Strong phylogenetic signal despite high phylogenomic complexity in an Andean plant radiation (Freziera,Pentaphylacaceae)

Abstract

The Andes mountains in South America are a biodiversity hotspot within a hotspot, the New World Tropics, for seed plants. Much of this diversity is concentrated at middle-elevations in cloud forests, yet the evolutionary patterns underlying this extraordinary diversity remain poorly understood. This is partially due to a paucity of resolved phylogenies for cloud forest plant lineages: the young age of the Andes and generally high diversification rates among Andean systems precludes robust phylogenetic inference, and remote populations, few genomic resources, and generally understudied organisms make acquiring high-quality data difficult. We present the first phylogeny of Freziera (Pentaphylacaceae), an Andean-centered, cloud forest radiation with potential to provide insight into some of the abiotic and extrinsic factors that promote the highest diversity observed on the globe. Our dataset, representing data for 50 of the ca. 75 spp. obtained almost entirely from herbarium specimens via hybrid-enriched target sequence capture with the universal bait set Angiosperms353, included a proportion of poorly assembled loci likely representing multi-copy genes, but with insufficient data to be flagged by paralog filters: cryptic paralogs. These cryptic paralogs likely result from limitations in data collection that are common in herbariomics combined with a history of genome duplication and are likely common in other plant phylogenomic datasets. Standard empirical metrics for identifying poor-quality genes, which typically focus on filtering for genes with high phylogenetic informativeness, failed to identify problematic loci in our dataset where strong but inaccurate signal was a greater problem. Filtering by bipartition support was the most successful method for selecting genes and resulted in a species tree with lower discordance, higher support, and a more accurate topology relative to a consensus tree. Using known paralogs, we investigate the utility of multi-copy genes in phylogenetic inference and find a role for paralogs in resolving deep nodes and major clades, though at the expense of gene tree concordance and support. With the first phylogeny, we infer the biogeographic history of Freziera and identify the northern Andes as a source region. We also identify distinct modes of diversification in the northern and central Andes, highlighting the importance of fine-scale biogeographic study in Andean cloud forest systems.

Utilizing paralogues for phylogenetic reconstruction has the potential to increase species tree support and reduce gene tree discordance in target enrichment data

“…As gene duplication and loss is one of the sources of gene tree discordance (Degnan & Rosenberg, 2009), discriminating between orthologues and paralogues is likely to reduce gene tree discordance, which we indeed observed in the case of our two Malinae data sets (Figure 3). ParalogWizard did, however, not increase the proportion of concordant nodes for the Oritrophium and Amomum data sets, which is in line with Gardner et al (2021) and Frost and Lagomarsino (2022), who found that gene tree discordance increased when paralogous loci were added.…”

“…We developed an approach to minimize the removal of entire loci and to, instead, use biologically real loci in target enrichment data sets. This is in line with Frost and Lagomarsino (2022), who promote the use of well curated data for phylogenomics. We first detect multicopy loci, then evaluate the pairwise sequence similarity of their exons, followed by a check for multimodal distribution, which is used to discriminate between a main and secondary copy (Figure 1).…”

“…Including paralogues for phylogenetic reconstruction has the potential to increase the support of phylogenetic trees (Hellmuth et al, 2015; Smith & Hahn, 2021; but see Frost & Lagomarsino, 2022). This was confirmed in the case of the species trees for our two Malinae data sets (Malinae‐optimized Angiosperms353 and Malinae481), but not for the Oritrophium and Amomum data sets (Figure 2).…”

“…Contrary to Yan et al (2022), our findings suggest that discriminating between orthologues and paralogues is not obsolete. Thus, using biologically real loci in target enrichment data sets is promising for phylogenomics (Frost & Lagomarsino, 2022).…”

The rapid radiation of Bomarea (Alstroemeriaceae: Liliales), driven by the rise of the Andes

“…Universal probe sets for target capture such as the GoFlag angiosperm 408 probes used in this study (Endara and Burleigh, 2022), Angiosperm 353 (Johnson et al, 2019), and ultraconserved elements (UCEs; McCormack et al, 2012) are designed to target sequences from an evolutionarily broad group of organisms and thus may be also more likely to recover contaminant sequences than clade-specific probe sets. Additionally, while the GoFlag angiosperm 408 probes (and most other universal probe sets) target single-copy nuclear genes, these regions were identified as single-copy at a broad evolutionary scale (all angiosperms), which cannot ensure that recent gene duplications have not occurred within our target group (Frost and Lagomarsino, 2021). While one option may be to design more clade-specific probe sets, this is not feasible for many researchers—including for us in this study—and may negate the potential data-sharing benefits of global probe sets (depending on the way clade-specific probe sets are generated).…”