Title: Are the genomes of royal ferns really frozen in time? Evidence for coinciding genome stability and limited evolvability in the royal ferns
Abstract: Progress in understanding genome evolution has recovered not only evidence for differences in genome dynamics between the major lineages of land plants but also putative links between genome dynamics, evolvability and the assembly of species diversity (Leitch & Leitch, 2012, 2013). A new, inspiring fossil discovery now provides us with a unique insight into the evolution of genome size in a species belonging to a remarkable group of ancient plants, the royal ferns (Osmundaceae). Based on the analysis of an Early Jurassic fossil with hypothesized close affinities to the extant Osmundastrum cinnamomeum, Bomfleur et al. (2014) argue that the exceptionally well preserved chromosomes and cell nuclei provide evidence for ‘genomic stasis’ over c. 180 million yr. Their argument coincides with previously reported morphological stasis over > 200 million yr in the same group of fossils (Phipps et al., 1998; Serbet & Rothwell, 1999), and Manton's (1950) suggestion that the geological longevity and morphological stability of Osmundaceae was, in part, due to the stability in chromosome number and structure she observed. Thus the fossil suggests a correlation between genotypic and phenotypic stasis over time in these ancient ferns. The hypothesis that royal ferns exhibit ‘genomic stasis’, might appear to be consistent with the conservation in chromosome number and near complete absence of polyploidy (Zhang et al., 2008; see Tsutsumi et al., 2011). Yet ‘genomic stasis’ implies that the underlying genomic processes have had no impact on genomic content over time, something that cannot be inferred solely from analysing fossils. Certainly, genomic turnover can take place without impacting genome size. For example, Medicago truncatula and Lotus japonicus are closely related angiosperm species with the same genome size (Cheng & Grant, 1973; Arumuganathan & Earle, 1991) but show many differences in genic content; several thousand gene families are present in one species but not the other (Varshney et al., 2012). Dynamic genome evolution is also reflected in changes in transposable element (TE) content and diversity; Aegilops cylindrica and A. geniculata are close grass relatives with similar genome sizes, but show contrasting patterns of TE amplification and deletion over time (Senerchia et al., 2013). These examples illustrate how relative stasis in genome size and chromosome number may belie genomic turnover. Clearly, fossil evidence of DNA content alone is not sufficient to conclude that royal fern genomes have become effectively ‘fossilized’. Here we explore new and existing genomic data to determine whether there is indeed evidence that royal fern genomes are frozen in time. Bomfleur et al. (2014) noted the similarity in nucleus size and hence, by proxy, genome size between the fossil and extant species Osmundastrum cinnamomeum. To examine if there is evidence of genome size stasis across the whole family we analysed genome size variation and reconstructed the ancestral genome size (Table 1; Fig. 1). By combining previously reported values with newly obtained data (Table 1), we now have genome sizes for half the extant species diversity in Osmundaceae, including at least one representative from each of the four recognized genera and three subgenera of Osmunda. Overall, genome sizes varied 1.34-fold with the smallest genome in Osmunda claytoniana (1C = 13.46 pg) and the largest in Todea barbara (1C = 19.46 pg) (Fig. 1). Using Bayesian approaches we inferred the ancestral genome size for the crown group of Osmundaceae to be 1C = 15.9 pg, with evidence of both increases (e.g. branch leading to Todea and Leptopteris) and decreases (e.g. branch leading to Osmunda subg. Osmunda) in genome size over time (Fig. 1). It is noted that if Osmundastrum is placed as sister to Osmunda, as suggested by morphological similarities, then the ancestral genome size of the crown group is estimated to be 1C = 17.1 pg. Nevertheless, even with this higher value, increases and decreases in genome size are still evident (data not shown). Overall, we have uncovered evidence for both evolutionary changes in genome size and limited genome size variation among extant species of royal ferns, albeit low compared with some other fern lineages (e.g. 1C-values in Salviniales range 5.29-fold based on just three genome size estimates, and 6.41-fold in Ophioglossales based on just five estimates) and ferns as a whole where genome sizes range 94-fold (based on data for 128 species, see Leitch & Leitch, 2013). An analysis of available chromosome data also reveals a similar story, with evidence of some, albeit limited, variation in the organization of DNA within the chromosomes, despite all but one of the 14 chromosome counts for Osmundaceae reporting 2n = 44. For example, recent karyological studies in Osmunda have provided evidence of chromosome rearrangements (Zhang et al., 2008), interspecific hybridization and allopolyploidy (Tsutsumi et al., 2011). Such rather limited genome size and chromosome diversity coincides with a notably low substitution rate of c. 1.1 × 10−4 substitutions per site per million years (8.0 × 10−5 to 1.4 × 10−4) for the rbcL region of the plastid genome, if the chronogram is calibrated using existing fossils. This contrasts with an average substitution rate of 5.0 × 10−4 substitutions per site per million years estimated for the majority of ferns and land plants in general (Villarreal & Renner, 2014; H. Schneider unpublished). Even older ages and thus lower substitution rates are estimated if the oldest fossils assigned to Osmundastrum and Todea (Wang et al., 2013; Bomfleur et al., 2014) are considered to belong to the crown group of these two extant genera rather than assigning the oldest royal fern fossil as an age constraint of the crown group of extant Osmundaceae (data not shown). Given the often arborescent sporophytes and the extended growing season of the gametophytes (Klekowski, 1973), it seems likely that royal ferns have long generation times. If so then these may contribute to the low substitution rates observed (as also suggested for tree ferns which have long generation times, Korall et al., 2010; Zhong et al., 2014), and perhaps also to the relatively limited genome size diversity (Beaulieu et al., 2010). The earlier data are also complemented by the finding that net diversification rates of royal ferns are 2.44–4.43 times lower than the average rate reported for ferns as a whole. Indeed Osmundaceae have the lowest reported diversification rate of any leptosporangiate fern lineage (Schuettpelz & Pryer, 2009). However, this low rate is not associated with a complete absence of recent speciation as some clades are seen to have diverged since the late Oligocene (e.g. Leptopteris, Osmunda subg. Plenasium in Fig. 1). It is notable that the co-occurrence of relatively large genomes in royal ferns with a low diversification rate is consistent with some theoretical predictions (Kraaijeveld, 2010). Overall, while these observations do suggest that the genomes of royal ferns are less dynamic than other fern lineages, they nevertheless challenge the hypothesis that their genomes are frozen in time. Instead of stasis, we find evidence for some variation suggesting rather slow evolution compared with other fern lineages. This is also reflected in various morphological features that are actually less static if all extant species of royal ferns are considered. For example, the lineage shows considerable variation in leaf morphology as illustrated by an absence of differentiation between sporangium-bearing and photosynthetically active parts in the leaves of Leptopteris and Todea, separation of these two functions within the leaf in Osmunda and between leaves in Osmundastrum. Once again, the explanation may lie in the long generation times of royal ferns, which have contributed to the limited speed of morphological evolution. Despite our disagreement with the overall interpretation of Bomfleur et al. (2014), we agree with the unique opportunity provided by these ferns for studies aiming to understand how changes in genome dynamics have contributed to the evolution of plants and, in turn, the importance of genomic changes for the evolution of lineages. By analysing genomic structure in species of Osmundaceae, this will allow the exploration of genomic changes in a group of plants with obvious limitations in their evolutionary capacity. To interpret these findings, it is crucial to understand the unique phylogenetic position of royal ferns as they comprise the sister lineage of the most species-rich lineage of ferns (Pryer et al., 2004). The royal ferns contain a unique combination of character states including those found only in these ferns (e.g. their stele) as well as characters that are intermediate between the plesiomorphic eusporangiate state and the apomorphic leptosporangiate state (Schneider et al., 2009). It therefore seems likely that the genomes of royal ferns will display some characteristics shared with all ferns as well as those that are unique to these ferns. The recent proposal (Sessa et al., 2014) to sequence two fern genomes is certainly welcomed, but it is noted that the genera selected (Ceratopteris and Azolla) occupy derived phylogenetic positions, well within the species-rich leptosporangiate clade. The addition of a member of Osmundaceae is strongly recommended, providing both an outgroup for comparative analyses with the Ceratopteris and Azolla data as well as the potential to gain fundamental insights into the genomics underpinning a lineage with limited evolutionary capacity.