Printer Friendly

A critical review of concepts and methods used in classical genome analysis.

Even accepting the fundamental premises, genome analysis is burdened by observational difficulties. Hence, chromosome pairing has been shown to be under genetic control and is also influenced by environmental conditions. Additionally, basic biological observations such as the distribution of meiotic configurations or the identity of the individual chromosomes are frequently neglected.

Data from chromosome pairing are captured as pair-wise comparisons and are amenable only to phenetic analysis, and hence are not suited for phylogenetic inferences. As currently perceived, genome analysis may have a role to play in plant breeding but it has no place in systematics. With an increased knowledge and understanding of the mechanism behind meiosis, data useful in a systematic context may eventually be produced.

II. Introduction

The term "genome analysis" has been used to describe a range of different methodologies (see Jauhar, 1996). Applied in its widest sense, genome analysis covers all techniques that describe aspects of the genome. Thus, the term has been applied to karyotype analysis, meiotic chromosome pairing, measurements of DNA content, in situ hybridization, restriction enzyme analysis, genome mapping, and so forth. Applied in a restricted sense, genome analysis is the study of meiotic metaphase I (MI) pairing of chromosomes in hybrids with the purpose of revealing phylogenetic relationships among a group of organisms. The methodology is based on the assumptions that only similar chromosomes pair during meiosis, that similar chromosomes are homologous, and hence, that the extent of chromosome pairing in hybrids reflects the degree of relationship between the parental species (Dewey, 1982: 52). The present paper deals exclusively with genome analysis in this context.

Genome analysis has been widely applied as a phylogenetic tool, especially in groups of plants that include economically important species, because chromosome pairing data also have potential value in plant breeding. In the following text we refer to a number of key papers dealing with genome analysis of various taxa. These references should be considered representative of a much larger number of conceptually similar papers. They have been selected in order to give an overview of the work done within selected taxa and to provide specific examples of how various authors have interpreted the basic concepts and applied the different methodologies of genome analysis.

The most comprehensive applications of genome analysis have undoubtedly been made within the grass tribe Triticeae, primarily in Triticum L./Aegilops L. (Kihara, 1924, 1930; Lilienfeld, 1951; Kimber & Feldman, 1987), Hordeum L. (Bothmer et al., 1983, 1985, 1986, 1987a, 1988a, 1988b, 1989a, 1989b; Bothmer & Jacobsen, 1991), and Elymus L. (Aung & Walton, 1990; Lu, 1993a, 1993b; Lu & Bothmer, 1993a, 1993b; Salomon, 1993; Salomon & Lu, 1994); additional studies have focused on almost all the usually recognized genera of Triticeae (Dewey, 1984; Barkworth, 1992), e.g., Agropyron Gaertn. (Dewey, 1969; Asay & Dewey 1979), Thinopyron A. Love (Jauhar, 1988; Liu & Wang, 1993), Psathyrostachys Nevski (Bothmer et al., 1987b; Baden et al., 1989), Dasypyrum (Cosson & Durieu) Durand (Frederiksen, 1991), Henrardia C. E. Hubb. (Sakamoto, 1972), Eremopyrum (Ledeb.) Jaub. & Spach (Sakamoto, 1991; Frederiksen & Bothmer, 1995), Taeniatherum Nevski (Frederiksen & Bothmer, 1986; Frederiksen, 1994), Heteranthelium Hochst. (Sakamoto, 1974), Elytrigia Desv. (Dewey, 1980; Assadi, 1996), Pseudoroegneria A. Love (Stebbins & Pun, 1953), Leymus Hochst. (Dewey, 1972a, 1972b), Pascopyrum A. Love (Dewey, 1975), and Hordelymus (Jessen) C. O. Harz (Bothmer et al., 1994). Data from genome analysis has influenced the taxonomy and classification of the Triticeae more profoundly than any other taxon (Dewey, 1984; Love, 1984; Barkworth & Dewey, 1985).

Among other grasses, genome analysis has been applied to Avena L. (Nishiyama et al., 1989; Rajhathy, 1991), Oryza L. (Nezu et al., 1960; Nayar, 1973; Lu et al., 1997), Bromus L. (Stebbins, 1947, 1981), Paspalum L. (Burson, 1978, 1981a, 1981b), Pennisetum Rich. (Jauhar, 1968), and Eleusine Gaertn. (Chennaveeraiah & Hiremath, 1974, 1991).

Among dicotyledons, studies using genome analysis have been done for Brassica L. and related genera (U, 1935; Harberd & McArthur, 1980; Mizushima, 1980; Prahash & Chopra, 1991), Gossypium L. (Beasley, 1942; Phillips, 1966; Endrizzi et al., 1984), Hibiscus L. (Menzel & Wilson, 1966; Menzel & Martin, 1970), Arachis L. (Wynne & Halward, 1989), Glycine Willd. (Singh & Hymovitz, 1985a; Singh et al., 1988; Hymovitz et al., 1991), Nicotiana L. (Goodspeed, 1954; Sfikas & Getstel, 1962), Solanum L. (Matsubayashi, 1991), Cucumis L. (Singh & Yadava, 1984), Carthamus L. (Estilai & Knowles, 1978; Kumar, 1991), Helianthus L. (Espinasse et al., 1995), and Mentha L. (Ikeda & Oho, 1991). In some but not all of these taxa, the prevailing ideas about phylogeny have been based on data from genome analysis.

Application of chromosome pairing data to studies of phylogeny is controversial for theoretical, biological, and methodological reasons (Darlington, 1937; Gaul, 1959; Baum et al., 1987; Kellogg, 1989; Seberg, 1989; Petersen & Seberg, 1996). Among the major points of criticism are 1) the lack of relationship between the concept of homology in genome analysis and the "classical" concept of homology in morphology and molecular systematics; 2) the possible influence on chromosome pairing of specific genes, B-chromosomes, the environment, and so forth; 3) the magnitude and pattern of variation of chromosome pairing in hybrids; and 4) the problem of identifying pairing chromosomes during meiosis.

Keeping these problems in mind, together with the extensive use of genome analysis and the wide-ranging conclusions drawn from its results, it is astonishing that no one since Gaul (1959) has thoroughly reviewed the methodology. Even recent treatments of genome analysis (Singh, 1993; Jauhar, 1996) have neglected some of the most important aspects of the critique. The present critical review is an attempt to fill this gap.

Most of our examples and data will be related to the Triticeae, but whenever possible, reference is made to other taxa - mostly plants, but a few references also from animals. We have chosen to focus on the Triticeae for several reasons; first, we have the greatest practical knowledge from this group of plants, but second, and more important, it is, as previously stated, the taxon in which genome analysis has been drawn to its present extreme.

III. Historical Outline of Genome Analysis

The inception of the field of genome analysis can be traced back to early papers of Kihara (1924, 1930). He introduced the term in 1930, describing the methodology precisely: "Ich will nun versuchen, den Gang und die Methoden einer genomanalytischen Untersuchung zu schildern" (Kihara, 1930: 265).

But a germ of the idea was already in existence: "A knowledge of the chromosome behavior would be of value both in the analysis of the origin and relationships of the various species and in the genetic analysis of partially sterile hybrids" (Sax, 1922:114).

Recent outlines of the origin of genome analysis (e.g., Love, 1982; Love & Connor, 1982; Kimber, 1984; Kimber & Yen, 1990; Singh, 1993) erroneously date the "invention" of genome analysis at Rosenberg's (1904a, 1904b, 1909) classic, turn-of-the-century study of Drosera L.

However, Rosenberg, like many of his contemporaries (e.g., Geerts, 1911; Digby, 1912; Pellew & Durham, 1916; Winge, 1917, 1925, 1932), was struggling with related but different problems: What is the explanation of the chromosome configurations seen at meiosis in polyploid hybrids and what is the mechanism behind the evolution of polyploids? The main emphasis of Rosenberg's (1904a, 1904b, 1909) studies was to explain the "origin" of "double" (bivalent) and "single" (univalent) chromosomes observed in meiosis of the hybrid D. longifolia L. x D. rotundifolia L., but without any explicit phylogenetic implications.

In a hybrid between D. rotundifolia (2n = 20) and D. longifolia (2n = 40), Rosenberg (1909) observed 10 unpaired and 20 paired (10 bivalent) chromosomes. He concluded that the unpaired chromosomes originated from D. longifolia and the paired chromosomes were derived from 10 each of the two parental species. He argued that if the paired chromosomes all originated from D. longifolia then one would also expect the 10 D. rotundifolia chromosomes to pair in the hybrid. Thus, Rosenberg was truly among the first to study chromosome pairing in hybrid organisms and to explain that observed paired chromosomes originated from each of two parental species. But Rosenberg (e.g., 1903, 1904a, 1904b, 1909, 1917) concentrated on cytology and never made any statements about relationships between species.

Winge was the first to suggest a hybrid origin of Drosera longifolia: "I could therefore imagine that Drosera longifolia was produced by hybridization between two species or forms with x = 10, the chromosomes of which had become added together in the zygote, and then suffered division. On further crossing with Drosera rotundifolia only half would then be capable of conjugation with those of the latter" (Winge, 1917:113). He went on to add: "We must from this suppose that 10 of D. longifolia's chromosomes are homologous with D. rotundifolia's" (Winge, 1917: 204).

However, it must be emphasized that neither this nor other statements by Winge (1917) touched on the identity of the other diploid ancestor. Nevertheless, Winge's paper (1917) is the first comprehensive treatment of genome evolution in polyploids, and it laid the foundation for later work by Kihara (1924 and on) and others.

The identification of both possible ancestral species of D. longifolia was not made until nearly half a century later by Wood (1955), who based his assumptions on a combination of cytological, morphological, and geographical evidence. Though Kihara (1930) was the first to formalize genome analysis, Federley (1913, 1914, 1915a, 1915b), studying the behaviour of the chromosomes at meiosis in a number of hybrids between species of Lepidoptera, was probably the first to indicate a possible correlation between chromosome pairing ("Chromosomenkonjugation") and taxonomic rank: "Konjugieren alle Chromosomen eines Mischlings, so gehoren die Eltern zweifellos zu derselben Art, konnen dabei aber verschiedene Biotypen, Mutationen oder Aberrationen - um den entomologischen Ausdruck zu gebrauchen - sein. Ist dagegen die Konjugation nur partielle, so haben wit es mit Varietaten, geographischen Rassen, oder nahe verwandten Arten zu tun. Bei vollstandig fehlender Affinitat zwischen den Chromosomen handelt es sich um selbstandige Arten oder grossere systematische Kategorien" (Federley, 1914: 20).

Using a theoretical example involving a tetraploid of unknown origin, Kihara (1930) originally described genome analysis as a two-step procedure. The first step is to decide whether the tetraploid under investigation is an autoploid (autopolyploid; Kihara & Ono, 1926: 475) or an alloploid (allopolyploid; Kihara & Ono, 1926: 475). In cases where only quadrivalents are observed at meiosis, these are considered evidence of autotetraploidy. Otherwise, artificial hybrids between the tetraploid and a number (in Kihara's example, three) of diploid species (p. 266: "die benutzten diploiden Analysatoren") with known genome constitution must be made, followed by an investigation of meiosis in the [F.sub.1]-hybrids. Given that there is no affinity (p. 265 footnote: "keine Affinitaten zwischen verschiedenen Genomen im Falle von Allopolyploidie") between the genomes of the diploids, two possible outcomes of meiosis in the [F.sub.1]-plants are possible; n bivalents and n univalents ([n.sub.II] + [n.sub.I]) or 3n univalents (3[n.sub.I]).

In the latter case, Kihara (1930) regarded it as solid evidence of allotetraploidy (p. 266: "in diesem Falle den sicheren Beweis fur Allotetraploidie"), and believed the genomes were nonhomologous (p. 266: "nichthomologen Genomen zusammengesetzt sein").

In the former case, both possibilities remain open and the decision must rely on indirect evidence. If the same meiotic pattern is shown in more than two of the hybrids with diploids having different genomic constitutions, the polyploid plant is an autotetraploid. Otherwise it becomes necessary to cross the diploids that are supposed to be the direct progenitors of the tetraploid with each other in order to synthesize the tetraploid and obtain evidence of its alloploidy. Apart from the tedium of this method, the best and most simple proof of alloploidy is the occurence of 3n univalents in crosses with different diploids with known genomic constitution: "Auf diesen Wegen kann der Genomanalytiker innerhalb einer Allopolyploidie aufweisenden Gattung oder, wenn wir die Gattungsgrenzen des Systematikers uberschreiten, innerhalb einer alloploide Beziehungen aufweisenden Artsgruppe ihren 'Genomschatz' feststellen" (Kihara, 1930: 267).

The second step in genome analysis is to investigate the relationships (p. 267: "ihre Affinitaten aufzudecken") between the different species-specific genomes. Different degrees of affinity are supposed to exist between genomes that normally behave as nonhomologous and are revealed only when the chromosomes are brought together in a hybrid. Chromosomes belonging to the same genomes that show more or less regular pairing during meiosis are considered "echt homolog" (p. 268; = homologous); chromosomes that show pairing, but only rarely, are considered "semihomolog" (p. 268; = homoeologous [Huskins, 1932:113]). Homologous genomes are genomes that are similar (p. 268) "in bez. auf die Anordnung und Beschaffenheit der Loci." Lilienfeld (1951) made Kihara's definitions of the various types of homology more precise: "Two genomes are strictly homologous, if the pairing chromosomes are identical as to the linear gene arrangement and freely interchangeable which garantees genomically conditioned full fertility. Two genomes are semihomologous, if all or a part of their pairing chromosomes have only similar segments in common" (Lilienfeld, 1951: 102). The affinity of the genomes is assessed thusly: "Das Mass der Affinitat zwischen zwei gegebenen Genomen wird besten dutch die Zahl der Bindungen (zu Bivalenten bzw. Trivalenten) die zwischen ihnen gebildet werden (im Verhaltmis zu der erwarteten Zahl von Bivalenten bei Annahme vollstandiger Homologie der betreffenden Genome ausgedruckt) angegeben" (Kihara, 1930: 269). In recent years a number of mathematical models have been developed to standardize calculations of genome "affinity." The models, which predominantly have been developed by Kimber and coworkers (e.g., Alonso & Kimber, 1981; Chapman & Kimber, 1992a), will be described in greater detail below.

Love (1982: 205-206) was perhaps the first to take the full consequences of Federley's idea and to suggest that taxonomic rank should be determined by genome analysis: "It is our opinion that the logically and biologically most consistent conclusion from these observations must inevitably be, that a genomically homogeneous taxon of this magnitude and distinction, based either on a single haplome [viz. the genome of the basic monoploid analyzers] with or without a ployploid series or on a distinct combination of two or more haplomes, would be most reasonably classified as a genus in its own right and delimited on basis of morphological traits strictly connected with its genomic constitution. Therefore, we recommend that each haplome or genome combination be given generic status" [our italics]. Moreover, in cases where crossing is possible, but the hybrids are completely sterile and devoid of even rudimentary meiotic pairing, "it is of importance as a sign of relationship at the tribal level only and not of generic affiliation" (p. 205).

Dewey's (1982) view, which adhered strongly to Love's (1982, 1984) interpretation of genome analysis, epitomized the principles underpinning the methodology: "The fundamental premise underlying genome analysis is that like (homologous) chromosomes pair during meiosis and unlike (nonhomologous) chromosomes do not. The corollary premise is that the level of chromosome pairing in a species-hybrid reflects the degree of relationship between the parental species" (Dewey, 1982: 52).

The application of chromosome pairing data to phylogeny had already been questioned by Darlington, who concluded that "a frequent lack of relationship between the 'affinity' of species and of their chromosomes is therefore to be expected, and is very often found" (Darlington, 1932: 39). Darlington's doubts about the usefulness of genome analysis were founded in his recognition of chromosome pairing being subject to genetic control. He stated that "frequency of pairing at pachytene is an uncertain indication of the genetic relationship of the parents. And since metaphase pairing is related to pachytene pairing through the formation of chiasmata, in the frequency of which species and varieties differ genetically, the indication becomes even more uncertain" (Darlington, 1937: 171).

Darlington further challenged the validity of the phylogenetic conclusions drawn from genome analysis by adding that "deductions from pairing must have an enormous margin of error. The extensive phylogenetic conclusions that have been made on observations of pairing in hybrids, especially in Triticum and Nicotiana, must therefore be regarded as little better founded than those based on chromosome number in Crepis" (Darlington, 1937: 172).

One would expect that scientists working with Triticum or Nicotiana at that time would respond to such criticism, but apparently no one did. Between 1930 and 1943 Kostoff published more than 100 papers on Nicotiana hybrids and some on Triticum, and even recognized pronounced variations in bivalent numbers. In a comprehensive survey of his earlier studies he wrote: "When a larger number of pollen mother cells are investigated and when the most frequent number of bivalents is found, we can estimate then with great certainty the degree of the chromosome homology, which gives some criteria concerning the relationship between the parental species" (Kostoff, 1943: 739). Kostoff (1941a, 1941b, 1943) frequently cited Darlington (1937), but never his critical remarks. Kihara, who continued with genome analysis of Triticum/Aegilops until the mid-1960s, seems never to have addressed either Darlington's (1932, 1937) or anyone else's criticism. One serious criticism focusing on several aspects of genome analysis was later addressed by Gaul (1959), who concluded that "knowledge of the number of chromosomes able to pair gives no information about the degree of homology involved" (p. 204). The paper was presented at the First International Wheat Genetics Symposium in 1958 and was published in the symposium proceedings (Gaul, 1959). Oral presentation should have ensured an effective appreciation of Gaul's ideas within the scientific community. At least wheat cytogenetists were confronted by his criticism. However, Gaul's paper (1959) seems to be cited rarely, except for a section concerning some numerical details (e.g., by Driscoll et al., 1979; Kimber et al., 1981; Singh et al., 1988; Singh, 1993). A few papers cite Gaul (1959) for other reasons (Chennaveeraiah & Hiremath, 1974; Fedak, 1980), but we have not been able to find any reference to his real message [even though Waines (1976), when reviewing genome analysis, cites Gaul (1959), the citation does not specifically address Gaul's critical ramarks]. The apparent neglect of Gaul's paper is inexplicable, especially when some authors citing Gaul (1959) used genome analysis extensively and even published reviews on the subject (Kimber, 1983; Kimber & Hulse, 1979; Kimber & Yen, 1990; Singh, 1993; Jauhar & Joppa, 1996). In the scientific community, criticism is usually met with counterarguments and the subject progresses in a healthy debate. With some hesitation we are forced to conclude that Gaul's (1959) paper has been deliberately neglected.

Whereas cytogeneticists, systematists, and plant breeders continued uncritically with genome analysis, it is worth noting that geneticists were perfectly aware of the shortcomings of the method. As Grell (1969) put it, "The danger inherent in such assumptions [about pairing equalling homology] is exemplified in interspecies studies in plants where chromosome pairing at the late prophase or metaphase of meiosis I has been interpreted as an indication of homology between the participating chromosomes, and such presumptive homology has been used as the basis for determining evolutionary pathways. It is now clear" (Grell, 1969: 362).

A weaker criticism addressing problems involving polyploids was raised by de Wet and Harlan (1972), and Waines (1976) questioned the validity of genome analysis on the basis of genetic interference with chromosome pairing: "Our understanding of genes that promote or suppress homoeologous chromosome pairing brings under question the ability of the technique of genome analysis, as practiced up to now, to elucidate phylogenetic relationships, at least among taxa in the wheat group" (Waines, 1976: 424).

Jackson (1982, 1984) added cytologic and genetic arguments against the use of chromosome pairing data for phylogenetic inference: "Classical genome analysis, as a means of showing lineage relationships, could provide data leading to a misinterpretation of such a relationship" (Jackson, 1982: 1521). Thus, Dewey (1984: 211) was hardly correct when he stated that "few, if any, will question the value of genomic relationships, as determined by genome analysis, as indicators of phylogeny and biological closeness."

Within the last decade criticism has intensified (Baum et al., 1987; Kellogg, 1989; Seberg, 1989; Jackson, 1991; Petersen & Seberg, 1996). The critique raised by Baum et al. (1987) was addressed by Jauhar and Crane (1989), who defended the methodology. But others still neglect the critique, and it is simply incorrect when Wang and coworkers state: "Today there is, however, no disagreement as to the conceptual ideas of genomes per se as defined by several authors" (Wang et al., 1996).

IV. Chromosome Pairing and the Concept of Homology

A. GENOME ANALYSIS, HOMOLOGY, AND CHROMOSOME PAIRING

The axioms of genome analysis have been most clearly stated by Dewey (1982: 52): "The fundamental premise underlying genome analysis is that like (homologous) chromosomes pair during meiosis and unlike (nonhomologous) chromosomes do not. The corollary premise is that the level of chromosome pairing in a species-hybrid reflects the degree of relationship between the parental species." Evidently, the first part of Dewey's fundamental premise is in complete agreement with the traditional cytogenetic definition of homology (Sutton, 1902: 34): "The numerical reduction (pseudo-reduction) is accomplished by the union of homologous members of the two series of a nucleus." Sutton originally based this statement on studies of meiosis in the grasshopper Brachystola magna, but generalised his findings into a formal definition a year later and was consequently the first to link the concept of homology with chromosome pairing (Sutton, 1903: 232): "The process of synapsis (pseudo-reduction) consists in the union in pairs of the homologous members (i.e., those that correspond in size) of the two series." The necessary proposition that the members of a chromosome series are derived by the union of paternal and maternal chromosomes was proposed earlier by Montgomery (1901: 223). This classic cytological definition of homology is entirely operational. It clearly defines the kind of observations that are necessary and sufficient in order to classify chromosomes as homologous: They must pair during meiosis. How and why pairing is brought about is considered to be irrelevant.

Doubtless most scientists will accept the first part of Dewey's fundamental premise, which in essence is nothing but a restatement of Sutton's definition. Although exceptions are known to occur, chromosomes that behave in accordance with Sutton's definition are homologous no matter which definition of homology is preferred. (Concerning the distinction between homology and homonomy, see below.)

However, in our opinion it is the second part of Dewey's fundamental premise, the notion that chromosomes that do not pair are nonhomologous, that is erroneous and has confounded the literature with confusion. For example, within any of the approximately 32 species of Hordeum (Bothmer et al., 1995) the two haploid sets of chromosomes are homologous [actually they violate Patterson's (1982) conjunction test for homology; see below] and will consequently pair during meiosis. The chromosomes belonging to different Hordeum species show, however, limited or no meiotic pairing in many interspecific hybrid combinations (Bothmer et al., 1995: 8-11). Irrespective of this, almost all of their traits (from gross morphology to genes and nucleotide sequences) must, provided the genus is monophyletic, be homologous, fulfilling even the strictest definitions of homology - except for Dewey's (1982). Much semantic and scientific confusion stems from this unfortunate mixing of lack of or impaired chromosome pairing and homology. Hence, when Bothmer et al. (1986: 528), concerning hybrid combinations involving Hordeum murinum L., states that "it can thus be concluded from the low pairing frequency observed with polyploids of this species that no homologies existed with any other species," it is only correct provided homology is interpreted in Dewey's sense, e.g., as pairing behaviour at metaphase I of meiosis, but is nonsensical if homology is defined in any other way. It would be highly surprising if most genes were not shared between H. murinum and other species of Hordeum.

Dewey's corollary principle, that the observed level of chromosome pairing in a hybrid is a reflection of the degree of relationship between the parental species, is also spurious. The link between the fundamental and the corollary premise uses the vague concept of similarity. Similar, homologous chromosomes pair whereas dissimilar, nonhomologous chromosomes do not. As a consequence, homology becomes quantifiable and measurable in degrees, as does relationship. Despite this notion, no indication is given of how the relationships between levels of chromosome pairing and relatedness of the parental species are to be quantified.

There is little doubt that in a normal diploid (2n) species the individual chromosomes (one from each parent) of the n different pairs that are united in the zygote at fertilisation are similar at some level. Mutually chromosomes may also in some way be more similar to each other than to the chromosomes of a second diploid species, but similarity per se is merely a first approximation of primary homology (Farris, 1979, 1981, 1983, 1985). Unfortunately, Dewey (1982:51) clearly lacks precision in his conception of relationships: "A taxonomic system should, whenever possible, reflect biological relationships and phylogeny," and "classification based on morphology is usually a reasonable approximation of genetic and phylogenetic relationships." The confusion is confounded on biological, genetic, and phylogenetic relationships, by other authors referring to biosystematic (Lu, 1993b: 4), genomic (Lu, 1993b: 8; Singh, 1993: 258), and chromosome relationships (Bothmer et al., 1986: 528).

To what extent and in what combinations the different types of relationships are used in constructing taxonomic systems or classifications (Dewey, 1982:51) or evolutionary (Love, 1982: 205) or taxonomic classifications (Lu, 1993b: 8) is unknown, as is their role in"applied evolution" (Love, 1982: 205). The relationship between systematics and genome analysis will be treated in a later section.

B. IS THE CONCEPT OF HOMOLOGY IN CYTOLOGY REALLY DIFFERENT?

Cytogenetics inherited its concept of homology from classical evolutionary biology, and its original use was a logical extension of its use in morphology. The subsequent terminological muddle, which lead Moritz and Hillis (1996: 7-9) to suggest that there are at least three different concepts of homology (common ancestry, chromosome pairing, and the use of molecular probes in study of the same species as the probe was derived from), has been reinforced by the unfortunate mix of levels of chromosome pairing and homology.

That chromosomes represent the borderline between morphology and molecular biology is perhaps the single most important cause of confusion. Viewed through a microscope, chromosomes may be described in morphological terms, but at the same time they carry the majority of the genetic material (excluding mtDNA and cpDNA). Although no other group has confounded the meaning of the term "homology" (Hillis, 1994) more than molecular biologists, modern treatments of the concept of homology emphasize the close relationships between morphological and molecular definitions of homology. However, pairing behaviour at metaphase I in meiosis is, as shown below, not an acceptable bridge (Hempel, 1966) between chromosomes as morphological entities and sequence information contained within them.

The presence of the chromosomes as pairs in all somatic cells of most eukaryotes is, however, a problem that is apparently easy to cope with. According to Patterson (1988: 611), "orthology is the molecular equivalent of classical homology and paralogy is the molecular equivalent of homonomy." Most orthologoues are, however, present in at least two copies in the organism and so fail Patterson's (1982) conjunction test (concerning tests of homology, see below). This applies to all genes, gene products, and fragments of the genome in multicellular organisms and in unicellular eukaryotes. To circumvent this problem, Patterson (1988) suggested that the haploid genome be used as the standard of comparison. However, strictly speaking, homonomy at the chromosomal level is an homology at the level of syngamy, as paralogy is homology at the level of duplication of the ancestral gene (or sequence) (Nelson, 1994).

C. CHROMOSOME PAIRING AND HOMONOMY

In crosses between eukaryotic organisms, two genomes, maternal and paternal, are brought together in the zygote. They remain distinct entities until meiosis, where they either recombine and produce viable gametes or perish. Hence, at one extreme chromosomes of the two individuals involved in the cross will recombine and viable gametes will be produced. At the other extreme no recombination takes place and no viable gametes are produced. Usually, the first situation is expected to occur in intraspecific crosses, the second in interspecific.

In the first scenario, it is to be anticipated that the multiple copies of genes, gene products, and fragments of the genome pass Patterson's (1982) similarity and congruence tests but not his conjunction test (see below). Hence, at the level of the haploid genome they must be viewed as homologous, and taken together they will be homonomous. This would also apply to the individual chromosomes of a pair, as suggested by Moritz and Hillis (1996: 9): "In cytogenetics, it is standard to refer to the respective chromosomes in a chromosome pair of a diploid organism as homologs and to refer to the homologous pair of chromosomes in another species as homeologs, even though this is quite different from the use of homology in classical morphology (where homonomy is used to refer to a repeated structure in a single organism)." In the second scenario, the most likely prediction is that different proportions of genes, gene products, and fragments of genome brought together in the same zygote will violate different combinations of the three tests of homology, the similarity, conjunction, and congruence tests (see below). Evidently, the individual chromosomes of the two haploid genomes need no longer be comparable and no overall statement about their homology is possible.

Therefore, the chromosome pairs within a species are usually homonomous, whereas the chromosomes brought together in interspecific crosses, even if they do pair, need be neither homonomous nor homologous when viewed as wholes.

This difference is further reflected by the fact that, when two chromosomes of a chromosome pair differ only by carrying different sequences at a single locus, they will be interpreted as two different alleles if they do not characterise a monophyletic group, but as different characters (synapomorphies) if they do. This also applies if the two chromosomes differ not at the sequence level but by, say, an inversion. The issue becomes more complicated when translocations are involved. If a piece of one chromosome in the pair is translocated onto a third chromosome, the same arguments as above are valid. These arguments are no longer restricted to whole chromosomes; they becomes valid for pieces of chromosomes.

In any hybrid the level of generality will vary between all possible comparisons of traits. In a hybrid between two Hordeum species, some traits (morphology, DNA sequences) will be homologous at the level of the angiosperms, some at the level of monocotyledons, some at the level of the Poaceae, and so on.

D. SIMILARITY AND HOMOLOGY

In science there is (or ought to be) a clear distinction between the theoretical definitions of terms (like "homology") established by convention or flat and the empirical observations used to determine ira particular definition has been met. As stated previously, Dewey (1982) very explicitely states that "like (homologous) chromosomes pair" and "unlike (nonhomologous) chromosomes do not." Therefore, homologous chromosomes are similar and pairing behaviour is the empirical observation needed to infer homology. The idea that a level of analysis different from the level of organization of the trait under consideration is a means of assessing homology stems from a desire to divorce homology from phylogeny (e.g., see Beer, 1980; Goodwin, 1984; Roth, 1991).

Though highly unfortunate, the misuse of similarity to mean homology has a long and tortuous past in comparative biology (Patterson, 1982, 1988) and is still the main reason for much confused writing (e.g., see Brandham & Bennett, 1995). Being empirical observations, measures of similarity can usually be objectively defined (e.g., percent similarity between two sequences); however, this does not apply to Dewey's concept of similarity, which is inferred. In genome analysis, pairing behaviour at meiosis is often thought to be a function of pairwise comparision of the whole or part of the DNA of the chromosomes. However, when Kimber discusses the relationships of allopolyploids in general (1986: 65), his comments are ad hoc and not based on observations: "It can be demonstrated that chromosome pairing in hybrids represents the comparison of some DNA distributed along the entire length of the nuclear DNA, and thus it should represent the most reliable method of determining genomic homology in allopolyploids." Presently, objective similarity between chromosomes exists only at the macroscopic level. Viewed through the light microscope, the karyotype and individual chromosomes may be described in traditional morphological terms, but the visible features of chromosomes are few, and the level of description has changed little since Sutton (1903) remarked that homologous chromosomes were "those that correspond in size."

Theoretical problems aside, the empirical observations deemed necessary to establish whether or not chromosomes, and eventually genomes, are similar (= homologous) is their pairing behaviour at metaphase I of meiosis. However, as stated above, chromosomes have only few morphological hallmarks, and the only features that can routinely be assessed at metaphase I of meiosis are the different chromosomal configurations (univalents, bivalents, trivalents, etc.), the number of chiasmata, and the larger structural rearrangements (e.g., translocations, inversions).

As we discuss later, this limited observational basis creates enormous problems in interpretations of data from genome analysis.

E. HOMOLOGY TESTING

The basis of both comparative morphology and comparative molecular biology is hypotheses of homology, and Patterson (1982, 1988) has convincingly shown that the same three tests (the similarity, congruence, and conjunction tests) apply to both fields of inquiry, though their role in each is different.

In both morphology and molecular biology, similarity is the first indicator of homology: "In the absence of a phylogenetic analysis, one can only propose homologies based on character similarity; one cannot test hypotheses of homology" (Lauder, 1994:188). Traditional morphological homologies are three-dimensional, or if ontogeny is taken into account, four-dimensional. Methods normally used to look at chromosomes only allow them to be perceived as two-dimensional, rather than three-dimensional, structures. Even though they take on different shapes during the cell cycle, they have no ontogeny. As morphological objects, chromosomes often fall in a few ill-defined groups (metacentrics, submetacentrics, etc.). However, our ability to recognise the individual chromosomes may be enhanced by three-dimensional reconstructions, in situ hybridization, and poorly understood methods such as C-banding. In contrast, molecular sequence homologies are uni-dimensional, and "discoverable" if the sequences can be aligned (positional homology). However, the acceptable level of mismatch between the sequences (governed by the penalty for introducing gaps) becomes the critical factor in deciding whether the two sequences are worthy of consideration as homologs. The cutting point can never be objectively defined, and even highly divergent sequences may still be homologous. Furthermore, it remains highly enigmatic as to what extent pairing behaviour is a reflection of sequence similarity. Even if chromosome pairing did reflect sequence similarity, this similarity may or may not be a result of common ancestry (Hillis, 1994). Consequently, to infer similarity at the molecular level from morphological observations of chromosome behaviour tends to obscure the issue beyond comprehension.

In morphology the conjunction test is used to distinguish between two structures orginally considered similar enough to be hypothesised as homologs. If both structures are found in the same organism they cannot be homologous. However, as previously stated, all eukaryotes have at least two copies of all genes and nucleotide sequences, which led Patterson (1988: 612) to suggest that the appropriate level of comparison in comparative molecular biology is the haploid genome. The performance of the observations made in genome analysis in the conjunction test has been treated above.

The decisive test of homology is the congruence test. Homologies are congruent with other homologies and specify groups that are rendered monophyletic by them.

The three tests of homology distinguish eight different relations between morphological and molecular characters as has been shown by Patterson (1988). Ultimately, common ancestry, and hence homology, is a theoretical concept, and what counts as a homology changes with our perception of relationships. In contrast to similarity, homology is not an absolute quantity that can be measured but a hypothesis, and unless one believes that the pairing behaviour at meiosis is a substitute for a cladistic analysis, this issue is not addressed in genome analysis.

Considering the above discussion, the value of the concept of homeology seems dubious. What is the need for a special, collective term to designate the almost infinite number of possible stages between complete pairing and non-pairing?

F. CHROMOSOME PAIRING AND HOMOLOGY - AGAIN

In any organism that is or behaves like a diploid, chromosome pairing is important for two reasons: It insures that crossing-over - i.e., the reciprocal exchange of parts of the chromatids of the homologous chromosomes - can take place, and it ensures that bivalents are formed from the homologous parental chromosomes and are held together by chiasmata until first meiotic division.

Chromosomal homology is, at some level, likely to be based on similarity of DNA. However, it is likely that homologous chromosomes or chromosome segments are identified and contact established between them during or before early meiotic prophase. The regions that have been recognized as homologous are brought within a critical distance and connected by the synaptonemal complex (SC), which is indifferent to homology (Wettstein et al., 1984). It is generally believed that less than 1% of the DNA of each homologue included in the SC is available for sequence matching, and of this DNA, only a tiny fraction consists of specialized sequences supposed to act as the substrate of the molecular events of biparental duplex formation, strand breakage, and rejoining (Wettstein et al., 1984; Loidl, 1994).

DNA sequences that serve the mutual recognition of homologous chromosomes probably reside in autonomously functioning individual sites (e.g., pairing sites, zygomeres, pairing centres, association sites) distributed along chromosomes (Loidl, 1994). It is unclear if recognition is mediated at the DNA or protein level.

If the above description of chromosome pairing is correct, it becomes exceedingly difficult to describe objectively degrees of pairing seen in hybrids and perhaps even complete lack of pairing. Although the number of association sites and their distribution on each chromosome is unknown (it is likely that there is more than one on each chromosome arm), their possible role in chromosome pairing raises a number of interesting questions: how many of the association sites must match before normal bivalents are formed in a species-hybrid? How much DNA mismatch is allowable before the association sites cannot function properly? Is it possible to distinguish different "types" of partial pairing caused by malfunction in different association sites on the same chromosomes? Are all association sites equally important, or is there a hierarchy of sites?

Loidl's (1994) model of meiosis opens up the possibility of viewing each individual association site as a character that may be subjected to all three tests of homology. At the chromosome level the association sites may be treated as simple presence/absence characters. In principle the function of each site could be viewed as a character, too. It is noteworthy that from a phylogenetic point of view it is not the association sites that function "properly" that are interesting, but rather the ones that do not, as the ability to pair by definition is a plesiomorphic character (Kellogg, 1989; Seberg, 1989). However, our present level of observation makes it impossible to take advantage of these possibilities and accordingly precludes their use in a phylogenetic context.

At the molecular level it is, at our present level of knowledge, impossible to know which of the three homology tests will be passed. It is not known whether the two individual association sites function by complimentary strand recognition at the DNA level or by producing different proteins that interact like lock and key in the recognition process. In the first case the conjuction test will most likely not be passed, but it may do so in the second case.

From a biological point of view there is, however, no reason to expect that the relationships between species necessarily are reflected in the pairing behaviour of chromosomes at meiosis. Reproductive isolation may arise in many different ways. If it evolves at the prezygotic stage it is difficult to envisage the need for a strong selection pressure to avoid chromosome pairing at meiosis. It is equally easy to see the strong selective advantage of avoiding multivalent formation at meiosis in an auto- or allopolyploid.

IV. Biological Concerns

A. GENETIC CONTROL OF CHROMOSOME PAIRING

Meiosis is a complex process somewhat arbitrarily divided into a number of stages or subprocesses that are, as all other biological processes, under genetic control (e.g., Dover & Riley, 1977; Kaul & Murthy, 1985; John, 1990; Loidl, 1990; Hawley & Arbel, 1993). In Pisum L., for instance, it has been suggested that up to 50 genes are involved in the different stages leading to microsporogenesis (Gottschalk, 1973). Whereas indirect knowledge usually derived from studies of chromosome pairing irregularities is abundant, direct evidence on the role of specific genes or gene products involved in recombination is scarce. Almost all direct evidence comes from studies of meiosis in yeast, Saccharomyces cerevisiae E. C. Hansen, where a number of genes that are either entirely meiosis-specific or functioning both during meiosis and mitosis have been characterized (e.g., DMC1: Bishop et al., 1992; HOP1: Hollingsworth et al., 1990; MEI4: Menees et al., 1992; MER1/MER2: Engebrecht et al, 1990; MRE11: Ajimura et al., 1992; Johzuka & Ogawa, 1995; RAD50: Alani et al., 1990; RED1: Rockmill & Roeder, 1990; Rockmill et al., 1995; SEP1: Tishkoff et al., 1995; XRS2: Iranov et al., 1992; ZIP1: Sym et al., 1993; see also Nag, 1995, for an overview). Products of the genes HOP1, required for synaptonemal complex formation and recombination, and ZIP1, essential for synapsis, have been localized to meiotic chromosomes and as a part of the central component of the synaptonemal complex, respectively (Hollingsworth et al., 1990; Sym et al., 1993). The genes RED1, MEI4, and DMC1 also seem to be active during synaptonemal complex formation (Rockmill & Roeder, 1990; Bishop et al., 1992; Menees et al., 1992). The MER2 product seems, however, not to be a structural component of the synaptonemal complex, but is suggested to play a role in the initiation of genetic exchange; in MER2 mutants homologous chromosomes fail to synapse and the alignment of homologous chromosomes is reduced (Rockmill et al., 1995). RAD50 has been suggested to be involved in chromosomal homology search (Alani et al., 1990) and SEP1 mutants fail to pass the pachytene stage of meiosis (Tishkoff et al., 1995).

In higher plants the DMC1 gene has been characterized from Arabidopsis thaliana Schur, and LIM15, a gene that is supposed to be homologous to DMC1, has been discovered in Lilium longiflorum Thunb. (Kobayashi et al., 1993; Terasawa et al., 1995). LIM15 is meiosis-specific and transcribed during the meiotic prophase. Additionally, the RAD51 gene, which is present in yeast and a member of the same epistasis group as RAD50, has been characterized from Arabidopsis Schur (Genbank acc. no. ATU43528), and the proteins encoded both by RAD51 and LIM15 have been localized to the meiotic chromosomes in Lilium L. (Terasawa et al., 1995). Both proteins can be found on the leptotene chromosomes, whereas only RAD51 is present at the later pachytene stage (Terasawa et al., 1995). Other meiosis-specific proteins found in Lilium have been described by Stern (1986). GenBank holds sequences of a gene possibly homologous to RAD51 from Lycopersicon Miller (LEU22441) and of genes possibly homologous to DMC1 from Glycine (GMU66836) and Oryza (OSU85613).

This very brief summary of genes and gene products active during meiosis is restricted to those affecting stages prior to or during metaphase I, as the preceding meiotic stages are rarely studied in genome analysis, but the list could easily be extended by genes affecting later stages. In light of ongoing discoveries and the continued characterization of meiosis-active genes, it seems naive to suggest that chromosome pairing until metaphase I is a consequence of "chromosomal factors" whereas the behaviour at subsequent stages is determined by "genetic" factors (Jauhar & Joppa, 1996: 14). It is to be expected that meiosis-controlling genes are present in all eukaryotes.

All other evidence of genetic regulation of chromosome pairing in higher plants is indirect. The best-documented "pairing control gene" is the Ph gene or linkage group located on the long arm of chromosome 5B of wheat and related species (Okamoto, 1957; Riley & Chapman, 1958; Sears & Okamoto, 1958; Riley, 1960; Dover, 1973). Presumably the major effect of the Ph gene is to prevent the occurrence of multivalents at MI in polyploid species. Ph mutants or plants lacking the gene entirely show meiotic irregularities by having an abundance of multivalents at MI. At earlier stages of meiosis, both normal plants and Ph mutants show multivalent formation, but correction of multivalents into bivalents takes place only in the former. It has been assumed that the Ph gene is activated during synaptonemal complex formation and delays crossing-over until the correction of multivalents into bivalents has been completed (Holm & Rasmussen, 1984; Holm, 1986). However, observations from genotypes with different dosage of the Ph gene do not support this idea (Holm & Wang, 1988). An alternative hypothesis suggests that the Ph gene might affect the process of length adjustment of the paired lateral components of the synaptonemal complex. In heteromorphic bivalents, in which a perfect equality in the length of the lateral components is achieved, length adjustments could result in a relocation of homologous loci along the synaptonemal complex, and crossing-over might be prevented in the adjusted segments (Holm & Wang, 1988). However, this hypothesis also lacks direct, supporting data (Holm & Wang, 1988). Apart from the Ph gene, the existence of numerous other genes influencing the chromosome pairing of wheat (e.g., Cuadrado et al., 1991; and reviews in Sears, 1976; Gale & Miller, 1987) and other polyploid species of Triticum/Aegilops (Farooq et al., 1990) has been suggested.

In a number of polyploid plants the presence of more or less similar mechanisms of"diploidization" has been suggested: e.g., Avena (Ladizinsky, 1973; Jauhar, 1977), Festuca L. (Jauhar, 1975b), Hordeum (Subrahmanyam, 1978; Gupta & Fedak, 1985), Glycine (Singh & Hymowitz, 1985b), Gossypium (Kimber, 1961), Chrysanthemum (Tourn.) L. (Watanabe, 1977, 1981 a, 198 lb), Solanum (Dvorak, 1983), Brassica (Harberd, 1972), and Tolmiea Hook. (Soltis & Rieseberg, 1986), and it has frequently been proposed that genetic diploidization of polyploids is a more or less universal phenomenon (e.g., Riley, 1960; Waines, 1976; Jauhar, 1977). At least for autopolyploids this is most likely. An example is the naturally occurring autotetraploid Tolmiea menziesii Torr. & Gray, that shows tetrasomic inheritance, yet at MI forms only bivalents (Soltis & Rieseberg, 1986). In artificially raised autotetraploid rye, in which the four chromosomes of each group must be considered nearly identical through years of inbreeding, Chatterjee and Jenkins (1993) demonstrated a significantly higher number of bivalents than would be expected through random pairing.

Indirect evidence for genetic influence on chromosome pairing has also been provided for a number of diploid species, mostly grasses: e.g., diploid species of Triticum/Aegilops (e.g., Avivi, 1976; Waines, 1976; Shang et al., 1989), Hordeum bulbosum L. and H. vulgare L. (Thomas & Pickering, 1985), Secale cereale L. (Fedotova et al., 1994), and Lophopyrum elongatum (Host) A. Love [syn. Agropyron elongatum Host ex Beauv., syn. Elytrigia elongata (Host) Nevski] (Dvorak, 1987; Charpentier et al., 1988). In hybrids with polyploid species, such as wheat, the genes from these diploid species are suggested to influence (suppress or promote) chromosome pairing between the genomes derived from the polyploid parent (Lelley, 1976; Dvorak, 1987; Charpentier et al., 1988). The presence of genes influencing chromosome pairing has mostly been invoked from studies of chromosome pairing in species or hybrids involving addition or deletion lines. Waines (1976) argued for the evolutionary significance of possessing chromosome pairing-regulating genes in diploid species, and suggested that genes preventing pairing between all but fully homologous chromosomes might function as a genetic isolation mechanism of taxa.

In addition to these examples of genetic control of chromosome pairing, most cases of asynapsis and desynapsis are best explained as genetic irregularities (mutation?). Hence, total asynapsis and desynapsis can be regarded as nothing but the extreme negative consequences of genetic regulation. Reviews of asynapsis and desynapsis have been given by, e.g., Darlington (1932, 1937), Prakken (1943), Rees (1961), Riley and Law (1965), Sjodin (1970), Gottschalk and Kaul (1980a, 1980b), Koduru and Rao (1981), Kaul and Murthy (1985), and Singh (1993).

Evidently any genetic influence on chromosome pairing must necessarily violate the postulated correlation between chromosome pairing and chromosome similarity. Evidence of genetic control of chromosome behaviour during meiosis has existed for more than 60 years, and its possible implications for phylogenetic inference were discussed already by Darlington (1937), as previously mentioned. Since then, several genome analysts have accepted that genetic interference is a likely source of error, but they justified their work by assuming that genetic influence is the exception rather than the rule: "Unfortunately, some of us have a tendency to focus on unusual and exceptional behavior and lose sight of the rule" (Dewey, 1982: 52).

However, it seems difficult to neglect the role and importance of genetic control over meiosis. Despite the interpretation of genetic influence, consequences for genome analysis vary among authors. Hence, to Jauhar and Joppa (1996: 21) it has no role to play, as chromosome pairing is a function of homology, whereas the chiasma frequency and distribution is under genetic control. However, to Jauhar (1977: 287) diploid-like meiosis in most, if not all, natural polyploids was previously considered genetically regulated. This change of attitude is difficult to justify in view of the investigations of the last two decades, but is paralleled by the idea that asynaptic and desynaptic genes occur rarely (Jauhar & Crane, 1989: 573) in opposition to the fact that numerous cases of desynapsis are known in grasses and other species (Jauhar, 1975a: 114).

To Wang (1992) there are two types of control: At one level chromosome pairing is under genetic control, but at the same time it is a measure of likeness between the genomes; at the other level special genes (promoters and suppressors) may interfere. But of course no evidence supports this idea.

A distinction should be made between normal meiosis-specific genes driving and regulating the meiotic process and mutated alleles leading to asynapsis or desynapsis. The question of their absolute frequency laid aside, mutations giving rise to total malfunction are probably the exception rather than the rule in natural species. However, in any hybrid organism it is entirely unknown how even closely related alleles of the same meiosis-specific genes may (or may not) cooperate. To invoke the presence of special genes for asynapsis or desynapsis or genes suppressing or promoting chromosome pairing seems to be premature and ill-founded, as judged from studies of hybrid organisms. Given the present state of our knowledge of the meiotic process in natural species, the outcome of meiosis in hybrids appears quite unpredictable.

B. PREFERENTIAL PAIRING

Genome analysts have debated whether genome analysis is best done on diploid or polyploid hybrids. Advocates for the superiority of using diploid hybrids argue that chromosome pairing in polyploid hybrids is less reliable due to increased genetic influence on pairing (de Wet & Harlan, 1972; Wang, 1989, 1990, 1992). Others claim that pairing assessed in diploid hybrids results in an overestimate of actual homology, because chromosomes will tend to pair with less-related chromosomes when homo(eo)logous chromosomes are not present. Therefore, pairing is to be studied in polyploid, preferably triploid, hybrids, in which competition among the genomes is supposed to take place. Hence, the two most similar (= closely related) genomes are supposed to have a preference to pair with each other (preferential pairing) rather than with the third, less closely related genome (Jauhar, 1975a; Jauhar & Crane, 1989; Kimber & Feldman, 1987; Gill & Kimber, 1989; Kimber & Yen, 1990; Jauhar & Joppa, 1996). As we have pointed out earlier, the ability of chromosomes to pair tells little if anything about the homology of the chromosomes. Accordingly, it is essentially irrelevant to decisions of homology whether studies are made on diploid or polyploid hybrids. Nevertheless, a few pertinent comments could be added to the debate.

It seemed difficult for Jauhar and Joppa (1996) to decide how much weight chromosome pairing data in diploid hybrids should be given. On one hand they state: "Conditions for preferential pairing are lacking in diploid hybrids, which, therefore, do not provide sound pairing data to valid inferences regarding intergenomic relationships" (Jauhar & Joppa, 1996: 15). But on the other hand they also suggest that when combining evidence from chromosome pairing with pollen and seed fertility in the hybrids being investigated, the level of pairing in diploid hybrids could be informative. A high level of pairing may always indicate genome homology irrespective of the level of fertility, but a low level of pairing combined with low levels of fertility would indicate nonhomology (Jauhar & Joppa, 1996). The rationale behind this idea seems likely to stem from accepting genetic control of meiosis only during stages after meiotic metaphase I (Jauhar & Joppa, 1996). Occurrence of genetic control prior to metaphase I invalidates the hypothesis.

In taxa that consist exclusively of diploids it has been suggested that artificial autotetraploids should be produced which then could be used in crosses with the original diploid species (e.g., Jauhar, 1975a). The resulting autoallotriploid hybrids may, however, behave very differently in reciprocal combinations. Synthetic autotetraploids of Lolium temulentum L. (genomic composition TT) and L. multiflorum L. (MM) have been backcrossed to the diploid parents and the meiotic behaviour of the autoallotriploid progeny (TTM and TMM) examined (Thomas, 1995). Whereas the chromosomes of the TMM hybrids paired mostly as M-M bivalents, leaving the T chromosomes as univalents, pairing in the TTM hybrids led predominantly to the formation of trivalents. Thus, different assumptions about the relatedness of the T and M genomes (and L. temulentum and L. multiflorum) would be made depending on which hybrids were analyzed. Even if both hybrid combinations are analyzed as in the study cited, an interpretation of the result is not straightforward (for "cytoplasmic effects" see below).

It is an important assumption for genome analysis based on preferential pairing that genes affecting pairing influence the overall pairing of all chromosomes and genomes and not the pattern or preferences of pairing (Gill & Kimber, 1989; Jauhar & Crane, 1989). However, this may not be so. Parker (1975) has shown that desynapsis of one specific chromosome pair in Hypochoeris radicata L. (2n = 8) was caused by a single recessive gene. Similarly Tease and Jones (1976) showed that desynapsis of the three chromosome pairs in Crepis capillaris (L.) Walk. (2n = 6) was individually determined. These observations were possible only because identification of the individual chromosome pairs was possible. Thus, in studies where chromosomes are not identified (which are the vast majority), only a slightly reduced level of pairing would be recognized.

Parker (1975) suggested that genetic control of chiasma formation is hierarchical, with some genes affecting the whole complement and others only specific chromosomes. Tease and Jones (1976) further added that even though chromosome-specific chiasma control had so far only been discovered in the Asteraceae, this may be due to their generally low chromosome numbers and the relative ease of identifying the chromosomes, rather than to taxonomically restricted occurrence. From the animal kingdom, evidence exists for the occurrence of different degrees of asynapsis among chromosome pairs in grasshoppers (Rees, 1957; John & Naylor, 1961), and Hewitt and John (1965) suggested some degree of autonomous control of each bivalent in grasshoppers.

If genes affect the pairing of specific chromosomes, is there no obvious reason why other genes may not affect pairing of specific genomes? In amphiploid Lolium L. hybrids, B-chromosomes reduce the number of multivalents, restricting chromosome pairing to form bivalents (Evan & Macefield, 1973; Taylor & Evans, 1977). The effect of B-chromosomes (considered in a later section) is most likely exerted by interaction with A-chromosome genes; thus, a change in the pattern of chromosome pairing among genomes is, in this case, likely to have a genetic basis. Perhaps the Ph gene of wheat offers better evidence for genetically induced changes in the pairing pattern. Although the Ph gene may affect pairing between the A, B, and D genomes similarly, it certainly has a different effect on, e.g., A-A pairing than it has on A-B pairing. Thus, the Ph gene changes the pattern rather than the overall level of pairing. This is generally recognized, but worded differently, as an effect exerted on homoeologous rather than homologous chromosome pairing (e.g., Sears, 1976; Dover & Riley, 1977). However, the distinction between homologous and homoeologous chromosomes is arbitrary, and therefore to postulate that the Ph or other genes always affect pairing among all genomes in hybrids in a similar manner is wrong in more than one way.

Additional evidence of genome-specific control of chromosome pairing comes from amphiploid hybrids between Aegilops ventricosa Tausch and Secale cereale (Orellana et al., 1985). In meiotic mutants asynapsis was restricted to the Aegilops chromosomes, whereas the Secale chromosomes paired normally. Further, it has been shown in hexaploid triticale (the amphiploid hybrid between tetraploid Triticum and Secale), that pairing of the rye chromosomes is much more reduced than pairing of the wheat chromosomes (Lelley, 1975; Galindo & Jouve, 1989). That this could be due to lesser "homology" between the rye genomes seems very unlikely.

A second comment on the debate about analysis of diploid vs. polyploid hybrids regards an apparent contradiction that only proponents of genome analysis in polyploids must face. If they disregard chromosome pairing in diploids as a measure of genome homology, then they must also neglect the fundamental basis of genome analysis itself. If pairing in diploid hybrids is considered to overestimate homology (because the chromosomes have an inherited urge to pair), pairing must be considered influenced by factors other than chromosome similarity alone. We fully agree with this, but we fail to see why introducing additional genomes should help alleviate the problem.

C. MATHEMATIC MODELS

A number of mathematical models for predicting preferential pairing in triploid and higher ploidy-level hybrids have been developed (Driscoll et al., 1980; Alonso & Kimber, 1981; Espinasse & Kimber, 1981; Kimber & Alonso, 1981; Sybenga, 1988, 1994; Crane & Sleper, 1989a, 1989b; Chapman & Kimber, 1992a, 1992b, 1992c, 1992d, 1992e). The theoretical basis and underlying assumptions are thoroughly discussed by Kimber et al. (1981), Alonso and Kimber (1981), Sybenga (1988, 1996), Crane and Sleper (1989a), Chapman and Kimber (1992a), and Crane (1996). Most models aim at predicting the most likely pattern of genomic pairing (e.g., 4:0 [= 1:1:1:1], 2:2, or 2:1:1 for a tetraploid hybrid) on the basis of detailed observations of MI pairing configurations, and at calculating the "relative affinity" among the genomes. To the extent that the underlying assumptions are fulfilled, the models probably predict this pattern rather precisely. However, one or several of the assumptions may be violated, and even if the pairing pattern is predicted exactly, this has no known relationship to homology or phylogeny.

At this point is should be noted that the mathematical models as well as all other calculations and interpretations in genome analysis are based on observations of chiasmata, although the claim is that these observations represent an estimate of pairing (and similarity) along the entire length of the chromosomes (e.g., Kimber, 1984; Kimber & Yen, 1990; Jauhar & Joppa, 1996). This seems a rough oversimplification of the meiotic process, because, as mentioned above, prophase pairing does not necessarily lead to chiasma formation.

The models constructed by Kimber and coworkers are based on a number of assumptions, the most important being the following: 1) Interference across the centromere does not occur - i.e., formation of a chiasma involving one ann of a chromosome has no influence on the probability of chiasma formation involving the other arm; 2) pairing of one arm with another precludes involvement of a third; 3) the frequency (c) with which two chromosome arms pair is the same for all arms - i.e., if the value of c for, e.g., the long arms of the chromosome pair 1A-1B is 0.4, then all other chromosome arms of the A complement pair with their B complement partner with the same frequency (Alonso & Kimber, 1981; Chapman & Kimber, 1992a); and 4) the distribution of the numbers of chiasmata is normal (Crane, 1996).

The first assumption implies that formation of chiasmata of the two arms of any chromosome is independent. Chapman and Kimber (1992a) note that autotetraploids that form an excess of bivalents at metaphase I violate this assumption, but might form an exception - the question being whether the formation of excess bivalents is due to interference or due to preferential pairing caused by "cryptic differences among homologues" (p. 98). It is evident from studies of synaptonemal complex formation that chromosomes in a number of species and hybrids are initially paired as multivalents though later these associations are resolved into bivalents (e.g., Jenkins, 1985, 1986; Jenkins et al., 1988; Holm, 1986; Davies et al., 1990; Stack & Roelofs, 1996). Hence, in these eases at least, there is initially no evidence of preferential pairing. The exact mechanism behind the process of subsequent bivalent formation is largely unknown and to what extent it is due to interference among the chromosome arms and to the formation of chiasmata has not been fully elucidated either. Crane (1996: 62-63) is aware that the assumption of interference is frequently violated and suggests that a solution to this problem (viz. having the chromosome divided into more than two domains) is the greatest challenge for the next generation of numerical meiotic models.

The second assumption may in general be valid, although cases of multiple synapsis (more than two chromosomes synapsed at the same location) are known from, e.g., the Solanaceae (Sherman et al., 1989).

The third assumption seems of doubtful value. In hybrids, when the identification of paired chromosomes is possible, the assumption has been invalidated. Jouve et al. (1985) showed differences in the pairing frequency among the seven B genome chromosome pairs in hybrids between different varieties of wheat, and similarly Gonzalez et al. (1993) have shown different pairing frequencies among the seven A genome chromosome pairs in a hybrid between Triticum aestivum L. and T. monococcum L. Further, it has recently been shown that the homologous chromosome arms of Hordeum marinum Hudson carrying rDNA segments have a lower chiasma frequency than all other chromosome arms (I. Linde-Laursen, pers. comm.). The above-cited studies of Hypochoeris L. (Parker, 1975) and Crepis (Vaill.) L. (Tease & Jones, 1976) provide additional evidence. Sybenga (1988), whose models for triploid hybrids are similar in many respects to those of Chapman and coworkers and are largely based on the same assumptions, admits that the x sets of three chromosomes do not necessarily behave identically, but suggests that this can be tested. If they behave similarly, "the critical configurations" (p. 747), i.e. the number of trivalents, should be distributed binominally (Sybenga, 1988). This is of course true, but it is an open question how often real data meet this criterion. Published data are very rarely detailed enough to allow testing. Are trivalent and other pairing configurations always composed of the same chromosomes, e.g., 1A-1B-1C, or, for instance, does 1A-1B-SC pairing sometimes take place? Another, more fundamental question is whether the postulated "sets" of chromosomes are real and not merely artifacts.

The fourth assumption requires that the number of chiasmata is normal distributed with fully populated upper and lower ends (Crane, 1996). Apart from the fact that the observed variable (number of chiasmata) is discrete, and hence the possibility that the expected distribution would be binominal rather than normal, the assumption is certainly not always valid for interspecific hybrids. Very rarely are data published in a manner that allows for testing the distribution of chiasmata, but the requirement of fully populated upper and lower ends are most likely not, e.g., in hybrids with virtually no pairing or in hybrids with almost full pairing, and distributions may even be bimodal ([ILLUSTRATION FOR FIGURE 3D OMITTED]; Petersen & Seberg, 1996). The distribution of chiasmata is dealt with in detail below.

It is an additional weakness of the models that pairing involving translocated segments are indistinguishable from pairing caused by "homoeology," and consequently, when translocations occur, the calculated "affinity" will be misleading. Fernandez-Calvin et al. (1995) tested the effect of translocations in hybrids between Aegilops variabilis Eig (U and S genomes) and Secale cereale (R genome). Using the models of Alonso and Kimber (1981), a 3:0 model, i.e., random pairing between the three genomes, was proposed, mainly due to the occurrence of trivalents. However, C-banding of meiotic metaphases revealed that only S and U genome chromosomes were involved in the pairing due to translocations between some of the S and U chromosomes. Thus, when using the mathematical models it has to be assumed that translocations play a negligible role, which seems an oversimplification.

In triploid hybrids only two values of relative "affinity" are allowed by the models. Thus, with a pattern of pairing described as 2:1, it is assumed that one of the genomes is equally "related" (i.e., shows equal affinity) to the remaining two, which in turn are equally related to each other. Again there are hardly any biological reasons for assuming this to be true. Actually, we would expect this to be true only when the two equally related genomes are perfectly similar. And after all, why should they be, given divergence in different species with unique histories? In tetra- and pentaploid hybrids, the problem of assuming equal "affinities" increases even more (Kimber & Alonso, 1981; Chapman & Kimber, 1992b), but still only two different values of relative "affinity" are allowed. Chapman and Kimber (1992e) tried to permit one more value of relative "affinity" in their so-called "second-order" models for tetra- and pentaploids, but because in tetraploids there are six and in pentaploids ten potentially different relationships to explore, three variables are evidently insufficient.

The models constructed by Crane and Sleper (1989a, 1989b) differ substantially from the above in terms of their mathematical approach to the problem. Statistically the models have too many parameters, i.e., they contain more variables than there are degrees of freedom in the data. Thus, rather than providing just one solution, the models provide a range of solutions, which makes interpretation less straightforward. The rationale behind increasing the number of parameters is to decrease the number of a priori assumptions that have to be made. Crane and Sleper (1989a: 83) recognize the weakness of assumption 3 above by stating: "A priori assumptions of equality or uniformity tend to be unrealistic for species hybrids." Consequently, they allow individual behaviour to exist with regard to synapsis and chiasma formation in each combination of chromosome arms and genomes, though "the pattern of genomic affinity" must be the same for the two arms belonging to the same chromosome (Crane & Sleper, 1989a). We agree with Chapman and Kimber (1992a: 102) that, given the quality of meiotic data, this added complexity is unjustified. The rest of the assumptions dealt with above must still be invoked.

Sybenga's (1994) models for tetraploids apply only to amphidiploids and autoploids. Thus, by making a priori assumptions about relationships of genomes, the number of pairing parameters to be estimated is reduced to one. However, in genome analysis, such a priori knowledge usually does not exist unless amphidiploids or autotetraploids have been artificially raised. Other models for autopolyploids have been published (e.g., Jackson & Casey, 1982; Jackson & Hauber, 1982), but, given their minor importance to genome analysis, they will not be treated here.

Many problems related to the use of the above models stem from a lack in identification of those chromosomes involved in chiasmata formation. Both Crane and Sleper (1989a) and Sybenga (1988) are clearly aware of the problem, and we can only agree with the latter's conclusion: "It is clear that the information available in unmarked MI analyses is limited and only justifies rather restricted conclusions without previous knowledge about the relationships between the genomes involved and the possibilities of upsetting pairing regulation systems" (Sybenga, 1988: 756). As the "previous knowledge" most frequently will be derived through genome analysis of other hybrid combinations, even this is of restricted value.

However, a fundamental problem in the use of mathematical models for estimating the preferential abilities (relative affinity) of chromosomes or genomes to pair is that no matter how advanced, the models can never tell whether chromosome pairing or lack of pairing is due to some degree of DNA similarity, to interference, or to other (e.g., genetic) factors. As elegantly described by Sybenga (1996: 95): "If it would be possible to derive meaningful affinity estimates from mathematical meiotic models, this would not necessarily mean that they give more than a superficial indication of DNA correspondence."

The mere fact that the Ptolemaic system could describe the motion of the celestial bodies does not make the underlying theory correct.

D. THE INFLUENCE OF B-CHROMOSOMES

A general review of B-chromosomes can be found in Jones and Rees (1982) and the effects of B-chromosomes upon the behaviour of the A-chromosome complement during meiosis in grass species and hybrids has been considered previously by Jones (1991). In the present context the most important effect of B-chromosomes is their apparent ability to change the number of chiasmata, but their ability to change the position of chiasmata may also be noteworthy in this respect (Parker et al., 1990; Jones, 1991). Whereas B-chromosomes have been shown to influence chiasma formation in some species and hybrids, they have no apparent effect on meiosis in others.

The effects of Secale cereale B-chromosomes have been studied both in some B-chromosome carrying cultivated rye varieties and in various hybrids, primarily with Triticum aestivum. Data from the rye varieties are inconsistent, as both increases and decreases in chiasma frequency have been reported as well as no effects at all (see Diez et al., 1993). Though a significant positive correlation exists between the number or B-chromosomes and the number of pachytene abnormalities, e.g., persistent interlockings, the effect could not be seen at metaphase I (Diez et al., 1993). In wheat x rye hybrids the influence of the B-chromosomes on pairing between the three wheat genomes is likewise equivocal, and genes located on the rye B-chromosomes have been postulated to interact with the Ph and other pairing regulating genes of wheat (Jones, 1991). In Pennisetum typhoides (Burro.) Stapf. & Hubb., the effects of B-chromosomes on meiosis appear as unclear as those in Secale (Jauhar, 1981; Pantulu & Rao, 1982).

Within the genus Aegilops, B-chromosomes have been found in Ae. speltoides Tausch and Ae. mutica Boiss. (Jones & Rees, 1982). In contrast to the B-chromosomes of Secale and Pennisetum, those of Aegilops seem to have a more clear-cut influence on meiotic behaviour, at least in hybrids with Triticum and other species of Aegilops. Their general effect is a marked reduction in chiasma frequency in hybrids carrying B-chromosomes compared to hybrids lacking them (Ohta & Tanaka, 1983; Ohta, 1991). In diploid hybrid combinations involving Ae. mutica ([+ or -] B-chromosomes) chiasma frequency could be reduced from around 11 chiasmata per cell to less than 1 by increasing the number of B-chromosomes (Ohta, 1991). In normal hybrids between hexaploid wheat and Ae. mutica or Ae. speltoides (+ B's) the B-chromosomes were not found to have any effect on chiasma frequency, but in hybrids lacking the Ph gene the B-chromosomes were found to compensate for the missing gene, resulting in almost complete asynapsis (Dover & Riley, 1972, 1977; Vardi & Dover, 1972). Hence, the B-chromosomes of Aegilops are supposed to suppress pairing of homoeologous chromosomes but not homologous ones (Dover & Riley, 1977; Sano & Tanaka, 1983). Rather similar patterns of B-chromosome effects and possible interactions with A-chromosome genes have been described in Lolium (Evans & Macefield, 1973; Taylor & Evan, 1977) and Agropyron (Chen et al., 1993).

It has been suggested that B-chromosomes might provide a useful tool for genome analysis in the Triticum/Aegilops complex (Sano & Tanaka, 1983; Ohta, 1991). If pairing in hybrids without B-chromosomes is compared with that in similar hybrids carrying B-chromosomes, the magnitude of the reduction of chromosome pairing, i.e., chiasma formation, caused by B-chromosomes should be correlated to the degree of homology. If the level of chromosome pairing is high in non-B-carrying hybrids, and no reduction occurs in the presence of B-chromosomes, then genomes are considered homologous. If the presence of B-chromosomes reduces an otherwise high level of pairing, then the genomes are considered homoeologous, the degree of homoeology being determined by the degree of reduction. Non-pairing in either hybrid indicates non-homology.

This means of assessing genome homology, of course, suffers from all the same flaws as traditional genome analysis, but it is additionally based on dubious assumptions about B-chromosome behaviour and effects. The fundamental idea that the Aegilops B-chromosomes affect only homoeologous pairing is without any meaning, as the distinction between homology and homoeology is completely arbitrary. If the assumption is correct that the effect of B-chromosomes is due to some interaction with Ph or other genes involved in the regulation of chromosome pairing (Jones, 1991), any possible correlation between chromosome pairing and similarity is just further obscured. Thus, differences in the response to the presence of B-chromosomes may only tell us something about an unquantifiable genetic difference - and nothing about homology.

A cytological explanation of the effects of B-chromosomes could be that the addition of chromatin to the nucleus prolongs prophase and, hence, prolongs the time available for recombination (Loidl, 1994). Alternatively, as B-chromosomes in Crepis have been shown to increase the length of the synaptonemal complex (Parker et al., 1990), the looser packing of DNA along the synaptonemal complex might positively affect the ability of crossing over by increasing the number of available sites (Loidl, 1994). However, these theories would only explain an increased amount of chromosome pairing, which obviously is not the only effect of B-chromosomes.

Although the actual mechanisms are unknown through which the B-chromosomes interact with the A-chromosome complement, they cannot possibly change the homology or similarity (structurally or in base composition) of the A-chromosomes. In Zea mays L., breakage of A-chromosomes caused by B-chromosomes was observed at the second microspore division (Rhoades & Dempsey, 1973), but there exists no evidence for B-chromosome-induced structural changes in the A-complement during meiosis. Thus, B-chromosomes offer another line of evidence that chromosome pairing in species as well as hybrids is genetically determined to an extent that totally obscures chromosome pairing as a measure of similarity (at whatever level) of the chromosomes.

E. OTHER CYTOLOGICAL FACTORS INFLUENCING CHROMOSOME PAIRING

Apart from B-chromosomes, both euchromatic and heterochromatic supernumerary chromosome segments have been shown to affect chromosome pairing especially in various insect species (e.g., John & King, 1985; Suja et al., 1994). Like B-chromosomes, this kind of extra chromatin does not change the actual homology of chromosomes, but may interfere by adding an extra set (or more) of homologous chromatin to the chromosome complement. For example, an extra chromosome arm would give rise to three copies of the chromosome arm in question, not just two. However, rather than changing the chiasma frequency systematically, the presence of the supernumerary segments, irrespective of their eu- or heterochromatic nature, influences the distribution of chiasmata. Interestingly, the redistribution of chiasmata may be more severe in the homozygous condition than in the heterozygous one, in which chromosome similarity has actually been slightly reduced (Suja et al., 1994). Although such supernumerary chromosome segments do not play a significant role in genome analysis, we are again faced with events other than similarity that determine the final outcome of the chromosome pairing process.

Chromosomal rearrangements, natural or induced ones, have also been found to affect chiasma frequency (e.g., Hewitt, 1967; Parker et al., 1982). Although some types of rearrangements change the homology of the involved chromosomes regarded as whole entities, the overall similarity at the DNA level is the same. Some rearrangements may not only influence the chiasma frequency of the chromosomes directly involved in the rearrangement but also have a more generally positive or negative effect on some or all of the chromosomes in the cell (Hewitt, 1967). These cytologically induced changes are likely to have a genetic basis, and gene expression may possibly be altered as a result of positional effects.

F. ENVIRONMENTAL INFLUENCE ON CHROMOSOME PAIRING

Certain environmental factors have been shown to have an influence on chromosome pairing and chiasma formation. How this influence is elicited is largely unknown. Most likely it is a matter of gene expression, and a clear distinction between genetic and environmental influence on chromosome pairing is evidently not always possible. However, in the present discussion this distinction is of minor relevance. The important issue here is that changes in chromosome pairing are brought about without the slightest DNA substitutional change in the chromosomes.

Among the environmental factors proposed to affect meiosis, most attention has been paid to temperature. Some studies have shown an influence on chiasma frequency caused by seasonal temperature fluctuations (see Loidl, 1989), but more exact knowledge comes from controlled temperature experiments (e.g., Elliott, 1955; Bayliss & Riley, 1972; Bennett et al., 1972; Lin, 1982; Loidl, 1989). In studies on Hyacinthus orientalis L. and Endymion nonscriptus (L.) Garke. it was shown that temperature-induced changes of chiasma frequencies were due to failure of chromosome pairing at pachytene (Elliott, 1955). Bayliss and Riley (1972) concluded that temperature-induced chiasma frequency changes in wheat were caused by pairing failure at zygotene. Loidl (1989) showed more exactly that the reduced chiasma frequency in Allium ursinum L. exposed to a high temperature (35 [degrees] C) was caused by abnormalities in synaptonemal complex formation. Apart from changing the frequency of chiasmata, elevated temperature has also been demonstrated to affect their localization on the chromosomes (Henderson, 1962, 1963).

Obviously, it is not possible to describe a single effect of non-optimal temperatures on chromosome pairing or chiasma formation. The effects and causes may be very different depending on the actual organism, on how long the organism is exposed to the sub-optimal temperature, and on the meiotic or premeiotic stages during which the organism is exposed. The optimal temperatures for different genes or proteins acting during meiosis need not even be the same. Very little is known about temperature-induced protein changes, but studies in Lilium and Mus L. demonstrated that elevated temperatures selectively reduced the activity of a Rec protein possibly involved in crossing-over (Stem, 1986).

The magnitude of changes in MI chromosome pairing in hybrids can be significant even within limited temperature ranges. In hybrids between a number of varieties of Hordeum vulgare and H. bulbosum (2x) grown at 15 [degrees] C and 21 [degrees] C, respectively, changes in average chiasma frequency varied from statistically insignificant to over 100% (3.35-8.15 chiasmata per cell) (Pickering, 1990). Minor but still significant differences have been reported in, for example, amphihaploids of Poa annua L. (Hovin, 1958). However, other studies have shown virtually no effects of temperature on meiotic pairing; e.g., several full-sib genotypes of Festuca mairei St.-Yves (4x) x F. arundinacea Schreb. (4x) grown at 14 [degrees] C, 21 [degrees] C, and 28 [degrees] C had almost the same level of pairing (Kopyto et al., 1989).

Nutrition or ionic uptake are other factors suggested to be able to influence chromosome pairing. However, very few studies and none of them involving hybrid organisms have been performed. Law (1963) showed an effect of potassium concentrations on chiasma frequency in Lolium temulentum, and Bennett and Rees (1970) applying different concentrations of phosphate to two genotypes of rye, observed a significant increase in chiasma frequency correlated with an increasing availability of phosphate. Similar results were obtained by Fedak (1973) in a desynaptic barley mutant. In autotetraploid rye a significant change in the pattern of pairing was found in response to different nitrogen treatments, but the chiasma frequency remained the same (Hossain, 1978). Magnesium deficiency has been shown to induce chromosome stickiness and breakage during meiosis in Tradescantia Rupp. ex L. (Steffensen, 1953).

Oehlkers (1940) recorded the frequency of crossing-over in Oenothera L. and Antirrhinum Tourn. ex L. and observed pronounced differences between plants grown under different levels of water-stress. Whether the observed effects were actually caused by the accessibility of water, through a changed ionic uptake, or by some other factor(s) is uncertain.

A number of investigators have observed significant differences in the number of chiasmate chromosomes between fixations made at different times from the same plant. Silicas and Gerstel (1962) reported minor but significant differences in the number of bivalents between fixations made on different days of some Nicotiana hybrids, and similarly Menzel and Wilson (1966) observed changes in the pattern of chromosome pairing in a Hibiscus hybrid. Kihara (1929) demonstrated quite dramatic differences in the pattern of chromosome pairing in some Triticum/Aegilops hybrids, and ascribed the variation to environmental conditions. However, the data shown by Kihara (1929) are based on bulk fixations from five plants, and as such may not offer any conclusive evidence. The investigators explained the above differences by environmental conditions, possibly temperature (Kihara, 1929; Sfikas & Gerstel, 1962; Menzel & Wilson, 1966). However, as no measurements have been made on any environmental factors apart perhaps from daily temperatures, explaining the different results remains a matter of speculation. Whatever the causes, it must, however, be disturbing for the genome analysts that "homology," and thus relationships between species, may change overnight!

If chiasma frequency also depends on external, physical factors, the practitioners of genome analysis need to explain under which conditions "homology" is to be assessed, which raises the obvious question: To what extent are the experiments reproducible?

VI. Observational Limitations in Genome Analysis

Correct identification of the paired chromosomes in diploid as well as polyploid hybrids would seem to be of crucial importance for genome analysis and a minimum requirement for assessing homology. The previously cited studies of Hypochoeris (Parker, 1975) and Crepis (Tease & Jones, 1976) are examples showing the importance of proper chromosome identification, and that failure in living up to this may lead to incorrect conclusions and generalizations. Identification of the parental origin of the paired chromosomes is almost always insufficient, and in the vast majority of studies using genome analysis, identification of the individual chromosomes has never been attempted.

Direct identification of all chromosomes during meiosis is only possible in extremely rare cases, but hybrids between species of Crepis do offer an example (Togby, 1943). Some species of Crepis have very low chromosome numbers (e.g., C. neglecta L., 2n = 8; C. fuliginosa Sibth. & Sm., 2n = 6) and the individual chromosomes can readily be identified by their length, centromere position, and the presence or absence of satellites. Studies of chromosome pairing at MI in their interspecific hybrid showed that three of the chromosomes were able to pair with two different chromosomes (Togby, 1943). In this case the chromosome pairing behaviour may be caused by translocations, but in general it shows that in hybrids where identification of chromosomes is not possible, the paired chromosomes observed in one cell may be different from the paired chromosomes in another, even when the same numbers and types of configurations are observed.

Hybrids between in- and outbreeding species of Lolium (Jenkins, 1985), between some species of Bromus (Stebbins, 1947), between Pennisetum typhoides and P. purpureum Schum. (Jauhar, 1969), and between Secale and Hordeum (Petersen, 1991) offer examples of instances where it is possible directly to recognize the parental origin of the chromosome during meiosis because of their size differences, allowing separation of homo- and heteromorphic bivalents. In hybrids where chromosome identification is not possible, the standard and only rarely verified assumption is that pairing takes place only between chromosomes of different parental origin. However, such ad hoc assumptions have often been falsified by observations from hybrids where the parental origins of chromosomes have been identified. For example, in Pennisetum, chromosome pairing was in earlier studies interpreted as purely allosyndetic between chromosomes of P. typhoides (2n = 14) and P purpureum (2n = 28), but Jauhar (1968) showed that a large number of the bivalents observed in the hybrid was actually formed between the two sets of chromosomes of P. purpureum.

In hybrids between Bromus marginatus Nees (2n = 8x = 56; genomic composition ABCL) and B. laevipes Shear (2n = 2x = 14; genome L) the large chromosomes belonging to the L genome can be distinguished from the remaining smaller chromosomes (Stebbins, 1947). Analysis of meiosis in the hybrid revealed that the large genomes formed an average of 6.2 bivalents per cell, whereas the smaller genomes only formed an average of 2.0 bivalents per cell. If identification of the paired chromosomes had not been possible, one would have been left with an average bivalent frequency of 8.2. Considering that the hybrid contains no less than five genomes the interpretation would hardly have been correct [ILLUSTRATION FOR FIGURE 1 OMITTED].

Hordeum depressum (Scribn. & Sm.) Rydb. (2n = 4x = 28) was originally considered an allotetraploid (Bothmer et al., 1988b; Bothmer & Jacobsen, 1989) and pairing in hybrids with other species belonging to the Triticeae was always interpreted as allosyndetic. Thus, the high level of pairing in the hybrid between H. depressum and Hordelymus europaeus (L.) C. O. Harz (2n = 4x = 28) was considered evidence that Hordelymus shared the H genome with Hordeum (Bothmer & Jacobsen, 1989). However, in hybrids between Hordeum depressum and species of Secale (2n = 14), which has chromosomes considerably larger than the Hordeum chromosomes, almost complete pairing of the two sets of chromosomes of H. depressum was observed (Petersen, 1991). Hordeum depressum was accordingly considered an autotetraploid, and pairing in the hybrid with Hordelymus was reinterpreted as autosyndesis between the H. depressum genomes leaving the Hordelymus genomes unpaired and without a known "relationship" (Petersen, 1991; Bothmer et al., 1994). This example is an excellent illustration of how easily chromosome pairing in hybrids may be misinterpreted when the parental chromosomes cannot be distinguished. Until direct evidence exists, pairing in H. depressum x Hordelymus hybrids cannot be assessed, no matter how many new hybrid combinations are being studied. It is still possible that the observed pairing in this hybrid is between the Hordeum and the Hordelymus chromosomes, that it involves only the Hordeum chromosomes, that it involves only the Hordelymus chromosomes, or that it is mixture of all possible types. The more genomes that are combined in one hybrid, the more ad hoc assumptions are needed in the attempts to interpret the observed pairing, and the less reliable are the conclusions that are drawn.

To solve the problem of identification of the individual paired chromosomes in interspecific hybrids, some investigators have used either C-banding (e.g., Lelley, 1975; Jouve et al., 1980, 1982, 1985; Hutchinson et al., 1983; Benavente & Orellana, 1986; Cunado et al., 1986; Orellana et al., 1985, 1989; Galindo & Jouve, 1989; Cunado, 1992; Gonzalez et al., 1993; Fernandez-Calvin & Orellana, 1994; Fernandez-Calvin et al., 1995) or in situ hybridization (e.g., Hutchinson et al., 1980; King et al., 1993, 1994; Reyes-Valdes et al., 1996) with the meiotic metaphase chromosomes. This may solve the practical problems of chromosome identification and make it possible to eliminate some errors, but it cannot solve the conceptual problems.

With problems of chromosome idenfication come problems of distinguishing chiasmatic MI association from achiasmatic (= secondary) association. Lack of distinction, which may be extremely difficult, may lead to an overestimation of chiasma frequency. However, the absolute level of achiasmatic association is generally low (e.g., Majisu & Jones, 1971; Orellana, 1985) and hence, probably only a minor problem for genome analysis.

VII. Miscellaneous Points

Another serious problem that ought to be adequately addressed by practitioners of genome analysis is the often pronounced variation in chiasma frequency that can be observed between cells, progeny from taxonomically identical hybrid combinations, and reciprocal hybrids, etc. Some aspects of these problems have been dealt with before by Gaul (1953, 1959) and us (Petersen & Seberg, 1996).

In nearly all published papers using genome analysis, the conclusions are based on the average chiasma frequencies calculated from observations of preferably 50 pollen mother-cells of the hybrids (Jauhar & Joppa, 1996). It remains enigmatic why the number 50 was chosen as giving a reliable estimate of the mean chiasma frequency. The number of observations made in any given case to calculate the mean with a certain level of confidence is a function of dispersion. Hence, in some cases 50 cells may be more than enough and in others hardly enough at all. (In the example in Fig. 2, observations from at least 143 cells are required to estimate the mean value within 95% confidence limits.) Kimber, in a personal communication to Crane (1996), argued that observations from 20-30 cells would be appropriate input data to calculate "genome affinity" using mathematical models. Uniform data could not be retrieved from greater numbers of cells from a single anther or floret of Aegilops, and other florets or inflorescences were likely to produce significantly different results! based on the mean chiasma frequency alone, distributions are often - without mathematical tests - considered different or similar. This is hardly a justifiable scientific approach.

The only other parameter used to describe the often peculiar distributions [ILLUSTRATION FOR FIGURES 2, 3 OMITTED] is the range, although it is rarely used in the arguments. The observed range may be considerable, even when only two genomes are brought together (Petersen & Seberg, 1996): e.g., in the diploid hybrid (2n = 14) between Hordeum brachyantherum Nevski and H. muticum Presl, the range is 1-13, with all figures in between represented (Bothmer et al., 1986). Ranges like this are not restricted to either Hordeum or the Triticeae. Hybrids of Brassica, for example, and closely related taxa (Harberd & McArthur, 1980), Nicotiana (Goodspeed, 1954), Carthamus (Estilai & Knowles, 1978), and Cucumis (Singh & Yadava, 1984) offer substantial evidence along the same lines. In polyploid hybrids the range of variation is equally high, but more difficult to interpret [ILLUSTRATION FOR FIGURES 2, 3 OMITTED]. Gaul (1959) showed that the maximum pairing value observed even had to be considered a minimum of the number of chromosomes able to pair because analysis of more cells might broaden the range of observed variation. Had the distributions been normal (actually binominal, as the variable is discrete) it would be justifiable to describe it using only two parameters, but this is rarely the case.

But what, then, is the biological meaning of this large variation? For all practical considerations, all cells in these hybrids will contain the same DNA. Thus, within the limits of variation of all observations, the actual homology (degree of similarity?) between the chromosomes of one cell must be very close to or identical to the homology (degree of similarity?) between the chromosomes of another. Nevertheless, according to the fundamental premise of genome analysis, the genomes in cells with no or very little pairing would be interpreted as non-homologous, whereas in cells with full pairing they would be regarded as completely homologous. Considering the range in chromosome pairing between cells (e.g., from almost non-homology to almost complete homology, as in the above-mentioned hybrid between Hordeum brachyantherum and H. muticum), it seems rather obvious that chromosome pairing tells little if anything about homology (and similarity?). Rather it illustrates our pronounced lack of knowledge about the processes leading to pairing of chromosomes.

Using average chiasma frequencies does not solve the problem, because how can such values be understood biologically? As measures of some sort of "average homology"? Only Gaul (1953, 1959) appears to have addressed the problem, but rhetorically rather than analytically. Searching the literature has not revealed attempts to explain or justify why exactly the average is more important than any other value, e.g., the minimum, the observed or expected maximum, or the mode. In older literature the modal value was actually more frequently used than the average (e.g., Kihara & Lilienfeld, 1932; Kostoff, 1941b). A gradual shift from using the mode to using the average can be observed, but we have failed to find any justification either for the use of any of these values or for the shift in use.

When discussing the possible importance of different values describing chromosome pairing it might be of interest to look at the actual distribution of chiasmata in hybrids (Petersen & Seberg, 1996). The intuitive expectation might be that chiasmata would be either binominally distributed with the mode equal to or close to the mean ([ILLUSTRATION FOR FIGURE 3E OMITTED]: Aegilops geniculata Roth x Triticum durum Desf.; 3F: HH10183-1) or form distributions sloping steeply from either zero (in hybrids with virtually no pairing) [ILLUSTRATION FOR FIGURE 3H OMITTED]: HH10339-2 or the maximally observed number of chiasmata (in hybrids or species with normal, full pairing). Hybrids with such chiasma distributions do occur, but deviating distributions are common. The distributions may be skewed [ILLUSTRATION FOR FIGURES 2, 3A OMITTED], more or less flattened [ILLUSTRATION FOR FIGURE 3H OMITTED]: HH10339-1, without any apparent top-point at all [ILLUSTRATION FOR FIGURE 3B OMITTED], or bimodal [ILLUSTRATION FOR FIGURE 3C OMITTED]: BB7271A; 3D. Whether the mode or the average is used has limited biological significance, but when the distribution is skewed, then the mode may mathematically represent the distribution of chiasmata better than the average. When all observed numbers of chiasmata occur with almost the same frequency, the average seems hardly more representative than any other value. And when the distribution is bimodal then the two modal values may be more representative than the average, which will often be located close to the trough between the modes. The latter type of distribution may possibly be restricted to polyploid hybrids including more than three genomes and may reflect different patterns of pairing between different sets of genomes (Petersen & Seberg, 1996). We believe that this can only be demonstrated by identification of the paired chromosomes, not by any mathematical calculations. When distributions are bimodal, assumption 3 in the mathematical models is probably violated.

Among the above examples, only the chiasma distribution of the triploid hybrid between the diploid Pseudoroegneria cognata (Hackel) A. Love (S genome) and the tetraploid Elymus semicostatus (Nees ex Steud.) Melderis (SY genomes) [ILLUSTRATION FOR FIGURE 2 OMITTED] may precisely represent the second step of genome analysis (crossing with a diploid analyzer species) as outlined by Kihara (1930). However, there is no particular reason to believe that chiasma distributions in this type of hybrid should be governed by other rules than in other triploid hybrids. In the hybrids between diploid species of Secale (R genome) and tetraploid species of Hordeum, believed to be segmental allotetraploids (modified H genomes) [ILLUSTRATION FOR FIGURE 3A, B, G, H OMITTED], it can be directly observed that only the chromosomes from the Hordeum genomes are involved in chiasma formation (Petersen, 1991). It hardly matters whether pairing and chiasma formation occur between two genomes of different parental origin (as the S genomes from Elymus and Pseudoroegneria) or between two genomes from the same parent (as the H genomes from Hordeum).

The conclusions drawn from genome analysis are further burdened when data from chromosome pairing are available for more comparable hybrid combinations. These may be offspring from just one cross. Such offspring will (again for all practical considerations) contain the same DNA, thus, the actual degree of similarity (homology?) between the chromosomes in one plant must be very close to identical to the degree of similarity (homology?) between the chromosomes in another plant, though of course plants may differ genetically according to the level of heterozygosity of the parents. However, pronounced differences between patterns of pairing as well as average chiasma frequencies of such offspring exist [ILLUSTRATION FOR FIGURE 3F, G, H OMITTED]. Average chiasma frequencies in offspring from a single cross between Hordeum brachyantherum (4x) and Secale cereale varied from 4.4 to 10.5 chiasmata (involving only Hordeum chromosomes) per cell (Petersen, 1991) [ILLUSTRATION FOR FIGURE 3G OMITTED]. Offspring from a cross between a variety of Triticum aestivum and Agropyron intermedium Reichb. varied in chiasma frequency from 1.4 to 12.5, and offspring from a cross involving another wheat variety varied from 9.2 to 20.5 chiasmata per cell (Gaul, 1953). As soon as the comparison of chiasma frequencies is broadened to include all taxonomically identical hybrids, the range of variation increases still further [ILLUSTRATION FOR FIGURE 3F OMITTED]. In the case of the Triticum/Agropyron combination we are now faced with a range of 1.4 to 20.5 chiasmata per cell. Examples of broad ranges of chromosome pairing are numerous and can be found among hybrids of, e.g., Elymus (Lu & Bothmer, 1993a), Hordeum/Secale (Petersen, 1991), and Nicotiana (Kostoff, 1943; Goodspeed, 1954). Gaul (1953, 1954, 1959) has pointed out that wide ranges of chromosome pairing are most likely, and even expected, when multiple hybrid plants derived from genetically different parents are analyzed. For genome analysis this has the implication that when only one or a few hybrids of a given combination are analyzed, they may not represent the possible range of variation. If, on the other hand, many hybrids are analyzed and a wide range of chromosome pairing is observed, then which pairing value has the highest importance? The lowest, the highest, or the average? And is the range totally irrelevant?

Another set of taxonomically but not necessarily genetically identical hybrids that must be compared are reciprocal crosses. Meiotic data for reciprocal hybrids are scarce, probably because of both convenience (hybridization is usually easier in one of the directions) and conviction (there is a general belief that the data will be similar). However, very different chiasma frequencies can be observed, as Kihara (1929) noted. In a comparison of reciprocal crosses between Triticum durum Desf. and Aegilops geniculata Roth (syn. Aegilops ovata L.), he observed virtually no pairing and approximately 4 chiasmata per cell in hybrids, depending on which species functioned as pollen donor ([ILLUSTRATION FOR FIGURE 3E OMITTED]; Kihara, 1929). Significant differences in chiasma frequency are also observed in reciprocal hybrids of, e.g., Elymus (Aung & Walton, 1990; Lu & Bothmer, 1993b), Brassica and Brassica/Sinapis L. (Harberd & McArthur, 1980), and Gossypium (Phillips, 1966). These differences in chiasma frequency are hardly surprising, considering the above variation in range among offspring from one single cross or taxonomically similar crosses. There are no reasons to assume that reciprocal hybrids will behave differently than any other types of hybrids. Of the above reciprocal hybrids, only the cross between Elymus trachycaulus (Link) Gould ex Shinners and E. canadensis L. involved genotypically well-characterized and presumably highly homozygotic parents (Aung & Walton, 1990). In one direction an average of approximately 22 chiasmata per cell were formed, whereas in the other 26-27 chiasmata per cell were formed. Aung and Walton (1990) interpreted the difference as attributable to influence from cytoplasmic genes, i.e., mitochondrial or chloroplast genes. If this is correct, genome analysis is faced with yet another disturbing factor.

Less specifically, "cytoplasmic effects" or differences in "genetic background" are often used as explanations when the same sets of chromosomes have been found to behave differently under various hybrid and non-hybrid conditions. The influence of "genetic background" can be illustrated by comparisons between pairing in a polyploid species, pairing in the dihaploid or polyhaploid derived from it, and pairing in different hybrid combinations. The genus Hordeum offers a good example, because a large number of both dihaploids, polyhaploids, and hybrids have been synthesized. In some cases data from polyhaploids and hybrids exist for the same accession, but mostly comparisons can only be made between different accessions. However, even when the same accession is used, it is most likely a collection of different genotypes, each in themselves heterozygous. In general, the polyploid species have a normal meiosis with regular bivalent formation (e.g., Bothmer et al., 1983; Bothmer & Subrahmanyam, 1988), whereas in the dihaploids and polyhaploids only little pairing can be observed (Subrahmanyam, 1978; Bothmer & Subrahmanyam, 1988), and in different hybrid combinations pairing may vary substantially (Petersen, 1991). For example, one tetraploid accession of H. brachyantherum (2n = 28) formed an average of 13.9 bivalents per cell (Bothmer et al., 1983; Bothmer & Subrahmanyam, 1988), but its dihaploid showed an average of 0.3 bivalents per cell (Bothmer & Subrahmanym, 1988). The same accession has been included in a large number of interspecific and intergeneric hybrids, but only the data from hybrids with Secale are conclusive with regard to pairing of the Hordeum chromosomes (see above). However, in hybrid combinations with different taxa of Secale the chromosomes of the same H. brachyantherum accession formed from an average of 0.8 to an average of 6.4 bivalents per cell (Petersen, 1991). Comparable data exist for an accession of the hexaploid H. procerum Nevski that shows slight irregularities at meiosis with an average of only 15.6 bivalents and 2.5 multivalents per cell (Bothmer et al., 1983; Bothmer & Subrahmanym, 1988). In polyhaploids, no multivalents and only an average of 0.7 bivalents per cell have been observed (Bothmer & Subrahmanym, 1988), whereas hybrid combinations with taxa of Secale exhibit an average of 3.2-6.0 Hordeum bivalents per cell and up to an average of 1.2 Hordeum multivalents per cell (Petersen, 1991). Considering the likely heterogeneity of the Hordeum material and the many cited examples of pairing variability, these differences are unsurprising. However, if genome analysis is to be applied, it must be decided which pairing value estimates "homology" best. The usual way of explaining unexpected problems is to suggest the presence of pairing promoting or pairing suppressing genes, as, e.g., in Lolium and Festuca (Jauhar, 1975a), Hordeum (Subrahmanyam, 1978; Bothmer et al., 1987a; Bothmer & Subrahmanyam, 1988), Elymus (Lu et al., 1993; Motsny & Simonenko, 1996), Avena (Nishiyama et al., 1989), and Brassica (Harberd & McArthur, 1980). Multiple genes for regulating chromosome pairing do exist, but how can it be determined when all these have been balanced or neutralized in such a way that the observed pairing reflects nothing but chromosome similarity? And after all, what does chromosome similarity mean?

VIII. Genome Analysis, Phylogeny, and Classification

A. CHROMOSOME PAIRING AS A CHARACTER

The data used in genome analysis are usually derived from pairwise comparisons between two specimens (representing two species) and are most often presented as the number of chiasmata observed in MI in a specified number of cells, frequently 50 (on the adequacy of this figure, see above). Hence, data from genome analysis are inherently comparative and never related to observations made on individual specimens. Despite the fact that in each instance they are observed as discrete values (e.g., number of chiasmata per cell), they are converted into continuous variables. As other data captured as continuous variables, e.g., DNA-DNA hybridization data, the data from genome analysis share all the shortcomings of this type of data (Swofford et al., 1996; Werman et al., 1996). In general, the arithmetic mean is supposed to be an adequate representation of the data, and any other statistical measurement that could be related to them (e.g., the range) is neglected. The mean is either used directly or the observations are scaled between 0 and 1 by calculation of the c-value, i.e. the mean arm-pairing frequency, in order to infer relationships. Rarely are the data analyzed in a deeper, rigorous manner (see, however, below), but occasionally they are more or less arbitrarily divided into a number of categories and summarized in "pairing polygons" (e.g., Bothmer & Jacobsen, 1991; Salomon & Lu, 1992; Lu, 1993a). It has been shown that the grouping of continuous data into discrete categories cannot be done in a non-arbitrary manner (Gift & Stevens, 1997) by simply inspecting the data, and assignment of any observation to a group entirely neglects the tremendous scatter found in the original observations. Even though the data are not actual morphometric data, applying similar coding methods (e.g., see Thiele, 1993) would have led to a more rigorous and non-arbitrary data analysis.

The underlying rationale behind the exercise is, however, in either case the same and may simply be viewed as a straightforward extension of the biological species concept - i.e., the higher the mean or c-value, the closer the relationships between the entities, as mirrored in classical biosystematic investigations. Thus, for example, the higher the fertility of interspecific hybrids the closer their relationships.

B. CHROMOSOME PAIRING AND TAXONOMIC PHILOSOPHIES

Data obtained through genome analysis are, like all other data acquired as pairwise measures, only amenable to treatment within a phenetic framework. However, some attempts have been made in this direction (Lu, 1993a; Lu & Bothmer, 1993b; Lu & Salomon, 1993). Methods for analysing distance data minimally require that the data are additive in expectation, but it has never been shown how data from genome analysis behave under this assumption. Additionally, it seems peculiar that data from genome analysis are grouped by a method that is indifferent to any statement about homology (Sober, 1993: 164).

Although it is beyond the scope of this paper, we have little confidence in ability of phenetic methods to depict relationships (Farris, 1979, 1983).

Whereas the level of pairing between the chromosomes of two different species in general is a continuous character, the endpoints of the continuum, complete pairing and complete lack of pairing, may reasonably be defined objectively as a discrete character (Seberg, 1989). In any group of organisms that show such all-or-none behaviour of chromosome pairing, this character may in turn be used in phylogenetic analyses together with other characters of the same organisms.

Normally the chromosomes in a given species pair during meiosis, but upon divergence and speciation, this ability may be lost. Hence, the inability to pair has variable categorical significance (Brady, 1983; Mishler & Brandon, 1987; Cracraft, 1989; Kellogg, 1989; Seberg, 1989) and it is the ability to pair that, by definition, is the plesiomorphic character state and consequently is phylogenetically uninformative (Kellogg, 1989; Seberg, 1989). If pairing ability is used as the only or main indicator of relationships, as is often the case in genome analysis or in classical biosystematic investigations (Grant, 1981), non-monophyletic groups are hypothesized and the relationships between groups cannot be determined (Seberg, 1989).

At best, data from genome analysis are measures of similarity between taxa. Hence, classifications based on data from genome analysis must be regarded as purely phenetic, whether they are based on arbitrarily defined limits of levels of pairing or the result of distance-based tree-building algorithms.

In no other taxon have arbitrarily defined limits between levels of pairing been used as rigidly as in the Triticeae, where genera are defined solely by possessing different genomes or combinations of genomes, defined by the level of chromosome pairing in interspecific/intergeneric hybrids (Dewey, 1984; Love, 1984). Recently it was formally decided that to merit recognition as separate genomes, chromosome pairing between them must be less than 50% (Wang et al., 1996).

However, there is far from universal agreement on when genomes should be considered similar or different. In taxa other than the Triticeae, where genome definition is based on frequencies of chiasmata, the genomes and taxonomic categories have been defined through frequencies of meiotic chromosome configurations (e.g., in Mentha: Ikeda & Oho, 1991; Glycine: Hymovitz et al., 1991; Avena: Rajharty, 1991; and Gossypium: Endrizzi et al., 1985).

Singh (1993) generally prefers the Genome Affinity Index, or GAI (Menzel & Martin, 1970), measured as the average frequencies of bivalents (and multivalents) divided by the basic chromosome number (hence it varies between 0 and 1) to define genomes and estimate their relatedness. Genomes with a GAI above 0.9 are considered similar, hence, apparently a stronger requirement than that used in the Triticeae - though the two measures are not directly comparable.

Although mostly used to define groups of species, genome analysis has also been used to make decisions on species delimitation. Lucas and Jahier (1987), though observing an absolutely normal meiosis in hybrids between Triticum urartu Tum. and T. boeoticum Boiss., maintained the taxa as separate species because different levels of chromosome pairing were obtained in hybrids with other species. Feldman (1977) considered both taxa conspecific with T. monococcum L.

It is obvious that a classification based on the ability of chromosomes to pair may serve as a useful tool for breeders or others interested in the possibilities of transfer of genetic material between species. However, a such special-purpose classification has no place in a broader scientific context.

C. GENOMES AND GENERA

The definition of genera used in the Triticeae appears to be unique (Love, 1982: 204). Applying the same approach (one genome or combination of genomes = one genus) to other taxa would often significantly increase the number of genera - e.g., Gossypium should probably be split into eight genera, Glycine into seven, and in the classical example from Drosera (see above), the three species would have to be assigned to three different genera (Seberg, 1989). However, in other taxa, such as the Orchidaceae, the number of genera would most likely be strongly reduced.

The rationale behind the idea of defining genera by their genomic composition stems from the recognition of the genome as a basic unit of evolution (Love & Connor, 1982: 180). The distinct genome or combination of genomes serves as a barrier to natural hybridization (Love, 1982; Love & Connor, 1982). Following this point of view, the genus must also be considered a basic evolutionary unit (Seberg, 1989). Although a very controversial issue that will not be dealt with in details here, only the species (or less inclusive wholes) are generally accepted as evolutionary units (Hull, 1980), whereas higher taxonomic categories are only historical entities. However, the possible existence of a hierarchy of evolutionary processes, as hinted at by Love (1986: 43) has been suggested before (Eldredge & Gould, 1972; Stanley, 1979; Vrba, 1980). Some authors, however, consider species to be no different from other taxa and view the species problem as a red herring (e.g., Nelson, 1989, 1994; Rieppel, 1994).

IX. Conclusion

Whereas genome analysis may give an indication, albeit imprecise, of the ability of chromosomes to pair, and hence may be of potential value for plant breeding, it has no role to play in taxonomy and is of limited value in understanding evolution. Genome analysis suffers from limited knowledge of the mechanisms behind meiosis and an imprecise handling of data; it is pervaded by numerous ad hoc assumptions, lacks a clear connection to the concept of homology, and cannot be given a meaningful biological interpretation.

Hence, continued use of genome analysis in systematics seems futile. However, given that our knowledge of the mechanisms behind meiosis is steadily increasing, new genome-specific characters suitable for phylogenetic analysis may emerge.

X. Acknowledgments

We thank Bjorn Salomon and Bao-Rong Lu for letting us have access to their original data for Elymus and Pseudoroegneria hybrids. We are indebted to Chris Humphries, Victor Albert, and Ib Linde-Laursen for their valuable comments on the manuscript. The Danish Natural Science Foundation is gratefully acknowledged for financial support (grant 9601340). Flemming Sarup provided skillful technical assistance.

XI. Literature Cited

Ajimura, M., S.-H. Leem & H. Ogawa. 1992. Identification of new genes required for meiotic recombination in Saccharomyces cerevisiae. Genetics 133: 51-66.

Alani, E., R. Padmore & N. Kleckner. 1990. Analysis of wild-type and rad50 mutants of yeast suggests an intimate relationship between meiotic chromosome synapsis and recombination. Cell 61: 419-436.

Alonso, L. C. & G. Kimber. 1981. The analysis of meiosis in hybrids. II. Triploid hybrids. Canad. J. Genet. Cytol. 23: 221-234.

Asay, K. H. & D. R. Dewey. 1979. Bridging ploidy differences in crested wheatgrass with hexaploid x diploid hybrids. Crop Sci. 19: 519-523.

Assadi, M. 1996. Experimental hybridization and genome analysis in Elymus L. sect. Caespitosae and sect. Elytrigia (Poaceae: Triticeae). Pp. 23-38 in R. R.-C. Wang et al. (eds.), Proceedings of the 2nd International Triticeae Symposium. Utah State University, Salt Lake City.

Aung, T. & P. D. Walton. 1990. Morphology and cytology of the reciprocal hybrids between Elymus trachycaulus and Elymus canadensis. Genome 33: 123-130.

Avivi, L. 1976. The effect of genes controlling different degrees of homoeologous pairing on quadrivalent frequency in induced autotetraploid lines of Triticum longissimum. Canad. J. Genet. Cytol. 18: 357-364.

Baden, C., R. von Bothmer, J. Flink & N. Jacobsen. 1989. Intergeneric hybridization between Psathyrostachys and Hordeum. Nordic J. Bot. 9: 333-342.

Barkworth, M. E. 1992. Taxonomy of the Triticeae: a historical perspective. Hereditas 116: 1-14.

----- & D. R. Dewey. 1985. Genomically based genera in the perennial Triticeae of North America: identification and membership. Amer. J. Bot. 72: 767-776.

Baum, B. R., J. R. Estes & P. K. Gupta. 1987. Assessment of the genomic system of classification in the Triticeae. Amer. J. Bot. 74: 1388-1395.

Bayliss, M. W. & R. Riley, R. 1972. An analysis of temperature-dependent asynapsis in Triticum aestivum. Genet. Res., Camb. 20: 193-200.

Beasley, J. O. 1942. Meiotic chromosome behavior in species, species hybrids, haploids, and induced polyploids of Gossypium. Genetics 27: 25-54.

Beer, C. G. 1980. Perspectives on animal behavior comparisons. Pp. 17-64 in M. H. Bornstein (ed.), Comparative methods in psychology. Erlbaum, Hillsdale.

Benavente, E. & J. Orellana. 1986. Pairing competition between identical and homologous chromosomes in diploid and tetraploid cells of rye telotrisomic plants. Canad. J. Genet. Cytol. 28: 568-573.

Bennett, M. D. & H. Rees. 1970. Induced variation in chiasma frequency in rye in response to phosphate treatments. Genet. Res., Camb. 16: 325-331.

-----, J. B. Smith & R. Kemble. 1972. The effect of temperature on meiosis and pollen development in wheat and rye. Canad. J. Genet. Cytol. 14: 615-624.

Bishop, D. K., D. Park, L. Xu & N. Kleckner. 1992. DMC1: a meiosis-specific yeast homolog or E. coli recA required for recombination, synaptonemal complex formation, and cell cycle progression. Cell 69: 439-456.

Bothmer, R. von & N. Jacobsen. 1989. Intergeneric hybridization between Hordeum and Hordelymus (Poaceae). Nordic J. Bot. 9: 113-117.

----- & N. Jacobsen. 1991. Interspecific hybrids within the genus Hordeum. Pp. 411-431 in P. K. Gupta & T. Tsuchiya (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part A. Elsevier, Amsterdam.

----- & N. C. Subrahmanyam. 1988. Assessment of chromosome associations in haploids and their parental accessions in Hordeum. Genome 30: 204-210.

-----, J. Flink, N. Jacobsen, M. Kotimaki & T. Landstrom. 1983. Interspecific hybridization with cultivated barley (Hordeum vulgare L.). Hereditas 99: 219-244.

-----, M. Kotimaki & Z. Persson. 1985. Genome relationships between eight diploid Hordeum species. Hereditas 103: 1-16.

-----, J. Flink & T. Landstrom. 1986. Meiosis in interspecific Hordeum hybrids. I. Diploid combinations. Canad. J. Genet. Cytol. 28: 525-535.

-----, ----- & -----. 1987a. Meiosis in Hordeum interspecific hybrids. II. Triploid hybrids. Evol. Trends Pl. 1: 41-50.

-----. M. Kotimaki & L. Linde-Laursen. 1987b. Genome relationships between Psathyrostachys huashanica and P. fragilis (Poaceae). Pl. Syst. Evol. 156: 183-188.

-----, J. Flink & T. Landstrom. 1988a. Meiosis in interspecific Hordeum hybrids. III. Tetraploid (2x x 6x) hybrids. Hereditas 108: 141-148.

-----, ----- & -----. 1988b. Meiosis in interspecific Hordeum hybrids. IV. Tetraploid (4x x 4x) hybrids. Genome 30: 479-485.

-----, ----- & -----. 1989a. Meiosis in interspecific Hordeum hybrids. VI. Hexaploid hybrids. Evol. Trends Pl. 3: 53-58.

-----, -----, ----- & H. M. Thomas. 1989b. Meiosis in interspecific Hordeum hybrids. V. Pentaploid hybrids. Hereditas 110: 217-226.

-----, B.-R. Lu, I. & Linde-Laursen. 1994. Intergeneric hybridization and C-banding patterns in Hordelymus (Triticeae, Poaceae). Pl. Syst. Evol. 189: 259-266.

-----, N. Jacobsen, C. Baden, R. B. Jorgensen & I. Linde-Laursen. 1995. An Ecogeographical study of the genus Hordeum. Ed. 2. IPGRI, Rome.

Brady, R. H. 1983. Parsimony, hierarchy, and biological implications. Pp. 49-60 in N. I. Platnick & V. A. Funk (eds.), Advances in cladistics. Vol. 2. Columbia University Press, New York.

Brandham, P. E. & M. D. Bennett (eds.). 1995. Kew Chromosome Conference IV. Royal Botanic Gardens, Kew.

Burson, B. L. 1978. Genome relations between Paspalum conspersum and two diploid Paspalum species. Canad. J. Genet. Cytol. 20: 365-372.

-----. 1981a. Cytogenetic relationships between Paspalum jurgensii and P. intermedium, P. vaginatum, and P. setaceum var. ciliatifolium. Crop Sci. 21: 515-519.

-----. 1981b. Genome relations among four diploid Paspalum species. Bot. Gaz. 142: 592-596.

Chapman, C. G. D. & G. Kimber. 1992a. Developments in the meiotic analysis of hybrids. I. Review of theory and optimization in triploids. Heredity 68: 97-103.

----- & -----. 1992b. Developments in the meiotic analysis of hybrids. II. Amended models for tetraploids. Heredity 68: 105-113.

----- & -----. 1992c. Developments in the meiotic analysis of hybrids. III. Amended models for pentaploids. Heredity 68: 193-200.

----- & -----. 1992d. Developments in the meiotic analysis of hybrids. IV. Utilizing datasets with merged figure classes. Heredity 68: 201-204.

----- & -----. 1992e. Developments in the meiotic analysis of hybrids. V. Second-order models for tetraploids and pentaploids. Heredity 68: 205-210.

Charpentier, A., M. Feldman & Y. Cauderon. 1988. The effect of different Agropyron elongatum chromosomes on pairing in Agropyron-common wheat hybrids. Genome 30: 978-983.

Chatterjee, R. & G. Jenkins. 1993. Meiotic chromosome interactions in inbred autotetraploid rye (Secale cereale). Genome 36: 131-138.

Chen, Q., J. Jahier & Y. Cauderon. 1993. The B chromosome system of Inner Mongolian Agropyron Gaertn. 2. Effects of the Bs on homologous and homoeologous meiotic chromosome pairing. Caryologia 46: 293-299.

Chennaveeraiah, M. S. & S. C. Hiremath. 1974. Genome analysis of Eleusine coracana (L.) Gaertn. Euphytica 23: 489-495.

----- & -----. 1991. Cytogenetics of minor millets. Pp. 613-627 in P. K. Gupta & T. Tsuchiya (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part A. Elsevier, Amsterdam.

Cracraft, J. 1989. Speciation and its ontology: the empirical consequences of alternative species concepts for understanding patterns and processes of differentiation. Pp. 28-58 in D. Otte & J. A. Endler (eds.), Speciation and its consequences. Sinauer Associates, Sunderland, MA.

Crane, C. F. 1996. Numerical meiotic models for the inference of genomic affinity in polyploids. Pp. 61-79 in P. P. Jauhar (ed.), Methods of genome analysis in plants. CRC Press, Boca Raton, FL.

----- & D. A. Sleper. 1989a. A model of meiotic chromosome association in triploids. Genome 32: 82-98.

----- & -----. 1989b. A model of meiotic chromosome association in tetraploids. Genome 32: 691-707.

Cuadrado, C., C. Romero & J. R. Lacadena. 1991. Meiotic pairing control in wheat-rye hybrids. I. Effect of different wheat chromosome arms of homoeologous group 3 and 5. Genome 34: 72-75.

Cunado, N. 1992. Analysis of metaphase I chromosome association in species of the genus Aegilops. Theor. Appl. Genet. 85: 283-292.

-----, M. C. Cermeno & J. Orellana. 1986. Interactions between wheat, rye and Aegilops ventricosa chromosomes on homologous and homoeologous pairing. Heredity 56: 219-226.

Darlington, C. D. 1932. The control of the chromosomes by the genotype and its bearing on some evolutionary problems. Amer. Naturalist 66:25-51.

-----. 1937. Recent advances in cytology. Ed. 2. J. & A. Churchill, London.

Davies, A., G. Jenkins & H. Rees. 1990. Diploidisation of Lotus corniculatus L. (Fabaceae) by elimination of multivalents. Chromosoma 99: 289-295.

Dewey, D. R. 1969. Hybrids between tetraploid and hexaploid crested wheatgrasses. Crop Sci. 9: 787-791.

-----. 1972a. Cytogenetic and genomic relationships of Elymus giganteus with E. dasystachys and E. junceus. Bull. Torrey Bot. Club 99: 77-83.

-----. 1972b. Cytogenetics of Elymus angustus and its hybrids with Elymus giganteus, Elymus cinireus, and Agropyron repens. Bot. Gaz. 133: 57-64.

-----. 1975, The origin of Agropyron smithii. Amer. J. Bot. 62: 524-530.

-----. 1980. Hybrids and induced amphiploids of Agropyron repens x A. curvifolium. Crop Sci. 20: 473-478.

-----. 1982. Genomic and phylogenetic relationships among North American perennial Triticeae. Pp. 51-81 in J. R. Estes et al. (eds.), Grasses and grasslands. University of Oklahoma Press, Norman.

-----. 1984. The genomic system of classification as a guide to intergeneric hybridization with the perennial Triticeae. Pp. 209-279 in J. P. Gustafson (ed.), Gene manipulation in plant improvement. Plenum Press, New York.

Diez, M., M. M. Jimenez & J. L. Santos. 1993. Synaptic patterns of rye B chromosomes. II. The effect of the standard B chromosomes on the pairing of the A set. Theor. Appl. Genet. 87: 17-21.

Digby, L. 1912. The cytology of Primula Kewensis and of other related Primula-hybrids. Ann. Bot. 26: 357-388.

Dover, G. A. 1973. The genetics and interactions of 'A' and 'B' chromosomes controlling meiotic chromosome pairing in the Triticinae. Pp. 653-666 in E. R. Sears & L. M. S. Sears (eds.), Proceedings of the 4th International Wheat Genetics Symposium. Missouri Agriculture Experiment Station, Columbia, MO.

----- & R. Riley. 1972. Prevention of pairing of homoeologous meiotic chromosomes of wheat by an activity of supernumerary chromosomes of Aegilops. Nature 240: 159-161.

----- & R. Riley. 1977. Inferences from genetical evidence on the course of meiotic chromosome pairing in plants. Phil. Trans. Roy. Soc. London, ser. B 277: 313-326.

Driscoll, C. J., L. M. Bielig & N. L. Darvey. 1979. An analysis of frequencies of chromosome configurations in wheat and wheat hybrids. Genetics 91: 755-767.

-----, G. H. Gordon & G. Kimber. 1980. Mathematics of chromosome pairing. Genetics 95: 159-169.

Dvorak, J. 1983. Evidence for genetic suppression of heterogenetic chromosome pairing in polyploid species of Solanum, sect. Petota. Canad. J. Genet. Cytol. 25: 530-539.

----- 1987. Chromosomal distribution of genes in diploid Elytrigia elongata that promote or suppress pairing of wheat homoeologous chromosomes. Genome 29: 34-40.

Eldredge, N. & S. Gould. 1972. Punctuated equilibria, an alternative to phyletic gradualism. Pp. 82-115 in T. M. J. Schopf (ed.), Models in paleobiology. Freeman Cooper, San Francisco.

Elliott, C. G. 1955. The effect of temperature on chiasma frequency. Heredity 9: 385-398.

Endrizzi, J. E., E. L. Turcotte & R. J. Kohel. 1984. Qualitative genetics, cytology, and cytogenetics. Pp. 81-129 in R. J. Kohel & C. F. Lewis (eds.), Cotton. Agronomy Monograph no. 24. American Society of Agronomists, Madison, WI.

-----, ----- & -----. 1985. Genetics, cytology, and evolution of Gossypium. Adv. Genet. 23: 271-375.

Engebrecht, J., J. Hirsch & G. S. Roeder. 1990. Meiotic gene conversion and crossing over: their relationship to each other and to chromosome synapsis and segregation. Cell 62: 927-937.

Espinasse, A. & G. Kimber. 1981. The analysis of meiosis in hybrids. IV. Pentaploid hybrids. Canad. J. Genet. Cytol. 23: 627-638.

-----, J. Foueillassar & G. Kimber. 1995. Cytogenetical analysis of hybrids between sunflower and four wild relatives. Euphytica 82: 65-72.

Estilai, A. & P. F. Knowles. 1978. Relationship of Carthamus leucocaulos to other Carthamus species (Compositae). Canad. J. Genet. Cytol. 20: 221-233.

Evans, G. M. & A. J. Macefield. 1973. The effect or B chromosomes on homoeologous pairing in hybrid species. I. Lolium temulentum x Lolium perenne. Chromosoma 41: 63-73.

Farooq, S., N. Iqbal & T. M. Shah. 1990. Promotion of homoeologous chromosome pairing in hybrids of Triticum aestivum x Aegilops variabilis. Genome 33: 825-828.

Farris, J. S. 1979. The information content of the phylogenetic system. Syst. Zool. 28: 483-519.

-----. 1981. Distance data in phylogenetic analysis. Pp. 3-23 in V. A. Funk & D. R. Brooks (eds.), Advances in cladistics. Proceedings of the first meeting of the Willi Hennig Society. New York Botanical Garden, Bronx.

-----. 1983. The logical basis of phylogenetic analysis. Pp. 1-36 in N. I. Platnick & V. A. Funk (eds.), Advances in cladistics. Vol. 2. Columbia University Press, New York.

-----. 1985. Distance data revisited. Cladistics 1: 67-85.

Fedak, G. 1973. Increased chiasma frequency in desynaptic barley in response to phosphate treatments. Canad. J. Genet. Cytol. 15: 647-649.

-----. 1980. Production, morphology and meiosis of reciprocal barley-wheat hybrids. Canad. J. Genet. Cytol. 22: 117-123.

Federley, H. 1913. Das Verhalten der Chromosomen bei der Spermatogenese der Schmetterlinge Pygaera anachoreta, curtula und picra sowie einiger ihrer Bastarde. Z. Indukt. Abstammungs-Vererbungsl 9: 1-100.

-----. 1914. Ein Beitrag zur Kenntnis der Spermatogenese bei Mischlingen zwischen Eltern verschiedener systematischer Verwandtschaft. Oefvers. Forh. Finska Vetensk.-Soc. 56 Afd. A. No. 13: 1-28.

-----. 1915a. Chromosomenstudien an Mischlingen. Oefvers. Forh. Finska Vetensk.-Soc. 57 Afd. A. No. 26: 1-36.

-----. 1915b. Chromosomenstudien an Mischlingen. II. Die Spermatogenese des Bastards Dicranura erminea ([female]) x D. vinula ([male]). Oefvers. Forh. Finska Vetensk.-Soc. 57 Afd. A. No. 30: 1-26.

Fedotova, Yu. S., Yu. F. Bogdanov, S. A. Gadzhiyeva, S. A. Sosnikhina, V. G. Smirnov & E. I. Mikhailova. 1994. Meiotic mutants of rye Secale cereale L. II. The nonhomologous synapsis in desynaptic mutants sy7 and sy10. Theor. Appl. Genet. 88: 1029-1036.

Feldman, M. 1977. Historical aspects and significance of the discovery of wild wheats. Stadler Genet. Symp. 9: 121-146.

Fernandez-Calvin, B. & J. Orellana. 1994. Metaphase I-bound arms frequency and genome analysis in wheat-Aegilops hybrids. 3. Similar relationships between the B genome of wheat and the S or [S.sup.1] genomes of Ae. speltoides, Ae. longissima and Ae. sharonensis. Theor. Appl. Genet. 88: 1043-1049.

-----, ----- & D. Pignone. 1995. Genome analysis of triploids using mathematical models. I. Effects of translocations not detected by conventional staining techniques. Hereditas 122: 41-45.

Frederiksen, S. 1991. Taxonomic studies in Dasypyrum (Poaceae). Nordic J. Bot. 11: 135-142.

-----. 1994. Hybridization between Taeniatherum caput-medusae and Triticum aestivum (Poaceae). Nordic J. Bot. 14: 3-6.

----- & R. von Bothmer. 1986. Relationships in Taeniatherum (Poaceae). Canad. J. Bot. 64: 2343-2347.

----- & -----. 1995. Intergeneric hybridizations with Eremopyrum (Poaceae). Nordic J. Bot. 15: 39-47.

Gale, M. D. & T. E. Miller. 1987. The introduction of alien genetic variation in wheat. Pp. 173-210 in F. G. H. Lupton (ed.), Wheat breeding. Chapman & Hall, London.

Galindo, C. & N. Jouve. 1989. C-banding in meiosis: an approach to the study of wheat and rye genome interactions in triticale. Genome 32: 1074-1078.

Gaul, H. 1953. Genomanalytische Untersuchungen bei Triticum x Agropyrum intermedium unter Berucksichtigung von Secale cereale x A. intermedium. Z. Indukt. Abstammungs-Vererbungsl. 85: 505-546.

-----. 1954. Asynapsis und ihre Bedeutung fur die Genomanalyse. Z. Indukt. Abstammungs-Vererbungsl. 86: 69-100.

-----. 1959. A critical survey of genome analysis. Pp. 194-206 in B. C. Jenkins (ed.), Proceedings of the 1st International Wheat Genetics Symposium. Public Press Limited, Winnipeg.

Geerts, J. M. 1911. Cytologische Untersuchungen einiger Bastarde von Oenothera gigas. Ber. Deutsch. Bot. Ges. 29: 160-166.

Gift, N. & P. F. Stevens. 1997. Vagaries in the delimitation of character states in quantitative variation - an experimental study. Syst. Biol. 46: 112-125.

Gill, B. S. & G. Kimber. 1989. Genome analysis in higher plants. Genome 31: 1087.

Gonzalez, J. M., S. Bernard & M. Bernard. 1993. Metaphase-I analysis of a Triticum aestivum x T. monococcum hybrid by the C-banding technique. Euphytica 68: 187-192.

Goodspeed, T. H. 1954. The genus Nicotiana. Origin, relationships and evolution of its species in the light of their distribution, morphology and cytogenetics. Chron. Bot. Co., Waltham, Mass.

Goodwin, B. C. 1984. Changing from an evolutionary to a generative paradigm in biology. Pp. 99-120 in J. W. Pollard (ed.), Evolutionary theory: paths into the future. Wiley, New York.

Gottschalk, W. 1973. The genetic control of meiosis. Genetics 74, suppl. 2: 99.

----- & M. L. H. Kaul. 1980a. Asynapsis and desynapsis in flowering plants. I. Asynapsis. Nucleus 23: 1-15.

----- & -----. 1980b. Asynapsis and desynapsis in flowering plants. II. Desynapsis. Nucleus 23: 97-120.

Grant, V. 1981. Plant speciation. Ed. 2. Columbia University Press, New York.

Grell, R. F. 1969. Meiotic and somatic pairing. Pp. 361-492 in E. W. Caspari & A. W. Ravin (eds.), Genetic organization. Academic Press, New York.

Gupta, P. K. & G. Fedak. 1985. Genetic control of meiotic chromosome pairing in polyploids in the genus Hordeum. Canad. J. Genet. Cytol. 27: 515-530.

Harberd, D. J. 1972 Bivalent-forming natural autotetraploids in the Brassiceae. Heredity 29: 394.

----- & E. D. McArthur. 1980. Meiotic analysis of some species and genus hybrids in the Brassiceae. Pp. 65-87 in S. Tsunoda et al. (eds.), Brassica crops and wild allies. Biology and breeding. Japan Scientific Societies Press, Tokyo.

Hawley, R. S. & T. Arbel. 1993. Yeast genetics and the fall of the classical view of meiosis. Cell 72: 301-303.

Hempel, C. G. 1966. Philosophy of natural sciences. Prentice-Hall, Totowa, NJ.

Henderson, S. A. 1962. Temperature and chiasma formation in Schistocerca gregaria. II. Cytological effects at 40 [degrees] C and the mechanism of beat-induced univalence. Chromosoma 13: 437-463.

-----. 1963. Temperature and chiasma formation in Schistocerca gregaria. I. An analysis of the response at a constant 40 [degrees] C. Heredity 18: 77-94.

Hewitt, G. M. 1967. An interchange which raises chiasma frequency. Chromosoma 21: 285-295.

----- & B. John. 1965. The influence of numerical and structural chromosome mutations on chiasma conditions. Heredity 20: 123-135.

Hillis, D. M. 1994. Homology in molecular biology. Pp. 339-368 in B. K. Hall (ed.), Homology. The hierarchical basis of comparative biology. Academic Press, San Diego.

Hollingsworth, N. M., L. Goetsch & B. Byers. 1990. The HOP1 gene encodes a meiosis-specific component of yeast chromosomes. Cell 61: 73-84.

Holm, P. B. 1986. Chromosome pairing and chiasmata formation in allohexaploid wheat, Triticum aestivum analyzed by spreading of meiotic nuclei. Carlsberg Res. Commun. 51: 239-294.

----- & S. W. Rasmussen. 1984. The synaptonemal complex in chromosome pairing and disjunction. Chromosomes Today 8: 104-116.

----- & X. Wang. 1988. The effect of chromosome 5B on synapsis and chiasma formation in wheat, Triticum aestivum cv. Chinese Spring. Carlsberg Res. Commun. 53: 191-208.

Hossain, M. G. 1978. Effects of external environmental factors on chromosome pairing in autotetraploid rye. Cytologia 43: 21-34.

Hovin, A. W. 1958. Meiotic chromosome pairing in amphihaploid Poa annua L. Amer. J. Bot. 45: 131-138.

Hull, D. L. 1980. Individuality and selection. Ann. Rev. Ecol. Syst. 11: 311-332.

Huskins, C. L. 1932. A cytological study of Vilmorin's unfixable dwarf wheat. J. Genet. 25: 113-124.

Hutchinson, J., V. Chapman & T. E. Miller. 1980. Chromosome pairing at meiosis in hybrids between Aegilops and Secale species: a study by in situ hybridisation using cloned DNA. Heredity 45: 245-254.

-----, T. E. Miller & S. M. Reader. 1983. C-banding at meiosis as a means of assessing chromosome affinities in the Triticeae. Canad. J. Genet. Cytol. 25: 319-323.

Hymovitz, T., R. G. Palmer & R. J. Singh. 1991. Cytogenetics of the genus Glycine. Pp. 53-63 in T. Tsuchiya & P. K. Gupta (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part B. Elsevier, Amsterdam.

Ikeda, N. & S. Ono. 1991. Cytogenetics of the genus Mentha. Pp. 565-580 in T. Tsuchiya & P. K. Gupta (eds), Chromosome engineering in plants: genetics, breeding, evolution. Part B. Elsevier, Amsterdam.

Ivanov, E. L., V. G. Korolev & F. Fabre. 1992. XRS2, a DNA repair gene og Saccharomyces cerevisiae, is needed for meiotic recombination. Genetics 132: 651-664.

Jackson, R. C. 1982. Polyploidy and diploidy: new perspectives on chromosome pairing and its evolutionary implications. Amer. J. Bot. 69: 1512-1523.

-----. 1984. Chromosome pairing in species and hybrids. Pp. 67-86 in W. F. Grant (ed.), Plant biosystematics. Academic Press, Toronto.

-----. 1991. Cytogenetics of polyploids and their diploid progenitors. Pp. 159-180 in P. K. Gupta & T. Tsuchiya (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part A. Elsevier, Amsterdam.

----- & J. Casey. 1982. Cytogenetic analyses of autopolyploids: models and methods for triploids to octoploids. Amer. J. Bet. 69: 487-501.

----- & D. P. Hauber. 1982. Autotriploid and autotetraploid analyses: correction coefficients for proposed binomial models. Amer. J. Bet. 69: 644-646.

Jauhar, P. P. 1968. Inter-and intra-genomal chromosome pairing in an interspecific hybrid and its bearing on basic chromosome number in Pennisetum. Genetica 39: 360-370.

-----. 1975a. Chromosome relationships between Lolium and Festuca (Gramineae). Chromosoma (Berlin) 52: 103-121.

-----. 1975b. Genetic control of diploid-like meiosis in hexaploid tall fencue. Nature 254: 595-597.

-----. 1977. Genetic regulation of diploid-like chromosome pairing in Arena. Theor. Appl. Genet. 49: 287-295.

-----. 1981. Cytogenetics of pearl millet. Adv. Agron. 34: 407-479.

-----. 1988. A reassessment of genome relationships between Thinopyrum bessarabicum and T. elongatum of the Triticeae. Genome 30: 903-914.

-----. 1996. Methods of genome analysis in plants. CRC Press, Boca Raton, FL.

----- & C. F. Crane. 1989. An evaluation of Baum et al.'s assessment of the genomic system of classification in the Triticeae. Amer. J. Bet. 76: 571-576.

----- & L. R. Joppa. 1996. Chromosome pairing as a tool in genome analysis: merits and limitations. Pp. 9-37 in P. P. Jauhar (ed.), Methods of genome analysis in plants. CRC Press, Boca Raton, FL.

Jenkins, G. 1985. Synaptonemal complex formation in hybrids of Lolium temulentum x Lolium perenne (L.). I. High chiasma frequency diploid. Chromosoma 92: 81-88.

-----. 1986. Synaptonemal complex formation in hybrids of Lolium temulentum x Lolium perenne (L.). III. Tetraploid. Chromosoma 93: 413-419.

-----, J. White & J. S. Parker. 1988. Elimination of multivalents during meiotic prophase in Scilla autumnalis. II. Tetraploid. Genome 30: 940-946.

John, B. 1990. Meiosis. Cambridge University Press, Cambridge.

----- & M. King. 1985. The inter-relationship between heterochromatin distribution and chiasma distribution. Genetica 66: 183-194.

----- & B. Naylor. 1961. Anomalous chromosome behaviour in the germ line of Schistocerca gregaria. Heredity 16: 187-198.

Johzuka, K. & H. Ogawa. 1995. Interaction of Mre11 and Rad50: two proteins required for DNA repair and meiosis-specific double-strand break formation in Saccharomyces cerevisiae. Genetics 139: 1521-1532.

Jones, R. N. 1991. Cytogenetics of B-chromosomes in crops. Pp. 141-157 in P. K. Gupta & T. Tsuchiya (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part A. Elsevier, Amsterdam.

----- & H. Rees. 1982. B-chromosomes. Academic Press, London.

Jouve, N., N. Diez & M. Rodriguez. 1980. C-banding in 6x-Triticale x Secale cereale L. hybrid cytogenetics. Theor. Appl. Genet. 57: 75-79.

-----, D. Montalvo & C. Soler. 1982. C-banding in cytogenetics of 6x-Triticale x Triticum aestivum L. hybrids. Z. Pflanzenzuchtg. 88: 311-321.

-----, J. M. Gonzalez, A. Fominaya & E. Ferrer. 1985. The analysis of meiosis of the B genome in common wheat. Canad. J. Genet. Cytol. 27: 17-22.

Kaul, M. L. H. & T. G. K. Murthy. 1985. Mutant genes affecting higher plant meiosis. Theor. Appl. Genet. 70: 449-466.

Kellogg, E. A. 1989. Comments on genomic genera in the Triticeae (Poaceae). Amer. J. Bet. 76: 796-805.

Kihara, H. 1924. Cytologische und genetische Studien bei wichtigen Getreidearten mit besonderer Rucksicht auf dan Verhalten der Chromosomen und die Sterilitat in den Bastarden. Mem. Coll. Sci. Kyoto Imp. Univ., ser. B 1: 1-200.

-----. 1929. Conjugation of homologous chromosomes in the genus hybrids Triticum x Aegilops and species hybrids of Aegilops. Cytologia 1: 1-15.

-----. 1930. Genomanalyse bei Triticum und Aegilops, Cytologia 1: 263-284.

----- & F. Lilienfeld. 1932. Genomanalyse bei Triticum und Aegilops. IV. Untersuchungen an Aegilops x Triticum- und Aegilops x Aegilops-Bastarden. Cytologia 3: 384-456.

----- & T. Ono. 1926. Chromosomenzahlen und Systematische Gruppierung der Rumex-Arten. Z. Zellforsch. Mikrosk. Anat. 4: 475-481.

Kimber, G. 1961. Basis of the diploid-like meiotic behaviour of polyploid cotton. Nature 191: 98-100.

-----. 1983. Genome analysis in the genus Triticum. Pp. 23-28 in S. Sakamoto (ed.), Proceedings of the 6th International Wheat Genetics Symposium. Plant Germplasm Institute, Kyoto University, Kyoto.

-----. 1984. Evolutionary relationships and their influence on plant breeding. Pp. 281-293 in J. P. Gustafson (ed.), Gene manipulation in plant improvement. Plenum Press, New York.

-----. 1986. The use of patterns in the study of the evolution of allopolyploids. Pp. 61-70 in C. Barigozzi (ed.), The origin and domestication of cultivated plants. Elsevier, Amsterdam.

----- & L. C. Alonso. 1981. The analysis of meiosis in hybrids. III. Tetraploid hybrids. Canad. J. Genet. Cytol. 23: 235-254.

----- & M. Feldman. 1987. Wild wheat. An introduction. Special Report 353, College of Agriculture. University of Missouri, Columbia.

----- & M. M. Hulse. 1979. The analysis of chromosome pairing in hybrids and the evolution of wheat. Pp. 63-72 in S. Ramanujam (ed.), Proceedings of the 5th International Wheat Genetics Symposium. Indian Society for Genetics and Plant Breeding, New Delhi.

----- & Y. Yen. 1990. Genomic analysis of diploid plants. Proc. Natl. Acad. Sci. U.S.A. 87: 3205-3209.

-----, L. C. Alonso & P. J. Sallee. 1981. The analysis of meiosis in hybrids. I. Aneuploid hybrids. Canad. J. Genet. Cytol. 23: 209-219.

King, I. P., K. A. Purdie, S. E. Orford, S. M. Reader & T. E. Miller. 1993. Detection of homoeologous chiasma formation in Triticum durum x Thinopyrum bessarabicum hybrids using genomic in situ hybridization. Heredity 71: 369-372.

-----, S. M. Reader, K. A. Purdie, S. E. Orford & T. E. Miller. 1994. A study of the effect of a homoeologous pairing promoter on chromosome pairing in wheat/rye hybrids using genomic in situ hybridization. Heredity 72: 318-321.

Kobayashi, T., Y. Hotta & S. Tabata. 1993. Isolation and characterization of a yeast gent that is homologous with a meiosis-specific cDNA from a plant. Molec. Gen. Genet. 237: 225-232.

Koduru, P. R. K. & M. K. Rao. 1981. Cytogenetics of synaptic mutants in higher plants. Theor. Appl. Genet. 59: 197-214.

Kopyto, R., C. F. Crane & D. A. Sleper. 1989. Effect of temperature on meiosis and fertility in Festuca mairei x Festuca arundinacea vat. glaucescens. Genome 32: 708-718.

Kostoff, D. 1941a. The problem of haploidy. (Cytogenetic studies on Nicotiana haploids and their bearings to some other cytogenetic problems.). Bibl. Genet. 13: 1-148.

-----. 1941b. Wheat phylesis and wheat breeding from a cytogenetic point of view. (Cytogenetic indices for the role of interspecific hybridization in the origin of wheat species and for applying interspecific hybridization in producing valuable wheat forms.). Bibl. Genet. 13: 149-224.

-----. 1943. Cytogenetics of the genus Nicotiana. State's Printing House, Sofia.

Kumar, H. 1991. Cytogenetics of safflower. Pp. 251-277 in T. Tsuchiya & P. K. Gupta (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part B. Elsevier, Amsterdam.

Ladizinsky, G. 1973. Genetic control of bivalent pairing in the Avena strigosa polyploid complex. Chromosoma 42: 105-110.

Lauder, G. V. 1994. Homology, form, and function. Pp. 151-196 in B. K. Hall (ed.), Homology. The hierarchical basis of comparative biology. Academic Press, San Diego.

Law, C. N. 1963. An effect of potassium on chiasma frequency and recombination. Genetica 33: 313-329.

Lelley, T. 1975. Identification of univalents and rod bivalents in Triticale with Giemsa. Z. Pflanzenzuchtg. 75: 252-256.

-----. 1976. Induction of homoeologous pairing in wheat by genes of rye suppressing chromosome 5B effect. Canad. J. Genet. Cytol. 18: 485-489.

Lilienfeld, F. A. 1951. H. Kihara: genome-analysis in Triticum and Aegilops. X. Concluding review. Cytologia 16: 101-123.

Lin, Y. J. 1982. Temperature and chiasma formation in Rhoeo spathacea var. variegata. Genetica 60: 25-30.

Liu, Z.-W. & R. R.-C. Wang. 1993. Genomic constitutions of Thinopyrum curvifolium, T. scirpeum, T. distichum, and T. junceum (Triticeae: Gramineae). Genome 36: 641-651.

Loidl, J. 1989. Effects of elevated temperature on meiotic chromosome synapsis in Allium ursinum. Chromosoma 97: 449-458.

-----. 1990. The initiation of meiotic chromosome pairing: the cytological view. Genome 33: 759-778.

-----. 1994. Cytological aspects of meiotic recombination. Experientia 50: 285-294.

Love, A. 1982. Generic evolution of the wheatgrasses. Biol. Zentralbl. 101: 199-212.

----- 1984. Conspectus of the Triticeae. Feddes Repert. 95: 425-521.

----- 1986. Some taxonomical adjustments in eurasiatic wheatgrasses. Veroff. Geobot. Inst. Rubel Zurich 87: 43-52.

----- & H. E. Connor. 1982. Relationships and taxonomy of New Zealand wheatgrasses. New Zealand J. Bot. 20: 169-186.

Lu, B.-R. 1993a. Genomic relationships within the Elymus parviglumis group (Triticeae: Poaceae). Pl. Syst. Evol. 187: 191-211.

-----. 1993b. Biosystematic investigations of Asiatic wheatgrasses - Elymus L. (Triticeae: Poaceae). Ph.D. dissertation, Swedish University of Agricultural Science, Svalov.

----- & B. Salomon. 1993. Two new Tibetan species of Elymus (Poaceae: Triticeae) and their genomic relationships. Nordic J. Bot. 13: 353-367.

----- & R. von Bothmer. 1993a. Genomic constitution of four Chinese endemic Elymus species: E. brevipes, E. yangii, E. anthosachnoides, and E. altissimus (Triticeae, Poaceae). Genome 36: 863-876.

----- & -----. 1993b. Meiotic analysis of Elymus caucasicus, E. longearistatus, and their interspecific hybrids with twenty-three Elymus species (Triticeae, Poaceae). Pl. Syst. Evol. 185: 35-53.

-----, K. B. Jensen & B. Salomon. 1993. Biosystematic study of hexaploids Elymus tschimganicus and E. glaucissimus. II. Interspecific hybridization and genomic relationship. Genome 36: 1157-1168.

-----, M. E. B. Naredo, A. B. Juliano & M. T. Jackson. 1997. Hybridization of AA genome rice species from Asia and Australia. II. Meiotic analysis of Oryza meridionalis and its hybrids. Genet. Res. Crop Evol. 44: 25-31.

Lucas, H. & J. Jahier. 1987. Differences in meiotic pairing induction between the A genomes of Triticum boeoticum Boiss. and T. urartu Tum. using diploid species of the subtribe Triticineae as analysers. Genome 29: 891-893.

Majisu, B. N. & J. K. Jones. 1971. Aegilops x Secale hybrids: the production and cytology of diploid hybrids. Genet. Res., Cambridge 17: 17-31.

Matsubayashi, M. 1991. Phylogenetic relationships in the potato and its relatives. Pp. 93-118 in T. Tsuchiya & P. K. Gupta (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part B. Elsevier, Amsterdam.

Menees, T. M., P. B. Ross-MacDonald & G. S. Roeder. 1992. MEI4, a meiosis-specific yeast gene required for chromosome synapsis. Molec. Cell. Biol. 12: 1340-1351.

Menzel, M. Y. & D. W. Martin. 1970. Genome affinities of four African diploid species of Hibiscus sect. Furcaria. J. Hered. 61: 179-184.

----- & F. D. Wilson. 1966. Hybrids and genome relations of Hibiscus sabdariffa, H. meeusei, H. radiatus and H. acetosella. Amer. J. Bot. 53: 270-275.

Mishler, B. D. & R. N. Brandon. 1987. Individuality, pluralism, and the phylogenetic species concept. Biol. Philos. 2: 37-54.

Mizushima, U. 1980. Genome analysis in Brassica and allied genera. Pp. 89-106 in S. Tsunoda et al. (eds.), Brassica crops and wild allies. Biology and breeding. Japan Scientific Societies Press, Tokyo.

Montgomery, T. H. 1901. A study of the chromosomes of the germ cells of Metazoa. Trans. Amer. Phil. Soc. Philadelphia 20: 154-236.

Moritz, C. & D. M. Hillis. 1996. Molecular systematics: context and controversies. Pp. 1-13 in D. M. Hillis et al. (eds.), Molecular systematics. Sinauer Associates, Sunderland, MA.

Motsny, I. I. & V. K. Simonenko. 1996. The influence of Elymus sibiricus L. genome on the diploidization system of wheat. Euphytica 91: 189-193.

Nag, D. K., H. Scherthan, B. Rockmill, J. Bhargava & G. S. Roeder. 1995. Heteroduplex DNA formation and homologous pairing in yeast meiotic mutants. Genetics 141: 75-86.

Nayar, N. M. 1973. Origin and cytogenetics of rice. Adv. Genet. 17: 153-292.

Nelson, G. 1989. Species and taxa: systematics and evolution. Pp. 60-81 in D. Otte & J. A. Endler (eds.), Speciation and its consequences. Sinauer Associates, Sunderland, MA.

-----. 1994. Homology and systematics. Pp. 101-149 in B. K. Hall (ed.), Homology: the hierarchical basis of comparative biology. Academic Press, San Diego.

Nezu, M., T. C. Katayama & H. Kihara. 1960. Genetic study of the genus Oryza. I. Crossability and chromosomal affinity among 17 species. Seiken Ziho 11: 1-11.

Nishiyama, I., T. Yabuno & T. Taira. 1989. Genomic affinity relationships in the genus Avena. Pl. Breed. (New York) 102: 22-30.

Oehlkers, F. 1940. Meiosis und crossing over. Biol. Zentralbl. 60: 337-348.

Ohta, S. 1991. Phylogenetic relationships of Aegilops mutica Boiss. with the diploid species of congeneric Aegilops-Triticum complex, based on the new method of genome analysis using its B-chromosomes. Mem. Coll. Agric. Kyoto Univ. 137: 1-116.

----- & M. Tanaka. 1983. Genome relationships between Ae. mutica and the other diploid Aegilops and Triticum species, based on the chromosome pairing in the hybrids with or without B-chromosomes. Pp. 983-991 in S. Sakamoto (ed.), Proceedings of the 6th International Wheat Genetics Symposium. Plant Germplasm Institute, Kyoto University, Kyoto.

Okamoto, M. 1957. Asynaptic effect of chromosome V. Wheat Inf. Serv. 5: 6.

Orellana, J. 1985. Most of the homoeologous pairing at metaphase I in wheat-rye hybrids is not chiasmatic. Genetics 111: 917-931.

-----, N. Cunado & M. C. Cermeno. 1985. Genome-specific control of meiotic pairing evidenced in mutant Aegilops ventricosa-Secale cereale amphidiploids. Theor. Appl. Genet. 71: 532-535.

-----, J. F. Vazquez & J. M. Carrillo. 1989. Genome analysis in wheat-rye-Aegilops caudata trigeneric hybrids. Genome 32: 169-172.

Pantulu, J. V. & M. K. Rao. 1982. Cytogenetics of pearl millet. Theor. Appl. Genet. 61: 1-17.

Parker, J. S. 1975. Chromosome-specific control of chiasma formation. Chromosoma 49: 91-406.

-----, R, W. Palmer, M. A. F. Whitehorn & L. A. Edgar. 1982. Chiasma frequency effects of structural chromosome change. Chromosoma 85: 673-686.

-----, G. H. Jones, L. A. Edgar & C. Whitehouse. 1990. The population cytogenetics of Crepis capillaris. III. B-chromosome effects on meiosis. Heredity 64: 377-385.

Patterson, C. 1982. Morphological characters and homology. Pp. 21-74 in K. A. Joysey & A. E. Friday (eds.), Problems of phylogenetic reconstruction. Academic Press, London.

-----, 1988. Homology in classical and molecular biology. Molec. Biol. Evol. 5: 603-625.

Pellew, C. & F. M. Durham. 1916. The genetic behaviour of the hybrid Primula kewensis, and its allies. J. Genet. 5: 159-182.

Petersen, G. 1991. Intergeneric hybridization between Hordeum and Secale. II. Analysis of meiosis in hybrids. Hereditas 114: 141-159.

----- & O. Seberg. 1996. Genomes, chromosomes, and genes and the concept of homology. Pp. 13-19 in R. R.-C. Wang et al. (eds.), Proceedings of the 2nd International Triticeae Symposium. Utah State University, Salt Lake City.

Phillips, L. L. 1966. The cytology and phylogenetics of the diploid species of Gossypium. Amer. J. Bot. 53: 328-335.

Pickering, R. 1990. The influence of temperature on chromosome pairing in diploid and triploid hybrids between Hordeum vulgare L. and H. bulbosum L. Hereditas 113: 221-226.

Prakash, S. & V. L. Chopra. 1991. Cytogenetics of crop Brassicas and their allies. Pp. 161-180 in T. Tsuchiya & P. K. Gupta (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part B. Elsevier, Amsterdam.

Prakken, R. 1943. Studies of asynapsis in rye. Hereditas 29: 475-495.

Rajhathy, T. 1991. The chromosomes of Avena. Pp. 449-467 in P. K. Gupta & T. Tsuchiya (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part A. Elsevier, Amsterdam.

Rees, H. 1957. Distribution of chiasmata in an 'asynaptic' locust. Nature 180: 559.

-----. 1961. Genotypic control of chromosome form and behaviour. Bot. Rev. (Lancaster) 27: 288-318.

Reyes-Valdes, M. H., Y. Ji, C. F. Crane, J. F. Taylor, M. N. Islam-Faridi, H. J. Price & D. M. Stelly. 1996. ISH-facilitated analysis of meiotic bivalent pairing. Genome 39: 784-792.

Rhoades, M. M. & E. Dempsey. 1973. Chromatin elimination induced by the B chromosome of maize. J. Hered. 64: 12-18.

Rieppel, O. 1994. Species and history. Pp. 31-50 in R. W. Scotland et al. (eds.), Models in phylogeny reconstruction. Clarendon Press, Oxford.

Riley, R. 1960. The diploidisation of polyploid wheat. Heredity 15: 407-429.

----- & V. Chapman. 1958. Genetic control of the cytologically diploid behaviour of hexaploid wheat. Nature 182: 713-715.

----- & C. N. Law. 1965. Genetic variation in chromosome pairing. Adv. Genet. 13: 57-114.

Rockmill, B. & G. S. Roeder. 1990. Meiosis in asynaptic yeast. Genetics 126: 563-574.

-----, J. Engebrecht, H. Scherthan, J. Loidl & G. S. Roeder. 1995. The yeast MER2 gene is required for chromosome synapsis and the initiation of meiotic recombination. Genetics 141: 49-59.

Rosenberg, O. 1903. Das Verbalten der Chromosomen in einer hyhriden Pflanze. Ber. Deutsch. Bot. Ges. 21: 110-119.

-----. 1904a. Uber die Tetradenteilung eines Drosera-Bastardes. Bet. Deutsch. Bot. Ges. 22: 47-53.

-----. 1904b. Uber die Reduktionsteilung in Drosera. Meddeland. Stockholms Hogskolas Bot. Inst. Kungl. Hofboktryckeriet, Stockholm.

-----. 1909. Cytologische und morphologische Studien an Drosera longifolia x rotundifolia. Kongl. Svenska Vetenskapsakad. Handl. 43: 1-65.

-----. 1917. Die Reduktionsteilung und ihre Degeneration in Hieracium. Svensk Bot. Tidskr. 11: 145-206.

Roth, V. L. 1991. Homologies and hierarchies: problems solved and unresolved. J. Evol. Biol. 4: 167-194.

Sakamoto, S. 1972. Intergeneric hybridisation between Eremopyrum orientale and Henrardia persica, an example of polyploid species formation. Heredity 28: 109-115.

-----. 1974. Intergeneric hybridization among three species of Heteranthelium, Eremopyrum and Hordeum, and its significance for the genetic relationships within the tribe Triticeae. New Phytol. 73: 341-350.

-----. 1991. The cytogenetic evolution of Triticeae grasses. Pp. 469-481 in P. K. Gupta & T. Tsuchiya (eds.), Chromosome engineering in plants: genetics, breeding, evolution. Part A. Elsevier, Amsterdam.

Salomon, B. 1993. Interspecific hybridizations in the Elymus semicostatus group (Poaceae). Genome 36: 899-905.

----- & B.-R. Lu. 1992. Genomic groups, morphology, and sectional delimitations in Eurasian Elymus (Poaceae, Triticeae). Pl. Syst. Evol. 180: 1-13.

----- & -----. 1994. Interspecific hybridization among species of the Elymus semicostatus and Elymus tibeticus groups (Poaceae). Pl. Syst. Evol. 189: 1-13.

Sano, J. & M. Tanaka. 1983. Aegilops speltoides as a useful tool for genome analysis in wheat. Pp. 1095-1101 in S. Sakamoto (ed.), Proceedings of the 6th International Wheat Genetics Symposium. Plant Germplasm Institute, Kyoto University, Kyoto.

Sax, K. 1922. Sterility in wheat hybrids. II. Chromosome behavior in partially sterile hybrids. Genetics 7: 513-552.

Sears, E. R. 1976. Genetic control of chromosome pairing in wheat. Ann. Rev. Genet. 10: 31-51.

----- & M. Okamoto. 1958. Intergenomic chromosome relationships in hexaploid wheat. Proc. 10th Int. Congr. Genet. 2: 258-259.

Seberg, O. 1989. Genome analysis, phylogeny, and classification. Pl. Syst. Evol. 166: 159-171.

Sficas, A. G. & D. U. Gerstel. 1962. Statistical analysis of chromosome pairing in interspecific hybrids. II. Applications to some Nicotiana hybrids. Genetics 47: 1171-1185.

Shang, X. M., R. C. Jackson, H. T. Nguyen & J. Y. Huang. 1989. Chromosome pairing in the Triticum monococcum complex: evidence for pairing control genes. Genome 32: 216-226.

Sherman, J. D., S. M. Stack & L. K. Anderson. 1989. Two-dimensional spreads of synaptonemal complexes from solanaceous plants. IV. Synaptic irregularities. Genome 32: 743-753.

Singh, A. K. & K. S. Yadava. 1984. An analysis of interspecific hybrids and phylogenetic implications in Cucumis. Pl. Syst. Evol. 147: 237-252.

Singh, R. J. 1993. Plant cytogenetics. CRC Press, Boca Raton, FL.

----- & T. Hymovitz. 1985a. The genomic relationships among six wild perennial species of the genus Glycine subgenus Glycine Willd. Theor. Appl. Genet. 71: 221-230.

----- & -----. 1985b. Diploid-like meiotic behavior in synthesized amphiploids of the genus Glycine Willd. subgenus Glycine. Canad. J. Genet. Cytol. 27: 655-660.

-----, K. P. Kollipara & T. Hymovitz. 1988. Further data on the genomic relationships among wild perennial species (2n = 40) of the genus Glycine. Genome 30: 166-167.

Sjodin, J. 1970. Induced asynaptic mutants in Vicia faba L. Hereditas 66: 215-232.

Sober, E. 1993. Philosophy of biology. Oxford University Press, Oxford.

Soltis, D. E. & L. H. Rieseberg. 1986. Autopolyploidy in Tolmiea menziesii (Saxifragaceae): genetic insights from enzyme electrophoresis. Amer. J. Bot. 73: 310-318.

Stack, S. M. & D. Roelofs. 1996. Localized chiasmata and meiotic nodules in the tetraploid onion Allium porrum. Genome 39: 770-783.

Stanley, S. 1979. Macroevolution: pattern and process. W. H. Freeman, San Francisco.

Stebbins, G. L. 1947. The origin of the complex of Bromus carinatus and its phytogeographic implications. Contrib. Gray Herb. 165: 442-54.

-----. 1981. Chromosomes and evolution in the genus Bromus (Gramineae). Bot. Jahrb. Syst. 102: 359-379.

----- & F. T. Pun. 1953. Artificial and natural hybrids in the Gramineae, tribe Hordeae. V. Diploid hybrids of Agropyron. Amer. J. Bot. 40: 444-449.

Steffensen, D. 1953. Induction of chromosome breakage at meiosis by a magnesium deficiency in Tradescantia. Proc. Natl. Acad. Sci. U.S.A. 39: 613-620.

Stern, H. 1986. Meiosis: some considerations. J. Cell Sci. Suppl. 4: 29-43.

Stevens, P. F. 1984. Homology and phylogeny: morphology and systematics. Syst. Bot. 9: 395-409.

Subrahmanyam, N. C. 1978. Meiosis in polyhaploid Hordeum: Hemizygous ineffective control of diploid-like behaviour in a hexaploid? Chromosoma (Berlin) 66: 185-192.

Suja, J. A., C. Antonio, C. G. de la Vega & J. S. Rufas. 1994. Supernumerary chromosome segments and intrabivalent chiasma redistribution in Pyrgomorpha conica (Orthoptera). Heredity 73: 1-10.

Sutton, W. S. 1902. On the morphology of the chromosome group of Brachystola magna. Biol. Bull. 4: 24-39.

-----. 1903. The chromosomes in heredity. Biol. Bull. 4: 231-248.

Swofford, D. L., G. J. Olsen, P. J. Waddell & D. M. Hillis. 1996. Phylogenetic inference. Pp. 407-514 in D. M. Hillis et al., (eds.), Molecular systematics. Ed. 2. Sinauer Associates, Sunderland, MA.

Sybenga, J. 1988. Mathematical models for estimating preferential pairing and recombination in triploid hybrids. Genome 30: 745-757.

-----. 1994. Preferential pairing estimates from multivalent frequencies in tetraploids. Genome 37: 1045-1055.

-----. 1996. Limitations and pitfalls in the use of quantitative polyploid meiotic models for genome analysis. Pp. 81-97 in P. P. Jauhar (ed.), Methods of genome analysis in plants. CRC Press, Boca Raton, FL.

Sym, M., J. Engebrecht & G. S. Roeder. 1993. ZIP1 is a synaptonemal complex protein required for meiotic chromosome synapsis. Cell 72: 365-378.

Taylor, I. B. & G. M. Evans. 1977. The genotypic control of homoeologous chromosome association in Lolium temulentum x Lolium perenne interspecific hybrids. Chromosoma 62: 57-67.

Tease, C. & G. H. Jones. 1976. Chromosome-specific control of chiasma formation in Crepis capillaris. Chromosoma 57: 33-49.

Terasawa, M., A. Shinohara, Y. Hotta, H. Ogawa & T. Ogawa. 1995. Localization of RecA-like recombination proteins on chromosomes of the lily at various meiotic stages. Genes Develop. 9: 925-934.

Thiele, K. 1993. The Holy Grail of the perfect character: the cladistic treatment of morphometric data. Cladistics 9: 275-304.

Thomas, H. M. 1995. Meiosis of triploid Lolium. II. Discrepancies between the analyses of chromosome congigurations at metaphase I in inverse autoallotriploid combinations. Heredity 75: 446-452.

----- & R. A. Pickering. 1985. The influence of parental genotype on the chromosome behaviour of Hordeum vulgare x H. bulbosum diploid hybrids. Theor. Appl. Genet. 71: 437-442.

Tishkoff, D. X., B. Rockmill, G. S. Roeder & R. D. Kolodner. 1995. The sep1 mutant of Saccharomyces cerevisiae arrests in pachytene and is deficient in meiotic recombination. Genetics 139: 495-509.

Togby, H. A. 1943. A cytological study of Crepis fuliginosa, C. neglecta, and their [F.sub.1] hybrid, and its bearing on the mechanism of phylogenetic reduction in chromosome number. J. Genet. 45: 67-111.

U, N. 1935. Genome analysis in Brassica with special reference to the experimental formation of B. napus and peculiar mode of fertilization. Japan. J. Bot. 7: 389-452.

Vardi, A. & G. A. Dover. 1972. The effects of B chromosomes on meiosic and pre-meiotic spindles and chromosome pairing in Triticum/Aegilops hybrids. Chromosoma 38: 367-385.

Vrba, E. 1980. Evolution, species, and fossils: how does life evolve? South African J. Sci. 76: 61-84.

Waines, J. G. 1976. A model for the origin of diploidizing mechanisms in polyploid species. Amer. Naturalist 110: 415-430.

Wang, R. R.-C. 1989. An assessment of genome analysis based on chromosome pairing in hybrids of perennial Triticeae. Genome 32: 179-189.

-----. 1990. Comparative pairing in triploids and diploids of perennial Triticeae. Genome 33: 89-94.

-----. 1992. Genome relationships in the perennial Triticeae based on diploids hybrids and beyond. Hereditas 116: 133-136.

-----, R. von Bothmer, J. Dvorak, G. Fedak, I. Linde-Laursen & M. Muramatsu. 1996. Genome symbols in the Triticeae. Pp. 29-34 in R. R.-C. Wang et al., (eds.), Proceedings of the 2nd International Triticeae Symposium. Utah State University, Salt Lake City.

Watanabe, K. 1977. The control of diploid-like meiosis in polyploid taxa of Chrysanthemum (Compositae). Japan. J. Genet. 52: 125-131.

-----. 1981a. Studies on the control of diploid-like meiosis in polyploid taxa of Chrysanthemum. I. Hexaploid Ch. japonense Nakai. Cytologia 46: 459-498.

-----. 1981b. Studies on the control of diploid-like meiosis in polyploid taxa of Chrysanthemum. II. Octoploid Ch. ornatum Hemsley. Cytologia 46: 499-513.

Werman, S. D., M. S. Springer & R. J. Britten. 1996. Nucleic acids I: DNA-DNA hybridization. Pp. 169-203 in D. M. Hillis et al. (eds.), Molecular systematics. Ed. 2. Sinauer Associates. Sunderland, MA.

Wet, J. M. J. de & J. R. Harlan. 1972. Chromosome pairing and phylogenetic affinities. Taxon 21: 67-70.

Wettstein, D. von, S. W. Rasmussen & P. B. Holm. 1984. The synaptonemal complex in genetic segregation. Ann. Rev. Genet. 18: 331-413.

Winge, O. 1917. The chromosomes. Their number and general importance. Compte Rendue Tray. Labor. Carlsberg. 13: 131-206.

-----. 1925. Contributions to the knowledge of chromosome numbers in plants. Cellule 35: 305-324.

-----. 1932. On the origin of constant species-hybrids. Svensk Botanisk Tidskrift 26: 107-122.

Wood, C. E., Jr. 1955. Evidence for the hybrid origin of Drosera anglica. Rhodora 57: 103-130.

Wynne, J. C. & T. Halward. 1989. Cytogenetics and genetics of Arachis. Crit. Rev. Pl. Sci. 8: 189-220.
COPYRIGHT 1998 New York Botanical Garden
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Seberg, Ole; Petersen, Gitte
Publication:The Botanical Review
Date:Oct 1, 1998
Words:25382
Previous Article:The question of cotyledon homology in angiosperms.
Next Article:Seed and seedling ecology of pinon and juniper species in the pygmy woodlands of western North America.
Topics:

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters