thus both one subspecies is dysploid and the other has dysploid populations.The basic karyotype in the section recalls that in subg.Homeria, evidently the result of convergence.
The karyotype in sect.Gynandriris is the most distinctive in subg.Polyanthes, consisting of one long metacentric pair, one small metacentric pair and four moderate-sized acrocentrics, one or sometimes two of which have large satellites on their short arms.The polyploid M. monophylla and M. sisyrinchium have the same karyotype with two sets of chromosome pairs otherwise matching the diploid karyotypes.Evidence is consistent with an independent origin of x = 6 in sect.Gynandriris and a shared origin of that base in sect.Polyanthes and sect.Pseudospicatae, but possibly not for sect.Hexaglottis.

Dysploidy
There are six known dysploid species in Moraea: M. indecora and M. papilionacea (subg.Moraea), M. variabilis and M. minima (subg.Galaxia), and M. demissa and M. fl avescens (subg.Homeria).Several more have euploid and dysploid populations.In addition, subg.Galaxia is itself dysploid, as is ser.Eurystigma of subg.Galaxia.Subgenera Grandifl ora, Vieusseuxia and Homeria, all x = 12, are dysploid as are four of six sections of subg.Polyanthes.Dysploidy in populations of otherwise euploid species is associated with no visible morphological changes except in M. virgata.Here the taller, larger-fl owered subsp.karooica has an extra pair of small chromosomes, thus 2n = 14 (vs.2n = 12 in 5 populations of subsp.virgata and 2n = 10 in four more) and is suffi ciently distinct in morphology to merit taxonomic separation.All but one of these examples clearly represents descending dysploidy.
Among the dysploid species two different patterns of chromosome rearrangement are evident.The fi rst is exemplifi ed by M. indecora (2n = 16), which has two large metacentric pairs (plus one long acrocentric pair bearing a small satellite on the short arm and fi ve short ± acrocentric pairs) whereas its relatives with 2n =20 have only acrocentric chromosomes.The reduction in base number is best explained by fusion of small chromosome pairs, the probable result of Robertsonian translocation (the product of unequal reciprocal translocation and loss of a pair of centromeres, fi de Jones 1970).In M. papilionacea, which has populations with 2n =18 and 2n =16, there are one or two pairs of large metacentric chromosomes respectively (Goldblatt 1971 and unpublished).
In subg.Galaxia both this and a second pattern, the absence of metacentric chromosomes in the dysploid derivatives, are evident.The subgenus has a basic karyotype with the derived 2n = 18 and consists of 9 pairs of acrocentric chromosomes.Base number in ser.Eurystigma is x = 8, again derived, and consisting only of acrocentric chromosome pairs.However, dysploid populations of M. versicolor with n = 7 have a pair of large metacentric chromosomes.M. variabilis is also dysploid, n = 7, but the karyotype consists only of acrocentric pairs (Goldblatt 1979a).In the most striking example of descending dysploidy, Moraea fugax displays a sequence of diploid chromosome numbers from the ancestral 2n = 20 to 10 (Goldblatt 1986b).In the blue-or white-or pink-fl owered morph, karyotypes with n = 10, 9 and 8 have exclusively acrocentric chromosomes but the n = 7 morphs have the largest pair metacentric and one n = 6 morph has two large metacentric chromosome pairs.In the yellow-fl owered morph of the species with n = 8, 7 and 5, the n = 8 population has the largest pair metacentric, the n = 7 population has two large metacentric pairs and the n = 5 pair has the three largest pairs metacentric.
In these examples, the dysploid populations or taxa thus exhibit both patterns of karyotype change with about equal frequency.

DISCUSSION
In Moraea, only narrowly endemic species can be confi dently predicted to have a single chromosome number.In more widespread species, multiple counts for different populations across their ranges are needed to establish chromosome number.Of 166 species counted, no less than 14 species have diploid and polyploid populations and another seven have euploid and dysploid populations at the diploid level and one at the polyploid level (M.inclinata).Chromosome number and karyotype in Moraea are unusually variable compared to most other genera of Iridaceae in sub-Saharan Africa, which mostly display very conservative karyotypes.However, in Lapeirousia and Romulea (subfam.Crocoideae) at least, dysploidy and subsequent polyploidy have been involved in their evolution and speciation (Goldblatt & Takei 1993, 1997).Although many species of Moraea remain to be examined cytologically, uncounted species are mostly in the species-rich subg.Grandifl orae and subg.Vieusseuxia, both of which to date have exhibited no variation in chromosome base number and, in subg.Vieusseuxia, only moderate variation in karyotype.

Polyploidy
Although believed to be a signifi cant mode of speciation in many fl oras (Stebbins 1950(Stebbins , 1971; and see review by Soltis et al. 2010), polyploidy appears relatively unimportant in the rich geophytic fl ora of sub-Saharan Africa, and of particularly the Greater Cape fl ora.Just 11 of the 166 species of Moraea for which we have counts are exclusively polyploid, thus less than 7%, and 15 more (9%) have diploid and polyploid populations.Of the polyploid species, two are Eurasian, thus only 9 of 164 sub-Saharan (African) species of Moraea counted (5.5%), are exclusively polyploid.This is consistent with a low frequency of neopolyploidy in other sub-Saharan geophytic monocot families.
For example, in Hyacinthaceae, which are particularly well represented in southern Africa, available counts in subfam.Hyacinthoideae show that 16 of 101 counted species (16%) of Eucomis L'Herit, Lachenalia J.Jacq.ex Murray and Massonia Thunb.ex Houtt.are exclusively polyploid, and a further 18 (18%) have diploid and polyploid populations (Goldblatt et al. 2012).However, fi ve genera, including Eucomis, have polyploid base numbers, and are thus evidently palaeopolyploid.In sub-Saharan members of subfam.Ornithogaloideae (Goldblatt & Manning 2011), only one of the 24 species of Ornithogalum L. and three of 23 species of Albuca L. subg.Albuca counted are exclusively polyploid.In subfamily Urgineoideae only one of 14 counted sub-Saharan species of Drimia Jacq. is exclusively polyploid (7%).In contrast, polyploidy is relatively frequent in Eurasian species of Ornithogalum, Drimia and most of the larger genera of Hyacinthoideae.Among southern African Amaryllidaceae and Colchicaceae, both smaller geophyte families, there are no recorded polyploid species.
Dysploidy provides a means of reproductive isolation in populations in the same way as polyploidy; and thus appears to have played almost as much a role in evolution and, by inference, speciation in Moraea as has polyploidy.Five species and one subspecies are what may be called neodysploids (dysploid at and below species rank) vs. 11 neopolyploids.More important, four subgenera (Galaxia, Grandifl orae, Homeria and Vieusseuxia) are palaeodysploid, as are ser.Eurystigma of subg.Galaxia and also sections Hexaglottis, Gynandriris, Polyanthes and Pseudospicatae of subg.Polyanthes.Dysploidy appears to be largely unidirectional in Moraea, with all but one example of dysploidy most parsimoniously inferred as descending.There is just one instance of ascending dysploidy, in M. virgata subsp.karooica, in which the karyotype has one small additional chromosome pair in both of the two populations examined (of three currently known).
Moraea is therefore one of the relatively few genera of fl owering plants in which polyploid changes in chromosome number and chromosome rearrangement leading to dysploidy have been established as important factors in its early diversifi cation and subsequent more recent speciation.Considerations of the determinants of speciation need to take these phenomena into account as much as geography and ecological factors such as shifts in soil, habitat and climate preferences, and reproductive and pollination biology.Chromosome rearrangement and polyploidy have played important roles in the evolution of the genus as well as in recent speciation.