The Considerable Genome Size Variation of Hordeum Species (Poaceae) Is Linked to Phylogeny, Life Form, Ecology, and Speciation Rates
http://www.100md.com
分子生物学进展 2004年第5期
Institute of Plant Genetics and Crop Plant Research, Gatersleben, Germany
E-mail: blattner@ipk-gatersleben.de.
Abstract
Genome size variation in plants is thought to be correlated with cytological, physiological, or ecological characters. However, conclusions drawn in several studies were often contradictory. To analyze nuclear genome size evolution in a phylogenetic framework, DNA contents of 134 accessions, representing all but one species of the barley genus Hordeum L., were measured by flow cytometry. The 2C DNA contents were in a range from 6.85 to 10.67 pg in diploids (2n = 14) and reached up to 29.85 pg in hexaploid species (2n = 42). The smallest genomes were found in taxa from the New World, which became secondarily annual, whereas the largest diploid genomes occur in Eurasian annuals. Genome sizes of polyploid taxa equaled mostly the added sizes of their proposed progenitors or were slightly (1% to 5%) smaller. The analysis of ancestral genome sizes on the base of the phylogeny of the genus revealed lineages with decreasing and with increasing genome sizes. Correlations of intraspecific genome size variation with the length of vegetation period were found in H. marinum populations from Western Europe but were not significant within two species from South America. On a higher taxonomical level (i.e., for species groups or the entire genus), environmental correlations were absent. This could mostly be attributed to the superimposition of life-form changes and phylogenetic constraints, which conceal ecogeographical correlations.
Key Words: DNA content ? flow cytometry ? genome size evolution ? Hordeum ? life form ? phylogeny
Introduction
Since the 1950s, large differences in the nuclear DNA content of different organisms were detected (Swift 1950; Price, Chambers, and Bachmann 1981; Laurie and Bennett 1985; Ceccarelli, Falistocco, and Cionini 1992; Bennett and Leitch 1995) and drew the attention of scientists to possible explanations. Because no correlation with organismal complexity could be found, these differences were described as "C value paradox" (Thomas 1971). During the following years, it became clear that not gene content, but the proportion of repetitive DNA, primarily influences nuclear DNA content (Flavell, Rimpau, and Smith 1977; Barakat, Carels, and Bernardi 1997). Particularly retrotransposon copy numbers can vary to a large extent among genomes and contribute markedly to genome size differences found in plants (Arumuganathan and Earle 1991; Vicient et al. 1999; Kalendar et al. 2000). The specific mode of retroelement mobilization, which transpose without excision, should, together with polyploidization, drive increase of genome size in plants. Because no comparably efficient mechanism of genome size reduction was identified, Bennetzen and Kellogg (1997) proposed the possibility of a "one-way ticket" to "genomic obesity" in plants. The rather large genomes of several derived groups of angiosperms (e.g., Liliaceae within monocots, Triticeae within grasses, and Santalales within eudicots) seemingly supported this hypothesis (Kellogg 1998; Leitch, Chase, and Bennett 1998).
Analyses of genome size distribution in various organisms found correlations of genome sizes with cytological, physiological, and ecological characters (Goin, Goin, and Bachmann 1968; Bennett 1976, 1987; Grime and Mowforth 1982; Jockusch 1997; Grime 1998; Ohri 1998; Gregory 2001, 2002; Knight and Ackerly 2002), even within single species (Price, Chambers, and Bachmann 1981; Laurie and Bennett 1985; Cullis and Clearly 1986; Bennett and Leitch 1995; Nevo 2001). The conclusions drawn in these studies are often contradictory. In the light of intraspecific DNA content changes under different environmental conditions (Turpeinen, Kulmala, and Nevo 1999; Kalendar et al. 2000; Nevo 2001), it seems questionable if analyses of genome size data in distantly related taxa are meaningful for the understanding of the mechanisms influencing genome size variation (Jockusch 1997). Furthermore, some authors reported possible phylogenetic constraints of genome sizes (Cox et al. 1998; Wendel et al. 2002). An explicit phylogenetic view on grass genomes resulted in clear genome size differences among the monophyletic grass subfamilies (Kellogg 1998; Gaut 2002), although for most groups, a considerable overlap of DNA contents existed. These studies analyzed, however, either variation on an intraspecific scale or variation among distantly related taxa, and the samples were often far from being representative for the groups. Data on genome size changes within clearly circumscribed plant groups that could fill the gap between analyses on the population level (Turpeinen, Kulmala, and Nevo 1999) and meta-analyses of entire plant families (Kellogg 1998), floras (Grime and Mowforth 1982; Knight and Ackerly 2002), life forms (Bennett and Leitch 1995), or of the plant kingdom (Bennett 1987) are scarce.
The Triticeae genus Hordeum L. comprises 31 species (including barley), which occur under a wide variety of climates in Eurasia and the New World (von Bothmer et al. 1995; Blattner, in press). Genome size of several Hordeum species and the contributions of BARE-1 retrotransposon activities to DNA content variation were analyzed by A. Schulman's group (Manninen and Schulman 1993; Kankanp??, Mannonen, and Schulman 1996; Suoniemi et al. 1996; Vicient et al. 1999; Kalendar et al. 2000). Their interpretation of genome size data was impeded by the lack of reliable phylogenetic information on Hordeum and resulted in the conclusion that genome size is not correlated with phylogenetic relationships and that DNA content may change faster than speciation takes place (Kankanp??, Mannonen, and Schulman 1996). Here, we report DNA content for 30 Hordeum species, determined by flow cytometry, and analyzed in a phylogenetic framework. We used these data to test for possible ecological and phylogenetical constraints, intraspecific genome size changes, and genome size evolution through time and to see how life form and polyploidization influenced DNA content variation in Hordeum.
Materials and Methods
Plant Material
Seeds from 134 accessions of 30 Hordeum species (46 taxa [table 1]) were germinated in petri dishes and then grown in pots containing standard soil in a greenhouse. This material belongs mostly to the Barley Core Collection (BCC [Knüpffer and van Hintum 2003]), a reference collection of Hordeum species and lineages, complemented by our own material from Patagonia and Eurasia. Accession numbers, the origins of the plant material, and basic climate data of the collection sites are shown in table A in Supplementary Material online. Up to 30 mg of fresh leaf tissue was used for flow cytometry analysis. At least three individuals per accession and one to 10 accessions per species were measured. For a more detailed analysis of intraspecific genome size variation, we analyzed 10 accessions of H. marinum subsp. marinum from Western Europe and 13 accessions of H. pubiflorum and 33 accessions of H. lechleri from southern Argentina. In total, 394 samples were measured.
Table 1 Hordeum Species Analyzed in This Study.
Genome Size Estimation
The preparation of nuclear suspensions and the determination of DNA content essentially followed the protocol given by Barow and Meister (2002, 2003), using a FACStarPlus flow cytometer (Becton Dickinson, San José, Calif.) equipped with two argon lasers INNOVA 90-5 (Coherent, Palo Alto, Calif.) and the analysis program CellQuest. Usually 10,000 nuclei per sample were analyzed.
Nuclear DNA content was estimated as a ratio between the fluorescence of the nuclei of the samples stained with propidium iodide (excitation at 514 nm and emission at 630 nm) and internal standards. Pisum sativum cv. "Viktoria, Kifejt? Borsó" (IPK GenBank accession number PIS 630) with a 2C genome size of 9.09 pg (Doleel et al. 1998) was used as the primary standard. For diploid and hexaploid Hordeum species, we included Secale cereale (IPK GenBank accession number R 737) as standard. The 2C genome size of S. cereale was calculated as 16.8 pg with regard to P. sativum.
Data Analyses
The average genome sizes of the Hordeum taxa were correlated with their phylogenetic position in the genus (fig. 1), as derived from the analysis of the nuclear rDNA internal transcribed spacer (ITS) region (Blattner, in press). This study involved cloned ITS sequences from multiple accessions of all Hordeum species, analyzed with cladistic, phenetic, and model-based approaches. The results of these analyses were summarized with an Adams consensus tree approach (Swofford 1991), integrating taxa with ambiguous clade affiliation in a polytomy at the last unambiguous branching point. Up to now, this is the only Hordeum phylogeny covering all diploid and polyploid species. The measured DNA contents of polyploids were compared with the sum of the genome sizes from their proposed parental species. Ancestral genome sizes within Hordeum were estimated for diploid lineages with the phylogenetic generalized least squares method (PGLS) in Compare version 4.4 (Martins 2003). For this calculation, the polytomies in the tree had to be resolved into bifurcations by introducing minute branch lengths. This procedure results in more or less equal size estimations at this artificial branching point, as along short branches, nearly no size changes take place. Genome size values for the outgroups Psathyrostachys Nevski and Dasypyrum T. Durand were converted from Caceres et al. (1998) and Vogel, Arumuganathan, and Jensen (1999). Age estimates for the Hordeum clades were taken from the ITS analysis by Blattner (in press), where a penalized likelihood approach was used to calculate the ages of internal nodes in the diploids' tree under an assumption of a split between wheat and barley about 13 MYA (Gaut 2002).
FIG. 1. Scheme of the nuclear rDNA internal transcribed spacer phylogeny of Hordeum (Blattner, in press) with average 2C genome size (in pg) of the species plotted to the right or below the taxon names. Diploid taxa were drawn directly to the branches of the tree, tetraploids and hexaploids are mapped to the right. Lines connect the taxon names with their respective parental species. Dots refer to annual life form of the respective species. Numbers along the branches depict posterior probabilities of the clades calculated by Bayesian inference. Bold numbers to the right give rough age estimations of the branching points. H, I, Xa, and Xu refer to genome composition of Hordeum species, for the H clade divided in the Asian (H1) and New World (H2) subgroups. The dashed line points to the second progenitor of H. depressum, possibly H. brachyantherum subsp. californicum (Salomon and Bothmer 1998), which was not found in the ITS analysis. Asterisk (*) indicates the DNA value of diploid H. brevisubulatum that was not measured but calculated from the polyploid accessions of this species. For H. guatemalense, no genome size value could be obtained
For statistical analyses SPSS version 10.0 was used. The entire data set did not show normal distribution, but some data subsets were normally distributed. When normal distribution and homogeneity of variance were satisfied, we used the t-test or the unifactorial ANOVA Tukey-HSD test. For data lacking homogeneity of variance, we used Dunnett-T3, and for data without normal distribution, we used the nonparametric Mann-Withney U test. Correlations between diploid genome sizes and environmental parameters were tested with a bivariate correlation using Pearson's product moment correlation.
For all diploid accessions, and additionally for some polyploids, average monthly temperatures and precipitation were determined using the coordinates of the collection sites together with two climate data bases (CLIMATE DATABASE Version 2.1, W. Cramer, Potsdam Institute of Climatic Impact Research, and the internet-based compilation of B. Mühr, available under http://www.klimadiagramme.de/all.html) and data from Walter and Lieth (1960–1967). The data were used to ascertain the respective local climate types according to K?ppen (1923), which, thus, reflect altitudinal as well as local geographical differences.
To reveal the influence of the factors phylogeny (represented by the Hordeum karyotype groups H1, H2, I, Xa, and Xu), life form (annuals versus perennials), climate, and, ages of species, a three-way analysis of variance with covariate (ANCOVA) was applied to the data from diploid Hordeum species. According to Lindman (1974), this is legitimate because the F-test is robust against deviations from normal distribution. Because some of the cells in the data scheme were empty, we analyzed only the main effects (no interaction effects, which are especially sensitive against missing combinations) and used model type IV of sums of squares, which is suitable for this situation. As the entire data set did not show homogeneity of variance, the level of significance was increased from 95% to 99% (Bühl and Z?fel 2000). All karyotype groups used for ANCOVA were statistically supported by bootstrap values and Bayesian posterior probabilities in the phylogenetic data.
On an intraspecific scale along latitudinal transects of collection sites, the respective genome sizes were analyzed in H. marinum subsp. marinum, H. pubiflorum, and H. lechleri. Here we tested genome size variation against ecogeographical components such as the average growing season, summer temperature and precipitation (July and January, respectively), and geographical latitude.
Results
Genome size measurements with flow cytometry revealed conspicuous variation among Hordeum species. Replicate measurements from individual Hordeum species were highly repeatable with a standard deviation of up to 4% of the measured genome size (table 1). Here, we describe the results in the order of the phylogenetic relationships within the genus.
Genome Size in Extant Hordeum Species
The I-genome clade comprises the perennial species H. bulbosum, with its diploid (2x) and tetraploid (4x) forms, and the annual species H. vulgare subsp. vulgare (barley), along with its wild progenitor H. vulgare subsp. spontaneum. The 2C DNA contents in this group are in a range from 8.88 pg in diploid H. bulbosum to 10.67 pg in H. vulgare. In the autotetraploid H. bulbosum, the average 2C DNA content (17.3 pg) is slightly smaller than twice the genome size (17.76 pg) of the diploid form. However, within the two analyzed tetraploid accessions, we ascertained a difference of 0.66 pg, and the H. bulbosum accession from Greece nearly contains the expected DNA amount (17.63 pg) calculated on the base of the adjacent Italian diploid accession. The more easterly accession from Tajikistan had a smaller genome (16.97 pg).
The Xu-genome clade consists only of the annual H. murinum, divided into three subspecies with varying ploidy levels from diploids to hexaploids (6x). The polyploid forms of H. murinum (4x subspp. murinum, 19.68 pg, and leporinum, 19.2 pg; 6x subsp. leporinum, 29.85 pg) are proposed to be of either autoployploid or segmental allopolyploid origin (von Bothmer et al. 1995). In all three polyploids, the respective tetraploid and hexaploid genome sizes are slightly larger than the corresponding multiple of the diploid subsp. glaucum (9.11 pg).
The Xa-genome clade is the sister group of the large H clade. It consists only of the annual H. marinum, with the subspecies marinum (2x) and gussoneanum (2x and 4x). The difference in genome size between the two diploid subspecies (9.10 pg vs. 10.41 pg) is highly significant. The DNA content of the tetraploid subsp. gussoneanum (19.64 pg) is slightly larger than the sum of the genomes of its proposed diploid progenitors (19.51 pg).
The H-genome taxon group consists of 27 diploid, tetraploid, and hexaploid species, with annual or perennial life form (fig. 1). 2C DNA amounts between the diploid Hordeum species differ by 1.42-fold, ranging from 6.85 pg in H. euclaston to 9.69 pg in H. roshevitzii. The group of tetraploid species combining two H genomes revealed genome sizes between 15.52 pg (H. depressum) and 18.57 pg (H. brachyantherum subsp. brachyantherum). Tetraploid H. guatemalense, a member of this group, was not analyzed, and no data were available from the literature. The genome sizes of the putative autotetraploid forms of H. brevisubulatum were 17.81 pg (subsp. brevisubulatum) and 18.16 pg (subsp. turkestanicum). European H. secalinum (19.98 pg) and the South African H. capense (19.74 pg) are phylogenetically closely related. Both consist maternally of the large Xa genome of diploid H. marinum subsp. gussoneanum (10.41 pg). The other parental genome stems from H. comosum (8.97 pg) or a close relative. Their genome sizes are, therefore, clearly larger than that of the other tetraploids in the clade, combining two H genomes.
The genome size values of the hexaploid species consisting of three H genomes were in a range between 24.68 pg (H. arizonicum) and 27.18 pg (H. procerum). One ancestor of all South American hexaploids is tetraploid H. fuegianum or H. tetraploidum. The second parental genomes stem from different diploid species. In the case of H. procerum (27.18 pg), the parental DNA amounts are perfectly additive. In H. parodii (26.19 pg) and H. lechleri (26.31 pg) the expected values were slightly larger (26.5 and 26.7 pg, respectively) than the measured ones. Californian H. brachyantherum subsp. brachyantherum revealed a conspicuously larger genome size (28.9 pg) than the other hexaploids from the H-genome clade, which is consistent with the origin of this taxon by hybridization of tetraploid subsp. brachyantherum with H. marinum subsp. gussoneanum (2x). The significantly smaller hexaploid genome of H. arizonicum resulted from the hybridization of the small-genome annual H. pusillum with H. jubatum.
Correlations of Phylogeny, Life Form, and Environmental Parameters with Genome Size
The ANCOVA confirmed the main impact of annual or perennial life form on genome size as well as the lesser, but also significant, influence of the phylogeny ("karyotype"). Climate types at the growing site and age of species had no significant effect on genome size variation in Hordeum (table 2). Species, with both small and large genomes were found in very stressful, nearly arid climates or at high altitudes, as well as under moderate climatic conditions. The 2C values of the perennials range between 8.51 pg (H. flexuosum) and 9.69 pg (H. roshevitzii), revealing 1.14-fold differences, whereas in annual species, 1.55-fold size differences occur (H. euclaston, 6.85 pg versus H. vulgare, 10.67 pg). Perennials and annuals co-occur only in the H-karyotype and the I-karyotype groups. Within these two groups, life form contributes markedly to the explanation of genome size variation. In the New World H-genome clade, the annuals revealed significantly smaller genomes than the perennials (mean 7.00 pg vs. mean 9.27 pg; Mann-Withney U test, P < 0.001), whereas in the I-genome group, we found, conspicuously, the contrary. The annual species shows significant larger genome sizes than the perennial (10.67 pg vs. 8.88 pg; t-test, P < 0.001). This result clearly excludes life form as a single parameter determining genome size.
Table 2 Three-Way Analysis of Variance with Covariate (ANCOVA) for the Contribution of Factors to the Measured Genome Size of the Hordeum Accessions.
Intraspecific Genome Size Variation
Within species, genome size differences along a latitudinal transect was tested for H. marinum subsp. marinum in Western Europe. Ten populations of H. marinum subsp. marinum were analyzed, occurring over a geographical range of 12° latitude. Genome size data were contrasted with different ecogeographical factors such as the mean July temperature and mean July precipitation. A negative correlation (Pearson = –0.862, P = 0.001) between the mean genome size and the mean July temperature was found (fig. 2). A positive correlation with smaller significance value (P = 0.027) was also assessed with mean July precipitation. Both parameters correlate reciprocally with northern latitude and influence markedly the length of the vegetation period for this annual species.
FIG. 2. Correlations of DNA content with ecogeographical parameters. DNA content versus mean July temperature in H. marinum subsp. marinum along a transect in Western Europe and DNA content versus latitudinal position of collection sites in H. pubiflorum and H. lechleri along a transect in southern South America
The same approach was used for H. pubiflorum (31 measurements of 13 populations) and H. lechleri (94 measurements of 33 populations) occurring over a geographical range of 12° latitude in South America. Instead of the mean January temperatures, which were not available for all collection sites, we used the southern latitudes as a rough estimation of the duration of vegetation period. We found no significant correlation between genome size variation and latitudinal distribution of the analyzed populations (fig. 2).
Ancestral Genome Size Estimations
Ancestral genome sizes, calculated with PGLS on the base of diploid Hordeum species confirmed the observation that within Hordeum, increasing and decreasing DNA contents existed along the lineages leading to the extant species (fig. 3). The basal 2C genome size of the genus was calculated as 10.8 pg DNA, a value close to the 10.7 pg assumed for ancient Triticeae by Kellogg (1998). The ancient DNA values steadily decreased in Hordeum and are significantly smaller in H. bulbosum, H. flexuosum, and the New World annuals than the initial Hordeum genome. Significant secondary genome size increase occurred in the lineage leading to H. marinum subsp. gussoneanum.
FIG. 3. Genome size evolution in diploid Hordeum species as inferred from extant genome size (average value/standard error) by PGLS. Ancient 2C genome size estimations (in pg) are given above the branches, the standard errors are given below. Asterisk (*) indicates genome sizes taken either from the literature (outgroups) or calculated from polyploid species. In these cases, the standard error values were increased to account for uncertainties in conversion of the data. Dashed lines in the tree indicate positions were the polytomies of the ITS tree were resolved in bifurcations by the introduction of short branch lengths for PGLS calculations of ancient genome sizes
Discussion
Genome size analysis of 134 accessions (394 individuals) representing all but one Hordeum species revealed up to 1.55-fold 2C DNA amount differences, laying in a range between 6.85 and 10.67 pg in diploid species, 15.52 to 19.98 pg in tetraploids, and 24.68 to 29.85 pg in hexaploids. Our results are compatible with the results of Kankanp??, Mannonen, and Schulman (1996) and generally compatible with Vogel, Arumuganathan, and Jensen (1999), although some systematical differences were detected. Our 2C DNA values were always 1.1 to 1.6 pg higher than these reported by Kankanp??, Mannonen, and Schulman (1996) and 0.3 to 0.9 pg lower compared with Vogel, Arumuganathan, and Jensen (1999). This clearly points to differences caused by the use of different internal DNA standards and/or fluorescent dyes, (i.e., DAPI versus propidium iodide [Doleel et al. 1998; Barow and Meister 2002]). However, we analyzed all accessions with the same procedure and with at least three individuals per accession and usually multiple accessions per species. Thus, we are confident that our data are internally comparable.
Genome Size Changes in Relation to Phylogeny
The conspicuous size differences found in Hordeum are not evenly distributed within the genus, but are primarily related to specific phylogenetic groups (fig. 1). Whereas in four clades, here informally named after the genome groups (I, H1, Xa, and Xu), the diploid 2C values were in a range between (8.88) 9.10 and 10.67 pg, only within the H2 subclade were genomes smaller than 8.8 pg found. The smallest genomes in this clade are restricted to three species that became secondarily annual, whereas the perennials vary between 8.51 and 9.61 pg. This correlation was also found in the polyploids, where the tetraploid annual H. depressum has about 2 pg less DNA than the smallest perennial taxon, and the hexaploid taxon with the smallest genome, H. arizonicum, is a perennial species that facultatively can also grow as a biennial or annual. As all three annual diploids are closely related and the annual polyploids include one of them as progenitor (fig. 1), it remains unclear if a smaller genome is a prerequisite for the annual life form or if both traits (ability for shorter generation time and the smaller genome) are inherited from a common ancestor. Because nearly 30 pg DNA per nucleus is compatible with the annual life habit of hexaploid H. murinum, simple phylogenetic constraints are probable. If phylogenetic relationships determine the genome size, only within close monophyletic groups should DNA contents show correlations to environmental parameters.
Genome Size Changes in Relation to Environmental Parameters
The narrowest monophyletic groups in our analyses were three intraspecific population samples. Hordeum marinum subsp. marinum was collected along a latitudinal transect in Western Europe, and H. pubiflorum and H. lechleri were colected over a comparable distance in Patagonia. Genome size analysis in the 10 H. marinum subsp. marinum populations revealed increasing genome sizes from south to north (fig. 2). Northern distribution here is linked with a prolonged vegetation period for these plants. This corroborates the results of Bennett (1987), Reeves et al. (1998), and Turpeinen, Kulmala, and Nevo (1999), who found positive correlations between the duration of vegetation period and genome size in plants. However, the genome sizes differ about 3.3% (0.30 pg) along our transect and thus are clearly different from the DNA values of H. marinum subsp. gussoneanum, where even in the southernmost populations the diploid genomes are more than 0.7 pg (8%) larger than in subsp. marinum. This indicates again that phylogenetic constraints, here depicted by taxonomic units, might be more important than ecological influences for genome sizes on a higher taxonomic level.
Within the South American species H. pubiflorum and H. lechleri, we found no correlations, neither with the climate types, with mean January temperatures, nor with increasing southern latitude. However, the assumption that for these species, southern latitude is directly correlated with vegetation period or other gradually changing ecoclimatic parameters may not be warranted. The rough landscape of Patagonia might present more complex local influences on plant habitats than expressed by the large-scale climate parameters we have taken into account.
Although on the world scale, correlations between climate zones, that is, habitats at higher latitudes, and genome sizes of plants were found (Bennett 1976, 1987; Levin and Funderburg 1979; Grime and Mowforth 1982), we could not confirm such a correlation among all species of the genus Hordeum. ANCOVA found no significant influence of local climate type for the explanation of genome size variation, which can be explained by the strong influence of phylogenetic relationships, concealing environmental correlations even among closely related taxa.
The Influence of Life Form on Genome Size Variation in Hordeum
Bennett and Leitch (1995) proposed that small genomes occur in annual taxa, whereas perennials could possess larger genomes. This correlation is well established for ephemeral species (Bennett 1972). In Hordeum, life form contributes significantly to the explanation of the genome sizes (table 2). However, the largest (H. vulgare) and the smallest (H. euclaston) diploid genomes were found in annuals. Whereas the annual species of the H2 clade seemingly support the conclusions of Bennett and Leitch (1995), the annuals of the Xa and Xu clade have medium-sized to large genomes (9.1 to 10.41 pg DNA), and in the I-genome clade, the relation is inversed. The perennial H. bulbosum has a much smaller genome (8.88 pg) than its annual sister taxon H. vulgare. Thus, low DNA content is obviously no prerequisite for annuality in Hordeum.
Genome size differences among annuals are likewise not related to summer or winter annuality. In H. marinum, we found that subsp. marinum, which partly grows as a winter annual, has a significantly smaller genome than subsp. gussoneanum, which is exclusively a summer annual. This contradicts the hypothesis of Grime and Mowforth (1982), who proposed that temporal detachment of cell proliferation and cell extension will allow larger genome sizes. Although analyses on the world scale seemingly support the fact that small genomes preferentially occur in annual species, among the evolutionarily more closely linked species of Hordeum, this ratio does not generally hold and is even inverted in the I-genome group.
Genome Size Evolution Through Time in Hordeum
Bennetzen and Kellogg (1997) proposed the possibility of "genomic obesity" in plants caused by polyploidization and the activity of retrotransposons, which should keep increasing the genome size. Our data and data by Wendel et al. (2002) provide information that plant genomes are capable of DNA gain and loss in closely related lineages. DNA loss seems to be the predominant process in Hordeum. Starting with the estimated diploid DNA amount of 10.8 pg (±1.8) for ancestral Hordeum taxa, all extant species now have smaller genomes (fig. 3). Ambiguities in the phylogenetic tree are restricted to the species relationships in the H2 clade. Topological changes in this part of the tree, however, do not change the estimation of the basal genome size and, thus, do not alter the general tendency of genome decrease in Hordeum. A significant secondary DNA content increase was found in H. marinum subsp. gussoneanum. Our data clearly indicate that a "one-way ticket" (Bennetzen and Kellogg 1997) for ever increasing genomes of plants does not exist. This result is not unexpected, as the analysis of Leitch, Chase, and Bennett (1998) proposed a small ancient angiosperm genome but also pointed out that several derived plant families have rather small genomes. Either genomic obesity, thus, would be found only in some lineages, possibly because of the loss of effective retroelement suppression, or mechanisms to jettison superfluous DNA are necessary. These mechanisms are not well understood in plants up to now. However, the biased DNA loss in Arabidopsis compared with tobacco during DNA double-strand break repair (Kirik, Salomon, and Puchta 2000; Orel and Puchta 2003) and unequal intrastrand recombination (Petrov 1997, 2001; Shirasu et al. 2000) indicate that possible mechanisms exist. Our analysis defines taxa that can be used to study these molecular mechanisms of genome size variation (Bennetzen 2002; Gregory 2003), as genome size changes in Hordeum can be traced in a phylogenetic framework.
Constancy of Genome Sizes Within Hordeum Species
The measurement of genome size in diploids revealed low intraspecific genome size variation (3%) among individuals and accessions from single species. Furthermore, the genome size of several polyploid species were nearly the added values of their progenitors. This holds particularly true for evolutionarily very young taxa, such as, for example, hexaploid H. brachyantherum, that originated during the past 200 years, after H. marinum was introduced to California, but also in H. arizonicum, H. procerum, H. marinum, and the western accessions of H. bulbosum. Differences between the summed genome sizes of the proposed progenitors and the polyploids indicate mostly a loss of about 0.5 pg DNA. Only in the closely related species H. secalinum and H. capense (+0.5 pg) and in the putative autopolyploid H. murinum subsp. murinum (+1.5 pg), larger than additive genomes were found (fig. 1). This could be either because of a larger genome in a parental taxon of these species or because DNA proliferation occurred in these lineages. These would, however, be the only cases in Hordeum where genome size increases instead of decreases after a polyploidization event. Constancy of genome size was also found in cases where we compared accessions grown for several generations ex situ with material collected directly from the wild. Many analyzed species were derived from germ plasm collections and, thus, underwent several cycles of propagation in cultivation. Comparisons of their genome size with the values derived from wild material of the same species resulted in values that all fell in the range of the natural intraspecific variation of the respective taxa. For newly collected plant material (>50 accessions) from Patagonia, we were even able to correctly predict the species affiliation of young seedlings, knowing their genome size and their geographical location. The constancy of DNA amounts in Hordeum species clearly contradicts the proposed fast intraspecific changes of genome size (Bennett and Bennett 1992; Kankanp??, Mannonen, and Schulman 1996; Price, Morgan, and Johnston 1998) but supports the view of genome sizes being fairly stable when a narrow species concept is used (Greilhuber 1988; Ohri 1998).
As intraspecific genome size changes were usually continuous, discontinuous changes were associated with (sub)species boundaries. If pronounced genome size changes occur preferentially during or after speciation, we would expect the largest differences in clades with high species numbers. This correlation was found for all clades of diploid Hordeum species (fig. 4), where particularly the H2 clade (13 diploid species) shows the largest (1.4-fold) genome size difference. However, we also found clear differences in the I-genome clade where only two species differ 1.2-fold. Within the I-genome clade, the oldest split between two single species in Hordeum occurs, representing about 7 Myr of independent evolution (fig. 1). Thus, time can also play a role in shaping genome size, although in the rapidly speciating groups, larger genome differences originated in shorter time (2 Myr in the H2 clade). ANCOVA accordingly showed that "age of species" explained few of the genome size differences. The low and uneven coverage of genome size values of angiosperms (Bennett 1998a) does not allow safe assertion about the relationship between taxonomically highly structured clades (i.e., comprising many species, genera, and tribes) and genome size variation. Nevertheless, in the analysis by Leitch, Chase, and Bennett (1998), ratios higher than 50:1 between maximum and minimum DNA amounts in angiosperm families were only found in Poaceae (104:1), Fabaceae (73:1), Iridaceae (66:1), Asteraceae (62:1), and Orchidaceae (57:1), which are all relatively species-rich families. Genome size changes during or shortly after speciation events would also be consistent with an often reduced effective population size in newly arising species, promoting rapid fixation of genome size changes (Lynch and Conery 2003).
FIG. 4. Correlation of the taxon number per clade with the percentage genome size differences within these clades for all diploid Hordeum taxa. The clades used in this calculation (I, Xa, H1, H2, H, H+Xa, I+Xu, and I+Xu+Xa+H) were chosen to represent monophyletic units according to figure 1
Our analysis of genome sizes in Hordeum species resulted in the paradoxical phenomenon that although meta-analyses of thousands of DNA values in angiosperms found correlations and trends in genome sizes with ecogeographical, physiological, and life history parameters (Bennett 1998b), these relationships could not be reliably established in the analysis of this one genus. For nearly all parameters proposed to influence genome sizes, we found examples in Hordeum that support these correlations but always with at least some exceptions. Thus, neither the knowledge of a species' life form, its ecogeographical distribution, nor its phylogenetic relationship allows a prediction of its genome size. Overlapping patterns of all three parameters influence the genomic composition of Hordeum species. However, the knowledge of these parameters allows safer posteriori interpretations of the observed DNA value patterns. Both earlier works reporting a certain number of genome size values for Hordeum species (Kankanp??, Mannonen, and Schulman 1996; Vogel, Arumuganathan, and Jensen 1999) struggled with the interpretation of the results. Kankanp??, Mannonen, and Schulman (1996) found no clear environmental or evolutionary patterns influencing genome size variation in Hordeum. However, they did not take into account life form as a third important parameter and argued on the basis of the wrong assumption that currently accepted sections in Hordeum represented natural phylogenetic groups. Moreover, the only hexaploid accession in their analysis seems, in the light of our data, clearly to be a misidentified tetraploid H. murinum. Vogel, Arumuganathan, and Jensen (1999), in an attempt to determine basic genome size values in Triticeae, included H. bulbosum in the H-genome taxa but excluded the respective annual species. Thus, their sample was not representative for the Hordeum H genomes and might also not be representative for the H genomes included in the heterogenomic Triticeae genera. As we have shown in H. depressum and H. arizonicum, the inclusion of an annual H genome in allopolyploids results in significantly smaller genome sizes compared with polyploids derived from perennial H-genome species, only. This can explain the smaller than expected 2C values of, for example, Elymus L. species with StH genome combinations (Vogel, Arumuganathan, and Jensen 1999).
To conclude, we can point out that although plant genome sizes might respond reasonably predictably to environmental parameters at the population level (Nevo 2001), with increasing taxonomic distance and on higher taxonomic levels, predictability rapidly diminishes through interacting influences of phylogenetic constraints, life form changes, and possibly also the number of speciation events occurring along the specific lineages of the analyzed plants. Correlations found in meta-analyses, therefore, represent trends that cannot be generalized and might even be inverted when looking into specific plant groups.
Supplementary Material
A table listing all 134 Hordeum accessions, their origins, climate data, and the measured genome size is available as Supplementary Material online linked to this article on the M.B.E. home page (http://mbe.oupjournals.org/).
Acknowledgements
We thank M. Arriaga and R. Gomez Cadret (Museo Argentino de Ciencias Naturales, Buenos Aires) and P. Cichero (Parque National Argentina, Buenos Aires) for kindly providing help in the organization of the field work in Patagonia. We also thank K. Bachmann, M. Barow, and M. A. Lysak, for critical reading of the manuscript, M. H. Hoffmann, who provided climate data, and A. Ihlow, for help with the statistics. Financial support from the Deutsche Forschungsgemeinschaft within priority program SPP 1127 (to F.R.B.) and from the Fond der Chemischen Industrie (to A.M.) is acknowledged.
Literature Cited
Arumuganathan, K., and E. D. Earle. 1991. Nuclear DNA content of some important plant species. Plant Mol. Biol. Rep. 9:208-218.
Barakat, A., N. Carels, and G. Bernardi. 1997. The distribution of genes in the genomes of Gramineae. Proc. Natl. Acad. Sci. USA 94:6857-6861.
Barow, M., and A. Meister. 2002. Lack of correlation between AT frequency and genome size in higher plants and the effect of nonrandomness of base sequences on dye binding. Cytometry 47:1-7.
Barow, M., and A. Meister. 2003. Endopolyploidy in seed plants is differently correlated to systematics, organ, life strategy and genome size. Plant Cell Environ. 26:571-584.
Bennett, M. D. 1972. Nuclear DNA content and minimum generation time in herbaceous plants. Proc. R. Soc. Lond. B. Biol. Sci. 181:109-135.
Bennett, M. D. 1976. DNA amount, latitude, and crop plant distribution. Environ. Exp. Bot. 16:93-108.
Bennett, M. D. 1987. Variation in genomic form in plants and its ecological implications. New Phytol. 106:177-200.
Bennett, M. D. 1998a. DNA amounts in two samples of angiosperm weeds. Ann. Bot. 82:(suppl. A): 121-134.
Bennett, M. D. 1998b. Plant genome values: how much do we know? Proc. Natl. Acad. Sci. USA 95:2011-2016.
Bennett, M. D., and I. J. Leitch. 1995. Nuclear DNA amounts in angiosperms. Ann. Bot. 76:113-176.
Bennett, S. T., and M. D. Bennett. 1992. Variation in nuclear DNA amount between wild and cultivated populations of Milium effusum (2n = 28). Genome 35:1050-1053.
Bennetzen, J. L. 2002. Mechanisms and rates of genome expansion and contraction in flowering plants. Genetica 115:29-36.
Bennetzen, J. L., and E. A. Kellogg. 1997. Do plants have a one-way ticket to genomic obesity? Plant Cell 9:1509-1514.
Blattner, F. R. 2004. Phylogenetic analysis of Hordeum (Poaceae) as inferred by nuclear rDNA ITS sequences. Mol. Phylogenet. Evol. (in press).
Bühl, A., and P. Z?fel. 2000. Einführung in die moderne Datenanalyse unter Windows. Addison-Wesley, Munich.
Caceres, M. E., C. De Pace, G. T. Scarascia Mugnozza, P. Kotsonis, M. Ceccarelli, and P. G. Cionini. 1998. Genome size variations within Dasypyrum villosum: correlations with chromosomal traits, environmental factors and plant phenotypic characteristics and behaviour in reproduction. Theor. Appl. Genet. 96:559-567.
Ceccarelli, M., E. Falistocco, and P. G. Cionini. 1992. Variation of genome size and organization within hexaploid Festuca arundinacea. Theor. Appl. Genet. 83:273-278.
Cox, A. V., G. J. Abdelnour, M. D. Bennett, and I. J. Leitch. 1998. Genome size and karyoptype evolution in the slipper orchids (Cypripedioideae: Orchidaceae). Am. J. Bot. 85:681-687.
Cullis, C. A., and W. Cleary. 1986. Rapidly varying DNA sequences in flax. Can. J. Genet. Cytol. 28:252-259.
Doleel, J., J. Greilhuber, S. Lucretti, A. Meister, M. A. Lysak, L. Nardi, and R. Obermayer. 1998. Plant genome size estimation by flow cytometry: inter-laboratory comparison. Ann. Bot. 82:(Suppl. A): 17-26.
Flavell, R. B., J. Rimpau, and D. B. Smith. 1977. Repeated sequence DNA relationships in four cereal genomes. Chromosoma 63:205-222.
Gaut, B. S. 2002. Evolutionary dynamics of grass genomes. New Phytol. 154:15-28.
Goin, O. B., C. J. Goin, and K. Bachmann. 1968. DNA and amphibian life history. Copeia 3:532-540.
Gregory, R. T. 2001. Genome size and developmental complexity. Genetica 115:131-146.
Gregory, R. T. 2002. A bird's-eye view of the C-value enigma: genome size, cell size, and metabolic rate in the class Aves. Evolution 56:121-130.
Gregory, R. T. 2003. Is small indel bias a determinant of genome size? Trends Genet. 19:485-488.
Greilhuber, J. 1988. Critical reassessment of DNA content variation in plants. Pp. 39–40 in P. E. Brandham, ed. Kew chromosome conference III. HMSO, London.
Grime, J. P. 1998. Plant classification for ecological purposes: is there a role for genome size? Ann. Bot. 82:(Suppl A): 117-120.
Grime, J. P., and M. A. Mowforth. 1982. Variation in genome size: an ecological interpretation. Nature 299:151-153.
Jockusch, E. L. 1997. An evolutionary correlate of genome size change in plethodontid salamanders. Proc. R. Soc. Lond. B Biol. Sci. 264:597-604.
Kalendar, R., J. Tanskanen, S. Immonen, E. Nevo, and A. H. Schulman. 2000. Genome evolution of wild barley (Hordeum spontaneum) by BARE-1 retrotransposon dynamics in response to sharp microclimatic divergence. Proc. Natl. Acad. Sci. USA 97:6603-6607.
Kankanp??, J., L. Mannonen, and A. H. Schulman. 1996. The genome sizes of Hordeum species show considerable variation. Genome 39:730-735.
Kellogg, E. A. 1998. Relationships of cereals crops and other grasses. Proc. Natl. Acad. Sci. USA 95:2005-2010.
Kirik, A., S. Salomon, and H. Puchta. 2000. Species-specific double-strand break repair and genome evolution in plants. EMBO J. 19:5562-5566.
Knight, C. A., and D. D. Ackerly. 2002. Variation in nuclear DNA content across environmental gradients: a quantile regression analysis. Ecol. Lett. 5:66-76.
Knüpffer, H., and T. van Hintum. 2003. Summarised diversity—the Barley Core Collection. Pp. 259–267 in R. von Bothmer, T. van Hintum, H. Knüpffer, and K. Sato, eds. Diversity in barley (Hordeum vulgare). Elsevier, Amsterdam.
K?ppen, W. 1923. Die Klimate der Erde: Grundri? der Klimakunde. de Gruyter, Berlin.
Laurie, D. A., and M. D. Bennett. 1985. Nuclear DNA content in the genera Zea and Sorghum. Intergeneric, interspecific and intraspecific variation. Heredity 55:307-313.
Leitch, I. J., M. W. Chase, and M. D. Bennett. 1998. Phylogenetic analysis of DNA C-values provides evidence for a small ancestral genome size in flowering plants. Ann. Bot. 82:(suppl. A): 85-94.
Levin, D. A., and S. W. Funderburg. 1979. Genome size in angiosperms: temperate versus tropical species. Am. Nat. 114:784-795.
Lindman, H. R. 1974. Analysis of variance in complex experimental designs. Freeman, San Francisco.
Lynch, M., and J. S. Conery. 2003. The origins of genome complexity. Science 302:1401-1404.
Manninen, I., and A. H. Schulman. 1993. BARE-1, a copia-like retroelement in barley (Hordeum vulgare L.). Plant Mol. Biol. 22:829-846.
Martins, E. P. 2003. COMPARE: Computer programs for the statistical analysis of comparative data. Version 4.4. Distributed by the author, http://compare.bio.indiana.edu/, Indiana University, Bloomington.
Nevo, E. 2001. Evolution of genome-phenome diversity under environmental stress. Proc. Natl. Acad. Sci. USA 99:6233-6240.
Ohri, D. 1998. Genome size variation and plant systematics. Ann. Bot. 82:(suppl. A): 75-83.
Orel, N., and H. Puchta. 2003. Differences in the processing of DNA ends in Arabidopsis thaliana and tobacco: possible implications for genome evolution. Plant Mol. Biol. 51:523-531.
Petrov, D. A. 1997. Slow but steady: reduction of genome size through biased mutation. Plant Cell 10:1900-1901.
Petrov, D. A. 2001. Evolution of genome size: new approaches to an old problem. Trends Genet. 17:23-28.
Price, H. J., K. L. Chambers, and K. Bachmann. 1981. Genome size variation in diploid Microseris bigelovii (Asteraceae). Bot. Gaz. 142:156-159.
Price, H. J., P. W. Morgan, and J. S. Johnston. 1998. Environmentally correlated variation in 2C nuclear DNA content measurements in Helianthus annuus. L. Ann. Bot. 82:(suppl. A): 95-98.
Reeves, G., D. Francis, M. S. Davies, H. J. Rogers, and T. R. Hodkinson. 1998. Genome size is negatively correlated with altitude in natural populations of Dactylis polygama. Ann. Bot. 82:(suppl. A): 99-105.
Salomon, B., and R. von Bothmer. 1998. The ancestry of Hordeum depressum (Poaceae, Triticeae). Nordic J. Bot. 18:257-265.
Shirasu, K., A. H. Schulman, T. Lahaye, and P. Schulze-Lefert. 2000. A contiguous 66-kb barley DNA sequence provides evidence for reversible genome expansion. Genome Res. 10:908-915.
Suoniemi, A., K. Anamthawat-Jónsson, T. Arna, and A. H. Schulman. 1996. Retrotransposon BARE-1 is a major, dispersed component of the barley (Hordeum vulgare L.) genome. Plant Mol. Biol. 30:1321-1329.
Swift, H. 1950. The constancy of deoxyribose nucleic acid in plant nuclei. Proc. Natl. Acad. Sci. USA 36:643-654.
Swofford, D. L. 1991. When are phylogenetic estimates from molecular and morphological data incongruent? Pp. 295–333 in M. M. Miyamoto and J. Cracraft, eds. Phylogenetic analysis of DNA sequences. Oxford University Press, New York.
Thomas, C. A. 1971. The genetic organizations of chromosomes. Annu. Rev. Genet. 5:237-256.
Turpeinen, T., J. Kulmala, and E. Nevo. 1999. Genome size variation in Hordeum spontaneum populations. Genome 42:1094-1099.
Vicient, C. M., A. Suoniemi, K. Anamthawat-Jónsson, J. Tanskanen, A. Beharav, E. Nevo, and A. H. Schulman. 1999. Retrotransposon BARE-1 and its role in genome evolution in the genus Hordeum. Plant Cell 11:1769-1784.
Vogel, K. P., K. Arumuganathan, and K. B. Jensen. 1999. Nuclear DNA content of perennial grasses of the Triticeae. Crop Sci. 39:661-667.
von Bothmer, R., N. Jacobsen, C. Baden, R. B. J?rgensen, and I. Linde-Laursen. 1995. An ecogeographical study of the genus Hordeum. 2nd Edition. IPGRI, Rome.
Walter, H., and H. Lieth. 1960–1967. Klimadiagramm-Weltatlas. Fischer-Verlag, Jena.
Wendel, J. F., R. C. Cronn, J. S. Johnston, and H. J. Price. 2002. Feast and famine in plant genomes. Genetica 115:37-47.(Sabine S. Jakob, Armin Me)
E-mail: blattner@ipk-gatersleben.de.
Abstract
Genome size variation in plants is thought to be correlated with cytological, physiological, or ecological characters. However, conclusions drawn in several studies were often contradictory. To analyze nuclear genome size evolution in a phylogenetic framework, DNA contents of 134 accessions, representing all but one species of the barley genus Hordeum L., were measured by flow cytometry. The 2C DNA contents were in a range from 6.85 to 10.67 pg in diploids (2n = 14) and reached up to 29.85 pg in hexaploid species (2n = 42). The smallest genomes were found in taxa from the New World, which became secondarily annual, whereas the largest diploid genomes occur in Eurasian annuals. Genome sizes of polyploid taxa equaled mostly the added sizes of their proposed progenitors or were slightly (1% to 5%) smaller. The analysis of ancestral genome sizes on the base of the phylogeny of the genus revealed lineages with decreasing and with increasing genome sizes. Correlations of intraspecific genome size variation with the length of vegetation period were found in H. marinum populations from Western Europe but were not significant within two species from South America. On a higher taxonomical level (i.e., for species groups or the entire genus), environmental correlations were absent. This could mostly be attributed to the superimposition of life-form changes and phylogenetic constraints, which conceal ecogeographical correlations.
Key Words: DNA content ? flow cytometry ? genome size evolution ? Hordeum ? life form ? phylogeny
Introduction
Since the 1950s, large differences in the nuclear DNA content of different organisms were detected (Swift 1950; Price, Chambers, and Bachmann 1981; Laurie and Bennett 1985; Ceccarelli, Falistocco, and Cionini 1992; Bennett and Leitch 1995) and drew the attention of scientists to possible explanations. Because no correlation with organismal complexity could be found, these differences were described as "C value paradox" (Thomas 1971). During the following years, it became clear that not gene content, but the proportion of repetitive DNA, primarily influences nuclear DNA content (Flavell, Rimpau, and Smith 1977; Barakat, Carels, and Bernardi 1997). Particularly retrotransposon copy numbers can vary to a large extent among genomes and contribute markedly to genome size differences found in plants (Arumuganathan and Earle 1991; Vicient et al. 1999; Kalendar et al. 2000). The specific mode of retroelement mobilization, which transpose without excision, should, together with polyploidization, drive increase of genome size in plants. Because no comparably efficient mechanism of genome size reduction was identified, Bennetzen and Kellogg (1997) proposed the possibility of a "one-way ticket" to "genomic obesity" in plants. The rather large genomes of several derived groups of angiosperms (e.g., Liliaceae within monocots, Triticeae within grasses, and Santalales within eudicots) seemingly supported this hypothesis (Kellogg 1998; Leitch, Chase, and Bennett 1998).
Analyses of genome size distribution in various organisms found correlations of genome sizes with cytological, physiological, and ecological characters (Goin, Goin, and Bachmann 1968; Bennett 1976, 1987; Grime and Mowforth 1982; Jockusch 1997; Grime 1998; Ohri 1998; Gregory 2001, 2002; Knight and Ackerly 2002), even within single species (Price, Chambers, and Bachmann 1981; Laurie and Bennett 1985; Cullis and Clearly 1986; Bennett and Leitch 1995; Nevo 2001). The conclusions drawn in these studies are often contradictory. In the light of intraspecific DNA content changes under different environmental conditions (Turpeinen, Kulmala, and Nevo 1999; Kalendar et al. 2000; Nevo 2001), it seems questionable if analyses of genome size data in distantly related taxa are meaningful for the understanding of the mechanisms influencing genome size variation (Jockusch 1997). Furthermore, some authors reported possible phylogenetic constraints of genome sizes (Cox et al. 1998; Wendel et al. 2002). An explicit phylogenetic view on grass genomes resulted in clear genome size differences among the monophyletic grass subfamilies (Kellogg 1998; Gaut 2002), although for most groups, a considerable overlap of DNA contents existed. These studies analyzed, however, either variation on an intraspecific scale or variation among distantly related taxa, and the samples were often far from being representative for the groups. Data on genome size changes within clearly circumscribed plant groups that could fill the gap between analyses on the population level (Turpeinen, Kulmala, and Nevo 1999) and meta-analyses of entire plant families (Kellogg 1998), floras (Grime and Mowforth 1982; Knight and Ackerly 2002), life forms (Bennett and Leitch 1995), or of the plant kingdom (Bennett 1987) are scarce.
The Triticeae genus Hordeum L. comprises 31 species (including barley), which occur under a wide variety of climates in Eurasia and the New World (von Bothmer et al. 1995; Blattner, in press). Genome size of several Hordeum species and the contributions of BARE-1 retrotransposon activities to DNA content variation were analyzed by A. Schulman's group (Manninen and Schulman 1993; Kankanp??, Mannonen, and Schulman 1996; Suoniemi et al. 1996; Vicient et al. 1999; Kalendar et al. 2000). Their interpretation of genome size data was impeded by the lack of reliable phylogenetic information on Hordeum and resulted in the conclusion that genome size is not correlated with phylogenetic relationships and that DNA content may change faster than speciation takes place (Kankanp??, Mannonen, and Schulman 1996). Here, we report DNA content for 30 Hordeum species, determined by flow cytometry, and analyzed in a phylogenetic framework. We used these data to test for possible ecological and phylogenetical constraints, intraspecific genome size changes, and genome size evolution through time and to see how life form and polyploidization influenced DNA content variation in Hordeum.
Materials and Methods
Plant Material
Seeds from 134 accessions of 30 Hordeum species (46 taxa [table 1]) were germinated in petri dishes and then grown in pots containing standard soil in a greenhouse. This material belongs mostly to the Barley Core Collection (BCC [Knüpffer and van Hintum 2003]), a reference collection of Hordeum species and lineages, complemented by our own material from Patagonia and Eurasia. Accession numbers, the origins of the plant material, and basic climate data of the collection sites are shown in table A in Supplementary Material online. Up to 30 mg of fresh leaf tissue was used for flow cytometry analysis. At least three individuals per accession and one to 10 accessions per species were measured. For a more detailed analysis of intraspecific genome size variation, we analyzed 10 accessions of H. marinum subsp. marinum from Western Europe and 13 accessions of H. pubiflorum and 33 accessions of H. lechleri from southern Argentina. In total, 394 samples were measured.
Table 1 Hordeum Species Analyzed in This Study.
Genome Size Estimation
The preparation of nuclear suspensions and the determination of DNA content essentially followed the protocol given by Barow and Meister (2002, 2003), using a FACStarPlus flow cytometer (Becton Dickinson, San José, Calif.) equipped with two argon lasers INNOVA 90-5 (Coherent, Palo Alto, Calif.) and the analysis program CellQuest. Usually 10,000 nuclei per sample were analyzed.
Nuclear DNA content was estimated as a ratio between the fluorescence of the nuclei of the samples stained with propidium iodide (excitation at 514 nm and emission at 630 nm) and internal standards. Pisum sativum cv. "Viktoria, Kifejt? Borsó" (IPK GenBank accession number PIS 630) with a 2C genome size of 9.09 pg (Doleel et al. 1998) was used as the primary standard. For diploid and hexaploid Hordeum species, we included Secale cereale (IPK GenBank accession number R 737) as standard. The 2C genome size of S. cereale was calculated as 16.8 pg with regard to P. sativum.
Data Analyses
The average genome sizes of the Hordeum taxa were correlated with their phylogenetic position in the genus (fig. 1), as derived from the analysis of the nuclear rDNA internal transcribed spacer (ITS) region (Blattner, in press). This study involved cloned ITS sequences from multiple accessions of all Hordeum species, analyzed with cladistic, phenetic, and model-based approaches. The results of these analyses were summarized with an Adams consensus tree approach (Swofford 1991), integrating taxa with ambiguous clade affiliation in a polytomy at the last unambiguous branching point. Up to now, this is the only Hordeum phylogeny covering all diploid and polyploid species. The measured DNA contents of polyploids were compared with the sum of the genome sizes from their proposed parental species. Ancestral genome sizes within Hordeum were estimated for diploid lineages with the phylogenetic generalized least squares method (PGLS) in Compare version 4.4 (Martins 2003). For this calculation, the polytomies in the tree had to be resolved into bifurcations by introducing minute branch lengths. This procedure results in more or less equal size estimations at this artificial branching point, as along short branches, nearly no size changes take place. Genome size values for the outgroups Psathyrostachys Nevski and Dasypyrum T. Durand were converted from Caceres et al. (1998) and Vogel, Arumuganathan, and Jensen (1999). Age estimates for the Hordeum clades were taken from the ITS analysis by Blattner (in press), where a penalized likelihood approach was used to calculate the ages of internal nodes in the diploids' tree under an assumption of a split between wheat and barley about 13 MYA (Gaut 2002).
FIG. 1. Scheme of the nuclear rDNA internal transcribed spacer phylogeny of Hordeum (Blattner, in press) with average 2C genome size (in pg) of the species plotted to the right or below the taxon names. Diploid taxa were drawn directly to the branches of the tree, tetraploids and hexaploids are mapped to the right. Lines connect the taxon names with their respective parental species. Dots refer to annual life form of the respective species. Numbers along the branches depict posterior probabilities of the clades calculated by Bayesian inference. Bold numbers to the right give rough age estimations of the branching points. H, I, Xa, and Xu refer to genome composition of Hordeum species, for the H clade divided in the Asian (H1) and New World (H2) subgroups. The dashed line points to the second progenitor of H. depressum, possibly H. brachyantherum subsp. californicum (Salomon and Bothmer 1998), which was not found in the ITS analysis. Asterisk (*) indicates the DNA value of diploid H. brevisubulatum that was not measured but calculated from the polyploid accessions of this species. For H. guatemalense, no genome size value could be obtained
For statistical analyses SPSS version 10.0 was used. The entire data set did not show normal distribution, but some data subsets were normally distributed. When normal distribution and homogeneity of variance were satisfied, we used the t-test or the unifactorial ANOVA Tukey-HSD test. For data lacking homogeneity of variance, we used Dunnett-T3, and for data without normal distribution, we used the nonparametric Mann-Withney U test. Correlations between diploid genome sizes and environmental parameters were tested with a bivariate correlation using Pearson's product moment correlation.
For all diploid accessions, and additionally for some polyploids, average monthly temperatures and precipitation were determined using the coordinates of the collection sites together with two climate data bases (CLIMATE DATABASE Version 2.1, W. Cramer, Potsdam Institute of Climatic Impact Research, and the internet-based compilation of B. Mühr, available under http://www.klimadiagramme.de/all.html) and data from Walter and Lieth (1960–1967). The data were used to ascertain the respective local climate types according to K?ppen (1923), which, thus, reflect altitudinal as well as local geographical differences.
To reveal the influence of the factors phylogeny (represented by the Hordeum karyotype groups H1, H2, I, Xa, and Xu), life form (annuals versus perennials), climate, and, ages of species, a three-way analysis of variance with covariate (ANCOVA) was applied to the data from diploid Hordeum species. According to Lindman (1974), this is legitimate because the F-test is robust against deviations from normal distribution. Because some of the cells in the data scheme were empty, we analyzed only the main effects (no interaction effects, which are especially sensitive against missing combinations) and used model type IV of sums of squares, which is suitable for this situation. As the entire data set did not show homogeneity of variance, the level of significance was increased from 95% to 99% (Bühl and Z?fel 2000). All karyotype groups used for ANCOVA were statistically supported by bootstrap values and Bayesian posterior probabilities in the phylogenetic data.
On an intraspecific scale along latitudinal transects of collection sites, the respective genome sizes were analyzed in H. marinum subsp. marinum, H. pubiflorum, and H. lechleri. Here we tested genome size variation against ecogeographical components such as the average growing season, summer temperature and precipitation (July and January, respectively), and geographical latitude.
Results
Genome size measurements with flow cytometry revealed conspicuous variation among Hordeum species. Replicate measurements from individual Hordeum species were highly repeatable with a standard deviation of up to 4% of the measured genome size (table 1). Here, we describe the results in the order of the phylogenetic relationships within the genus.
Genome Size in Extant Hordeum Species
The I-genome clade comprises the perennial species H. bulbosum, with its diploid (2x) and tetraploid (4x) forms, and the annual species H. vulgare subsp. vulgare (barley), along with its wild progenitor H. vulgare subsp. spontaneum. The 2C DNA contents in this group are in a range from 8.88 pg in diploid H. bulbosum to 10.67 pg in H. vulgare. In the autotetraploid H. bulbosum, the average 2C DNA content (17.3 pg) is slightly smaller than twice the genome size (17.76 pg) of the diploid form. However, within the two analyzed tetraploid accessions, we ascertained a difference of 0.66 pg, and the H. bulbosum accession from Greece nearly contains the expected DNA amount (17.63 pg) calculated on the base of the adjacent Italian diploid accession. The more easterly accession from Tajikistan had a smaller genome (16.97 pg).
The Xu-genome clade consists only of the annual H. murinum, divided into three subspecies with varying ploidy levels from diploids to hexaploids (6x). The polyploid forms of H. murinum (4x subspp. murinum, 19.68 pg, and leporinum, 19.2 pg; 6x subsp. leporinum, 29.85 pg) are proposed to be of either autoployploid or segmental allopolyploid origin (von Bothmer et al. 1995). In all three polyploids, the respective tetraploid and hexaploid genome sizes are slightly larger than the corresponding multiple of the diploid subsp. glaucum (9.11 pg).
The Xa-genome clade is the sister group of the large H clade. It consists only of the annual H. marinum, with the subspecies marinum (2x) and gussoneanum (2x and 4x). The difference in genome size between the two diploid subspecies (9.10 pg vs. 10.41 pg) is highly significant. The DNA content of the tetraploid subsp. gussoneanum (19.64 pg) is slightly larger than the sum of the genomes of its proposed diploid progenitors (19.51 pg).
The H-genome taxon group consists of 27 diploid, tetraploid, and hexaploid species, with annual or perennial life form (fig. 1). 2C DNA amounts between the diploid Hordeum species differ by 1.42-fold, ranging from 6.85 pg in H. euclaston to 9.69 pg in H. roshevitzii. The group of tetraploid species combining two H genomes revealed genome sizes between 15.52 pg (H. depressum) and 18.57 pg (H. brachyantherum subsp. brachyantherum). Tetraploid H. guatemalense, a member of this group, was not analyzed, and no data were available from the literature. The genome sizes of the putative autotetraploid forms of H. brevisubulatum were 17.81 pg (subsp. brevisubulatum) and 18.16 pg (subsp. turkestanicum). European H. secalinum (19.98 pg) and the South African H. capense (19.74 pg) are phylogenetically closely related. Both consist maternally of the large Xa genome of diploid H. marinum subsp. gussoneanum (10.41 pg). The other parental genome stems from H. comosum (8.97 pg) or a close relative. Their genome sizes are, therefore, clearly larger than that of the other tetraploids in the clade, combining two H genomes.
The genome size values of the hexaploid species consisting of three H genomes were in a range between 24.68 pg (H. arizonicum) and 27.18 pg (H. procerum). One ancestor of all South American hexaploids is tetraploid H. fuegianum or H. tetraploidum. The second parental genomes stem from different diploid species. In the case of H. procerum (27.18 pg), the parental DNA amounts are perfectly additive. In H. parodii (26.19 pg) and H. lechleri (26.31 pg) the expected values were slightly larger (26.5 and 26.7 pg, respectively) than the measured ones. Californian H. brachyantherum subsp. brachyantherum revealed a conspicuously larger genome size (28.9 pg) than the other hexaploids from the H-genome clade, which is consistent with the origin of this taxon by hybridization of tetraploid subsp. brachyantherum with H. marinum subsp. gussoneanum (2x). The significantly smaller hexaploid genome of H. arizonicum resulted from the hybridization of the small-genome annual H. pusillum with H. jubatum.
Correlations of Phylogeny, Life Form, and Environmental Parameters with Genome Size
The ANCOVA confirmed the main impact of annual or perennial life form on genome size as well as the lesser, but also significant, influence of the phylogeny ("karyotype"). Climate types at the growing site and age of species had no significant effect on genome size variation in Hordeum (table 2). Species, with both small and large genomes were found in very stressful, nearly arid climates or at high altitudes, as well as under moderate climatic conditions. The 2C values of the perennials range between 8.51 pg (H. flexuosum) and 9.69 pg (H. roshevitzii), revealing 1.14-fold differences, whereas in annual species, 1.55-fold size differences occur (H. euclaston, 6.85 pg versus H. vulgare, 10.67 pg). Perennials and annuals co-occur only in the H-karyotype and the I-karyotype groups. Within these two groups, life form contributes markedly to the explanation of genome size variation. In the New World H-genome clade, the annuals revealed significantly smaller genomes than the perennials (mean 7.00 pg vs. mean 9.27 pg; Mann-Withney U test, P < 0.001), whereas in the I-genome group, we found, conspicuously, the contrary. The annual species shows significant larger genome sizes than the perennial (10.67 pg vs. 8.88 pg; t-test, P < 0.001). This result clearly excludes life form as a single parameter determining genome size.
Table 2 Three-Way Analysis of Variance with Covariate (ANCOVA) for the Contribution of Factors to the Measured Genome Size of the Hordeum Accessions.
Intraspecific Genome Size Variation
Within species, genome size differences along a latitudinal transect was tested for H. marinum subsp. marinum in Western Europe. Ten populations of H. marinum subsp. marinum were analyzed, occurring over a geographical range of 12° latitude. Genome size data were contrasted with different ecogeographical factors such as the mean July temperature and mean July precipitation. A negative correlation (Pearson = –0.862, P = 0.001) between the mean genome size and the mean July temperature was found (fig. 2). A positive correlation with smaller significance value (P = 0.027) was also assessed with mean July precipitation. Both parameters correlate reciprocally with northern latitude and influence markedly the length of the vegetation period for this annual species.
FIG. 2. Correlations of DNA content with ecogeographical parameters. DNA content versus mean July temperature in H. marinum subsp. marinum along a transect in Western Europe and DNA content versus latitudinal position of collection sites in H. pubiflorum and H. lechleri along a transect in southern South America
The same approach was used for H. pubiflorum (31 measurements of 13 populations) and H. lechleri (94 measurements of 33 populations) occurring over a geographical range of 12° latitude in South America. Instead of the mean January temperatures, which were not available for all collection sites, we used the southern latitudes as a rough estimation of the duration of vegetation period. We found no significant correlation between genome size variation and latitudinal distribution of the analyzed populations (fig. 2).
Ancestral Genome Size Estimations
Ancestral genome sizes, calculated with PGLS on the base of diploid Hordeum species confirmed the observation that within Hordeum, increasing and decreasing DNA contents existed along the lineages leading to the extant species (fig. 3). The basal 2C genome size of the genus was calculated as 10.8 pg DNA, a value close to the 10.7 pg assumed for ancient Triticeae by Kellogg (1998). The ancient DNA values steadily decreased in Hordeum and are significantly smaller in H. bulbosum, H. flexuosum, and the New World annuals than the initial Hordeum genome. Significant secondary genome size increase occurred in the lineage leading to H. marinum subsp. gussoneanum.
FIG. 3. Genome size evolution in diploid Hordeum species as inferred from extant genome size (average value/standard error) by PGLS. Ancient 2C genome size estimations (in pg) are given above the branches, the standard errors are given below. Asterisk (*) indicates genome sizes taken either from the literature (outgroups) or calculated from polyploid species. In these cases, the standard error values were increased to account for uncertainties in conversion of the data. Dashed lines in the tree indicate positions were the polytomies of the ITS tree were resolved in bifurcations by the introduction of short branch lengths for PGLS calculations of ancient genome sizes
Discussion
Genome size analysis of 134 accessions (394 individuals) representing all but one Hordeum species revealed up to 1.55-fold 2C DNA amount differences, laying in a range between 6.85 and 10.67 pg in diploid species, 15.52 to 19.98 pg in tetraploids, and 24.68 to 29.85 pg in hexaploids. Our results are compatible with the results of Kankanp??, Mannonen, and Schulman (1996) and generally compatible with Vogel, Arumuganathan, and Jensen (1999), although some systematical differences were detected. Our 2C DNA values were always 1.1 to 1.6 pg higher than these reported by Kankanp??, Mannonen, and Schulman (1996) and 0.3 to 0.9 pg lower compared with Vogel, Arumuganathan, and Jensen (1999). This clearly points to differences caused by the use of different internal DNA standards and/or fluorescent dyes, (i.e., DAPI versus propidium iodide [Doleel et al. 1998; Barow and Meister 2002]). However, we analyzed all accessions with the same procedure and with at least three individuals per accession and usually multiple accessions per species. Thus, we are confident that our data are internally comparable.
Genome Size Changes in Relation to Phylogeny
The conspicuous size differences found in Hordeum are not evenly distributed within the genus, but are primarily related to specific phylogenetic groups (fig. 1). Whereas in four clades, here informally named after the genome groups (I, H1, Xa, and Xu), the diploid 2C values were in a range between (8.88) 9.10 and 10.67 pg, only within the H2 subclade were genomes smaller than 8.8 pg found. The smallest genomes in this clade are restricted to three species that became secondarily annual, whereas the perennials vary between 8.51 and 9.61 pg. This correlation was also found in the polyploids, where the tetraploid annual H. depressum has about 2 pg less DNA than the smallest perennial taxon, and the hexaploid taxon with the smallest genome, H. arizonicum, is a perennial species that facultatively can also grow as a biennial or annual. As all three annual diploids are closely related and the annual polyploids include one of them as progenitor (fig. 1), it remains unclear if a smaller genome is a prerequisite for the annual life form or if both traits (ability for shorter generation time and the smaller genome) are inherited from a common ancestor. Because nearly 30 pg DNA per nucleus is compatible with the annual life habit of hexaploid H. murinum, simple phylogenetic constraints are probable. If phylogenetic relationships determine the genome size, only within close monophyletic groups should DNA contents show correlations to environmental parameters.
Genome Size Changes in Relation to Environmental Parameters
The narrowest monophyletic groups in our analyses were three intraspecific population samples. Hordeum marinum subsp. marinum was collected along a latitudinal transect in Western Europe, and H. pubiflorum and H. lechleri were colected over a comparable distance in Patagonia. Genome size analysis in the 10 H. marinum subsp. marinum populations revealed increasing genome sizes from south to north (fig. 2). Northern distribution here is linked with a prolonged vegetation period for these plants. This corroborates the results of Bennett (1987), Reeves et al. (1998), and Turpeinen, Kulmala, and Nevo (1999), who found positive correlations between the duration of vegetation period and genome size in plants. However, the genome sizes differ about 3.3% (0.30 pg) along our transect and thus are clearly different from the DNA values of H. marinum subsp. gussoneanum, where even in the southernmost populations the diploid genomes are more than 0.7 pg (8%) larger than in subsp. marinum. This indicates again that phylogenetic constraints, here depicted by taxonomic units, might be more important than ecological influences for genome sizes on a higher taxonomic level.
Within the South American species H. pubiflorum and H. lechleri, we found no correlations, neither with the climate types, with mean January temperatures, nor with increasing southern latitude. However, the assumption that for these species, southern latitude is directly correlated with vegetation period or other gradually changing ecoclimatic parameters may not be warranted. The rough landscape of Patagonia might present more complex local influences on plant habitats than expressed by the large-scale climate parameters we have taken into account.
Although on the world scale, correlations between climate zones, that is, habitats at higher latitudes, and genome sizes of plants were found (Bennett 1976, 1987; Levin and Funderburg 1979; Grime and Mowforth 1982), we could not confirm such a correlation among all species of the genus Hordeum. ANCOVA found no significant influence of local climate type for the explanation of genome size variation, which can be explained by the strong influence of phylogenetic relationships, concealing environmental correlations even among closely related taxa.
The Influence of Life Form on Genome Size Variation in Hordeum
Bennett and Leitch (1995) proposed that small genomes occur in annual taxa, whereas perennials could possess larger genomes. This correlation is well established for ephemeral species (Bennett 1972). In Hordeum, life form contributes significantly to the explanation of the genome sizes (table 2). However, the largest (H. vulgare) and the smallest (H. euclaston) diploid genomes were found in annuals. Whereas the annual species of the H2 clade seemingly support the conclusions of Bennett and Leitch (1995), the annuals of the Xa and Xu clade have medium-sized to large genomes (9.1 to 10.41 pg DNA), and in the I-genome clade, the relation is inversed. The perennial H. bulbosum has a much smaller genome (8.88 pg) than its annual sister taxon H. vulgare. Thus, low DNA content is obviously no prerequisite for annuality in Hordeum.
Genome size differences among annuals are likewise not related to summer or winter annuality. In H. marinum, we found that subsp. marinum, which partly grows as a winter annual, has a significantly smaller genome than subsp. gussoneanum, which is exclusively a summer annual. This contradicts the hypothesis of Grime and Mowforth (1982), who proposed that temporal detachment of cell proliferation and cell extension will allow larger genome sizes. Although analyses on the world scale seemingly support the fact that small genomes preferentially occur in annual species, among the evolutionarily more closely linked species of Hordeum, this ratio does not generally hold and is even inverted in the I-genome group.
Genome Size Evolution Through Time in Hordeum
Bennetzen and Kellogg (1997) proposed the possibility of "genomic obesity" in plants caused by polyploidization and the activity of retrotransposons, which should keep increasing the genome size. Our data and data by Wendel et al. (2002) provide information that plant genomes are capable of DNA gain and loss in closely related lineages. DNA loss seems to be the predominant process in Hordeum. Starting with the estimated diploid DNA amount of 10.8 pg (±1.8) for ancestral Hordeum taxa, all extant species now have smaller genomes (fig. 3). Ambiguities in the phylogenetic tree are restricted to the species relationships in the H2 clade. Topological changes in this part of the tree, however, do not change the estimation of the basal genome size and, thus, do not alter the general tendency of genome decrease in Hordeum. A significant secondary DNA content increase was found in H. marinum subsp. gussoneanum. Our data clearly indicate that a "one-way ticket" (Bennetzen and Kellogg 1997) for ever increasing genomes of plants does not exist. This result is not unexpected, as the analysis of Leitch, Chase, and Bennett (1998) proposed a small ancient angiosperm genome but also pointed out that several derived plant families have rather small genomes. Either genomic obesity, thus, would be found only in some lineages, possibly because of the loss of effective retroelement suppression, or mechanisms to jettison superfluous DNA are necessary. These mechanisms are not well understood in plants up to now. However, the biased DNA loss in Arabidopsis compared with tobacco during DNA double-strand break repair (Kirik, Salomon, and Puchta 2000; Orel and Puchta 2003) and unequal intrastrand recombination (Petrov 1997, 2001; Shirasu et al. 2000) indicate that possible mechanisms exist. Our analysis defines taxa that can be used to study these molecular mechanisms of genome size variation (Bennetzen 2002; Gregory 2003), as genome size changes in Hordeum can be traced in a phylogenetic framework.
Constancy of Genome Sizes Within Hordeum Species
The measurement of genome size in diploids revealed low intraspecific genome size variation (3%) among individuals and accessions from single species. Furthermore, the genome size of several polyploid species were nearly the added values of their progenitors. This holds particularly true for evolutionarily very young taxa, such as, for example, hexaploid H. brachyantherum, that originated during the past 200 years, after H. marinum was introduced to California, but also in H. arizonicum, H. procerum, H. marinum, and the western accessions of H. bulbosum. Differences between the summed genome sizes of the proposed progenitors and the polyploids indicate mostly a loss of about 0.5 pg DNA. Only in the closely related species H. secalinum and H. capense (+0.5 pg) and in the putative autopolyploid H. murinum subsp. murinum (+1.5 pg), larger than additive genomes were found (fig. 1). This could be either because of a larger genome in a parental taxon of these species or because DNA proliferation occurred in these lineages. These would, however, be the only cases in Hordeum where genome size increases instead of decreases after a polyploidization event. Constancy of genome size was also found in cases where we compared accessions grown for several generations ex situ with material collected directly from the wild. Many analyzed species were derived from germ plasm collections and, thus, underwent several cycles of propagation in cultivation. Comparisons of their genome size with the values derived from wild material of the same species resulted in values that all fell in the range of the natural intraspecific variation of the respective taxa. For newly collected plant material (>50 accessions) from Patagonia, we were even able to correctly predict the species affiliation of young seedlings, knowing their genome size and their geographical location. The constancy of DNA amounts in Hordeum species clearly contradicts the proposed fast intraspecific changes of genome size (Bennett and Bennett 1992; Kankanp??, Mannonen, and Schulman 1996; Price, Morgan, and Johnston 1998) but supports the view of genome sizes being fairly stable when a narrow species concept is used (Greilhuber 1988; Ohri 1998).
As intraspecific genome size changes were usually continuous, discontinuous changes were associated with (sub)species boundaries. If pronounced genome size changes occur preferentially during or after speciation, we would expect the largest differences in clades with high species numbers. This correlation was found for all clades of diploid Hordeum species (fig. 4), where particularly the H2 clade (13 diploid species) shows the largest (1.4-fold) genome size difference. However, we also found clear differences in the I-genome clade where only two species differ 1.2-fold. Within the I-genome clade, the oldest split between two single species in Hordeum occurs, representing about 7 Myr of independent evolution (fig. 1). Thus, time can also play a role in shaping genome size, although in the rapidly speciating groups, larger genome differences originated in shorter time (2 Myr in the H2 clade). ANCOVA accordingly showed that "age of species" explained few of the genome size differences. The low and uneven coverage of genome size values of angiosperms (Bennett 1998a) does not allow safe assertion about the relationship between taxonomically highly structured clades (i.e., comprising many species, genera, and tribes) and genome size variation. Nevertheless, in the analysis by Leitch, Chase, and Bennett (1998), ratios higher than 50:1 between maximum and minimum DNA amounts in angiosperm families were only found in Poaceae (104:1), Fabaceae (73:1), Iridaceae (66:1), Asteraceae (62:1), and Orchidaceae (57:1), which are all relatively species-rich families. Genome size changes during or shortly after speciation events would also be consistent with an often reduced effective population size in newly arising species, promoting rapid fixation of genome size changes (Lynch and Conery 2003).
FIG. 4. Correlation of the taxon number per clade with the percentage genome size differences within these clades for all diploid Hordeum taxa. The clades used in this calculation (I, Xa, H1, H2, H, H+Xa, I+Xu, and I+Xu+Xa+H) were chosen to represent monophyletic units according to figure 1
Our analysis of genome sizes in Hordeum species resulted in the paradoxical phenomenon that although meta-analyses of thousands of DNA values in angiosperms found correlations and trends in genome sizes with ecogeographical, physiological, and life history parameters (Bennett 1998b), these relationships could not be reliably established in the analysis of this one genus. For nearly all parameters proposed to influence genome sizes, we found examples in Hordeum that support these correlations but always with at least some exceptions. Thus, neither the knowledge of a species' life form, its ecogeographical distribution, nor its phylogenetic relationship allows a prediction of its genome size. Overlapping patterns of all three parameters influence the genomic composition of Hordeum species. However, the knowledge of these parameters allows safer posteriori interpretations of the observed DNA value patterns. Both earlier works reporting a certain number of genome size values for Hordeum species (Kankanp??, Mannonen, and Schulman 1996; Vogel, Arumuganathan, and Jensen 1999) struggled with the interpretation of the results. Kankanp??, Mannonen, and Schulman (1996) found no clear environmental or evolutionary patterns influencing genome size variation in Hordeum. However, they did not take into account life form as a third important parameter and argued on the basis of the wrong assumption that currently accepted sections in Hordeum represented natural phylogenetic groups. Moreover, the only hexaploid accession in their analysis seems, in the light of our data, clearly to be a misidentified tetraploid H. murinum. Vogel, Arumuganathan, and Jensen (1999), in an attempt to determine basic genome size values in Triticeae, included H. bulbosum in the H-genome taxa but excluded the respective annual species. Thus, their sample was not representative for the Hordeum H genomes and might also not be representative for the H genomes included in the heterogenomic Triticeae genera. As we have shown in H. depressum and H. arizonicum, the inclusion of an annual H genome in allopolyploids results in significantly smaller genome sizes compared with polyploids derived from perennial H-genome species, only. This can explain the smaller than expected 2C values of, for example, Elymus L. species with StH genome combinations (Vogel, Arumuganathan, and Jensen 1999).
To conclude, we can point out that although plant genome sizes might respond reasonably predictably to environmental parameters at the population level (Nevo 2001), with increasing taxonomic distance and on higher taxonomic levels, predictability rapidly diminishes through interacting influences of phylogenetic constraints, life form changes, and possibly also the number of speciation events occurring along the specific lineages of the analyzed plants. Correlations found in meta-analyses, therefore, represent trends that cannot be generalized and might even be inverted when looking into specific plant groups.
Supplementary Material
A table listing all 134 Hordeum accessions, their origins, climate data, and the measured genome size is available as Supplementary Material online linked to this article on the M.B.E. home page (http://mbe.oupjournals.org/).
Acknowledgements
We thank M. Arriaga and R. Gomez Cadret (Museo Argentino de Ciencias Naturales, Buenos Aires) and P. Cichero (Parque National Argentina, Buenos Aires) for kindly providing help in the organization of the field work in Patagonia. We also thank K. Bachmann, M. Barow, and M. A. Lysak, for critical reading of the manuscript, M. H. Hoffmann, who provided climate data, and A. Ihlow, for help with the statistics. Financial support from the Deutsche Forschungsgemeinschaft within priority program SPP 1127 (to F.R.B.) and from the Fond der Chemischen Industrie (to A.M.) is acknowledged.
Literature Cited
Arumuganathan, K., and E. D. Earle. 1991. Nuclear DNA content of some important plant species. Plant Mol. Biol. Rep. 9:208-218.
Barakat, A., N. Carels, and G. Bernardi. 1997. The distribution of genes in the genomes of Gramineae. Proc. Natl. Acad. Sci. USA 94:6857-6861.
Barow, M., and A. Meister. 2002. Lack of correlation between AT frequency and genome size in higher plants and the effect of nonrandomness of base sequences on dye binding. Cytometry 47:1-7.
Barow, M., and A. Meister. 2003. Endopolyploidy in seed plants is differently correlated to systematics, organ, life strategy and genome size. Plant Cell Environ. 26:571-584.
Bennett, M. D. 1972. Nuclear DNA content and minimum generation time in herbaceous plants. Proc. R. Soc. Lond. B. Biol. Sci. 181:109-135.
Bennett, M. D. 1976. DNA amount, latitude, and crop plant distribution. Environ. Exp. Bot. 16:93-108.
Bennett, M. D. 1987. Variation in genomic form in plants and its ecological implications. New Phytol. 106:177-200.
Bennett, M. D. 1998a. DNA amounts in two samples of angiosperm weeds. Ann. Bot. 82:(suppl. A): 121-134.
Bennett, M. D. 1998b. Plant genome values: how much do we know? Proc. Natl. Acad. Sci. USA 95:2011-2016.
Bennett, M. D., and I. J. Leitch. 1995. Nuclear DNA amounts in angiosperms. Ann. Bot. 76:113-176.
Bennett, S. T., and M. D. Bennett. 1992. Variation in nuclear DNA amount between wild and cultivated populations of Milium effusum (2n = 28). Genome 35:1050-1053.
Bennetzen, J. L. 2002. Mechanisms and rates of genome expansion and contraction in flowering plants. Genetica 115:29-36.
Bennetzen, J. L., and E. A. Kellogg. 1997. Do plants have a one-way ticket to genomic obesity? Plant Cell 9:1509-1514.
Blattner, F. R. 2004. Phylogenetic analysis of Hordeum (Poaceae) as inferred by nuclear rDNA ITS sequences. Mol. Phylogenet. Evol. (in press).
Bühl, A., and P. Z?fel. 2000. Einführung in die moderne Datenanalyse unter Windows. Addison-Wesley, Munich.
Caceres, M. E., C. De Pace, G. T. Scarascia Mugnozza, P. Kotsonis, M. Ceccarelli, and P. G. Cionini. 1998. Genome size variations within Dasypyrum villosum: correlations with chromosomal traits, environmental factors and plant phenotypic characteristics and behaviour in reproduction. Theor. Appl. Genet. 96:559-567.
Ceccarelli, M., E. Falistocco, and P. G. Cionini. 1992. Variation of genome size and organization within hexaploid Festuca arundinacea. Theor. Appl. Genet. 83:273-278.
Cox, A. V., G. J. Abdelnour, M. D. Bennett, and I. J. Leitch. 1998. Genome size and karyoptype evolution in the slipper orchids (Cypripedioideae: Orchidaceae). Am. J. Bot. 85:681-687.
Cullis, C. A., and W. Cleary. 1986. Rapidly varying DNA sequences in flax. Can. J. Genet. Cytol. 28:252-259.
Doleel, J., J. Greilhuber, S. Lucretti, A. Meister, M. A. Lysak, L. Nardi, and R. Obermayer. 1998. Plant genome size estimation by flow cytometry: inter-laboratory comparison. Ann. Bot. 82:(Suppl. A): 17-26.
Flavell, R. B., J. Rimpau, and D. B. Smith. 1977. Repeated sequence DNA relationships in four cereal genomes. Chromosoma 63:205-222.
Gaut, B. S. 2002. Evolutionary dynamics of grass genomes. New Phytol. 154:15-28.
Goin, O. B., C. J. Goin, and K. Bachmann. 1968. DNA and amphibian life history. Copeia 3:532-540.
Gregory, R. T. 2001. Genome size and developmental complexity. Genetica 115:131-146.
Gregory, R. T. 2002. A bird's-eye view of the C-value enigma: genome size, cell size, and metabolic rate in the class Aves. Evolution 56:121-130.
Gregory, R. T. 2003. Is small indel bias a determinant of genome size? Trends Genet. 19:485-488.
Greilhuber, J. 1988. Critical reassessment of DNA content variation in plants. Pp. 39–40 in P. E. Brandham, ed. Kew chromosome conference III. HMSO, London.
Grime, J. P. 1998. Plant classification for ecological purposes: is there a role for genome size? Ann. Bot. 82:(Suppl A): 117-120.
Grime, J. P., and M. A. Mowforth. 1982. Variation in genome size: an ecological interpretation. Nature 299:151-153.
Jockusch, E. L. 1997. An evolutionary correlate of genome size change in plethodontid salamanders. Proc. R. Soc. Lond. B Biol. Sci. 264:597-604.
Kalendar, R., J. Tanskanen, S. Immonen, E. Nevo, and A. H. Schulman. 2000. Genome evolution of wild barley (Hordeum spontaneum) by BARE-1 retrotransposon dynamics in response to sharp microclimatic divergence. Proc. Natl. Acad. Sci. USA 97:6603-6607.
Kankanp??, J., L. Mannonen, and A. H. Schulman. 1996. The genome sizes of Hordeum species show considerable variation. Genome 39:730-735.
Kellogg, E. A. 1998. Relationships of cereals crops and other grasses. Proc. Natl. Acad. Sci. USA 95:2005-2010.
Kirik, A., S. Salomon, and H. Puchta. 2000. Species-specific double-strand break repair and genome evolution in plants. EMBO J. 19:5562-5566.
Knight, C. A., and D. D. Ackerly. 2002. Variation in nuclear DNA content across environmental gradients: a quantile regression analysis. Ecol. Lett. 5:66-76.
Knüpffer, H., and T. van Hintum. 2003. Summarised diversity—the Barley Core Collection. Pp. 259–267 in R. von Bothmer, T. van Hintum, H. Knüpffer, and K. Sato, eds. Diversity in barley (Hordeum vulgare). Elsevier, Amsterdam.
K?ppen, W. 1923. Die Klimate der Erde: Grundri? der Klimakunde. de Gruyter, Berlin.
Laurie, D. A., and M. D. Bennett. 1985. Nuclear DNA content in the genera Zea and Sorghum. Intergeneric, interspecific and intraspecific variation. Heredity 55:307-313.
Leitch, I. J., M. W. Chase, and M. D. Bennett. 1998. Phylogenetic analysis of DNA C-values provides evidence for a small ancestral genome size in flowering plants. Ann. Bot. 82:(suppl. A): 85-94.
Levin, D. A., and S. W. Funderburg. 1979. Genome size in angiosperms: temperate versus tropical species. Am. Nat. 114:784-795.
Lindman, H. R. 1974. Analysis of variance in complex experimental designs. Freeman, San Francisco.
Lynch, M., and J. S. Conery. 2003. The origins of genome complexity. Science 302:1401-1404.
Manninen, I., and A. H. Schulman. 1993. BARE-1, a copia-like retroelement in barley (Hordeum vulgare L.). Plant Mol. Biol. 22:829-846.
Martins, E. P. 2003. COMPARE: Computer programs for the statistical analysis of comparative data. Version 4.4. Distributed by the author, http://compare.bio.indiana.edu/, Indiana University, Bloomington.
Nevo, E. 2001. Evolution of genome-phenome diversity under environmental stress. Proc. Natl. Acad. Sci. USA 99:6233-6240.
Ohri, D. 1998. Genome size variation and plant systematics. Ann. Bot. 82:(suppl. A): 75-83.
Orel, N., and H. Puchta. 2003. Differences in the processing of DNA ends in Arabidopsis thaliana and tobacco: possible implications for genome evolution. Plant Mol. Biol. 51:523-531.
Petrov, D. A. 1997. Slow but steady: reduction of genome size through biased mutation. Plant Cell 10:1900-1901.
Petrov, D. A. 2001. Evolution of genome size: new approaches to an old problem. Trends Genet. 17:23-28.
Price, H. J., K. L. Chambers, and K. Bachmann. 1981. Genome size variation in diploid Microseris bigelovii (Asteraceae). Bot. Gaz. 142:156-159.
Price, H. J., P. W. Morgan, and J. S. Johnston. 1998. Environmentally correlated variation in 2C nuclear DNA content measurements in Helianthus annuus. L. Ann. Bot. 82:(suppl. A): 95-98.
Reeves, G., D. Francis, M. S. Davies, H. J. Rogers, and T. R. Hodkinson. 1998. Genome size is negatively correlated with altitude in natural populations of Dactylis polygama. Ann. Bot. 82:(suppl. A): 99-105.
Salomon, B., and R. von Bothmer. 1998. The ancestry of Hordeum depressum (Poaceae, Triticeae). Nordic J. Bot. 18:257-265.
Shirasu, K., A. H. Schulman, T. Lahaye, and P. Schulze-Lefert. 2000. A contiguous 66-kb barley DNA sequence provides evidence for reversible genome expansion. Genome Res. 10:908-915.
Suoniemi, A., K. Anamthawat-Jónsson, T. Arna, and A. H. Schulman. 1996. Retrotransposon BARE-1 is a major, dispersed component of the barley (Hordeum vulgare L.) genome. Plant Mol. Biol. 30:1321-1329.
Swift, H. 1950. The constancy of deoxyribose nucleic acid in plant nuclei. Proc. Natl. Acad. Sci. USA 36:643-654.
Swofford, D. L. 1991. When are phylogenetic estimates from molecular and morphological data incongruent? Pp. 295–333 in M. M. Miyamoto and J. Cracraft, eds. Phylogenetic analysis of DNA sequences. Oxford University Press, New York.
Thomas, C. A. 1971. The genetic organizations of chromosomes. Annu. Rev. Genet. 5:237-256.
Turpeinen, T., J. Kulmala, and E. Nevo. 1999. Genome size variation in Hordeum spontaneum populations. Genome 42:1094-1099.
Vicient, C. M., A. Suoniemi, K. Anamthawat-Jónsson, J. Tanskanen, A. Beharav, E. Nevo, and A. H. Schulman. 1999. Retrotransposon BARE-1 and its role in genome evolution in the genus Hordeum. Plant Cell 11:1769-1784.
Vogel, K. P., K. Arumuganathan, and K. B. Jensen. 1999. Nuclear DNA content of perennial grasses of the Triticeae. Crop Sci. 39:661-667.
von Bothmer, R., N. Jacobsen, C. Baden, R. B. J?rgensen, and I. Linde-Laursen. 1995. An ecogeographical study of the genus Hordeum. 2nd Edition. IPGRI, Rome.
Walter, H., and H. Lieth. 1960–1967. Klimadiagramm-Weltatlas. Fischer-Verlag, Jena.
Wendel, J. F., R. C. Cronn, J. S. Johnston, and H. J. Price. 2002. Feast and famine in plant genomes. Genetica 115:37-47.(Sabine S. Jakob, Armin Me)