|Home | About | Journals | Submit | Contact Us | Français|
The amount of DNA comprising the genome of an organism (its genome size) varies a remarkable 40 000-fold across eukaryotes, yet most groups are characterized by much narrower ranges (e.g. 14-fold in gymnosperms, 3- to 4-fold in mammals). Angiosperms stand out as one of the most variable groups with genome sizes varying nearly 2000-fold. Nevertheless within angiosperms the majority of families are characterized by genomes which are small and vary little. Species with large genomes are mostly restricted to a few monocots families including Orchidaceae.
A survey of the literature revealed that genome size data for Orchidaceae are comparatively rare representing just 327 species. Nevertheless they reveal that Orchidaceae are currently the most variable angiosperm family with genome sizes ranging 168-fold (1C = 0·33–55·4 pg). Analysing the data provided insights into the distribution, evolution and possible consequences to the plant of this genome size diversity.
Superimposing the data onto the increasingly robust phylogenetic tree of Orchidaceae revealed how different subfamilies were characterized by distinct genome size profiles. Epidendroideae possessed the greatest range of genome sizes, although the majority of species had small genomes. In contrast, the largest genomes were found in subfamilies Cypripedioideae and Vanilloideae. Genome size evolution within this subfamily was analysed as this is the only one with reasonable representation of data. This approach highlighted striking differences in genome size and karyotype evolution between the closely related Cypripedium, Paphiopedilum and Phragmipedium. As to the consequences of genome size diversity, various studies revealed that this has both practical (e.g. application of genetic fingerprinting techniques) and biological consequences (e.g. affecting where and when an orchid may grow) and emphasizes the importance of obtaining further genome size data given the considerable phylogenetic gaps which have been highlighted by the current study.
Across eukaryotes as a whole, genome size (the amount of DNA in the unreplicated gametic nucleus, also known as the 1C value) varies over 40 000-fold. The smallest genome so far reported is found in the tiny, single-celled parasitic microsporidian Encephalitozoon intestinalis. Its genome comprises just 0·003 pg of DNA or 2·9 Mbp DNA (Vivares, 1999. N.B. 1 pg = 978 Mbp; Doležel et al., 2003). At the other end of the scale, the largest genome measured to date is found in the marbled lung fish Protopterus aethiopicus with over 130 pg of DNA or 130 000 Mbp of DNA (Pedersen, 1971). Despite this large range of genome sizes in eukaryotes, most groups are characterized by much narrower ranges in genome size (e.g. mammals, 3- to 4-fold, gymnosperms, 14-fold; Bennett and Leitch, 2005a; Gregory, 2005b). Angiosperms are unusual in that their genomes range nearly 2000-fold from Genlisea margaretae and G. aurea (Lentibulariaceae) with just 0·065 pg to the gigantic genome in Fritillaria assyriaca (Liliaceae) with 127·4 pg. Nevertheless, most angiosperms have small genomes with a modal and median genome size of just 0·6 pg and 2·9 pg, respectively (based on data for 4427 species in the Plant DNA C-values database www.kew.org/genomesize/homepage.html). Species with large genomes are largely restricted to monocots, most notably in Alliaceae, Asparagaceae, Liliaceae, Melanthiaceae and Orchidaceae.
The biological significance and evolution of the genome size diversity encountered in angiosperms has received considerable attention over the years (reviewed in Bennett and Leitch, 2005a; Leitch et al., 2007; Leitch and Bennett, 2007), but little work has focused specifically on Orchidaceae. This is largely due to the poor representation of genome size data in this family, especially given the enormous number of species comprising Orchidaceae (approx. 25 000 species; World Checklist of Selected Plant Families, 2008). Indeed the first orchid to have its genome size estimated (a Cymbidium hybrid) was reported in 1975 (Capesius et al., 1975), nearly 25 years after the first genome size estimate for a plant was reported (Ogur et al., 1951). Since then only 13 out of the 466 sources of genome size data in the Plant DNA C-values database have contained C-value estimates for orchid species (213 species in 60 genera).
In part, the lack of new data may arise from difficulties in obtaining accurate genome size estimates for some orchids using flow cytometry. In particular, problems have been noted in Vanilla (subfamily Vanilloideae; Bory et al., 2008) and for many species in the subfamily Orchidoideae where the low percentage of cells in the unreplicated (G0/G1) phase can lead to difficulties in distinguishing the 2C and 4C peak in flow histograms and hence errors in estimating DNA amounts (Suda et al., 2007; I. Leitch et al., unpubl. obs.). Further, in some species, the lack of additivity between assumed 2C and 4C peaks has led to concerns about how an accurate genome size can be estimated for such species. Possible explanations for non-additivity include only partial DNA replication for cells entering endopolyploidy rather than the cell division cycle (Nagl, 1983; Bosco et al., 2007) and/or induced nuclear hypertrophy arising from mycorrhizal infections (e.g. Barroso and Pais, 1990). Whatever the explanation, such observations should alert researchers to be vigilant to potential pitfalls in interpreting flow histograms for orchids and to ensure that appropriate controls are carried out. Alternatively, the problems may, in some cases, be circumvented by using the technique of Feulgen microdensitometry (Greilhuber and Temsch, 2001; Greilhuber, 2008) as, in our experience, material that is recalcitrant to study by flow cytometry can give reproducible genome size data using Feulgen microdensitometry.
Most available data are in the Plant DNA C-values database (Bennett and Leitch, 2005b). However, for the purposes of this review additional data from Aagaard et al. (2005) of four species not yet included in the database, together with unpublished data from the authors were also included. Together the compiled dataset contained genome size estimates for 327 species, comprising 91 genera in 16 subtribes, 11 tribes and four of the five subfamilies (based on the classification of Chase et al., 2003).
By analysing these data, this paper aims to provide an overview of the diversity of genome sizes encountered in Orchidaceae, viewed within the increasingly robust phylogenetic framework available for the family. In addition, insights into the possible consequences and evolution of the diversity of genome sizes are discussed.
Overall, the data show that, although orchids do not possess species with the smallest or largest genomes, they are currently the most variable angiosperm family, with values ranging 168-fold from 1C = 0·33 pg in Trichocentrum maduroi to 55·4 pg in Pogonia ophioglossoides. Despite this large range, most species are characterized by small genomes with modal, median and mean genome sizes of just 1·2, 4·0 and 8·5 pg, respectively (Fig. 1). Species with genomes >20 pg are thus far restricted to 48 out of the 59 species in Cypripedioideae and one species in Vanilloideae.
Figure 2 shows the range and distribution of genome sizes for each subfamily and highlights differences between them. The most variable subfamily is Epidendroideae with values ranging over 60-fold (1C = 0·3–19·8 pg). Nevertheless, the vast majority are characterized by small genomes (median 1C = 3·2 pg), with outliers largely restricted to terrestrial species in tribe Neottieae and subtribe Collabiinae (Fig. 2B).
In contrast, Orchidoideae are characterized by a much narrower range of genomes (1C = 2·9–16·4 pg) varying just 6-fold (Fig. 2C), although the average genome size is bigger in this subfamily (mean 1C = 8·4 pg) than in Epidendroideae (mean 1C = 3·6 pg). It is, however, noted that nearly all data (40 out of 42 species) come from species in subtribe Orchidinae, so sampling is currently uneven.
Cypripedioideae show yet another pattern. Here genome sizes range just 10-fold (1C = 4·1–43·1 pg) but include several Cypripedium species possessing genomes more than twice the size of those found in Epidendroideae and Orchidoideae. The mean genome size for Cypripedioideae (1C = 25·8 pg) is the largest for any subfamily for which there are data.
To date, there are just four genome size estimates for Vanilloideae so data are sparse, yet they hint at a big dichotomy in genome sizes between the two tribes comprising this subfamily. The narrow range of intermediate genome sizes for three Vanilla species (1C = 7·3–8·0 pg) in tribe Vanilleae contrasts with the very large genome in Pogonia ophioglossoides (1C = 55·4 pg, the largest of any orchid reported to date) in tribe Pogonieae. The prediction that this will turn out to reflect a more general pattern comes from the more extensive cytological studies by Baldwin and Speese (1957) and Felix and Guerra (2005) who noted ‘exceptionally large chromosomes’ (up to approx. 10 µm long) in other genera of Pogonieae including Cleistes (both temperate North and tropical South American representatives) and Isotria (temperate North American). This contrasts with the much smaller chromosomes reported in tropical Vanilleae; i.e. generally <3 µm and often much smaller in Vanilla, although chromosome numbers can reach 2n = approx. 170 in Epistephium lucidum (Felix and Guerra, 2005). Although the relationship between genome size and total chromosome length in a karyotype is not precise (e.g. see Bennett and Rees, 1967, 1969), there is often a good correlation between these two parameters (typically exceeding r = 0·85; see Levin, 2002, and references within) enabling chromosome data to be used as rough proxies for genome size. Given this, it suggests that genome size evolution in subfamily Vanilloideae is dynamic and clearly merits further scrutiny.
Currently there are no genome size estimates for subfamily Apostasioideae, although again, chromosome data for three of the 16 species provide hints as to what might be expected. Reports by Okada (1988) and Larsen (1968) showed that these genera are characterized by numerous (2n = 48, 96, approx. 144), but small chromosomes (i.e. <2 µm in Apostasia and approx. 0·5 µm in Neuwiedia singapureana with 2n = approx. 144) suggesting that their genomes may be only small to intermediate in size.
At the genus level, genome sizes are available for 91 out of approx. 800 genera currently recognized in the family (Chase et al., 2003). Table 1 lists the minimum, maximum and mean 1C-value for each genus together with the percentage representation, which ranges from 0·09% (Epidendrum) to 100% in four monotypic genera (Ansellia, Gennaria, Mexipedium, Zelenkoa). It is clear that there are huge gaps, not only at the genus level but throughout the family. These are highlighted in Fig. 3, which shows the major subdivisions of Orchidaceae (subfamilies, tribes and subtribes) with at least one genome size estimate. Even species-rich tribes such as Diurideae (874 species) and Podochileae (1232 species) and subtribes including Pleurothallidinae (3999 species) still lack genome size data.
Despite limitations in the phylogenetic spread of data, it is already clear that genome sizes in Orchidaceae vary considerably, raising questions about the biological significance of this diversity. These questions have long intrigued biologists since it first became clear that the DNA amount of an organism did not reflect organismal complexity (Vendrely, 1955; Comings, 1972). Indeed in 1971 Thomas coined the term ‘C-value paradox’ to reflect this ongoing puzzle (Thomas, 1971). The confusion was largely based on the mistaken assumption that genome size was correlated with gene number and hence complexity, and it was not clear why, for example, an onion (Allium cepa) should have five times as much DNA as humans. Today this paradox has largely been resolved as it is now well understood that the huge variation in genome size reflects not differences in gene number (which do not vary extensively between plants) but differences in the amount of non-coding, often highly repeated, DNA sequences (Flavell et al., 1974, 1977). Nevertheless, many questions remain, relating to the types of non-coding DNA sequences that predominate in genomes of different sizes, the mechanisms and evolutionary forces driving genome size changes and the consequences of this variation, among others. This has led to the paradox being renamed the ‘C-value enigma’ (Gregory, 2001), and there is a large active research field addressing many aspects of this (e.g. Bennett and Leitch, 2005a; Gregory, 2005a; Leitch and Bennett, 2007; Francis et al., 2008; Knight and Beaulieu, 2008).
Understanding the consequences of genome size variation has received considerable attention over the years. In relation to Orchidaceae, such studies reveal that variation in genome size has both practical and biological consequences.
Genome size has been shown to play a role in determining the success of various genetic fingerprinting techniques such as nuclear microsatellites (Garner, 2002; Barbará et al., 2007; Leitch and Fay, 2008) and amplified fragment length polymorphisms (AFLP) (Fay et al., 2005), both of which have been used to assess genetic diversity for addressing, for example, conservation questions. For AFLPs, orchids with small genomes up to approx. 10 pg have been shown to produce acceptable traces with a sufficient number of bands to estimate genetic diversity (Fig. 4A for Liparis loeselii 1C = 6·8 pg), whereas for species with 1C values above 10–15 pg AFLP traces are usually unusable (Fay et al., 2005; Leitch and Fay, 2008; Fay et al., 2009). This is mainly because many of the loci are not amplified strongly enough to be scored, rather than because of a real decrease in the number of loci. For example, in Cypripedium calceolus (1C = 32·4 pg) the AFLP traces were uninformative, giving just a few strongly amplifying bands with no variation, not only between individuals of different populations, but also when compared with other species of Cypripedium such as C. macranthos (Fig. 4B).
Genome size is therefore an important consideration when embarking on such studies. Given that species with large genomes may be at greater risk of extinction (Vinogradov, 2003), orchids possessing such large genomes are thus doubly compromised – they are more likely to be at risk of extinction and yet the techniques available for analysing and assessing their genetic diversity for input into conservation strategies are more limited.
Over the years there have been many comparative studies that have shown genome size is correlated with a wide range of phenotypic characters at the nuclear, cellular and tissue level (Leitch and Bennett, 2007). Often these correlations are remarkably tight, and as a result they have considerable predictive value. For example, Fig. 5 shows the strong correlation (P < 0·001) between genome size and the duration of meiosis for 20 angiosperm species, illustrating how it can vary from <24 h in Petunia hybrida (Solanaceae; 1C = 1·68 pg) to over 2 weeks in Fritillaria meleagris (Liliaceae; 1C = 70·68 pg) (Bennett, 1977). Although we are unaware of any estimates for the duration of meiosis in Orchidaceae, such a tight relationship enables one to predict that this would also vary from much less than 24 h in Trichocentrum maduroi (1C = 0·33 pg) to nearly 2 weeks in Pogonia ophioglossoides (1C = 55·4 pg).
Correlations such as these have knock-on effects at the whole-plant level, and there are now a number of studies that suggest that as genome size increases these effects have an increasing impact on the plant. For example, it has been shown that with increasing genome size, species become more sensitive to radiation (Sparrow and Miksche, 1961; Abrahamson et al., 1973) and metal pollution (Vidic et al., 2003), show reduced speciation rates (Vinogradov, 2003; Knight et al., 2005) and have a more restricted ecological distribution (Knight and Ackerly, 2002; Knight and Beaulieu, 2008). Further, species with genomes larger than 1C = approx. 20 pg are restricted to being obligate perennials (Bennett, 1972), are at greater risk of extinction (Vinogradov, 2003) and show a markedly different relationship between cell cycle time and genome size compared with species with smaller genomes (Francis et al., 2008). Thus, having a large genome does have consequences for the plant, and whereas such studies have involved large-scale analyses across angiosperms, we believe they should apply to orchids in particular and the following examples illustrate this.
In orchids it appears that having a large genome may impose constraints on the life-style options available, restricting such species to a terrestrial rather than epiphytic habit. This is seen by comparing the distribution of available genome size data for these two life-style options (Fig. 6). Whereas terrestrial orchids are characterized by relatively large genomes (mean 1C = 18·3 pg, range 2·9–55·4 pg; Fig. 6A), those in epiphytic species are smaller (mean 1C = 3·0 pg; range 0·33–8·5 pg; Fig. 6B) and occupy the bottom approx. 15% of the entire range of genome sizes encountered in orchids. Why this should be is not clear, but it could arise, in part, due to selection on cell size, particularly guard cell size, which plays a role in determining the response of a plant to water stress. Epiphytes are considered to be under considerable water stress due to the intermittent availability of water, often displaying characteristics associated with drought such as thick cuticles and water storage organs. Since species with small guard cells (and hence small genomes) have a more rapid response to water stress (Aasamaa et al., 2001; Hetherington and Woodward, 2003) than those with large guard cells, it might be expected that there would be selection for small guard cells. Since cell size and genome size have been shown to be correlated (Beaulieu et al., 2008), selection for small guard cells would result in small genome sizes. Of course there will be other factors determining the life style an orchid adopts and indeed not all species with small genomes are epiphytic, as there are some terrestrial species with small genomes. However, it seems that once the genome size gets too big then this option is closed, and the orchid is restricted to a terrestrial habit.
A similar theme is observed when considering the relationship between genome size and the diverse array of life-history strategies in subtribe Oncidiinae (Chase et al., 2005). The majority of species in this subtribe, like many epiphytic orchids, take a relatively long time to reach maturity, possibly up to 5 years. These species generally occupy typical epiphytic sites such as the main axes of trees. However, there are some species that can develop extremely rapidly, reaching flowering within 1 year. They are often found on twigs, hence their colloquial name ‘twig epiphytes’, but sometimes individuals may even complete their life cycle on leaves of other plants including Coffea and Hibiscus (Chase, 1988; Chase and Palmer, 1997).
Although there is at least one genome size estimate for 34 out of the 55 genera comprising Oncidiinae, sampling is still sparse with genome size data available for just 103 out of an estimated 1700 species in the subtribe. Nevertheless, rapid cyclers have a much narrower range of genome sizes (1C = 0·8–1·9 pg) and include species with some of the smallest genomes for the subtribe. In contrast, more typical epiphytic species of Oncidiinae are characterized by a wider range of genome sizes (1C = 0·3–8·5 pg) including not only the species with the smallest genome in Oncidiinae (Trichocentrum maduroi, 1C = 0·33 pg) but also the species with the largest so far reported (1C = 8·5 pg for Rossioglossum grande).
It is perhaps easy to see that a small genome would be advantageous for species needing to grow quickly in nutrient-poor sites where minerals such as phosphorus needed for DNA synthesis would be in short supply and where the duration of both meiosis and mitosis would be short (cf. Fig. 5), but Chase et al. (2005) also examined the distribution of genome sizes in Oncidiinae within a phylogenetic context and noted that sister clades containing species that were not rapid cyclers also had small genomes. Thus from an evolutionary perspective, the smaller genome sizes in some members of Oncidiinae appear to have been under selection first for other reasons, which then facilitated further alteration of life history strategies giving rise to these rapid cyclers. Overall, although having a small genome does not mean that a species will be a rapid cycler, once the genome becomes too big then the option to evolve a rapid-cycling strategy is closed.
It summary, in Orchidaceae, as in other angiosperm families, the large variation in genome size does have consequences, and with increasing genome size the number of options available is reduced. Examples have been given for life-style and life-history strategies specifically for orchids, but as the number of genome size measurements grows, further examples are likely to be uncovered. This may indeed result in a shift in the balance of survival and lead to an increased risk of extinction as genome size increases, as noted in an analysis of 3036 species taken from the Plant DNA C-values database by Vinogradov (2003) (see also Leitch and Bennett, 2007). However, there are currently too few genome size data for orchids to show whether the striking relationship between genome size and threat of extinction is also found specifically within orchids.
Numerous approaches have been made to analyse genome size evolution within a phylogenetic context, including the use of various statistical programs to optimize genome size either as a discrete (Bennetzen and Kellogg, 1997; Soltis et al., 2003) or continuously varying character (Albach and Greilhuber, 2004; Weiss-Schneeweiss et al., 2005; Leitch et al., 2007). Such approaches have uncovered dynamic patterns of genome size evolution with both increases and decreases being identified (e.g. Bennetzen and Kellogg, 1997; Wendel et al., 2002; Soltis et al., 2003) and nodes associated with significant shifts in genome size being highlighted (Leitch et al., 2007). However, as Chase et al. (2005) argued, such approaches are flawed if sampling is sparse. Given the large gaps in our knowledge for Orchidaceae (see above, Fig. 3) attempts to examine patterns of genome size evolution across the family as a whole and to infer the ancestral genome size seem premature. Such insights will have to await significantly improved sampling.
Nevertheless, there are two studies that have examined genome size evolution in a phylogenetic framework within orchids. The first is by Chase et al. (2005) in Oncidiinae, already discussed above. The second is by Cox et al. (1998) and I. Kahandawala et al. (unpubl. res.) who analysed patterns of genome size evolution in the slipper orchids (subfamily Cypripedioideae).
Cypripedioideae are currently the subfamily with the best representation of data with genome size estimates available for 38. 1% of species. Superimposing these values with chromosome information onto a phylogenetic tree (based on the work of Cox et al., 1997a) has revealed that even within this relatively small orchid subfamily, there is considerable variation in genome size and how this DNA is packaged into chromosomes (Fig. 7).
Paphiopedilum is characterized by a relatively wide range of genome size (2·2-fold, 1C = 16·5–35·9 pg, mean 1C = 25·4 pg) and chromosome number (2n = 26–42), whereas Phragmipedium has a narrower range (1·5-fold, 1C = 6·1–9·2 pg, mean 1C = 7·3 pg) although it is still variable in terms of chromosome number (2n = 18–30). The monotypic genus Mexipedium, has a genome size of 1C = 6·73 pg, and 2n = 26 chromosomes which are similar in size to those found in the sister genus Phragmipedium (Fig. 8A, B). Cypripedium has the greatest range in genome size (10·5-fold, 1C = 4·1–43·1 pg) but a nearly constant chromosome number with the great majority of species possessing 2n = 20 chromosomes. Currently there are no genome size data for Selenipedium, although chromosome preparations of S. aequinoctiale by Karasawa et al. (2005) showed that it had 2n = 20 like Cypripedium. However the total length of its karyotype (40 µm) was shorter than those of the other genera [average chromosome lengths of 283 µm in Cypripedium (15 taxa), 202 µm in Paphiopedilum (73 taxa) and 86 µm in Phragmipedium (16 taxa)] suggesting that the genome of Selenipedium is also considerably smaller.
Cytologically, both Paphiopedilum and Phragmipedium have been well characterized (Karasawa, 1979, 1980, 1986; Karasawa and Tanaka, 1980, 1981; Karasawa and Saito, 1982; Karasawa and Aoyama, 1986, 1988). Such studies have revealed that the diversity of chromosome numbers reported in these two genera has arisen via the karyotypic processes of Robertsonian change, whereby alterations in chromosome number arise by the fission or fusion of chromosomes at or near the centromere to generate telocentric or metacentric chromosomes, respectively. Such processes conserve the total number of chromosome arms, termed the ‘nombre fundamental’ (n.f.) (Matthey, 1949).
In Paphiopedilum the n.f. is 52 with chromosome numbers ranging from 2n = 26, comprising all metacentrics to 2n = 42 with 10 metacentrics and 32 telocentrics (e.g. see Fig. 8C, D). Superimposing chromosome data onto a phylogenetic framework, Cox et al. (1997a) showed that the direction of chromosome evolution was predominantly through centric fission of metacentric chromosomes leading to an increase in chromosome number (Fig. 9). Superimposing genome size data onto the phylogenetic tree (Fig. 9A) or plotting chromosome number against genome size (Fig. 9B) it becames clear that the increase in chromosome number had been accompanied by an increase in genome size. The need to synthesize telomeric DNA sequences to stabilize the new chromosome ends following fission is one likely source of the additional DNA as observed in the ‘healing of broken chromosomes’ (e.g. Tsujimoto et al., 1997; Putnam et al., 2004), but most of the DNA is likely to be additional repetitive DNA including retrotransposons, which can lead to increased genetic diversity (Gregory, 2005c; Kidwell, 2005).
A similar but less marked trend was observed in Phragmipedium (given the narrower range of genome sizes) although here the n.f. was 36, with chromosome numbers ranging from 2n = 18 (all metacentrics) to 2n = 30 (six metacentrics and 24 telocentrics) (Cox et al., 1998).
The nature of selective forces that might cause this mode of chromosome evolution is presently unknown, although it has been suggested that an increase in chromosome number can increase the amount of genetic recombination and hence genetic diversity while minimizing the risk of deleterious rearrangements (Imai et al., 1986, 2001; Schubert, 2007). Indeed many species of Paphiopedilum with high chromosome numbers are narrow endemics on islands in South-East Asia, which might be expected to be under rigorous selective pressure. It is possible therefore that this mode of karyotype evolution might be under selection to provide a way by which these slipper orchids can adapt more rapidly to their environments. Similar trends have been observed in some other genera displaying this mode of chromosome evolution (e.g. Jones, 1998; Imai et al., 2001).
As noted above, Cypripedium is characterized by the largest range of genome sizes in the subfamily, but chromosome number is more or less constant, with most species having 2n = 20 (Brandham, 1999). Genome size evolution has therefore been accompanied by changes in chromosome size rather than number (Fig. 8E, F). Superimposing available genome size data onto the phylogenetic tree of Cox et al. (1997a) highlights the distinct division between the large (1C > 20 pg) genomes characterizing the majority of species and the small genomes in C. molle (1C = 4·1 pg) and C. irapeanum (1C = 4·6 pg), both members of section Irapeana (I. Kahandawala et al., unpubl. res.). As section Irapeana is one of the early branching groups in the genus, and may be sister to all other sections, the ancestral genome size of Cypripedium may have been small, as in C. molle and C. irapeanum, with all the other species experiencing an increase in genome size or alternatively it could have been large, with the species of section Irapeana having undergone genome downsizing. Resolution of this situation will require further sampling of Cypripedium species for genome size measurement and phylogenetic studies. However, some species are rare and material is not readily available.
One way around this problem is to exploit the relationship between stomatal guard cell size and genome size, using guard-cell size as a proxy for genome size (Masterson, 1994; Beaulieu et al., 2008). By measuring guard cells from herbarium sheets of some additional species, approximate estimates of genome size can be obtained. Further, extending such approaches to Selenipedium has the potential to shed light on the ancestral genome size for the whole subfamily.
Given the vast number of species in Orchidaceae, the likelihood of obtaining genome size data for all species is remote. Nevertheless, by careful targeting of taxa, as illustrated by the study of Oncidiinae by Chase et al. (2005), the opportunity to obtain meaningful data and gain a greater understanding of the true diversity of genome sizes in the family exists. In addition, if the relationship between genome size and guard cells can be shown to hold across all orchids then this approach provides an additional tool for filling gaps, particularly for species in which obtaining fresh material is difficult. Given the wealth of herbarium material available for orchids, the prospects for increasing our understanding of genome size diversity and evolution in orchids seem better than ever.
We thank J. Rauchová for her help with flow cytometric analyses and Jan Larsen for providing some of the orchid material. This work was partly supported by the Ministry of Education, Youth and Sports of the Czech Republic (MSM 0021620828) and the Academy of Sciences of the Czech Republic (AV0Z60050516).