Ferns were the last major lineage of land plants without a reference genome. With the coming publication of the Ceratopteris genome assembly and recent publication of two water fern genomes, it is time to review the past and future applications of Ceratopteris and fern research. Here I delve into the utilization of a Ceratopteris genome assembly for studying the alternation of generations, reproductive biology, genome biology, and the C-Fern Curriculum.

Ferns are one of the most speciose lineages of land plants, and occupy an important phylogenetic position sister to seed plants. Despite this, ferns remain one of the last groups of land plants that do not have a fully developed model system. Here we review the biology and status of each emerging fern model. While reference genomes have been completed for Azolla filiculoides and Salvinia cucullata, they lack transformation capability. Meanwhile, other ferns including Marsilea vestita, Adiantum capillus-veneris, Pteris vittata, and Ceratopteris richardii have transformation methods, but lack genomic resources. Nevertheless, as genome sequencing becomes increasingly more affordable, we believe that M. vestita and C. richardii can become powerful fern models, which represent heterosporous and homosporous ferns, respectively.

The large genomes of ferns have long deterred genome sequencing efforts. To date, only two heterosporous ferns with remarkably small genomes, Azolla filiculoides and Salvinia cucullata, have been sequenced. However, as sequencing technologies continue to improve and become more affordable, generating high-quality, “normal-sized” fern genomes is within reach. Here we provide new genome size data and discuss candidates for whole genome sequencing. In particular, we identified 18 species representing major branches in the fern phylogeny that are worth pursuing. We also review the current sequencing technologies and offer our opinions on the best sequencing approach for these fern species.

Allopolyploidization is a common mode of speciation in ferns with many taxa having formed recurrently from distinct hybridization events between the same parent species. Each hybridization event marks the union of divergent parental gene copies, or homeologs, and the formation of an independently derived lineage. Little is known about the effects of recurrent origins on the genomic composition and phenotypic variation of allopolyploid fern taxa. To begin to address this knowledge gap, we investigated gene expression patterns in two naturally formed, independently derived lineages of the allotetraploid fern Polypodium hesperium relative to its diploid progenitor species, Polypodium amorphum and Polypodium glycyrrhiza. Using RNA-sequencing to survey total gene expression levels for 19194 genes and homeolog-specific expression for 1073 genes, we found that, in general, gene expression in both lineages of P. hesperium was biased toward P. amorphum—both by mirroring expression levels of P. amorphum and preferentially expressing homeologs derived from P. amorphum. However, we recovered substantial expression variation between the two lineages at the level of individual genes and among individual specimens. Our results align with similar transcriptome profile studies of angiosperms, suggesting that expression in many allopolyploid plants reflects the dominance of a specific parental subgenome, but that recurrent origins impart substantial expression, or phenotypic, variation to allopolyploid taxa.

Nuclear genome size is highly variable in vascular plants. The composition of long terminal repeat retrotransposons (LTR-RTs) is a chief mechanism of long term change in the amount of nuclear DNA. Compared to flowering plants, little is known about LTR-RT dynamics in ferns and lycophytes. Drawing upon the availability of recently sequenced fern and lycophyte genomes we investigated these dynamics and placed them in the context of vascular plants. We found that similar to seed plants, median LTR-RT insertion times were strongly correlated with haploid nuclear genome size. Fern and lycophyte species with small genomes such as those of the heterosporous Selaginella and members of the Salviniaceae had recent median LTR-RT insertion times, whereas species with large genomes such as homosporous ferns had old median LTR-RT insertion times intermediate between angiosperms and gymnosperms. This pattern holds despite methylation and life history differences in ferns and lycophytes compared to seed plants, and our results are consistent with other patterns of structural variation in fern and lycophyte genomes.

Reticulate evolution, in which phylogenetic relationships are not strictly bifurcating (tree-like), is a common feature of fern evolution. Ferns are prone to hybridization and whole genome duplication, two processes that can make untangling phylogenetic relationships among species challenging. Next-generation sequencing technologies have greatly increased the amount of data available for analyzing various aspects of evolutionary history, and here we test the ability of one next-generation sequencing approach to identify the progenitors of allopolyploids. We produced and analyzed double-digest restriction-site-associated DNA (ddRAD) sequences from six species of North American Dryopteris, including two allopolyploids and their respective diploid parents. The relationships of these species have been confidently established in previous studies, and our goal was to determine the extent to which RAD data are capable of identifying these known relationships. Analyses of the genetic structure in our samples reliably separated the diploids from one other, but in general each polyploid sample resembled one or the other of its progenitors, or had genetic variation unassignable to either parent. None of the polyploid samples had unambiguous genetic contributions from both known parents, as we had expected. These results may have been influenced by small overall sample size, different numbers of samples from the two diploid parents in each pair, and the large divergence times between the diploids. These are all potentially important issues to consider when designing similar studies, and our results therefore have useful implications for researchers interested in using a RAD approach to study polyploid complexes.