Discussion
Although haploid induction was not successful, we obtained a more
juvenile, easily regenerable and transformable individual GM15, which
appears to be extremely similar to its parent tree LM50 based on ploidy,
genotype and genome size evidence, and thus was considered suitable for
sequencing. Here, we integrated advanced SMRT sequencing technology
(PacBio), Illumina correction and chromosome conformation capture (Hi-C)
to assemble a high quality haplotype-resolved genome. In comparion to
several published poplar genomes, including P. trichocarpa(Tuskan et al., 2006), P.
euphratica (T. Ma et al., 2013),P. pruinosa (Yang et al., 2017),
and P. alba var. pyramidalis(J. Ma et al., 2019), the assembly
quality of P. tomentosa was of higher or comparable quality .
Only for the P. alba genome was the contig N50 longer than forP. tomentosa (1.18 Mb vs. 0.96 MB); however, its contigs have not
been associated with specific chromosomes yet
(Y. J. Liu et al., 2019) (Table S7). The
whole genome size of P. tomentosa is 740.2 Mb, which is comprised
of the sum of subgenome A (P. alba var. pyramidalis ) and
subgenome D (P. adenopoda ). It obviously differs with those ofP. trichocarpa (422.9 Mb), P. euphratica (497.0 Mb), andP. pruinosa (479.3 Mb), P. alba var. pyramidalis(464.0 Mb) and P. alba (416.0 Mb), which respectively consist of
19 chromosomes as the allelic diversity in these diploids were subsumed
into a single haploid genome rather than into two diploid subgenomes
(Y. J. Liu et al., 2019;
J. Ma et al., 2019;
T. Ma et al., 2013;
Tuskan et al., 2006;
Yang et al., 2017). However, this case is
very similar to the genome of a hybrid poplar (84K) recently published,
which was subdivided into two subgenomes (P. alba and P.
tremula var. glandulosa ) with a total genome size of 747.5 Mb
(Qiu et al., 2019) (Table S7).
We presented evidence for divergence and duplication events inPopulus , as well as within the P. tomentosa lineage. Like
other many flowering plants (Otto, 2007),Salicaceae species underwent a common palaeohexaploidy event,
followed by a palaeotetraploidy event before the divergence ofSalix and Populus (Lin et
al., 2018; Y. J. Liu et al., 2019;
Tuskan et al., 2006). Subsequently,
poplar speciation occurred gradually. Section Populus andP. trichocarpa differentiated from each other approximately 13.44
Mya (Ks ≈ 0.035). The ancestors of P. tomentosa , P.
adenopoda and P. alba var. pyramidalis successively
diverged from section Populus approximately 9.3 Mya and 4.8 Mya.Populus tomentosa emerged from a hybridization event
approximately 3.9 Mya. This finding differs from previous proposals on
the origin of P. tomentosa (Z. Wang
et al., 2014). Unlike most other sequenced poplars
(T. Ma et al., 2013;
Tuskan et al., 2006;
Yang et al., 2017), the P.
tomentosa genome consists of subgenome A (P. alba var.pramidalis ) and subgenome D (P. adenopoda ) (Fig. 3 and
Table 1). Hi-C, as a chromosome conformation capture-based method, has
become a mainstream technique for the study of the 3D organization of
genomes (W. Ma et al., 2018). Based on
both Hi-C analysis (Figure 2) and phylogenetics analysis with P.
adenopoda and P. alba var. pramidalis , we were able to
partion the P. tomentosa genome into two subgenomes. Phylogenetic
analysis clearly revealed the relationships among three white poplars
(Fig. 4d, Fig. S6). Further, 19 chromosome-by-chromosome
phylogenetic trees all supported the same hybrid origin hypothesis (Fig.
S7). The phylogenetic analyses of the chloroplast genomes of P.
tomentosa showed that the female parental species was P.
adenopoda (Figure S8); thus, it appears that P. alba var.pramidalis was the paternal parent species. There also appears to
be variation within P. tomentosa with respect to its hybrid
origin. Based on a small number of marker genes,
Wang et al. (2019) suggested thatP. alba acted as the male parental species, but that the maternal
parent could be either P. adenopoda or P. davidiana (forP. tomentosa types mb1 and mb2, respectively)
(D. Wang et al., 2019). However, P.
tomentosa of Shandong provenance had not been collected in their
experimental materials, quite coincidentally, the elite P.
tomentosa clone LM50 in our study was from Shandong provenance, is
different with P. tomentosa types mb1 and mb2. Thus, P.
tomentosa may have a more complex evolutionary history than is fully
understood, including possibly multiple independent origins.
Our analysis of recombination events within genes showed that theP. tomentosa subgenomes have largely remained independent,
despite sharing the same nucleus for approximately 3.93 million years.
Comparision of 5,345 single copy orthologs from P. tomentosa ,P. alba var. pyramidalis and P. adenopoda showed
recombination was only observed in 0.87% of the genes studied (Fig. S5,
Table S13). To assess if this low rate of recombination would be
expected given the time since the species’ origin, we used recombination
data from a recent study in the closely related European aspen (P.
tremula ) (to generate an expected rate of recombination, assuming this
non-hybrid species shows normal recombination rates for Populus ).
They estimated the recombination rate to be 15.6-16.1 cM/Mbp/generation
(Apuli et al., 2020). In general, P.
tomentosa has a long life cycle, the seedlings begin flowering after at
least 7–8 years and thereafter annual flowering occurs during the
reproductive phase (Zhu, 1992). Assuming
a generation time of about 20 years, 31 recombinations per 1 kb gene
would be expected—several orders of magnitude below our observation.
This suggests that the two subgenomes of P. tomentosa have been
maintained largely intact over many thousands of genertions, despite
ample opportunity for recombination events to have occurred within the
studied genes. The subgenome integrity of P. tomentosa , where
there appears to be a low rate of normal meiotic products, is congruent
with observations of very low fertility in the species. In a study of
elite tree resourse of P. tomentosa , most of them showed weak
fertility, a low rate of seed setting, germination and seedling
surviving (Bai, 2015). Such
characteristics and recent genetic analysis of P. tomentosa(D. Wang et al., 2019) suggest thatP. tomentosa acts like the F1 generation of a
wide cross, with quite limited but not zero fertility.
SVs are increasingly being recognized as major factors underying
phenotypic variation in eukaryotic organisms
(Gabur, Chawla, Snowdon, & Parkin,
2019). In plants, SVs have been proved to be closely related to many
phenotypic variations such as of plant height
(Zhou et al., 2015), and biotic stress
resistance (Cook et al., 2012). In our
study, we detected 15,480 SVs across the genome of GM15 of which 12,885
were INDELS and accounted for the majority of SVs (83%). GO analysis
indicated INDELS are highly represented within genes with roles in
plant-pathogen interaction and carbohydrate metabolism. They may
therefore contribute to characteristics such as disease resistance and
fast growth, for which P. tomentosa is well known. A few INDELS
are also enriched in genes associated with meiotic DNA double-strand
break processing and repair, as well as inactivation of chromation and
histone methylation in telomeres. Perhaps such SVs contribute to
retaining independence of the two subgenomes and maintaining karyotype
stability in P. tomentosa —thus play a role in maintaining its
putative “fixed heterosis,” as discussed further below. We also found
299 CNVs, and GO analysis suggested an association with plant hormone
signal transduction, plant-pathogen interaction, and sugar metablism. In
sum, the many identified SVs in P. tomentosa provide logical
focal points for study of their biological roles and phenotypic effects
in relation to heterosis, evolution, breeding and biotechnology.
The mechanisms for the low recombination among sub-genomes are unknown.P. tomentosa is well known for having low sexual fertility
(K. Ma et al., 2013), likely a reflection
of meiotic difficulties that give rise to abnormal gametes. As suggested
for Cucurbita subgenomes (Sun et
al., 2017), the low recombination rate in P. tomentosa genome
could be due to the rapid divergence between the two parental species in
their repetitive DNA composition, which may have inhibited meiotic
pairing of homologous chromosomes and subsequent exchanges; as shown
above, the transposon compositions of the two genomes differ
significantly. In addition, TE activity can cause CNVs, INSs, TRANSs and
DELs due to their capacity to mobilize and recombine gene sequences
within and between chromosomes (Morgante,
De Paoli, & Radovic, 2007), both in the wild and in breeding processes
(Lisch, 2013). These SVs may further
inhibit normal meiosis. Karyotype stability and rare recombination among
sub-genomes has been observed in paleo-allotetraploid Cucurbitagenomes (Sun et al., 2017), and in newly
synthesized allotetraploid wheat genome
(H. Zhang et al., 2013). However, their
functional connection to recombination rate suppression is unclear. The
maintenance of subgenomes that we found in P. tomentosa may be
advantageous in providing a degree of “fixed heterosis”. This may help
to explain P. tomentosa ’s high productivity and wide distribution
in spite of its low sexual fertility.