Bridging the micro‐ and macroevolutionary levels in phylogenomics: Hyb‐Seq solves relationships from populations to species and above

Abstract

Reconstructing phylogenetic relationships at the micro- and macroevoutionary levels within the same tree is problematic because of the need to use different data types and analytical frameworks. We test the power of target enrichment to provide phylogenetic resolution based on DNA sequences from above species to within populations, using a large herbarium sampling and Euphorbia balsamifera (Euphorbiaceae) as a case study. Target enrichment with custom probes was combined with genome skimming (Hyb-Seq) to sequence 431 low-copy nuclear genes and partial plastome DNA. We used supermatrix, multispecies-coalescent approaches, and Bayesian dating to estimate phylogenetic relationships and divergence times. Euphorbia balsamifera, with a disjunct Rand Flora-type distribution at opposite sides of Africa, comprises three well-supported subspecies: western Sahelian sepium is sister to eastern African-southern Arabian adenensis and Macaronesian-southwest Moroccan balsamifera. Lineage divergence times support Late Miocene to Pleistocene diversification and climate-driven vicariance to explain the Rand Flora pattern. We show that probes designed using genomic resources from taxa not directly related to the focal group are effective in providing phylogenetic resolution at deep and shallow evolutionary levels. Low capture efficiency in herbarium samples increased the proportion of missing data but did not bias estimation of phylogenetic relationships or branch lengths.

A customized nuclear target enrichment approach for developing a phylogenomic baseline for Dioscorea yams (Dioscoreaceae)

“…In our study, the average target length of the Dioscorea‐ specific kit was 1699 bp. This is comparable to 1362 bp for a Heuchera L. kit (Folk et al., ), 1667 bp for a Leguminosae kit (Vatanparast et al., ), and ~1645 bp for a Euphorbia L. kit (Villaverde et al., ). However, average target length was shorter (820 bp) for Buddleja L.‐specific locus sets compared to three different non‐taxon‐specific locus sets (1267 bp; Chau et al., ).…”

“…Using transcriptomes from Dioscoreales genera Dioscorea and Tacca , and LSCN gene references from more distantly related Oryza and Xerophyta , we produced a versatile bait kit that successfully captures most target loci from the focus group ( Dioscorea ) and one other genus in the same order ( Trichopus Gaertn.). Our average capture efficiency (i.e., the proportion of reads that HybPiper mapped to gene target references) of 31.6% (5.4–60.7%) is comparable to or lower than that from other custom baits (26% in Zingiberales, Sass et al., ; 48.6% in Euphorbia , Villaverde et al., ; 54.7% in Heuchera , Folk et al., ; 59% in Oxalis L., Schmickl et al., ). Our average recovery of the total reference gene set length was relatively high at 78.6% compared to other studies using custom baits (e.g., 33.4% in Folk et al., ; 59.4% in Sass et al., ; 73% in Villaverde et al., ), but varied across taxa (24.2–96.6%).…”

“…Our average capture efficiency (i.e., the proportion of reads that HybPiper mapped to gene target references) of 31.6% (5.4–60.7%) is comparable to or lower than that from other custom baits (26% in Zingiberales, Sass et al., ; 48.6% in Euphorbia , Villaverde et al., ; 54.7% in Heuchera , Folk et al., ; 59% in Oxalis L., Schmickl et al., ). Our average recovery of the total reference gene set length was relatively high at 78.6% compared to other studies using custom baits (e.g., 33.4% in Folk et al., ; 59.4% in Sass et al., ; 73% in Villaverde et al., ), but varied across taxa (24.2–96.6%). Baits are expected to work well up to a 30% threshold of sequence divergence to targets (Johnson et al., ).…”

“…Therefore, variability in enrichment success in our study was likely not fully due to insufficient affinity between samples and baits (including for Trichopus , as its uncorrected p ‐distance to sampled Dioscorea was 0–8%), but rather to differences in sample pooling strategy, starting DNA quality and quantity, library complexity, and genome size (Hart et al., ; Sass et al., ; Johnson et al., ). For example, we expect that lower gene recovery for Trichopus reported here (151 of 260 genes; Table ) could have been improved by not pooling this outgroup taxon with Dioscorea taxa; baits were designed using Dioscorea transcriptomes and so a high proportion of them were likely sequestered by more similar ingroup samples (Villaverde et al., ). Dioscorea minima had similar gene recovery to outgroup Trichopus with 157 of 260 genes; the former ingroup taxon had the third lowest amount of starting DNA (1.88 ng/μL), which may have impacted enrichment success, especially if the quality of the available DNA was poor and the resulting genomic library had low complexity.…”

“…This is particularly important for groups such as Dioscorea that are large and geographically widespread (Wilkin et al., ), with few species in wide cultivation. Genomic targets of hybridization‐based enrichment approaches can be recovered using “universal” baits optimized across broad taxon sets (e.g., the Angiosperm v.1 kit of Buddenhagen et al., ; the Angiosperms‐353 probe set of Johnson et al., ; the fern bait kit of Wolf et al., ) and using “taxon‐specific” baits designed for a focus group, ranging in taxonomic scale from populations (e.g., Euphorbia balsamifera Aiton; Villaverde et al., ) to a variety of higher‐level taxa (e.g., Inga Mill., Nicholls et al., ; Asclepiadoideae, Weitemier et al., ; Asteraceae, Mandel et al., ). The increasing availability of public transcriptome data and pipelines for LSCN gene discovery and bait development that require relatively few genomic resources can enable clade‐specific bait design, making phylogenomic research in groups of interest more attainable (see McKain et al., for details).…”