Genome and chromosome sizes of the Poaceae
2C values. The analyzed species from 11 of the 12 subfamilies of the grasses had 2C values (holoploid diplophasic, i.e., sporophytic genome sizes of the non-replicated nuclear DNA) between 0.67 pg and 45.91 pg (Tables 1, 2; Figs. 1A, 2; Online Resource 1), thus spanning the size range from “very small” to “large” (Leitch et al. 1998). No data were available for the subfamily Puelioideae. The 2C values of most subfamilies ranged from about 2.5 pg to 8.0 pg, thus falling predominantly into the “small” category defined by ≥ 2.8 pg/2C and ≤ 7.0 pg (Leitch et al. 1998). This was true for the subfamilies Anomochlooideae, Aristidoideae, Arundinoideae, Bambusoideae, Danthonioideae, Micrairoideae, Oryzoideae and Pharoideae. The only estimate for the Chloridoideae was 1.55 pg/2C, placing it in the “small” category, as well as the Oryzoideae, which had 0.84–1.84 pg/2C. The greatest variation was found in the Pooideae, which alone accounted for the aforementioned range of variation in the entire Poaceae family, followed by the Panicoideae, where 0.9–8.12 pg/2C were found, all in all comparable to the previous review of monocot genome sizes (Leitch et al. 2010).
Table 2
Genome sizes (holoploid 2C and monoploid 1Cx values) and mean chromosome DNA content (MC) of the examined representatives of the grass subfamilies. The most frequent chromosome base numbers in a subfamily are in bold. For details on our data see Table 1 and Online Resource 1. For further data as specified in Material and Methods see the individual subfamilies in Results and Discussion. Data for the Pooideae are from Tkach et al. (in prep.) and Winterfeld et al. (in prep.). N/A = not available.
Subfamilies and chromosome base numbers | 2C value [pg] | 1Cx value [pg] | MC [pg] |
---|
Anomochlooideae (x = 9?, 11) | 2.63 | 1.32 | 0.12 |
Pharoideae (x = 12) | 2.48 | 1.24 | 0.10 |
Puelioideae (x = 12) | N/A | N/A | N/A |
BOP clade | | | |
Bambusoideae (x = 7, 9, 10, 11, 12) | 3.19–7.01 | 0.53–1.75 | 0.04–0.15 |
Oryzoideae (x = 12) | 0.84–1.84 | 0.42–0.50 | 0.04 |
Pooideae (x = 2, 4, 5, 6, 7, 8, 9, 10, 11, 12, 14) | 0.67–45.91 | 0.33–9.45 | 0.02–1.84 |
PACMAD clade | | | |
Aristidoideae (x = 11, 12) | 2.6–2.62 | 1.30–1.31 | 0.12 |
Arundinoideae (x = 6, 9) | 2.09–5.25 | 0.26–0.59 | 0.04–0.10 |
Chloridoideae (x = 6?, 7, 8, 9, 10, 12) | 1.55 | 0.39 | 0.04 |
Danthonioideae (x = 6, 7, 9) | 1.63–8.30 | 0.69–0.82 | 0.12–0.14 |
Micrairoideae (x = 10) | 3.64 | 0.61 | 0.06 |
Panicoideae (x = 3, 4, 5, 7, 8, 9, 10, 11, 12) | 0.9–8.12 | 0.45–1.91 | 0.05–0.19 |
1Cx values. The genome size of the monoploid non-replicated chromosome sets ranged from 0.26 pg to 9.45 in the grasses, but here, too, this was mainly due to the subfamily Pooideae (Tables 1, 2; Figs. 1B, 2; Online Resource 1). This was followed by variation in the Panicoideae (0.49–1.91 pg/1Cx) and Bambusoideae (0.53–1.71 pg/1Cx). Medium-sized monoploid genomes of about 1.2–1.3 pg/1Cx occurred in the Anomochlooideae, Aristidoideae and Pharoideae, while the smallest of about 0.3–0.8 pg/1Cx were found in Arundinoideae, Chloridoideae, Danthonioideae, Micrairoideae and Oryzoideae. The Oryzoideae, the subfamily of rice, thus belongs to the grasses with a small genome size, as has long been known. However, small monoploid grass genomes of < 0.4 pg/1Cx occurred also in some taxa of the Arundinoideae, Chloridoideae and Pooideae.
Mean chromosome DNA content (MC). The chromosomes sizes varied altogether between 0.02 pg and 1.84 pg, implying a 92-fold variation, which is due to the MCs of only a single subfamily, the Pooideae. Most subfamilies had MCs of 0.04–0.19 pg (Arundinoideae, Bambusoideae, Danthonioideae, Panicoideae). The examined representatives of the Chloridoideae, Micrairoideae and Oryzoideae were at the lower limit of MCs with 0.04–0.05 pg. Anomochlooideae, Pharoideae and Aristidoideae were medium-sized with MCs of 0.10–0,12 pg (Tables 1, 2; Figs. 1C, 2; Online Resource 1).
Characteristics of the subfamilies
Data on genome size (2C and 1Cx values) and chromosomal DNA content (MC) for the Poaceae subfamilies are presented in Tables 1 and 2, Online Resource 1, and illustrated in Figs. 1, 2.
Early diverging grass lineages
Anomochlooideae. Genome size data for this small neotropical subfamily of 4 species in 2 genera, which were not sampled in this study, are only available for Streptochaeta angustifolia. Its 2C value was estimated to be 2.60–2.66 pg, also using FCM + PI (Seetharam et al. 2021). The chromosome number is not known, but could be 2n = 22, as has been repeatedly found in two other Streptochaeta species (CCDB). The 1Cx value of S. angustifolia would be then ca. 1.32 pg and the MC ca. 0.12 pg.
The genome size of monospecific Anomochloa, the second genus of this subfamily, is yet unknown. Its chromosome number is n = 18, well documented by a microphotograph showing 18 bivalents at diakinesis (Hunziker et al. 1989; Judziewicz & Soderstrom 1989).
Pharoideae. This tropical African subfamily with 12 species and 3 genera was represented in this study only by Pharus latifolius with a 2C value of 2.48 pg, which is in good agreement with previous studies using FCM + PI as fluorescent dye that found 2,467 Mbp (≈ 2.52 pg/2C) and 2,270 Mbp (≈ 2.32 pg/2C), respectively (Šmarda et al. 2014; Ma et al. 2021). The P. latifolius genome assembled from sequencing data had a length of 1,002.88 Mb/1C (≈ 2.05 pg/2C), with an estimated genome completeness of approximately 92.6% (Ma et al. 2021). Therefore, the interpolated 2C value would be about 2.21 pg/2C. Pharus latifolius has 2n = 24 (CCDB), is probably based on x = 12, resulting in a 1Cx value of 1.24 pg and an MC of 0.10 pg.
Puelioideae. For this tropical African subfamily of 11 species in 2 genera, 2n = 24 has been found repeatedly in two Puelia species (CCDB), but their genome size is apparently still unknown.
BOP clade
The BOP clade comprises three subfamilies with a total of more than 5,900 species in 374 genera (Soreng et al. 2022), all of which are characterized by C3 photosynthetic pathway. The clade is distributed worldwide.
Bambusoideae. The mainly tropical to subtropical, partly warm temperate distributed subfamily Bambusoideae (bamboos) are the third largest subfamily of grasses and comprise about 1700 species in 120–140 genera (Clark et al. 2015; Soreng et al. 2022; Clark 2023). The studied members of this subfamily had 2C values between 3.19 pg and 7.01 pg.
The three studied species of the genera Bambusa and Gigantochloa, which belong to the paleotropical members of the consistently woody and polyploid bamboos of the tribe Bambuseae, had rather uniform 2C genome sizes of 3.19–3.42 pg. The values for B. multiplex, B. vulgaris and G. verticillata were in the same order of magnitude as those previously found for the same or other species of both genera (Zhou et al. 2017 and Chalopin et al. 2021, both using FCM + PI). The chromosome numbers of the species we studied are 2n = 68–72 (CCDB), suggesting a 6-fold ploidy probably based on x = 12. The resulting 1Cx values were to 0.53–0.57 pg and the MCs 0.04–0.05 pg. For neotropical Bambuseae species of the genera Guadua and Chusquea, genome sizes of 3.63 pg/2C and 3.99 pg/2C (both G. angustifolia), 3.98 pg/2C (G. chacoensis) and 4.77 pg/2C (C. tenella), all tetraploid, were found in estimates with FCM + PI (Guo et al. 2019; Zappellini et al. 2020). Their 1Cx values would be 0.91–1.19 pg and MCs 0.08– 0.11 pg. Genome sequencing of G. angustifolia found a genome size of 1,580 Mb (≈ 3.23 pg/2C) (Guo et al. 2019).
The sampled species of Arundinaria, Fargesia and Pseudodasa, which belong to the tribe Arundinarieae, the temperate woody and also consistently polyploid bamboos, had 5.49–7.01 pg/2C. The genome sizes of the Arundinarieae are thus significantly larger than those of the paleotropical Bambuseae, as previously noted (Zhou et al. 2017; Chalopin et al. 2021). Chromosome numbers are not available for our sampled species, but for many congeners, all of which consistently had 2n = 48 (CCDB), presumably also based on x = 12. Thus, we can assume that our Arundinarieae taxa have 1Cx genome sizes of 1.37–1.75 pg, about three times than those of the paleotropical woody Bambuseae studied. The MCs were 0.11–0.15 pg, which is also considerably larger (about 2–3 times). A difference to the neotropical woody bamboo species is less pronounced, but also recognizable.
The predominantly tropical New World herbaceous bamboos of the tribe Olyreae, which were not sampled in this study, are mostly diploids with typically 2n = 20 or 22, although lower numbers of 2n = 14 or 18 have rarely been found (Kellogg 2015; CCDB). Genome sizes of 1,265 Mbp/2C (≈ 1.29 pg/2C) and 1,384 Mbp/2C (≈ 1.42 pg/2C) for Olyra latifolia and 1,370 Mb/2C (≈ 1.40 pg/2C) for Raddia guianensis have been reported (Šmarda et al. 2014; Guo et al. 2019). Their 1Cx values would be 0.65–0.70 pg and MCs uniformly 0.06 pg, so Olyreae seem to have the smallest monoploid genomes and the smallest chromosomes of the whole Bambusoideae. Genome sizes of 681 Mb (≈ 1.39 pg/2C) and 629 (≈ 1.28 pg/2C) Mb were estimated for O. latifolia and R. guianensis, respectively, by genome sequencing (also Guo et al. 2019).
Oryzoideae. The worldwide distributed rice subfamily of 117 species in 19 genera (Soreng et al. 2022) was sampled using three taxa.
The diploid Hygroryza aristata (2n = 24) had a 2C DNA value of 0.84 pg, which is comparable to an unpublished previous estimate of 1.00 pg/2C for this species using Feulgen microdensitometry (Leitch et al. 2019).
A studied accession (subspecies and cultivar not known) of rice, Oryza sativa (2n = 24), had 0.99 pg/2C. For O. sativa, 0.87–1.20 pg/2C have been estimated in previous studies using FCM or Feulgen microdensitometry (e.g., Martinez et al. 1993; Kurata & Fukui 2003; Loureiro et al. 2007; Yamamoto et al. 2018; Panibe et al. 2021; Dai et al. 2022). The reference genome of O. sativa subsp. japonica cv. Nipponbare, which is often used also as standard in FCM studies, has been reported to be 384.2–386.5 Mbp and 375.1 ± 20.9 Mbp in genome sequencing projects (Kawahara et al. 2013; Wang et al. 2018), corresponding to 0.79 pg/2C and 0.77 ± 0.04 pg/2C, respectively. Gap-free reference genomes of two subsp. indica varieties were 392 Mbp (≈ 80.0 pg/2C) and 396 Mbp (≈ 80.1 pg/2C), respectively (Song et al. 2021).
The tetraploid Leersia oryzoides (2n = 48) had 1.84 pg/2C), which is consistent with previously found values of 1.84 pg/2C and 1.83 pg/2C also using FCM + PI (Bai et al. 2012; Zonneveld 2019). All three Oryzoideae taxa examined in this study had 1Cx genome sizes of 0.42–0.50 pg and their MCs were uniformly around 0.04 pg.
The genus Zizania with four species, used as wild rice for grain harvest in North America and as a vegetable in China due to its Ustilago-infected, enlarged stems, is characterized by a WGD that occurred after the Zizania-Oryza phylogenetic split. For East Asian Z. latifolia, genome sizes of 586 Mb (≈ 1.20 pg/2C) and of 604.1 Mb (≈ 1.24 pg/2C) were found by FCM + PI and by genome sequencing, respectively (Guo et al. 2015). Recent sequencing studies found 547.38 Mb and 545.36 Mb (both ≈ 1.12 pg/2C) (Yan et al. 2022; Xie et al. 2023). The holoploid genome size of Z. latifolia is therefore about 1.1–1.2 times larger than that of O. sativa. Conflicting chromosome numbers have been reported for this species (see CCDB). However, assuming that 2n = 34 is correct, although 2n = 30 has also been reported (Probatova & Sokolovskaya 1982; Tzvelev & Probatova 2019), the MCs would be 0.03–0.04 pg. The genome size of another Zizania species, North American Z. palustris (2n = 30), was estimated to be 3.68–3.87 pg/2C by FCM + PI and 1,289 Mb (≈ 2.63 pg/2C) by genome sequencing (Haas et al. 2021). Its MC is therefore 0.12–0.13 pg, about 3 times larger than that of O. sativa. The Z. palustris genome has strongly restructured chromosomes compared to rice and is characterized by a massive amplification of repetitive elements, comprising about 74% of the total genome, compared to about 50% in rice and 53% in Z. latifolia (Haas et al. 2021; Yan et al. 2022).
Pooideae. This is the largest subfamily of grasses, comprising nearly 220 genera with 4,130 species (Soreng et al. 2022), slightly more than one-third of all grass species. The Pooideae are most abundant in the temperate to cool regions of both hemispheres. The subfamily is taxonomically further subdivided into 10 to 16 tribes, depending on the width of the respective delineations (GPWG 2001; Schneider et al. 2009, 2011, 2012; GPWG II 2012; Kellogg 2015; Tkach et al. 2020; Soreng et al. 2022). The holoploid genome sizes found for the subfamily Pooideae ranged from the low estimates of 0.56 pg/2C and 0.67 pg/2C in Brachypodium stacei (Catalán et al. 2012; Winterfeld et al. in prep.) to 45.26 pg/2C in Thinopyrum ponticum (Vogel et al. 1999). The variation was therefore greater than in any other grass subfamily (Tables 1, 2; Figs. 1, 2), as already noted (Bennetzen & Kellogg 1997; Kellogg & Bennetzen 2004; Caetano-Anollés 2005; Leitch et al. 2010; Kellogg 2015). The 1Cx values varied widely from 0.33 pg in hexaploid Austrostipa scabra to 9.45 pg in diploid Secale montanum (Eilam et al. 2007) and the MCs from 0.02 pg to 1.84 pg in the same species (Table 2; Figs. 1, 2). The genome size data of the Pooideae also showed strong differences between the phylogenetic lineages and tribes of this subfamily and will be discussed in more detail elsewhere (Tkach et al. in prep.; Winterfeld et al. in prep.).
PACMAD clade
This clade comprises six subfamilies with over 5,800 species (Soreng et al. 2022) and is characterized by the frequent occurrence of the highly efficient C4 photosynthetic pathway. In this study, the clade was represented by 21 example taxa in 18 genera.
Panicoideae. This subfamily is the second-largest subfamily of grasses, with over 3,300 species in 242 genera (Soreng et al. 2022), and distributed from tropical to warm temperate regions. The DNA 2C values of the sampled species ranged from 0.9 pg to 8.12 pg, and their ploidy levels ranged from 2x to 8x (CCDB). The chromosome base numbers in the Panicoideae vary depending on the tribes to which the genera studied belonged:
x = 9 in Cenchrus, Digitaria, Oplismenus, Panicum and Setaria (tribe Paniceae), which had 1Cx values of 0.45–0.83 pg and MCs of 0.05–0.09 pg;
x = 10 in the tribe Andropogoneae genera Coix, Saccharum and Tripidium, which were characterized by distinctively larger 1Cx values of 1.02–1.91 pg and MCs of 0.10–0.19 pg. This also applies to Miscanthus from this tribe, which has 2n = 38 or 57 based on x = 19. This chromosome number is derived from ancestors with x = 9 and x = 10 through allopolyploidy/amphidiploidy (Adati & Shiotani 1962; Chramiec-Głąbik et al. 2012); and
x = 12 in Chasmanthium (tribe Chasmanthieae) with intermediate values of 1Cx and MC, specifically 0.88 pg and 0.07 pg, respectively.
There were no distinct differences between the perennial (1Cx of 0.50–1.81 pg in Cenchrus, Chasmanthium, Miscanthus, Oplismenus, Saccharum, Tripidium) and annual taxa (1Cx of 0.45–1.91 pg in Coix, Digitaria, Panicum, Setaria), nor between the taxa with C4 (0.45–1.91 pg in Cenchrus, Coix, Digitaria, Miscanthus, Panicum, Saccharum, Tripidium) and C3 photosynthesis (0.54–1.11 pg in Chasmanthium, Oplismenus, Setaria). Previous genome size estimates using FCM + PI in Coix, Digitaria, Miscanthus (both cytotypes), Panicum and Setaria, including the same species as used in this study, agree with our data (Rayburn et al. 2008; Nishiwaki et al. 2011; Chramiec-Głąbik et al. 2012; Zhang et al. 2013; Chae al. 2014; Zonneveld 2019: Table 5 electron. supplement). This also largely applies to the genome sizes estimated by sequencing, i.e. 1,560 Mb (≈ 3.19 pg/2C) in Coix lacryma-jobi, and 395.1 Mb and 397 Mb (both ≈ 0.81 pg/2C) in Setaria viridis (Kang et al. 2020; Mamidi et al. 2020; Thielen et al. 2020).
Arundinoideae. The subfamily Arundinoideae is distributed worldwide and has a consistently C3 photosynthetic pathway. According to its current narrow taxonomic definition, it includes only about 14 genera and 36 species (Hardion et al. 2017; Soreng et al. 2022). The most likely chromosome base number is x = 6 (see Hardion et al. 2011, 2013 with a review of previous literature data). The 2C value of 5.25 pg found in Arundo donax, most likely the 18x cytotype with 2n = 108, agrees with the previously recorded amounts of 5.6 pg and 4.5–4.8 pg, respectively, also estimated by using FCM + PI (Zonneveld et al. 2005; Hardion et al. 2011). The studied A. donax accession had a 1Cx value of 0.29 pg and an MC of 0.05 pg. The two subspecies of Phragmites australis investigated, both likely 2n = 48, had proportionally lower 2C values of 2.09–2.18 pg compared to Arundo, but similar 1Cx values of 0.26–0.27 pg and MCs of 0.04–0.05 pg. This estimate agrees with the genome sequence length of 1,140 Mbp (≈ 2.33 pg/2C) for a presumably tetraploid accession of subsp. australis, invasive in North America (Oh et al. 2022).
The holoploid genome size of Molinia caerulea (3.52 pg/2C) was found to be intermediate between those of the Arundo and Phragmites accessions. Assuming a chromosome number of 2n = 36, which occurs most frequently in M. caerulea (CCDB), the 1Cx value of our accession would be 0.59 pg and the MC 0.10 pg, which is larger than in the other Arundinoideae taxa studied. The 2C value of 3.52 pg, which was recalculated from previously reported data for the tetraploid cytotype of M. caerulea (Dančák et al. 2012), agrees with our findings and suggests that our accession is also tetraploid, although the chromosome number has not been determined. The estimated 2C values for Arundinoideae species are largely consistent also with the results of previous estimates using FCM + PI. Phragmites australis was recorded as having 1.89 pg and 2.26 pg, while M. caerulea had 3.04–3.13 pg and 3.51–3.54 pg (Šmarda et al. 2019; Zonneveld 2019).
Chloridoideae. The subfamily Chloridoideae, distributed mainly in the tropics to subtropics, rarely in temperate zones, with an almost uniform C4 photosynthetic pathway, has about 1,600 species and 120 genera. It is represented in this study only by the C4 species Cleistogenes mucronata. Its 2C value was 1.55 pg, but its chromosome number is unknown. According to CCDB, other Cleistogenes species usually have 2n = 40, which is probably based on x = 10. So, if we assume a 4-fold ploidy for C. mucronata, the 1Cx value would be 0.39 pg and the MC would be 0.04 pg. However, it has often been argued that x = 10 is already a polyploid number in Chloridoideae, which was originally based on x = 5, but reports of 2n = 10 are still extremely rare, as noted by Roodt & Spies (2003), and would need to be confirmed.
Danthonioideae. The mainly African to Australasian subfamily Danthonioideae (C3 throughout) with about 19 genera and 290 species had 2C values ranging from 1.63 pg in Schismus arabicus, 4.66 pg in Danthonia decumbens, 4.89 pg in D. alpina to 8.30 pg in Cortaderia selloana. Considering x = 6 as the established chromosome base number in this subfamily, the 1Cx values are quite uniform, namely 0.82 pg in S. arabicus (2x), 0.78 pg and 0.82 pg in the two sampled Danthonia species (both 4x) and 0.69 pg in the highly polyploid C. selloana (12x). The MCs of 0.12–0.14 pg were also quite uniform. Comparable 2C values were obtained for D. alpina (3.90 pg) and D. decumbens (3.873 Mbp ≈ 3.96 pg and 4.19 pg) in previous studies also using FCM + PI (Šmarda et al. 2014, 2019; Zonneveld 2019).
Aristidoideae and Micrairoideae. These tropical to subtropical subfamilies, not sampled in this study, each include both C3 and C4 grasses and have 3 and 9 genera, respectively, with a nearly cosmopolitan distribution. Genome size data using FCM + PI have been previously obtained for some species. Aristida purpurea (diploid with 2n = 22) and A. tuberculosa had 2.60 pg/2C and 2.62 pg/2C (Bai et al. 2012; Šmarda et al. 2014), suggesting that the latter species is also diploid and implying 1Cx values of 1.30 pg and 1.31 pg, respectively, and an MC of 0.12 pg each. The Micrairoideae species Isachne globosa with 2n = 6x = 60 had 3.64 pg/2C, a 1Cx of 0.61 pg and an MC of 0.06 pg (Murray et al. 2005). It should be noted that Murray et al. (2005: p. 1300) explicitly corrected previous estimates (Murray et al. 2003), stating that they were about 30% too low.
Comparison of the grass subfamilies
Holoploid genomes. The 2C values do not show an overall trend of increase or decrease across the sampled grass subfamilies (Table 2, Figs. 1A, 2). Comparatively small 2C values occur in the Oryzoideae, Panicoideae and Pooideae. While the Oryzoideae have consistently small 2C values, the Panicoideae and Pooideae are highly variable and also have the largest values found in our sample, followed by the Bambusoideae. The phylogenetically early diverging grass subfamilies Anomochlooideae and Pharoideae have small but not strikingly small genome sizes. They are therefore not characterized by conspicuous 2C values compared to the ‘core grasses’, but correspond to the average of the grass subfamilies of the BOP and PACMAD clades.
Monoploid genomes. Regarding the monoploid chromosome sets, the lowest values of less than 0.4 pg are found in some species of Arundinoideae and Pooideae, the highest in Chloridoideae, Panicoideae and also Pooideae (Table 2, Figs. 1B, 2). With values of about 1.2–1.3 pg/1Cx, the Anomochlooideae and Pharoideae do not even belong to the small genome species. Comparatively large monoploid genomes occur in Bambusoideae, Panicoideae and the (more extensively sampled) Pooideae. The monoploid genomes of the Oryzoideae, including that of cultivated rice, are therefore among the smaller, but not the smallest, of the grasses.
Chromosome sizes. The mean chromosome DNA content (MC values) is below 0.1 pg in most subfamilies, i.e. the chromosomes are relatively small (Table 2, Figs. 1C, 2). In the early diverging lineages such as the subfamilies Anomochlooideae and Pharoideae, it is even at or above 0.1 pg. Similarly high values are also found in the Aristidoideae and Panicoideae from the PACMAD clade and in the Bambusoideae and especially the Pooideae from the BOP clade. The latter subfamily, however, has much larger chromosomes in many cases.
Origin of grasses
The Anomochlooideae and Pharoideae, which have also been shown to be characterized by the ρ-WGD typical of all other grasses (McKain et al. 2016; Ma et al. 2021; Seetharam et al. 2021), thus have neither particularly small nor particularly large genomes (1Cx) or chromosome sizes (MC) compared to the other Poaceae, but are somehow intermediate (Table 2; Figs. 1B,C, 2). Both are also characterized by a more or less unspectacular content of repetitive sequences in the genome of 51% and 78.9%, respectively (Ma et al. 2021; Seetharam et al .2021), which is in the order of magnitude of grasses with medium-sized genomes such as Sorghum bicolor (62.8%) of the subfamily Panicoideae, but higher than in small-genome grasses such as Oryza sativa (32.3%) or Brachypodium distachyon (28.0%) of the subfamilies Oryzoideae and Pooideae, respectively.
The sister families of the Poaceae are Ecdeiocoleaceae and Joinvilleaceae, all of which form the ‘graminid clade’ within Poales, but Ecdeiocoleaceae and Joinvilleaceae lack the ρ-WGD of the Poaceae (McKain et al. 2016). Their genome sizes of 1.98–2.72 pg/2C (Winterfeld et al. 2024) are comparable to, but not half as large as that of the Anomochlooideae and Pharoideae (2.48–2.63 pg/2C), as might be expected in principle from the WGD event. However, assuming that their chromosome numbers of 2n = 36, ca. 38 and ca. 48 reflect a 4-fold ploidy based on x = 9 and 12, their 1Cx values are 0.50–0.68 pg (Winterfeld et al. 2024), about half that of Anomochlooideae and Pharoideae (1.24–1.32 pg). In addition, their MCs are 0.05–0.08 pg (Winterfeld et al. 2024), about half that of Anomochlooideae and Pharoideae (0.10–0.12 pg). This all implies that the ρ event would indeed still be reflected in the genome and chromosome size data of the ‘basal’ grass subfamilies compared to the closest sister families of Poaceae, suggesting a pre-ρ karyotype of 9–12 chromosomes as an intermediate stage between 7 protochromosomes (Murat et al. 2014; Pont et al. 2019) and the formation of the AGK.
Origin of the ‘spikelet clade’ and the ‘core grasses’
The ‘spikelet clade’, i.e. all grass subfamilies except for the Anomochlooideae, and the ‘core grasses’, which include the BOP and PACMAD clades as sister lineages, do not appear to be characterized by a consistent clear difference in 1Cx genome size or chromosome sizes (MC) compared to the Anomochlooideae or the Anomochlooideae and Pharoideae (Table 2; Figs. 1B,C, 2).
It is therefore conceivable that genome sizes of about 1.3 pg/1Cx, such as those of the studied representatives of the Anomochlooideae (Streptochaeta angustifolia) and Pharoideae (Pharus latifolius), can be considered as ancestral for the grasses. This value is much lower than the previous suggestion of 3.0 pg to 5.2 pg DNA per 2C nucleus for the genome size of the ancestor of the grass family (Caetano-Anollés 2005). The very small genomes found in the Oryzoideae and parts of the Bambusoideae could therefore be the result of genome shrinkage and thus of secondary origin. Within the BOP clade, this probably applies to some Pooideae as well (Winterfeld et al. 2024). Furthermore, within the PACMAD clade, secondarily reduced 1Cx genome sizes (Table 2) are also plausible for the Arundinoideae, parts of the Chloridoideae and the Panicoideae compared to the Aristidoideae.
In the opposite case, small genomes such as that of rice (Oryzoideae) would have been ancestral within the grasses, with a corresponding (apomorphic) genome enlargement already occurring within the early diverging lineages Anomochlooideae and Pharoideae, as well as within the lineages of the BOP clade except for the Oryzoideae, and additionally within the PACMAD clade. This hypothesis cannot be excluded in principle, but seems less plausible. Genome size data on the second genus of the Anomochlooideae, the monospecific genus Anomochloa, the other two genera of the Pharoideae (Leptaspis and Scrotochloa), and especially the third subfamily of the early diverging lineages, namely the subfamily Puelioideae (Guaduella, Puelia), which has not yet been investigated in this respect and which together with the ‘core grasses’ (BOP and PACMAD clades) forms the monophyletic ‘bistigmatic clade’, would be needed for a final clarification of this question.
Ancestral grass karyotype (AGK)
The paleogenomic reconstruction of the AKG with 12 chromosomes, which arose after the ρ event, a genome duplication that resulted in a chromosome set of most likely 14 chromosomes that was restructured to 12, is supported by the well preserved synteny at the chromosomal level the in studied species of the different grass subfamilies. Comparatively few chromosomal rearrangements occurred between Pharus latifolius and rice, with some more changes with respect to Phyllostachys edulis (Bambusoideae), for example, suggesting that the AGK remained evolutionarily rather static for a long time after the origin of grasses (Ma et al. 2021). Differences were larger for Sorghum bicolor and Cenchrus americanus (both Panicoideae) and particularly dramatic for representative taxa from other lineages of the ‘core Poaceae’, i.e. Oropetium thomaeum (Chloridoideae), Brachypodium distachyon and Aegilops tauschii (both Pooideae). Most rearrangements therefore were lineage-specific and occurred within subfamilies, with the AGK remaining largely unchanged in the lineage leading to the ‘core Pooideae’, after the split of Pharoideae (Ma et al. 2021) and, by implication, Puelioideae.
The base chromosome numbers of grasses show a prevalence of x = 12 in most subfamilies (Table 2; Fig. 3). This is true for the Pharoideae and Puelioideae within the ‘early diverging lineages’, while x = 11 was recorded for two Streptochaeta species of the Anomochlooideae. Anomochloa marantoidea, on the other hand, has 2n = 36, suggesting x = 18, which is supported by the occurrence of 18 bivalents in the meiotic prophase of this species (Hunziker et al. 1989). However, since nearly half of the sexually reproducing polyploid plants show bivalent chromosome pairing and are functionally diploid (Li et al. 2021), Anomochloa could also represent a diploidized polyploid species, making x = 9 a possible monoploid number for Anomochloa. It could have arisen, like x = 11 in Streptochaeta, by descending dysploidy from x = 12.
The number of x = 12 prevails in the subfamilies of the BOP clade, only in Bambusoideae lower numbers of x = 7, 9, 10, 11 are well documented and were most likely also derived by reductional dysploidy. This is also true for the Pooideae, where x = 7, the base chromosome number often considered to be characteristic of this subfamily, actually prevails only in its phylogenetically late diverging lineages, summarized as the ‘core Pooideae’, while its early diverging lineages mostly have x = 12 (Fig. 3) (Winterfeld et al. 2024).
The same might apply to the PACMAD clade, where the higher monoploid numbers of x = 11, 12 are represented in the Aristidoideae, which represents the earliest diverging lineage of this clade according to the nuclear DNA phylogenetic analyses (Fig. 3B). Comparatively high numbers of x = 9, 10 are also found in the Panicoideae and Chloridoideae, from which much lower chromosome base numbers are derived, similar to Pooideae (Table 2; Fig. 3).
As mentioned above, rice has largely preserved the AGK (Wang et al. 2015), but so have bamboos with few changes in genome structure after their split from other clades. The studied genomes of bamboos, including diploid, tetraploid and hexaploid species, show genome-wide collinearity with the rice genome (Guo et al. 2019; Ma et al. 2021), while major genomic repatterning processes such as chromosome fusions and subsequent chromosome base number reduction are widespread in grasses but concentrated in the phylogenetically late diverging lineages. The increasing number of genomic analyses in different grass clades shows that the AGK with 12 chromosomes is unexpectedly well conserved in grasses and has remained evolutionarily almost unchanged for almost 100 million years in some grass lineages (Fig. 3). The majority of major genome rearrangements, as seen in both the BOP and the PACMAD clades, are lineage-specific and occurred after the diversification of their subfamilies had begun.