Chromosome Numbers and Evolution in Anoda and Periptera (Malvaceae)

Relationships within and between the principally North American, malvaceous genera A noda and Periptera are assessed through analysis of chromosomal and hybridization data. Chromosome numbers are reported for ten species of Anoda and one of Periptera, and observations on meiosis in hybrid and non hybrid plants are presented . The results indicate: I) that Anoda and Periptera are closely related and occupy a relatively isolated position in the tribe Malveae, 2) that speciation in A noda has occurred primarily at the diploid level, n = IS , although A. crenatijlora is tetraploid and A. cristata includes diploids, tetraploids, and hexapJoids, and 3) that A. thurberi and Periptera form a lineage based on n = 13, which probably was derived from Anoda sect. Liberanoda or its progenitors.


INTRODUCTION
Anoda Cav., which includes 23 species primarily of Mexico (Fryxell1987), and Periptera DC. , which is composed of four Mexican species (Fryxell 1974(Fryxell , 1984(Fryxell , 1987)), are distinguished from each other primarily by characters that reflect modes of pollination.The flowers of Periptera have a red or orange-red, tubular corolla and markedly exserted staminal column.They are assumed to be hummingbird pollinated.The flowers of Anoda are white, yellow, lavender, or purple.The corolla tends to be rotate, and the petals exceed or at most are equalled by the staminal column.Pollination is insect mediated.
Cytological and hybridization studies were initiated in Anoda and Periptera with the expectation they would lead to a better understanding of the nature of the relationship between the two genera, perhaps would suggest relationships to other genera of Malveae, and would provide a framework for considering evolution in Anoda.Similar approaches elsewhere in the Malvaceae have proven useful in defining and grouping genera (Bates 1968;Bates and Blanchard 1970, for further discussion), and in giving insights concerning relationships and the course of evolution in a number of taxa, most notably Gossypium L. (Saunders 1961 ;Fryxell 1979) and Hibiscus sect.Furcaria DC. (Menzel and Wilson 1973).

METHODS
Chromosome counts were made from meioses in pollen mother cells of flower buds collected from plants growing in the wild or grown from seed at Ithaca, New York.Procedures followed those described by Bates and Blanchard (1970).Pollen fertility was measured by scoring percentages of pollen staining with cotton blue in lactophenol.Voucher specimens are deposited in the L. H. Bailey Hortorium (BH).The systematic treatments and nomenclature follow those of Fryxell (1974Fryxell ( , 1987)).Jackson (1982Jackson ( , 1984) ) questioned characterizations of chromosomal pairing during meiosis and assumptions concerning autoploidy and alloploidy that are widespread in the taxonomic literature.While the relevance of Jackson's comments to this study are recognized, the cytological data available for Anoda and Periptera are those traditionally presented.Interpretations of chromosomal pairing were made at diakinesis and metaphase 1.The terms allopolyploid and autopolyploid are used to distinguish different pathways of euploid development, although the equivocal nature of the concepts they represent is recognized.

PREVIOUS CHROMOSOME COUNTS
Other than numbers reported by Bates and Blanchard (1970), which are considered in the following section, chromosome numbers for Anoda have been reported by Davie (1935), Skovsted (1935Skovsted ( ,1941)), Ford (1938), Delay (1947), and Krapovickas (1957).With the exception of a single count of 2n = 30, reported by Delay (1947) for A. parviflora Cav.(=A.crenatiflora Ortega), all counts were of A. cristata (L.) Schlecht., although some were reported under the names of A .hastata Cav.(=A .cristata) or A. wrightii A. Gray (=A.lanceolata Hook.& Am.).Chromosome numbers of n = 15 or 30 or their somatic equivalents were given in all but one instance.The single exception was a somatic count of 2n = 36 given for A. wrightii (Davie 1935).Using plants grown from seed obtained from Davie, Ford (1938) confirmed the number in a meiotic count, reporting 18 bivalents.He pointed out, however, that these plants did not differ morphologically from those of two other accessions of A. cristata, each of which had a chromosome number of n = 15.The number n = 18 remains unexplained, although it might have arisen as a stabilized derivative of hybridization between diploid and tetraploid or hexaploid forms of A. cristata.

Chromosome Numbers
Chromosome numbers determined in this study, as well as eight chromosome counts that were reported by Bates and Blanchard (1970), are given in Table 1.The inclusion of these earlier counts corrects nomenclature and summarizes the findings made in these two related studies.
Among the counts are the first reported numbers for Periptera and the following species of Anoda: A. abutiloides A. Gray, A. leonensis Fryxell, A. palmata Fryxell, A. pubescens Schlecht., A. thurberi A. Gray, andA.zuccagnii (K.Sprengel) Fryxell.Anoda crenatiflora also is probably reported for the first time.The earlier report of that species made by Delay (1947) is at variance with the tetraploid counts consistently made for A. crenatiflora.Further, the identification of species from the revision of Hochreutiner (1916) must be regarded as questionable.Counts for A. reticulata S. Wats.(Bates and Blanchard 1970) are here referred to A. pedunculosa Hochr., while those of A. acerifolia (Zuccagni) DC. (=A.zuccagnii) belong to A. cristata.
The chromosome numbers reported by Bates and Blanchard (1970) established the presence of n = 15 in taxa other than A. cristata and the occurrence of hexaploids, as well as diploids and tetraploids, in that species.This study confirmed the widespread occurrence of n = 15 in Anoda and revealed a second euploid species, the tetraploid A .crenatijlora.Unanticipated were the chromosome numbers of n = 13 in P. punicea (Lag.)DC. and n = 13 and 14 in A .thurberi.

Meiosis in Nonhybrids
Meiosis in diploid Anoda and Periptera was unexceptional with consistently high bivalent formation and normal segregation through first and second divisions.Unambiguous counts were possible from diakinesis through telophase II and even into early pollen formation .Although not always fully analyzable, pairing in tetraploid and hexaploid plants also seemed normal and showed no evidence of multivalent formations.Pollen fertility, as determined for at least one individual of each species and each ploidy level of A. cristata, ranged from 92 to 100 percent.
As is characteristic of many Malveae, premature disjunction of chromosomes in metaphase I may lead to the presence of unassociated univalents.Although these univalents usually segregate normally, in some instances they do not, and their differential inclusion in daughter cells could lead to the addition or loss of chromosomes in progeny.This is a probable explanation for the presence of a univalent in addition to the normal diploid complement in A. cristata (Bates & Vivaldi 3442) and A .thurberi (Bates & Vivaldi 3436).

Hybridization in Anoda and Periptera
Natural interspecific hybridization within Anoda apparently is rare.Other than a putative hybrid between A. thurberi and A. crenatijlora, described below, none has been found, although A. cristata is sometimes sympatric with other species.Periptera punicea and P. macrostelis Rose are evidently sympatric in western Jalisco (Fryxell 1974), but no hybrids have been reported, nor are they known between Anoda and Periptera.
In the greenhouse artifical hybrids between diploid and tetraploid A. cristata and between A. thurberi and P. punicea have been made and are summarized in Table 2. Attempted hybridizations between A. cristata, at diploid and tetraploid levels, and A. crenatijlora and A. palmata produced no viable seed.
Self-compatibility in Anoda and Periptera also affects hybridization potential.Greenhouse-grown plants of A. crenatijlora, A. cristata, A. palmata, and A .thurberi, and those of P. punicea consistently set fruit following self-pollinations.Anoda thurberi proved to be obligately self-pollinated due to the fact that the stigmas did not extend beyond the dehiscing anthers.Small-flowered forms of A. crenatijlora were functionally cleistogamous with pollination taking place before the corolla opened.
Meiosis in this plant exhibited little pairing.In most cells only one or two bivalents were formed , although in one instance six bivalents were observed.The remaining chromosomes appeared as univalents.Pollen formation was low and anther dehiscence rare.* Parentage assumed, male and female contributors not known.
In A .cristata meiotic analyses of two hybrids resulting from crosses between diploid plants (Fryxell 561 & Fryxell, Bates & Blanchard 1647) and tetraploid plants (Fryxell, Bates & Blanchard 1611), showed rather different chromosome pairing.In one hybrid (72-379) up to 15 bivalents were formed, although as few as 12 were found in some cells.The remaining chromosomes were unassociated.In the other hybrid (72-377) maximum pairing was less, with as many as ten pairs found in only one cell.Although stainability was high, pollen of both hybrids varied markedly in size, presumably a reflection of their different chromosome numbers.
Chromosome pairing in F I hybrids between P. punicea and A. thurberi tended to be complete.Generally 13 bivalents were formed.Univalents, if present, were no more than four and seemed attributable to early disjunction.An expected extra chromosome in plants (73-163) derived from the cross involvingA.thurberi with n = 14 (Fryxell & Bates 2164) did not appear, although an extra chromosome was present in one F, (73-162) in which Fryxell, Bates & Blanchard 1694 was the A. thurberi parent.Despite pairing that is in accord with that observed in nonhybrid species, pollen fertility in two F,'s was measured at 65 and 68 percent.In a sample of five F 2 's, derived from selfed F,'s, pollen fertility ranged from 87 to 98 percent.

DISCUSSION AND CONCLUSIONS
The discussion is organized around three themes: 1) the relationships of Anoda to other genera ofMalveae, 2) evolution in Anoda, and 3) the relationship between Anoda and Periptera.

Relationships of Anoda
The morphology of Anoda places it among the abutiloid genera of the tribe Malveae, but none of these genera show particular morphological affinities to it (Fryxell 1987), and only Bakeridesia Hochr.(Bates 1973) and Bastardiastrum (Rose) Bates (Bates 1978) share with it a gametic number of n = 15.
In his systematic treatment Fryxell (1987) placed the 23 species of A noda in six sections.Chromosome numbers are now known for ten species, distributed through five of the six sections.The occurrence of species with a chromosome number of n = 15 in each of the five sections argues that this number is basic in the genus.Since base chromosome numbers of abutiloid genera are .x= 6, 7, and 8, n = 15 in Anoda apparently was derived secondarily.One possibility would be direct formation as an allotetraploid combining base numbers of seven and eight.Other possibilities are found in derivation from a stabilized polyploid ancestor with a gametic number of n = 14, 15, or 16.The origin and relationships of Anoda cannot be determined more precisely from existing data.

Evolution in Anoda
Speciation in Anoda has occurred primarily at the diploid level.Direct evidence concerning interspecific genetic isolation among the n = 15 diploids is limited, but does reveal genic/chromosomal barriers between A. cristata and both A. palmata and A. crenatif/ora and between A. thurberi and A .crenatif/ora.Similar barriers may exist between other species or species groups of the genus.
The tetraploid A. crenatif/ora ranges through Mexico to Texas and Arizona in the north.As a member of the relatively homogeneous section Cleistanoda A. Gray, either allo-or autoploid origin is possible.However, if the species is an allopolyploid, the identity of its diploid progenitors, if extant, is not obvious.
Anoda cristata is the most widely distributed and morphologically variable species of the genus.In view of the sample size, statements concerning morphological differentiation among the three ploidy levels are premature, although there are suggestions of certain trends.For example, diploids have mericarps in which the endocarp is absent, is present as a partial, dorsally reticulate hood, or encloses the seed in a reticulate covering.In tetraploids the endocarp is absent or present as partial hood; while in hexaploids the endocarp appears only as a partial hood.Further, there appears to be altitudinal distribution of ploidy levels.Tetraploids were collected from near sea level to about 460 meters, hexaploids between 675 and 1230 meters, and diploids between 1385 and 1850 meters.
When viewed broadly, however, the similar patterns of morphological variation found at each ploidy level in A. cristata and the closely related diploid A. z uccagnii suggest the complex is autoploid, as has been proposed for other Malveae.For example, Bates, Dorr and Blanchard (in press) have speculated autoploid origins for tetraploid Callirhoe involucrata (T.& G.) A. Gray and hexaploid C. papaver (Cav.) A. Gray.Arguing against autoploidy in A .cristata is the strong preferential pairing observed at all ploidy levels and in one of the diploid/tetraploid hybrids, the other being more equivocal.De Wet and Harlan (1972), however, have pointed out the rapidity with which autoploids can establish normal pairing, and Jackson (1982) raises questions concerning traditional interpretations of polyploidy based on the kinds of pairing data available in this study.Whatever the extent of genomic differentiation in A. cristata, it is probably genic in character and represents little structural chromosomal differentiation.column of Periptera.F 1 hybrids between P. punicea and A. thurberi essentially duplicate the subequal corolla and staminal column, and F/s segregate for a wide variety of floral types.
In an evolutionary context the derivation of A. thurberi and Periptera from section Liberanoda or its progenitors could have occurred as a single event or as independent events.The high level of chromosome pairing and relatively high levels of pollen fertility in hybrids between A .thurberi and P. punicea argue against independent events.As a single lineage, chromosomal reduction to n = 13 either has been accomplished stepwise, initially to n = 14, then to n = 13, or with more drastic chromosomal reorganization in a single step.In the latter case those plants of A. thurberi with n = 14 would be derived secondarily and perhaps only as sporadic variants, since the extra chromosome was not carried into hybrids between that form and P. punicea.
Morphology tends to favor reduction to n = 13 in a single step.If n = 14 were to represent the intermediate step at which A. thurberi was differentiated, then subsequent derivation of n = 13 A .thurberi without morphological change, followed by diversification of Periptera, is implied.When one considers the reduced floral parts and autogamous breeding systems of A. thurberi, this seems less probable than derivation of the lineage from an ancestor with a generalized floral form , followed by floral divergence, as it is now represented in A. thurberi and Periptera.The petal and staminal column characteristics of A. pubescens are perhaps representative of those hypothesized to be ancestral.The fact that Fryxell (1974) described collections of A. pubescens as P. grandiflora Fryxell strengthens that contention.
Regardless of sequence and details of change, acceptance of A. thurberi and P. punicea as members of a single lineage and more closely related to each other than to the other species of Anoda has taxonomic implications.Either Periptera should be returned to Anoda, its species placed either in the section Liberanoda or with A .thurberi in a new section, or A. thurberi should be transferred to Periptera.Morphology favors the first alternative; chromosomal discontinuity and presumed genetic isolation the second.

Table I .
Chromosome Counts in Anoda and Periptera.