Centre for Plant Biodiversity Research

Evolution of polyploid agamic complexes with examples from Antennaria (Asteraceae)

Bayer, R.J.

Centre for Plant Biodiversity Research, G.P.O. Box 1600, Canberra, Australia.

Abstract

Polyploidy is an evolutionary phenomenon that is a widespread mode of evolution among perennial flowering plants . It has also long been known that polyploidy is often accompanied by agamospermy, or asexual seed production. When the two phenomena, polyploidy and agamospermy, occur together in a single plant species the result is the polyploid agamic complex. Because polyploid agamic complexes are relatively common among flowering plants, it is important to understand how they evolve. Antennaria (Asteraceae: Gnaphalieae) is useful as a tool for studying polyploid agamic complexes because both phenomena are widespread in this genus. Antennaria are dioecious, perennial herbs that occur throughout temperate and arctic regions of the northern hemisphere, with the major center of diversity in the Rocky Mountains of North America. About 32 species are primarily sexual and diploid (2n = 28). These species are morphologically well-defined and usually have rather restricted ranges that are principally in unglaciated regions. The five polyploid complexes differ considerably from one another with respect to ploidy levels, morphological variation and reproductive modes. The ploidy levels in Antennaria range from tetraploid (2n = 56) to dodecaploid (2n = 168). The polyploid complexes generally have wider geographic ranges and possibly greater ecological amplitudes than the diploids. The polyploid complexes are the result of multiple hybrid origin from among the sexual diploid species. The agamic complexes may be morphologically confluent with one another because they share sexual progenitors. The diploids give rise to polyploids through the processes of interspecific hybridization and backcrossing among themselves and probably through hybridization with facultatively sexual apomicts. The triploid/tetraploid gametophytic apomict, A. rosea is one of the most thoroughly examined complexes in the genus and is used to illustrate the evolution of a typical Antennaria polyploid agamic complex.

Introduction

Investigations into the evolution of polyploid agamic complexes represent considerable challenges to evolutionary botanists. Groups such as Antennaria, Bouteloua, Crataegus, Crepis, Hieracium, Poa, Potentilla, Rubus, Taraxacum and Townsendia are well known for their taxonomic complexity and as a result they have been somewhat ignored by most modern biosystematists. Antennaria Gaertner (Asteraceae: Gnaphalieae) is a genus which has several interesting evolutionary and morphological features such as gametophytic apomixis, polyploidy, and sexual dioecism, making it an suitable tool to use when studying the evolution of polyploid agamic complexes. About 350 species of Antennaria have been described world-wide, but most of these are apomictic microspecies because in the past taxonomic practice has been to give each species status (Bayer & Stebbins 1993). This has led to unwieldy taxonomic classifications, which only a few experts on each group can use (Bayer & Stebbins 1993). Additionally, these types of classifications reveal little about the evolutionary relationships or phylogenies of the agamic complexes. The only way to study the evolution of agamic complexes is through a careful series of investigations including morphology, cytogenetics, phytogeography, ecology, enzyme electrophoresis, DNA sequence data, and RAPDs (random amplified polymorphic DNA). This example from Antennaria demonstrates how some of these data can be integrated to reveal clues to the phylogenetic origin of the polyploid agamic complex.

Evolution of the Typical Polyploid Agamic Complex

Agamic complexes typically include a series of polyploid sexual, facultatively or obigately agamospermous microspecies, which are largely the result of various hybridizations among sexual diploid and sexual polyploid members of the complex. If several diploids have contributed genes to the complex, then an enormous number of microspecies evolve because each microspecies can be of different multiple hybrid origin. The sexual diploid (and sometimes sexual autotetraploid) relatives are distinct morphologically from one another, but have a much greater amount of morphological variation than any single microspecies. The initial steps in the formation of the agamic complex involve hybridization among the sexual diploids which give rise to polyploids at many different ploidy levels. Generally, the polyploids at the lower levels (triploids and tetraploids) reproduce sexually and consequently there is a possibility for gene exchange among them. At higher ploidy levels, where agamospermy is primarily obligate, the microspecies are reproductively isolated from one another. The morphological variation among various obligately agamospermous microspecies is slightly discontinuous because of their reproductive isolation. Some clones tend to merge morphologically into their sexual relatives, especially if they have a significant number of genes from a single sexual taxon. Consequently, this obscures the morphological distinctness of the sexual diploids themselves. Perhaps genes from the sexual diploids can continue to be relocated, via introgressive gene flow, into the agamic complex through sexual triploid and tetraploid members of the complex. Unbalanced cytotypes such as triploids, pentaploids, septaploids, and various aneuploids can reproduce easily via agamospermy and therefore any hybrid has the potential to become fertile. The resultant polyploid agamic complex is comprised of a myriad clones that appear morphologically to represent a giant hybrid swarm, which is usually of multiple hybrid origin from among several sexual relatives.

The Genus Antennaria

Antennaria is a moderate-sized genus of about 40 recognized species and belongs to the tribe Gnaphalieae of the Asteraceae (Bayer, Soltis, & Soltis, in press). It is a peculiar member of the Gnaphalieae not only because of the predominance of features such as agamospermy, polyploidy and sexual dioecism, but also in its distribution. Most Gnaphalieae occur in southern Africa and Australia (see Anderberg 1989, 1991, for a systematic monographs of the Gnaphalieae), whereas Antennaria is confine largely to the northern hemisphere, especially North America, with several unstudied taxa from the Andes of South America (Cabrera, 1957).

Antennaria are dioecious, herbaceous, usually stoloniferous, perennials with simple, entire basal leaves and ˜ reduced cauline leaves. The flowering stalks are composed of one to many capitula and the pistillate and staminate capitula and florets are sexually dimorphic. The white or more often colored involucral bracts, imbricated in several series, are dry and scarious. The base chromosome number in Antennaria is x = 14 and taxa with 2n = 28 have been considered to be diploids by North American authors (Stebbins 1932a; Bayer & Stebbins 1981; Bayer 1984; Bayer & Stebbins 1987). The genus consists of 33 known sexual diploid/tetraploid species and at least five large polymorphic polyploid agamic complexes (Bayer 1990a).

A morphology-based cladistic analysis of 32 sexual diploid species (Bayer 1990a) provided the first phylogenetic hypothesis for the sexually reproducing species, revealing that the genus comprises five constitutive monophyletic groups, Geyeriae, Argenteae, Dimorphae, Pulcherrimae (= Carpaticae sensu Urbanska; Urbanska (1983a, 1983b)), and Catipes. Three of the groups, Argenteae, Dimorphae, and Geyeriae, retained relatively large numbers of plesiomorphic characters (Bayer 1990a). The members of these three groups are mostly non-stoloniferous and have not evolved characters often associated with sexual dimorphism that occur in the remainder of Antennaria. On the other hand, the Pulcherrimae and Catipes represent the most specialized groups. This cladistic analysis (Bayer 1990a) did not delineate Antennaria as a monophyletic group, but found that Antennaria would be monophyletic only if A. geyeri (Geyeriae) was withdrawn from the genus. Additionally, cpDNA restriction site data (unpublished data) indicated that Antennaria was monophyletic only with the inclusion of Anaphalis within Antennaria. Because the earlier cladistic analysis based on morphology (Bayer 1990a) did not fully resolve the phylogeny of Antennaria a different technique utilizing molecular data was investigated.

The internal transcribed spacer regions (ITS-1 & ITS-2) of nuclear ribosomal DNA were sequenced for 30 sexual diploid species of Antennaria and one species from each of the outgroup genera Anaphalis, Ewartia, Leontopodium, and Pseudognaphalium (Bayer, Soltis, & Soltis, in press). ITS sequences have been very useful in reconstructing the phylogeny of Antennaria and also in defining its relationship to the outgroup taxa (Bayer, Soltis, & Soltis, in press). Phylogenetic analysis indicates that, relative to the outgroups included, Antennaria is a well-supported monophyletic group (Bayer, Soltis, & Soltis, in press). Based on the genera surveyed, Leontopodium appears to be the sister genus of Antennaria (Bayer, Soltis, & Soltis, in press). The general topology of the molecular trees agrees with that based on previous morphological analyses (Bayer 1990a) and indicates that Antennaria is composed of six clades of equal rank, corresponding to the traditionally recognized informal groups, the Geyeriae, Argenteae, Dimorphae, Pulcherrimae, Catipes, as well as a new group, the Arcuatae (Fig 1). Phylogenetic analysis of ITS sequences (Bayer, Soltis, & Soltis, in press) suggested the placement of A. arcuata in its own group, hence six monophyletic clades are recognized instead of the five that were evident in the analysis of morphology (Bayer 1990a).

Fig. 1. Evolutionary and phylogenetic trends within the genus Antennaria are superimposed on a reduced cladistic analysis based on morphology and DNA sequences from the internal transcribed spacer regions of nuclear ribosomal DNA (ITS). The monotypic groups Geyeriae and Arcuateae are viewed as the most unspecialized groups within the genus are relatively primitive members of the genus. On the other hand the speciose Catipes group is regarded as being the most specialized lineage and it is from this group of sexual diploids that all the polyploid pillar complexes within the genus have been derived.

Phylogenetic trees based on morphology (Bayer 1990a) and ITS sequences (Bayer, Soltis, & Soltis) can be reduced to form a consensus of the six major lineages within the genus (Fig 1). Evolutionary and phytogeographical features can be superimposed on the tree, which show interesting trends within the genus (Fig 1). The least specialized groups, the Geyeriae, Arcuateae, and Argenteae all occur in arid habitats at low elevations in the western United States (Fig 1). They are all strictly diploid amphimicts in which sexual dimorphism is poorly developed. The Dimorphae and Pulcherrimae groups show moderate degrees of specialization, in that they are strictly amphimictic diploids to tetraploids and sexual dimorphism is well-developed (Fig 1). They have expanded their ranges beyond the western United States to other parts of the New and Old Worlds (Fig 1). The Catipes groups consists of sexual diploids and tetraploids and derived from them are all of the polyploid agamic complexes in the genus (Fig 1). Most of the species of the Catipes group have aggressive horizontal stolons, an effective means of asexual reproduction, and sexual dimorphism is well developed. Amphimixis, apomixis (agamospermy), and the very high levels of polyploidy (up to dodecaploid; Bayer & Minish 1993) are prevalent among polyploid pillar complexes that are derivatives of the sexual diploid taxa from the Catipes lineage (Fig 1). Five polyploid agamic complexes, A. alpina (L.) Gaertn., A. howellii E. L. Greene, A. parlinii Fern., A. parvifolia Nutt., and A. rosea E. L. Greene, have evolved via multiple hybridization from among 16 of the 17 members of the Catipes group (Bayer 1987a; Bayer, Soltis, & Soltis, in press). The great success of the Catipes group seems to be correlated with their ability to grow in a diversity of habitats (many of them are specialized as edaphic endemics, such as A. virginica Stebbins on Devonian-age shale barrens (Bayer & Stebbins 1987, 1993), A. suffrutescens Greene on serpentine (Bayer & Stebbins 1993), and A. aromatica Evert and A. densifolia A. E. Porsild on limestone talus (Bayer 1989a)) throughout their range from Great Britain across Eurasia and North America to Tierra del Fuego. It is also due to their acquisition of characters such as strong sexual dimorphism, aggressive vegetative reproduction (stolons), polyploidy, and agamospermy (Fig 1).

Clearly, the only way to comprehend the perplexing patterns of variation found in the agamic complex is to study the sexual progenitors of the complex. Once morphological, ecological, and genetic variation in sexual species is delimited then that of the polyploids can be better understood and explained. This is the approach that I have taken in my work with all the polyploid complexes in Antennaria. We shall now focus in on the multiple hybrid origin of one of these polyploid agamic complexes, the A. rosea complex, which is derived from sexual species of the Catipes groups.

Evolution of the Antennaria rosea Polyploid Agamic Complex

Antennaria rosea is one of the most morphologically diverse polyploid complexes in Antennaria (Bayer 1990b). The center of diversity for A. rosea is the Rocky Mountains of western North America and its range stretches from New Mexico and southern California, north to Alaska and the Northwest Territories, and east through Alberta, Saskatchewan, to the western shores of Hudson Bay (Bayer 1989b). It also occurs as disjunct populations around the shore of James Bay, the north shore of Lake Superior, the Gaspé Peninsula in eastern Quebec, and in western Newfoundland. It occurs in a broad scope of habitats from dry sagebrush steppe to moist alpine tundra (Bayer 1990b).

The species has triploid (2n = 42), tetraploid (2n = 56), and pentaploid (2n = 70) cytotypes, but the tetraploids are by far the most common (Bayer & Stebbins 1987). Antennaria rosea is a gametophytic apomictic, and because the genus is dioecious, populations of A. rosea consist entirely of pistillate clones (Bayer 1987a; Bayer & Stebbins 1987). Taxonomically, I have treated this compilospecies as a large polymorphic species, A. rosea s.l., because it is unrealistic to try to recognize all the microspecies that delineate the complex (Bayer 1989c).

I have hypothesized that the variation in the A. rosea complex is derived from several sexually reproducing species of Antennaria primarily from the southern, unglaciated, portions of the western cordillera (Bayer 1990b). Antennaria aromatica, A. corymbosa E. Nelson, ^åA. pulchella E. Greene, A. microphylla Rydb., A. racemosa Hook., and A. umbrinella Rydb. are sexual species that appear to be implicated in the parentage of A. rosea (Fig 2). Additionally, A. marginata E. Greene, and A. rosulata Rydb. may have contributed to the genetic composition of the A. rosea complex (Fig 2). Antennaria aromatica, A. corymbosa, A. racemosa, and A. umbrinella occur primarily in the unglaciated portions of the northern Rocky Mountains, whereas A. marginata and A. rosulata occur in the southern Rocky Mountains, mainly north of Mexico. Two of the sexual species, A. microphylla and A. pulchella/media, are more widespread throughout the western Cordillera from the southern portions in New Mexico and California to the Yukon. The ranges of the sexual species, for the most part, overlap completely with that of A. rosea. Populations that were surveyed electrophoretically along with maps outlining the distributions of these sexual species can be found in Bayer (1988).

Fig. 2. Reticulate evolution in Antennaria. Putative relationship of the Antennaria rosea complex to its eight sexual, primarily diploid, progenitors, A. aromatica, A. corymbosa, A. marginata, A. microphylla, A. pulchella (tetraploids = A. media), A. racemosa, A. rosulata, and A. umbrinella. Phylogenetic relationships of the sexual diploid species themselves is shown.

Morphological Evidence:

It was obvious to me, based on casual herbarium and field observations, that suites of morphological characters that are unique to a particular sexual species often occur in different combinations in the segregates of A. rosea (Bayer 1990b). Consequently, A. rosea is a compilospecies, having arisen through multiple hybridization and introgression from among several sexual species. In order to test this hypothesis morphological characters were measured on a set of representative specimens of A. rosea and its eight putative diploid progenitors (Bayer 1990b). Various ordination and clustering techniques (outlined in Bayer 1990b) were performed and a redraw version of one of these is shown in Fig 3. The phenetic analyses (Bayer 1990b) indicated that A. aromatica, A. corymbosa, A. microphylla, A. pulchella/media, and A. umbrinella are the major sexual progenitors of the A. rosea complex (Fig 3). Some white-phyllaried segregates of A. rosea bear a very close resemblance to A. microphylla and populations of these A. rosea clones can only be distinguished from A. microphylla in that they are more robust and are entirely pistillate (Bayer 1990b). Incidentally, the rose phyllary color in some A. rosea clones is not prevalent in any of the sexual taxa, but does occur sporadically in A. marginata, A. microphylla, and A. umbrinella and presumably was inherited from these sexual taxa by A. rosea (Bayer 1990b). Very few clones of A. rosea display morphological characteristics that can be attributed to A. marginata or A. rosulata (Fig 3), providing only weak evidence that may have contributed to the genetic composition of A. rosea, (Bayer 1990b). The likely sexual progenitors of the A. parvifolia polyploid complex are Antennaria marginata and A. rosulata. (Fig 3; Bayer 1987a).

Fig. 3. Three dimensional principal components analysis (PCA) composed of 55 OTUs of Antennaria. Factors 1 to 3 are labelled with arrows. The species composition of the OTUs is one centroid (Stars with specific epithet symbols inside) each of Antennaria aromatica (open, downpointing triangles), A. corymbosa (open circles), A. marginata (closed, downpointing triangles), A. microphylla (closed circles), A. pulchella/media (closed squares), A. rosulata (closed diamonds), and A. umbrinella (open squares) and 47 OTUs of A. rosea (open diamonds). Antennaria rosea OTUs are labelled with the specific epithet symbol of the species to which it has the highest similarity as indicated by Pearson product-moment correlation coefficients. The centroids are labelled as follows: A. aromatica (ARO), A. corymbosa (COR), A. marginata (MAR), A. microphylla (MIC), A. pulchella/media (PUL), A. racemosa (RAC), A. rosea (ROS), A. rosulata (ROZ), and A. umbrinella (UMB). This three-dimensional projection was obtained by rotation of the PCA by the Mac-Spin program (D² Software, Austin, Texas). Diagram redrawn from Fig 2 of Bayer (1990b).

Antennaria racemosa is a sexual diploid that is morphologically very different from the other sexual species (Bayer 1987b). Consequently, when included in a principal components analysis (PCA) it caused all the operational taxonomic units (OTUs) of the other species to mass together, therefore could not be incorporated into the formal analysis (See Bayer 1987b, pg. 2391, Fig 1). Evidence from observations in the field and herbarium indicate that some segregates of A. rosea have morphological features that indicate A. racemosa may have been a diploid progenitor (Bayer 1990b). Antennaria racemosa has large, glabrous basal leaves, an open, racemose arrangement of heads, and purple stalked glands on its cauline leaves and peduncles (Bayer 1987b). These characters are seen in A. rosea segregates that have often been identified as the microspecies A. alborosea A. E. Porsild. No other amphimictic species of Antennaria has this unique combination of characters, so they must be attributed to parentage involving A. racemosa (Bayer 1990b). In summary, morphometric analyses have shown that six sexually reproducing progenitors, A. aromatica, A. corymbosa, A. microphylla, A. pulchella/media, A. racemosa, and A. umbrinella are primary source of morphological variability found in the A. rosea complex (Fig 3). Additionally, two other sexually reproducing species, A. marginata and A. rosulata, may have donated to the genetic complexity of the A. rosea complex (Fig 3).

The origin of A. rosea is confounded as a consequence of the overwhelming morphological complexity of the group. Morphological characters that are unique to each of the sexual progenitor species can be detected in various combinations in the A. rosea segregates (Bayer 1990b). Based on morphology, the A. rosea polyploid agamic complex is a typical compilospecies, i.e. it is the result of hybridization from among several sexually reproducing diploids and tetraploids.

Isozyme Evidence:

Isozymes have been shown to be useful not only in looking at divergence among populations of plants, but also in looking at the origins of diploids and polyploids of hybrid origin (Crawford 1983, 1985). With regard to the A. rosea complex isozymes have been used to look at divergence among the diploid progenitors of the complex (Refs.), clonal diversity in A. rosea (Bayer 1990c; Bayer 1991) and the origins of the A. rosea complex (Bayer 1989b).

Morphological differences among the sexual species, such as those in Antennaria, may be due to relatively few genes, as has been suggested for other groups by Gottlieb (1984) in his review of the genetics of morphological differences. The sexual species of Antennaria follow a pattern that is comparable to that occurring in other genera such as Quercus (Manos & Fairbrothers 1987), and Heuchera (Saxifragaceae; Soltis 1985) in which only moderate amounts of divergence have occurred at the enzyme loci, whereas relatively large amounts of morphological divergence has taken place among the species.

As would be expected for intraspecific comparisons, average values of genetic identities, I, (average I = 0.944; range = 0.802 - 0.997) for all pairwise comparisons of 33 populations of A. rosea indicate that the populations within the species have diverged only slightly (Bayer 1989b). The cluster analysis (UPGMA) of the genetic distance matrix of 33 A. rosea populations and eight populations representing taxa that are related to A. rosea (Fig. 4) effectively portrays the relationship of the populations and species (Bayer 1989b). Most of the A. rosea populations are most similar to populations of A. corymbosa, A. microphylla, A. pulchella/media and A. umbrinella, whereas fewer populations are similar to A. aromatica, A. marginata, and A. rosulata (Fig. 4; Bayer 1989b). The segregates of A. rosea should contain alleles that have their origins in the sexual taxa (Fig 2), however only A. corymbosa, A. microphylla, and A. umbrinella have alleles that are unique (Bayer 1989b). Consequently, isozyme data can only be used as independent evidence to support the hypothesis that A. corymbosa, A. microphylla, and A. umbrinella are among the sexual progenitors of A. rosea (Bayer 1989b). Values of genetic identity (I) indicate that although the remaining species, A. aromatica, A. marginata, A. pulchella/media and A. rosulata, are genetically very similar to A. rosea segregates, they can be neither included nor excluded as ancestors of A. rosea based solely on isozyme data (Bayer 1989b). Similar results, where unique alleles of a few sexual progenitors could be used as independent evidence to substantiate the genomic composition of polyploids, were obtained (Bayer & Crawford 1986) for the A. parlinii and A. howellii (previously A. neodioica Greene) complexes.

Fig. 4. Distance phenogram (UPGMA) derived from Nei's genetic distances of all pairwise comparisons of 33 populations of Antennaria rosea with average identity values for seven sexual species of Antennaria, including A. aromatica (ARO), A. corymbosa (COR), A. marginata (MAR), A. microphylla (MIC), A. pulchella/media (PUL), A. racemosa (RAC), A. rosulata (ROZ), and A. umbrinella (UMR). Population designations for the A. rosea (ROS) populations are the same as those cited in Tab 1 of Bayer 1989b. Cophenetic correlation coefficient is 0.7946. Diagram redrawn from Fig 2 of Bayer (1989b).

Antennaria rosea probably arose fairly recently in the Rocky Mountains of the western United States because this is the region where it is still sympatric with its diploid progenitors (Bayer 1989b). This is also supported by the fact that the only novel allele (Pgm-1^a) found in A. rosea has not yet detected in the probable sexual progenitors (Bayer 1989b). It would be likely that if sufficient time would have elapsed such that more unique alleles would be found in the A. rosea complex, then the polyploids would be relatively old (Bayer 1989b). Similar results have been found in other relatively youthful polyploids, as Tragopogon (Roose & Gottlieb 1976).Edited for sentence changes up to here.....

In additional studies of the A. rosea complex (Bayer 1990c; Bayer 1991) the clonal diversity in populations within the complex were studied to look at the distribution and number of clones in those populations. Clonal diversity among 63 populations of A. rosea was studied over a large portion of its range (Bayer 1990c). Isozyme electrophoresis utilizing four polymorphic enzyme systems detected 192 multilocus genotypes among the populations (Bayer 1990c). Populations of A. rosea tend to be composed of one or a few genotypes (range 1 - 11; mean 3.5), and these genotypes usually occur in only one or a few localized populations (Bayer 1990c). Geographic patterns of clonal diversity may be a result of frequent genesis of new clones in populations that occur in areas where sexual relatives of A. rosea donate compatible pollen to facultatively sexual apomicts. Populations from previously glaciated regions tend to have fewer clones per population than those from unglaciated portions of the range. However, the evolution of new apomictic clones in A. rosea is paradoxical. If A. rosea is obligately agamospermous, then how do new clones arise? Perhaps some clones of A. rosea occasionally produce sexual embryo sacs (facultative sexuality) and these are subsequently fertilized by compatible pollen from nearby sexual progenitors (Bayer 1990c). As mentioned above, unique marker allozymes are supportive of the fact that A. corymbosa, A. microphylla, and A. umbrinella are among the parents of the A. rosea complex (Bayer 1989b). Fourteen of the 17 populations of A. rosea containing clones that have these unique alleles are from sites that overlap the range of the particular sexual species containing the same marker alleles, which further strengthens this argument (Bayer 1990c). Geographic patterns of clonal diversity in A. rosea may be a result of more frequent genesis of new clones in populations that co-occur in areas where sexual relatives of A. rosea contribute large amounts of compatible pollen to facultatively sexual clones (Bayer 1990c). Clausen's "Henry Ford" or "model T" hypothesis (Clausen 1954; Marshall & Brown 1981) concerning the adaptive significance of apomixis may apply to the case of A. rosea, where facultative sexuality continually supplies the necessary genotypes to be tested, but only the most successful ones are "reproduced authentically" and continuously via apomixis. Facultative amphimicts may play a very important role in the maintenance of clonal diversity in the A. rosea complex (Bayer 1990c).

This study (Bayer 1990c) also showed that clonal diversity declines with distance from the center of distribution. Lower numbers of apomicts in populations at the margins of the ranges could have several possible causes. First, these populations could be more recently established. Secondly, populations near the center of the range, where the largest numbers of sexual species occur, could be more clonally diverse because new apomicts arise more frequently as a result of facultative sexuality in some clones of A. rosea (Bayer 1990c). This might also account for the sporadic occurrence of staminate A. rosea plants near the center of the range (Bayer, unpubl. obs.). Also, pistillate triploid A. rosea occurs mostly near the center of the range of the species (Bayer 1987a; Bayer & Stebbins 1987; Bayer 1990c). Finally, even if rates of genesis of new apomicts are the same, populations in the center of the range could be more diverse if selection in these regions is less intense. Consequently, stronger directional selection in marginal sites along with balanced selection in central sites perhaps accounts for the observed patterns of clonal diversity.

These result of these studies (Bayer 1990c; Bayer 1991) also support morphological data (Bayer 1990b) indicating that A. rosea is a species complex composed of numerous apomicts. It is difficult to account for this clonal diversity because A. rosea is a gametophytic apomict and almost entirely pistillate. If A. rosea is an obligate apomict, then the diversity could only arise from mutations or from many independent origins. It seems unlikely that mutations could account for the large numbers of clones that have been encountered, because a previous study has demonstrated that only one Pgm-1 allele occurs in A. rosea that has not also been detected in its sexual progenitors (Bayer 1989b). If A. rosea is facultatively sexual, then when it comes in contact with compatible pollen donors, new apomicts might arise through sexual reproduction.

In summary, isozymes were useful in documenting the origins of the A. rosea complex from among several diploids, including A. corymbosa, A. microphylla, and A. umbrinella (Fig 2), but could neither confirm nor discount the others as ancestors of the complex. It is likely that A. rosea is a facultative apomict and that new clones are being continually introduced into the environment by hybridization back to their sexual progenitors. These new apomictic clones are then tested by natural selection.

Ecological Evidence:

Based on informal field observations, each of the eight sexual progenitor species that may have given rise to the A. rosea complex appears to be associated with rather specific habitats (Bayer 1987a). For example, in west-central Montana up to six species, A. aromatica, A. corymbosa, A. media, A. microphylla, A. racemosa, and A. umbrinella, can occur on a single mountain (Fig 5), but each appears to be associated with a different community (Bayer 1987a). Antennaria microphylla typically grows at the lower elevations along river terraces with sagebrush, whereas A. umbrinella is usually found at drier sites with sagebrush or Ponderosa pine (Fig 5; Bayer 1987a; Bayer, Purdy, & Lebedyk 1991). The montane zone in the understorey of moist coniferous forests is the typical habitat of A. racemosa (Fig 5; Bayer 1987a; 1990b). Both A. aromatica and A. corymbosa occur at timberline and marginally in the alpine, but the former is associated with limestone talus, while the latter is found in moist Salix thickets (Fig 5; Bayer 1987a; 1990b). Antennaria media is found on dry, alpine tundra (Fig 5; Bayer 1990d; Bayer, Ritland, & Purdy 1990). In the southern Rockies, Antennaria marginata and A. rosulata also appear to occupy distinct habitats (Fig 5; Bayer, Purdy & Lebedyk 1991). Sexual populations appear to be reproductively isolated by ecological mechanisms and as a result when two species grow in close proximity hybrids often occur (Bayer 1987a; Bayer, Purdy & Lebedyk 1991).

Fig. 5. Mountain slopes in the southern and northern Rocky Mountains of western North America. The boundary between the northern and southern Rockies is the Red Desert in southwestern Wyoming, U.S.A. On the slopes, alpine above treeline is gradually replaced by subalpine krummholz, followed by montane coniferous forests, and finally by pine savannah and sagebrush steppe at the lowest elevations. The relative elevational / ecological distribution of Antennaria rosea and each of its eight sexual, primarily diploid, progenitors, A. aromatica, A. corymbosa, A. marginata, A. microphylla, A. pulchella (tetraploids = A. media), A. racemosa, A. rosulata, and A. umbrinella, is indicated on the slopes. Species are labelled with their specific epithets.

Antennaria rosea appears to occupy a wide variety of habitats from dry steppe to alpine tundra (Fig 5; Bayer 1990a; Bayer, Purdy & Lebedyk 1991). It is often found sympatric with or in parapatric populations near its sexual relatives (Bayer, Purdy & Lebedyk 1991). Individual clones of A. rosea often display suites of morphological characters that are indications of their parentage, as these same features are characteristic of certain sexual taxa (Bayer 1990a; Bayer, Purdy & Lebedyk 1991). Casual field observations indicated that clones of A. rosea often resemble the sexual species that occur in habitats that are most similar to their own (Bayer, Purdy & Lebedyk 1991). It is likely that the ecological amplitude of A. rosea has perhaps increased over those of its sexual progenitors because the overall genetic composition of the species is more diverse as a result of heterogeneous gene combinations from many divergent sexual species (Bayer, Purdy & Lebedyk 1991). It was not clear whether A. rosea occupied only habitats that were very similar to those of its sexual progenitors or whether it sometimes occurred in hybrid, intermediate or unique habitats.

As a part of a community, a plant population occupies a niche, an environment composed of both biotic and abiotic elements, both of which must be evaluated to acquire a complete understanding of a species' habitat and niche. One purpose of the ecological study (Bayer, Purdy & Lebedyk 1991) of A. rosea and its progenitors was to investigate and quantify the habitat diversity in A. rosea and its sexual progenitors to determine whether the individual taxa are ecologically distinct. The need for studies of the adaptive significance of polyploidy, using modern statistical analyses of plant community association has been underscored by both Stebbins (1984) and Ehrendorfer (1980) and the ecological study (Bayer, Purdy & Lebedyk 1991) endeavored to contribute to our knowledge on that topic.

In this study (Bayer, Purdy & Lebedyk 1991) environmental and relative abundance of community associate at sites occupied by A. rosea and its putative progenitors were surveyed. Ordination of species and environmental variables from the 101 study sites was carried out using the canonical correspondence analysis (CCA) option of the Canonical Community Ordination (CANOCO) program (ter Braak 1985). This method of analysis allows the environmental variables to be related to species data simultaneously and is becoming the preferred ordination method for the analysis of data used in vegetation classification (ter Braak 1986, 1987, 1988). The CANOCO ordination arranges sites such that those occupying similar habitats appear in close proximity to each other. Each of the sexual species of Antennaria, including A. aromatica, A. corymbosa, A. marginata, A. media/pulchella, A. microphylla, A. racemosa, A. rosulata, and A. umbrinella, appears to occur in a distinct habitat (Bayer, Purdy & Lebedyk 1991). Both community associates (the biotic factors) and environmental variables (the abiotic factors) can be used to predict which Antennaria species will occur in a given habitat and vice-versa (Bayer, Purdy & Lebedyk 1991). In most cases, several variables can be used to accurately and effectively describe the habitat, although many of the species habitats can be confidently distinguished from those of the other Antennaria species by the use of single variables (Bayer, Purdy & Lebedyk 1991). For example, all A. aromatica sites are characterized by having soils with amounts of CaCO₃ that are several orders of magnitude higher than those of any other site (Bayer, Purdy & Lebedyk 1991).

Some sympatric species pairs have similar habitats, although spatially overlapping or conterminous ones are uncommon (Bayer, Purdy & Lebedyk 1991). The CCA indicates that the habitats of the following sympatric species pairs are the most likely to overlap: A. corymbosa - A. microphylla, A. corymbosa - A. media, A. microphylla - A. umbrinella, and A. media - A. umbrinella (Bayer, Purdy & Lebedyk 1991). These occur because very different mountain habitats can often occur side by side (Bayer, Purdy & Lebedyk 1991). For example, mountain stream beds running through a surrounding terrain that is by comparison very dry creates the situation where two very different communities exist juxtaposed (Bayer, Purdy & Lebedyk 1991). For example, both A. microphylla and A. umbrinella occur at similar elevations, but the habitat of A. microphylla is floodplains or moist alkaline depressions, while that of A. umbrinella are the drier sagebrush covered hillsides (Bayer, Purdy & Lebedyk 1991). In ecotonal areas the two species can co-occur such as in the Flint Mountains, Granite Co., Montana (See Bayer et al. MT-736/MT-738 at ALTA) (Bayer, Purdy & Lebedyk 1991). On the other hand, hybridizations are a necessary step in the evolution of polyploid agamic complexes. Interspecific F₁ hybrids are frequently found in communities where two or more sexual species occur sympatrically (Bayer 1987a). Such hybrids are usually easy to identify because they are morphologically intermediate between their parents (Bayer 1987a). However, later generation backcross progeny and introgressant segregates may be very difficult or impossible to distinguish morphologically from their recurrent parent (Bayer 1987a). Later generation hybrids become polyploidized and acquire agamospermy becoming the fully fertile agamic microspecies, which make up the large polyploid agamic complexes such as A. rosea (Bayer 1987a).

Habitat divergence tends to follow morphological divergence among the sexual taxa, which was discussed above (Bayer 1987a). Those taxa that tend to be morphologically most distinct, such as A. racemosa and A. marginata, are also ecologically discrete (Bayer, Purdy & Lebedyk 1991). Morphometrics indicated that A. aromatica, A. media/pulchella, and A. umbrinella have the most similar morphologies and introgressive hybridization is strongly indicated in some populations (Bayer 1987b). Antennaria aromatica however is very distinct ecologically (Fig 5) from A. media and A. umbrinella and the last two also occupy rather discrete habitats (Bayer, Purdy & Lebedyk 1991). Habitat divergence among the species has obviously played an important role in the evolution of these species, as was suggested in Bayer (1987a). Ecological isolation along with geographic isolation are the primary isolating mechanisms that help maintain species integrity and allow the species to continue to diverge (Bayer, Purdy & Lebedyk 1991).

The A. rosea sites occupy ordination space on the diagram (see Bayer, Purdy & Lebedyk (1991) Figs 5-7) that lies in the center of the ordination and overlaps at least parts of those of all the sexual species except A. aromatica and A. racemosa (Bayer, Purdy & Lebedyk 1991). Several sites are however close to some of the A. aromatica sites (Bayer, Purdy & Lebedyk 1991). Many A. rosea sites do not occur in distinct habitats themselves, but instead occupy habitats that are very similar to those of their sexual progenitors (Bayer, Purdy & Lebedyk 1991). Some sites do however fall in between the groups of sites of the sexual species and these habitats could be described as hybrid habitats since they have sets of conditions that are intermediate between or among groups of sexual taxa, as is indicated by CCA (Bayer, Purdy & Lebedyk 1991). None of the A. rosea sites cover the range of abiotic conditions associated with the sexual taxa, i.e. they don't completely overlap the habitat of any of their sexual relatives (Bayer, Purdy & Lebedyk 1991). The CCA supports the hypothesis for the multiple hybrid origin of A. rosea proposed in Bayer (1990a), in that many sites of A. rosea occupy similar habitats to their diploid progenitors, while others seem to occupy hybrid habitats that are intermediate between those of their sexual progenitors (Bayer, Purdy & Lebedyk 1991). The greatest amount of overlap between sites of A. rosea and those of the sexual progenitors occurs between A. rosea and A. microphylla and A. umbrinella and to a lesser degree A. corymbosa, A. marginata, A. media, and A. rosulata (Bayer, Purdy & Lebedyk 1991).

What are the evolutionary implications of habitat divergence in relationship to allopolyploidy? Several authors have pointed out that the pivotal studies that may lead us to learn more about the adaptive significance of polyploidy may be ecological studies of diploids and their polyploid derivatives. A need for the extension of studies of the adaptive significance of polyploidy, using modern statistical analyses of plant community associations, particularly in closely related diploids and polyploids, has been stressed in reviews by both Stebbins (1984) and Ehrendorfer (1980). Fowler & Levin (1984) stress the importance of ecological factors, especially habitat divergence, in establishment and persistence of polyploids. Stebbins (1984) has commended the studies of Johnson & Packer (1965, 1967, 1968) that emphasize the significance of ecological factors in delimiting the relative distribution of diploids and polyploids.

The "secondary contact hypothesis" (Stebbins 1984, 1985) places importance on secondary contact and subsequent hybridization between differentiated diploids to produce new, highly adaptive gene combinations which are subsequently maintained through polyploidy (and in the case of A. rosea, apomixis). Stebbins (1984) predicted that this secondary contact might happen most often during glacial episodes when populations of well differentiated diploids would be brought together as the result of the shifting of communities because of changing climatic conditions. In the A. rosea complex the situation is more complicated than in many other polyploid complexes because several progenitor species, including A. aromatica, A. marginata, A. pulchella, and A. umbrinella, have sexual autopolyploid cytotypes in addition to sexual diploids. Hybridization among all the cytotypes has led to an enormous amount of variation in morphology among the descendent A. rosea clones. The study of the ecology of A rosea and its sexual relatives (Bayer, Purdy & Lebedyk 1991) illustrates that it possible for the Antennaria species to come into contact to hybridize because their habitats overlap sporadically.

Ehrendorfer (1980) has emphasized the fact that in young allopolyploid complexes, the polyploid derivatives generally occupy habitats that are different from their progenitors. Using examples from two genera, Cruciata (Rubiaceae) and Knautia (Dipsacaceae), he contended that in these genera, the diploids are morphologically uniform and ecologically separated (Ehrendorfer 1980). The polyploid derivatives are morphologically diverse and occupy different habitats than their progenitors. He also asserts that the polyploids in these genera do not occur in environments that are more harsh than those occupied by their diploid ancestors, in fact they seem to occupy less extreme environments (Ehrendorfer 1980). Stebbins (1985) has also rejected the hypothesis that polyploids succeed because they have a superior tolerance of harsh ecological or climatic conditions. The results of the ecological study of habitat differentiation among species of Antennaria (Bayer, Purdy & Lebedyk 1991) supports these hypotheses, in that well over half of the A. rosea sites occupy habitats not occupied by their progenitors and also the habitats would seem to be moderate in comparison to those occupied by many of the sites of the progenitors, which seem to occupy more extreme sites with regard to a host of environmental parameters. Levin's conclusion that polyploidy may drive a population into a new "adaptive sphere", and allow it to occupy habitats that are beyond that of its progenitors (Levin 1983) is supported by the results of this investigation. Polyploidy, particularly when accompanied by hybridization between either different species or different ecotypes of the same taxon, is one of the quickest ways for new species or races to become adapted to different ecological circumstances (Stebbins 1984).

Ehrendorfer (1980) has suggested that in established floras polyploidy contributes intermediate "fill-in" taxa that contribute to a more intensive partitioning of areas and eventually leads to a saturation of most available habitats by the progenitor taxa and their polyploid derivatives. This may be the most adaptively significant role that polyploids perform. Antennaria rosea seems to fulfil this prediction in that many of the A. rosea populations exist in habitats that are both unoccupied by and intermediate to those of its progenitors. Populations of A. rosea can therefore exist parapatrically with those if its progenitors. The idea that habitat differentiation increases the probability of the coexistence of diploids and their polyploid derivatives (Fowler & Levin 1984) is therefore supported by this study. It would be interesting to investigate other polyploid agamic complexes in Antennaria, such as A. alpina, for similar patterns of habitat differentiation among related amphimictic taxa and their polyploid derivatives.

Conclusions

The most perplexing problem when considering the taxonomy of the polyploid agamic complexes in Antennaria is their apparently limitless amount of morphological variation, compared to that found in the sexual diploid species. In addition to variation in morphological characters, there is considerable variation in ploidy levels and mode of reproduction. If the basis for this variation is determined, then the patterns of variation become comprehensible. The source of the morphological variation in the polyploids resides with their sexual diploid and tetraploid ancestors (Fig 2). Consequently, the sexual species must first be delimited and studied before the variation within the polyploid agamic complexes can be explained and circumscribed.

Antennaria rosea is a model system for studying the evolution of polyploid agamic complexes and it closely follows the outline of the evolution of the typical complex that I presented at the beginning of this paper. Much is known about various aspects of the complex from its origin to the genetic structure of its populations, but much is still to be uncovered. All of the analyses support the relationship of the polyploid A. rosea to the diploids A. aromatica, A. corymbosa, A. marginata, A. microphylla, A. pulchella, A. racemosa, A. rosulata, and A. umbrinella. From a genetic and morphological point of view, A. rosea can be described as those segregates that are primarily tetraploid, contain genes from A. aromatica, A. corymbosa, A. marginata, A. microphylla, A. pulchella/media, A. racemosa, A. rosulata, and A. umbrinella, and have morphological characteristics that can be attributed to one or more of the eight diploids (Fig 3). The A. rosea agamic complex can be considered to be in a mature stage of development because the sexual diploids are still extant, sexual reproduction is evidently still prevalent among many of the facultative apomicts, and there are a large number of widely distributed agamospecies.

Acknowledgments ¡ This research was supported by grants from the Natural Sciences and Engineering Research Council of Canada (NSERC). Many persons have contributed in various ways to my work on Antennaria summarized in this article, I would like to thank Glenn Bilodeau, Ilse Breitwieser, Joe Bruner, Dan Crawford, Rob DeLuca, Mike Dillon, Shawn Francis, Mark Joncas, Bengt Jonsell, Ron Hartman, Klaus Lackchewitz, Dan Lebedyk, Pete Lesica, Shunguo Liu, LuDean Marvin, Helen Michaels, Travis Minish, David Murray, Brett Purdy, Doug & Pam Soltis, Rob Soreng, Ledyard Stebbins, Josephine Ward & Krystyna Urbanska for their contributions.

References

Anderberg, A. A. 1989. Phylogeny and reclassification of the tribe Inuleae (Asteraceae). Canad. J. Bot. 67: 2277-2296.

----- 1991. Taxonomy and phylogeny of the tribe Gnaphalieae (Asteraceae). Opera Bot. 104: 1-195.

Bayer, R. J. & Stebbins, G. L. 1981. Chromosome numbers of North American species of Antennaria Gaertner (Asteraceae; Inuleae). Amer. J. Bot. 68: 1342-49.

----- 1984. Chromosome numbers and taxonomic notes for North American species of Antennaria (Asteraceae: Inuleae). Syst. Bot. 9: 74-83.

----- & Crawford, D. J. 1986. Allozyme divergence among five diploid species of Antennaria and their allopolyploid derivatives. Amer. J. Bot. 73: 287-296.

------ & Stebbins, G. L. 1987. Chromosome numbers, patterns of distribution, and apomixis in Antennaria (Asteraceae: Inuleae). Syst. Bot. 12: 305-319.

------ 1987a. Evolution and phylogenetic relationships of the Antennaria polyploid agamic complexes. Biol. Zentral. 106: 683-698.

------ 1987b. Morphometric analysis of western North American Antennaria Gaertner (Asteraceae: Inuleae). I. Sexual species of sections Alpinae, Dioicae and Plantaginifoliae. Canad. J. Bot. 65: 2389-2395.

------ 1988. Patterns of isozyme variation in western North American Antennaria (Asteraceae: Inuleae). I. Sexual species of section Dioicae. Syst. Bot. 13: 525-537.

------ 1989a. A systematic and phytogeographic study of Antennaria aromatica and A. densifolia (Asteraceae: Inuleae) in the western North American cordillera. ¡ Madroño 36: 248-259.

------ 1989b. Patterns of isozyme variation in the Antennaria rosea (Asteraceae: Inuleae) polyploid agamic complex. Syst. Bot. 14: 389-397.

------ 1989c. A taxonomic revision of the Antennaria rosea (Asteraceae: Inuleae: Gnaphaliinae) polyploid complex. ¡ Brittonia 41: 53-60.

------ 1989d. Patterns of isozyme variation in western North American Antennaria (Asteraceae: Inuleae). II. Diploid and polyploid species of section Alpinae. Amer. J. Bot. 76: 679-691.

------ 1990a. A phylogenetic reconstruction of Antennaria Gaertner (Asteraceae: Inuleae). Canad. J. Bot. 68: 1389-1397.

------ 1990b. Investigations into the evolutionary history of the Antennaria rosea (Asteraceae: Inuleae) polyploid complex. Pl. Syst. Evol. 169: 97-110.

------ 1990c. Patterns of clonal diversity in the Antennaria rosea (Asteraceae) polyploid agamic complex. Amer. J. Bot. 77: 1313-1319.

------ 1990d. A systematic study of Antennaria media, A. pulchella, and A. scabra (Asteraceae: Inuleae) of the Sierra Nevada and Inyo Mountains. Madroño 37: 171-183.

------, Ritland, K. & Purdy, B. G. 1990. Evidence of partial apomixis in Antennaria media (Asteraceae: Inuleae) detected by the segregation of genetic markers. Amer. J. Bot. 77: 1078-1083.

------ 1991. Patterns of clonal diversity in geographically marginal populations of Antennaria rosea (Asteraceae: Inuleae) from subarctic Alaska and Yukon Territory. Bot. Gaz. (Crawfordsville) 152: 486-493.

------, Purdy, B. G. & Lebedyk, D. G. 1991. Niche differentiation among eight sexual species of Antennaria Gaertner (Asteraceae: Inuleae) and A. rosea, their allopolyploid derivative. Evol. Trends in Plants. 5: 109-123.

------ & Minish, T. M. 1993. Isozyme variation, ecology and phytogeography of Antennaria soliceps (Asteraceae: Inuleae), a geographically restricted alpine apomict from the Spring Mountains, Nevada. Madroño 40: 75-89.

------ & Stebbins, G. L. 1993. A synopsis with keys for the genus Antennaria (Asteraceae: Inuleae: Gnaphaliinae) for North America. Canad. J. Bot. 71: 1589-1604.

------, Soltis, D. E. & Soltis, P. S. Phylogenetic inferences in Antennaria (Asteraceae: Gnaphalieae) based on sequences from the nuclear ribosomal DNA internal transcribed spacers (ITS). Amer. J. Bot. (In press).

Cabrera, A. L. l957. Una nueva especie del genero Antennaria (Compositae). Notas Mus. De La Plata. l9: 74-79.

Clausen, J. 1954. Partial apomixis as an equilibrium system in evolution. Caryologia Supplement 6: 469-479.

Crawford, D. J. 1983. Phylogenetic and systematic inferences from electrophoretic studies. ¡ In: Tanksley and Orton (eds.), Isozymes in Plant Genetics and Breeding, Part A. Elsevier Publishers, Amsterdam, pp. 257-287.

------ 1985. Electrophoretic data and plant speciation. Syst. Bot. 10: 405-416.

Ehrendorfer, F. 1980. Polyploidy and distribution. In: Lewis, W.H. (ed.), Polyploidy: Biological Relevance. Plenum Press, New York, pp. 45-60.

Fowler, N. L. & Levin, D. A. 1984. Ecological constraints on the establishment of a novel polyploid in competition with its diploid progenitor. Amer. Naturalist 124: 703-711.

Gottlieb, L. D. 1984. Genetics and morphological evolution in plants. Amer. Naturalist 123: 681-709.

Johnson, A. W. & Packer, J. G. 1965. Polyploidy and environment in arctic Alaska. Science 148: 237-239.

------ 1967. Distribution, ecology and cytology of the Ogotoruk Creek flora and the history of Beringia. In: Hopkins, D. M. (ed.), The Bering Land Bridge. Stanford Univ. Press, Stanford, pp. 245-265.

------ 1968. Chromosome numbers in the flora of Ogotoruk Creek, N.W. Alaska. Bot. Not. 121: 403-456.

Levin, D. A. 1983. Polyploidy and novelty in flowering plants. Amer. Naturalist 122: 1-25.

Manos, P. S. & Fairbrothers, D. E. 1987. Allozymic variation in populations of six northeastern American red oaks (Fagaceae: Quercus subg. Erythrobalanus). Syst. Bot. 12: 365-373.

Marshall, D. R., & Brown, A. H. D. 1981. The evolution of apomixis. Heredity 47: 1-15.

Roose, M. L. & Gottlieb, L. D. 1976. Genetic and biochemical consequences of polyploidy in Tragopogon. Evolution 30: 818-830.

Soltis, D. E. 1985. Allozymic differentiation among Heuchera americana, H. parviflora, H. pubescens, and H. villosa (Saxifragaceae). Syst. Bot. 10: 193-198.

Stebbins, G. L. 1932a. Cytology of Antennaria. I. Normal species. Bot. Gaz. (Crawfordsville) 94: 134-151.

------ 1984. Polyploidy and the distribution of the arctic-alpine flora: new evidence and a new approach. Bot. Helv. 94: 1-13.

------ 1985. Polyploidy, hybridization, and the invasion of new habitats. Ann. Missouri. Bot. Gard. 72: 824-832.

ter Braak, C. J. F. 1985. CANOCO - A FORTRAN program for canonical correspondence analysis and detrended correspondence analysis. IWIS-TNO, Wageningen, The Netherlands.

------ 1986. Canonical correspondence analysis: A new eigenvector technique for multivariate direct gradient analysis. Ecology 67: 1167-1179.

------ 1987. The analysis of vegetation-environment relationships by canonical correspondence analysis. Vegetatio 69: 69-77.

------ 1988. CANOCO - an extension of DECORANA to analyze species-environment relationships. Vegetatio 75: 159-160.

Urbanska, K. M. l983a. Antennaria carpatica (Wahlb.) Bl. et Fing. s.l. in North America. I. Chromosome numbers, geographical distribution and ecology. Ber. Geobot. Inst. ETH, Stiftung Rübel 50: 33-66.

------ l983b. Cyto-geographical differentiation in Antennaria carpatica s.l. Bot. Helv. 93: 123-131.

Submitted to: Opera Botanica.

Updated