Dense infraspecific sampling reveals rapid and independent trajectories of plastome degradation in a heterotrophic orchid complex

Abstract

Heterotrophic plants provide excellent opportunities to study the effects of altered selective regimes on genome evolution. Plastid genome (plastome) studies in heterotrophic plants are often based on one or a few highly divergent species or sequences as representatives of an entire lineage, thus missing important evolutionary-transitory events. Here, we present the first infraspecific analysis of plastome evolution in any heterotrophic plant. By combining genome skimming and targeted sequence capture, we address hypotheses on the degree and rate of plastome degradation in a complex of leafless orchids (Corallorhiza striata) across its geographic range. Plastomes provide strong support for relationships and evidence of reciprocal monophyly between C. involuta and the endangered C. bentleyi. Plastome degradation is extensive, occurring rapidly over a few million years, with evidence of differing rates of genomic change among the two principal clades of the complex. Genome skimming and targeted sequence capture differ widely in coverage depth overall, with depth in targeted sequence capture datasets varying immensely across the plastome as a function of GC content. These findings will help to fill a knowledge gap in models of heterotrophic plastid genome evolution, and have implications for future studies in heterotrophs.

Keywords: Corallorhiza; hybrid capture; mycoheterotroph; parasite; plastid genome; pseudogenization.

Fig. 1

Sampling locations across North America for the Corallorhiza striata complex. Magenta, C. involuta; brown, C. bentleyi; blue, C. striata Sierra Nevadan accessions; green, C. striata var. vreelandii; red, C. striata var. striata.

Fig. 2

Relationships among members of the Corallorhiza striata species complex based on Maximum Likelihood analysis of whole aligned plastomes (unpartitioned GTR-GAMMA). Accession codes are followed by US, Canadian, or Mexican state/province. (a) Cladogram showing relationships and Maximum Likelihood Bootstrap Support (adjacent to branches; no number indicates 100%). (b) Phylogram showing branch lengths; scale bar units are substitutions per site. 350b OR (gs), genome skim dataset.

Fig. 3

Summary of putatively functional genes (black), pseudogenes (gray), and gene losses (white) among Corallorhiza and the Corallorhiza striata complex. Collection numbers are in the second column, including US, Canadian, or Mexican state/province. Green lines indicate breaks between ndh, photosynthesis-related, rpo, and atp genes; red line marks ‘housekeeping’ genes.

Fig. 4

(a) Divergence time estimates from BEAST2 under an uncorrelated lognormal clock model (internal transcribed Spacer (ITS), matK, psaB, and rbcL), calibrated with fossils from Conran et al. (2009) and Poinar & Rasmussen (2017). Scale axis indicates millions of years before present. Node bars indicate the 95% Highest Posterior Density estimates. (b) Substitution rates per branch (substitutions·per site·per year) from BEAST2. Branch widths are scaled by substitution rate, and mean estimates are given for each branch. Red branches indicate putatively non-photosynthetic, fully mycoheterotrophic lineages.

Fig. 5

Putative pseudogenes and gene losses in Corallorhiza, based on the chronogram in Fig. 4. Genes are categorized roughly by functional class. Closed circles, pseudogenes; open circles, gene losses. *, pseudogene in some but not all accessions. Red branches correspond to fully mycoheterotrophic lineages.

Fig. 6

(a) Comparison of read coverage depth distribution across the entire plastome as a result of genome skimming (above) and targeted sequence capture (below), for Corallorhiza striata var. striata accession 350b OR. LSC, large single copy; SSC, small single copy; IR, inverted repeat. (b) The relationship between mean GC content and coverage depth across the plastomes of the C. striata complex for target capture data, based on a sliding window analysis. Slope of the best-fit line indicates the Pearson Correlation Coefficient (R = 0.634, P < 0.0001).