Target sequence capture in orchids: Developing a kit to sequence hundreds of single-copy loci.

PREMISE: Understanding relationships among orchid species and populations is of critical importance for orchid conservation. Target sequence capture has become a standard method for extracting hundreds of orthologous loci for phylogenomics. Up-front cost and time associated with design of bait sets makes this method prohibitively expensive for many researchers. Therefore, we designed a target capture kit to reliably sequence hundreds of orthologous loci across orchid lineages.
METHODS: We designed an Orchidaceae target capture bait set for 963 single-copy genes identified in published orchid genome sequences. The bait set was tested on 28 orchid species, with representatives of the subfamilies Cypripedioideae, Orchidoideae, and Epidendroideae.
RESULTS: Between 1,518,041 and 87,946,590 paired-end 150-base reads were generated for target-enriched genomic libraries. We assembled an average of 812 genes per library for Epidendroideae species and a mean of 501 genes for species in the subfamilies Orchidoideae and Cypripedioideae. Furthermore, libraries had on average 107 of the 254 genes that are included in the Angiosperms353 bait set, allowing for direct comparison of studies using either bait set. DISCUSSION: The Orchidaceae963 kit will enable greater accessibility and utility of next-generation sequencing for orchid systematics, population genetics, and identification in the illegal orchid trade.

Targeted enrichment methods allow for the extraction of hundreds of low‐copy loci, which is useful for phylogenomic inference under the multispecies coalescent model (Lemmon et al., 2012; Heyduk et al., 2016). However, the bait sets required for target enrichment can be costly to produce and require genomic resources including reference genomes or transcriptome sequences (Hale et al., 2020). Universal bait sets, such as Angiosperms353 (Johnson et al., 2019), can ameliorate these challenges by enabling the generation of target gene sequences across a broad phylogenetic scale. Although this is an attractive option for some phylogenetic investigations, a “one‐size‐fits‐all” approach may not be sufficient for disentangling phylogenetic relationships at lower taxonomic levels in many groups. Rapid species radiations, low sequence divergence, and gene and genome duplications (all rampant within the angiosperm clade) are likely to reduce the efficacy of using a universal bait set for species‐level phylogeny reconstruction. Rather, it is likely that lineage‐specific bait sets will be necessary to obtain sufficient variation to address questions at lower phylogenetic scales.

In this study, we address this issue by generating a single‐copy gene bait set for the hyperdiverse Orchidaceae, the second‐largest family of flowering plants (behind Asteraceae) with over 27,000 recognized species (The Plant List, 2013). Numerous studies over the years have attempted to reconstruct relationships among species and lineages within the Orchidaceae, many with conflicting results (e.g., Freudenstein et al., 2004; Neubig et al., 2009; Chase et al., 2015). Phylogenetic support at lower taxonomic levels has been poor, primarily because of limited variation in genes commonly used for phylogenetic inference, e.g., ITS, rbcL, matK, chloroplast intergenic spacers (Freudenstein et al., 2004; Neubig et al., 2009). Hybrid enrichment has been used in orchids applying 517 nuclear genes (Granados Mendoza et al., 2020) and 446 nuclear, chloroplast, and mitochondrial genes (Bogarín et al., 2018), which have shown promise in resolving relationships in orchids of the genera Epidendrum L. and Lepanthes Sw., respectively.

In this study, we have generated an Orchidaceae‐specific bait set including exons of 963 putatively single‐copy nuclear genes (Unruh et al., 2018; Johnson et al., 2019), which we hypothesize will allow for more robust estimates of relationships with greater support. This bait set includes 754 genes from a previously generated data set (Unruh et al., 2018). Additionally, we identified 254 of the 353 Angiosperms353 genes (45 genes overlapping with the set of 754 genes) to be single copy in the Phalaenopsis equestris (Schauer) Rchb. f. genome (Cai et al., 2015; Johnson et al., 2019). This set, with overlap removed, was incorporated into the final Orchidaceae963 set. This allows for continuity among studies using either the Orchidaceae963 or Angiosperms353 bait sets. We tested the utility of the bait set in reconstructing relationships at the tribal level using a diverse sampling of orchid species from the three most speciose orchid subfamilies, Orchidoideae, Cyprepedioideae, and Epidendroideae. Furthermore, we used the genus Acineta Lindl. as a test case for resolving species‐level relationships. Species‐level taxonomy in Acineta is currently based on floral morphology and has been controversial, with different authorities recognizing different numbers of species (Gerlach, 1999; Christenson, 2006), and no hypothesis of species relationships using molecular phylogenetics exists to date. This represents a general challenge in the Orchidaceae. Of the vast number of species resulting from a rapid radiation, many orchid clades have gone understudied, and the genes typically used for phylogenetic reconstruction in other plant groups do not contain sufficient variation to resolve species‐level relationships in orchids (Neubig et al., 2009). Lineage‐specific bait sets such as the Orchidaceae963 have the potential to advance our ability to resolve relationships at both deep and shallow scales and will facilitate species delimitation in the speciose orchid family.

METHODS

Probe design

Starting from an existing set of 775 genes that were found to be single copy across orchids (Unruh et al., 2018), we retrieved exon sequences for each gene from the Phalaenopsis equestris whole genome assembly (Cai et al., 2015). We removed any sequences that were less than 50 bp in length. For sequences ranging from 50–99 bp, we added flanking intron sequences on both sides to reach a minimum length of 100 bp needed for bait construction, called “padded” exons. After this filtering step, we were left with 5940 exon sequences for 754 genes. To identify additional single‐copy loci, we performed a local BLAST search (Altschup et al., 1990) to find hits of the Angiosperms353 targets to the P. equestris genome (Cai et al., 2015). For each Angiosperms353 target, we filtered out any loci mapping to multiple contigs and loci with hit alignments with high‐scoring segment pairs spanning a distance of more than 250 kbp on a single contig to account for potential duplicates. With the remaining hits, we used the P. equestris genomic coordinates, filtered and padded exons as previously described, and used the BEDTools getfasta function (Quinlan and Hall, 2010) to extract the sequences. This resulted in 763 exons for 254 genes. To compare the overlap between the set of 754 genes and the 254 single‐copy Angiosperms353 targets identified in P. equestris, we ran a local BLAST search (Altschup et al., 1990). This resulted in a 45‐gene overlap between both sets. We combined both exon sets and dropped the overlap from the set of 254 genes from the Angiosperms353 set. Finally, this resulted in a total of 6005 padded exons across 963 genes. This exon set was sent to Arbor Biosciences for synthesis of 100‐bp tiled bait sets. The Orchidaceae963 bait set is available from Arbor Biosciences (Ann Arbor, Michigan, USA), and the exons used to generate the bait set are provided as Appendix S1.

Library preparation and sequencing

Utilizing an existing Orchidaceae taxonomy (Pridgeon et al., 1999, 2001, 2003, 2005, 2009, 2014), we sampled 28 species, including three species within the genus Acineta, to test the efficacy of this bait set at resolving species‐level relationships (Appendix 1). Fresh leaf material was collected from plants in the Atlanta Botanical Garden Living Collection and dried on silica desiccant. Dried leaf tissue was ground in a FastPrep‐24 (MP Biomedicals, Santa Ana, California, USA), and DNA was isolated using the sorbitol cetyltrimethylammonium bromide (CTAB) method (Doyle, 1987; Storchova et al., 2000). DNA fragment size distribution was visually assessed on an agarose gel, and DNA concentration was quantified with the Qubit 3.0 fluorometer broad‐range assay (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Total genomic DNA was sheared to an average fragment size of 550 bp using a Covaris sonicator (Covaris, Woburn, Massachusetts, USA) at either the Georgia Genomics and Bioinformatics Core at the University of Georgia (Athens, Georgia, USA; https://dna.uga.edu) or the Emory Integrated Genomics Core (Atlanta, Georgia, USA). Dual‐indexed libraries were generated using the KAPA LTP Library Preparation kit for Illumina (Roche, Basel, Switzerland) using dual‐index primers from the EHS DNA Laboratory at the University of Georgia (Athens, Georgia, USA; http://www.baddna.uga.edu). Genomic libraries were amplified using 12 cycles and were quantified using the Qubit fluorometer high‐sensitivity assay (Thermo Fisher Scientific) as well as using real‐time quantitative PCR (qPCR). Average fragment size of the genomic DNA libraries was assessed using the Agilent Bioanalyzer (Agilent Technologies, Santa Clara, California, USA). Prior to hybridization, eight dual‐indexed genomic DNA libraries were pooled in equal concentrations to include 100 ng of each library for a final 800‐ng total pool. The pooled library was then used in the hybridization reaction with the custom orchid‐specific bait set (Arbor Biosciences myBaits Target Capture Kit). Hybridization reactions were performed at 65°C for 24 h. Enriched libraries were amplified for 10 cycles, and PCR products were cleaned using Sera‐Mag SpeedBeads (GE Healthcare Life Sciences, Pittsburgh, Pennsylvania, USA). Final hybridized libraries were quality checked on the Agilent Bioanalyzer, and concentration was measured with qPCR and the Qubit fluorometer. Multiple enriched pooled libraries were sequenced on an Illumina NextSeq (300 cycles) PE150 High Output flow cell at the Georgia Genomics and Bioinformatics Core.

Data analysis

Gene assembly used Perl scripts in the Reads2Trees pipeline (Heyduk, 2014), except SPAdes was used in place of Trinity for contig assembly. Trimmomatic was used to trim adapter sequences, remove reads with a quality score below 10, and remove reads less than 40 bp long (Bolger et al., 2014). Only reads where both members of the pair passed this quality filtering were used in downstream analysis. Genes were assembled de novo using SPAdes version 3.13.0 (Nurk et al., 2013), and contigs with greater than 98% similarity were merged using CAP3 (Huang and Madan, 1999). BLASTN (Altschup et al., 1990) was then used to query contigs against the reference exon set, and when more than one contig had a match with an E‐value of 1E‐20 to an exon, those contigs were removed from further analysis as potential duplicate genes. Contigs with no evidence of duplication were then merged into genes according to the exon order and orientation in the P. equestris reference exon set (Cai et al., 2015; Unruh et al., 2018). To assess exon coverage, filtered reads were mapped to the assembled exons using Bowtie2 (Langmead and Salzberg, 2012), and coverage was calculated using the “genomeCoverageBed” program in BEDTools (Quinlan and Hall, 2010). Genes were aligned using PRANK (Löytynoja and Goldman, 2005, 2008; Löytynoja, 2014). Multiple sequence alignments were cleaned with Gblocks using the default settings (Castresana, 2000; Talavera and Castresana, 2007). Maximum likelihood phylogenetic analysis was performed using RAxML on each Gblocks‐filtered multiple sequence alignment. Clade support was assessed from 100 bootstrap replicates using the rapid bootstrapping algorithm in RAxML (Stamatakis, 2014). A species tree was estimated in ASTRAL‐III (Zhang et al., 2018) to account for possible incomplete lineage sorting.

Comparison of the Angiosperms353 and orchid‐specific bait sets

To compare the utility of the general angiosperm‐wide (Angiosperms353) bait set with the Orchidaceae lineage‐specific bait set, we generated phylogenetic trees using the 254 genes from the Angiosperms353 bait set that were included in the Orchidaceae963 bait set to compare to the species tree generated from the total set of 963 genes. Species trees were generated, and multi‐locus bootstrapping and polytomy tests were run on the species tree as implemented in ASTRAL‐III, which tests whether a node differs significantly from the null hypothesis of a polytomy (Sayyari and Mirarab, 2018; Zhang et al., 2018).

Assessing the Orchidaceae963 bait set on species‐level relationships

To test the utility of the Orchidaceae bait set on resolving species relationships, we sequenced genes from three species in the genus Acineta, including two individuals of the species A. superba (Kunth) Rchb. f. The related genus Stanhopea Frost ex Hook. was used as an outgroup; the leaf sample was taken from a division of the plant used for the type specimen of S. nicaraguensis G. Gerlach (Table 1). Separate alignments of these five taxa were made in PRANK, and alignments were cleaned using Gblocks (see above). Cleaned alignments of these five taxa were concatenated using concatenation scripts (https://github.com/ODiogoSilva/ElConcatenero) to assess the number of variable and parsimony informative sites across all genes in PAUP (Swofford, 1993). Individual gene trees were generated in RAxML, and clade support was assessed from 100 bootstrap replicates using the rapid bootstrapping algorithm in RAxML (Stamatakis, 2014). A species tree was reconstructed in ASTRAL‐III (Zhang et al., 2018).

TABLE 1

Taxonomy, read counts, and target gene and exon recovery information for species included in this study.

Subfamily	Tribe	Subtribe	Species	Authority	Sequenced reads	Target exons recovered	Genes recovered
Cypripedioideae			Paphiopedilum exul	(Ridl.) Rolfe	2,753,445	653	430
Cypripedioideae			Phragmipedium longifolium	(Warsz. & Rchb. f.) Rolfe	87,946,590	890	572
Orchidoideae	Cranchideae	Goddyerinae	Anoectochilus chapaensis	Gagnep.	72,034,725	988	559
Orchidoideae	Cranchideae	Spiranthinae	Spiranthes sp.		1,792,508	682	429
Orchidoideae	Orchideae	Orchidinae	Platanthera blephariglottis	(Willd.) Lindl.	2,967,114	889	514
Epidendroideae	Arethuseae	Arethusinae	Calopogon barbatus	(Walter) Ames	33,176,407	1985	849
Epidendroideae	Arethuseae	Coelogyninae	Coelogyne trinervis	Lindl.	4,809,455	2339	875
Epidendroideae	Collabieae		Gastrorchis humblotii	(Rchb. f.) Schltr.	26,652,164	2305	884
Epidendroideae	Cymbideae	Catasetinae	Catasetum integerrimum	Hook.	5,454,392	1948	812
Epidendroideae	Cymbideae	Coeliopsidinae	Lycomormium squalidum	(Poepp. & Endl.) Rchb. f.	20,258,435	1767	781
Epidendroideae	Cymbideae	Cymbidiinae	Cymbidium ensifolium	(L.) Sw.	2,447,334	2145	836
Epidendroideae	Cymbideae	Eulophiinae	Eulophia streptopetala	Lindl.	2,368,908	2033	825
Epidendroideae	Cymbideae	Maxillariinae	Anguloa uniflora	Ruiz & Pav.	3,225,178	2081	847
Epidendroideae	Cymbideae	Maxillariinae	Maxillaria tenuifolia	Lindl.	4,572,350	2057	827
Epidendroideae	Cymbideae	Oncidiinae	Tolumnia bahamensis	(Nash ex Britton & Millsp.) Braem	6,407,944	1724	776
Epidendroideae	Cymbideae	Stanhopeinae	Acineta mireyae	G. Gerlach & M. H. Weber	34,362,615	1961	841
Epidendroideae	Cymbideae	Stanhopeinae	Acineta sella‐turcica	Rchb. f.	1,832,152	2095	834
Epidendroideae	Cymbideae	Stanhopeinae	Acineta superba	(Kunth) Rchb. f.	22,716,176	1980	837
Epidendroideae	Cymbideae	Stanhopeinae	Acineta superba	(Kunth) Rchb. f.	46,435,181	2005	836
Epidendroideae	Cymbideae	Stanhopeinae	Stanhopea nicaraguensis	G. Gerlach	1,518,041	2081	841
Epidendroideae	Cymbideae	Zygopetalinae	Pescatoria lehmannii	Rchb. f.	30,265,873	1923	807
Epidendroideae	Epidendreae	Coeliinae	Coelia bella	(Lem.) Rchb. f.	53,909,915	2302	895
Epidendroideae	Epidendreae	Laeliinae	Encyclia polybulbon	(Sw.) Dressler	33,891,917	2148	855
Epidendroideae	Malaxideae		Liparis rhombea	J. J. Sm.	19,827,922	1975	824
Epidendroideae	Podochileae	Eriinae	Eria ornata	Lindl.	13,544,574	2062	842
Epidendroideae	Sobralieae		Sobralia citrea	Dressler	18,923,228	2017	847
Epidendroideae	Vandeae	Aeridinae	Phalaenopsis aphrodite	Rchb. f.	25,942,519	3005	926
Epidendroideae	Vandeae	Angraecinae	Angraecum breve	Schltr.	8,345,929	2691	892
Epidendroideae	Vandeae	Polystachyinae	Polystachya concreta	(Jacq.) Garay & H. R. Sweet	26,626,748	2101	828

RESULTS

Between 1,518,041 and 87,946,590 paired‐end 150‐base reads were generated for target‐enriched genomic libraries (Table 1). The number of genes recovered from each species ranged from 430 in Paphiopedilum exul (Ridl.) Rolfe to 926 in Phalaenopsis aphrodite Rchb. f., and recovery was much higher in Epidendroideae taxa than other orchid subfamilies tested (Fig. 1). Most genes had relatively high occupancy; 826 genes were recovered for half or more taxa. However, 137 genes had fairly low capture success, capturing fewer than half of the tested taxa (Fig. 1). Additionally, of the 254 genes included from the Angiosperms353 bait set, we were able to recover between 18 genes in Spiranthes Rich. sp. and 186 genes in Phalaenopsis aphrodite (Fig. 2). The mean total sequence assembly length (including targeted exons and adjacent non‐coding sequence) ranged from 583 in Paphiopedilum exul to 1914 in Coelia bella (Lem.) Rchb. f., with a mean exon sequence length of 311 bp and a mean intron length of 907 bp (Fig. 3). Unlike capture success, variation in sequence assembly lengths did not correspond to taxon relationships (Fig. 3).

FIGURE 1

Presence and absence of target genes in 28 orchid species. For each row, a colored vertical line indicates presence of the gene, and a white vertical line represents absence. Genes were ordered from left to right by the number of taxa with a gene present. The capture rate is high among most genes, and a low number on the right indicates a very low capture rate. Rows are grouped and colored by Orchidaceae subfamily, with Orchidoideae species in yellow, Cypripedioideae in green, and Epidendroideae in blue.

FIGURE 2

The number of genes captured for each taxon in the test set, showing both the total number captured (blue) and the number that are part of the Angiosperms353 universal probe set (yellow). Species in the Epidendroideae tended to have a greater number of genes captured than species in the subfamilies Cypripedioideae and Orchidoideae. Species are grouped by Orchidoideae subfamily.

FIGURE 3

Mean sequence length of exons and introns for each taxon. Mean exon length for a sample is shown in blue, and mean intron length is indicated in yellow. Species are grouped by Orchidoideae subfamily.

Species trees generated using the Angiosperms353 genes and the Orchidaceae963 gene set exhibited consistent subfamily and tribal relationships with one exception (Fig. 4). The Angiosperms353 tree exhibited support for a clade including the tribes Cymbideae and Vandeae that was sister to Epidendreae. However, for the node uniting the Cymbideae and Vandeae tribes, we were not able to reject the null hypothesis that the node is a polytomy. In contrast, we were able to reject a polytomy for the three tribes in the species tree generated using all of the Orchidaceae963 targets, and the species tree analysis yielded strong support for a clade uniting the tribes Cymbideae and Epidendreae, which was placed sister to the tribe Vandeae.

FIGURE 4

Summary trees illustrating tribal relationships. Orchidaceae species trees were generated using 254 genes from the Angiosperms353 bait set and the 963 genes from the Orchidaceae‐wide bait set. The support value denoting bootstrap support is shown in black, and the P value for the polytomy test is shown in red. Branches without support values have 100% bootstrap support and P = 0 for the polytomy test. The pie charts show quartet frequencies for the branch leading to the Cymbideae (C), Epidendreae (E), and Vandeae (V) tribes in each tree. Green, red, and blue indicate the main, first alternative, and second alternative topologies, respectively (36%: [(C,E),V]; 35%: [(C,V),E]; 29%: [(E,V),C]).

For the test on the three species in the genus Acineta and related Stanhopea nicaraguensis, we recovered more than 800 genes for four of the five taxa, including the outgroup S. nicaraguensis. The Gblocks‐filtered concatenated alignment for these five taxa consisted of 2,090,511 sites, 69,675 variable sites, and 5934 parsimony informative sites. The ASTRAL‐III species tree for Acineta + Stanhopea nicaraguensis showed 100% multilocus bootstrap support for all nodes, and polytomy tests rejected polytomies in the tree (Fig. 5).

FIGURE 5

Relationships among four Acineta individuals including the outgroup Stanhopea nicaraguensis. Numbers above nodes indicate bootstrap support values from multilocus bootstrapping in ASTRAL‐III.

DISCUSSION

In this study, we designed and tested an Orchidaceae‐specific bait set that has the potential to provide more resolution in the case of rapid radiations compared to the “one‐size‐fits‐all” approach taken by the angiosperm‐wide bait set (Johnson et al., 2019). The lower number of genes recovered from the Cypripedioideae and Orchidoideae clades and the 137 genes with low capture rates across the Orchidaceae can potentially be addressed with future reference‐genome sequences for these subfamilies and improvements to target capture bioinformatic pipelines. For example, capture rates can be improved by using subfamily‐specific exon files for read mapping or clustering methods to select multiple sequences per loci representative of sequence diversity across the family (Johnson et al., 2019; McLay et al., 2021).

Previous analyses of tribal relationships within the Epidendroideae subfamily have found a polytomy among the Cymbideae, Vandeae, and Epidendreae tribes (Freudenstein et al., 2004; Neubig et al., 2009). Other studies that are able to resolve the polytomy show three alternative results: (1) the Vandeae and Epidendreae tribes are closely related and together are sister to the Cymbideae (Cameron et al., 1999; Chase et al., 2015), (2) the Cymbideae and Epidendreae tribes are most closely related and together are sister to the Vandeae (Freudenstein and Chase, 2015), and (3) the Cymbideae and Vandeae tribes are most closely related and together are sister to the Epidendreae (Givnish et al., 2015). Interestingly, the relationship shown using the 963 genes from the Orchidaceae‐specific bait set show the same relationships between the Vandeae, Epidendreae, and Cymbideae tribes as an analysis of seven nuclear, chloroplast, and mitochondrial genes for 312 taxa (Freudenstein and Chase, 2015). Despite nearly equal quartet frequencies for the three alternative tribal relationships (36%: [(C,E),V]; 35%: [(C,V),E]; 29%: [(E,V),C]), we were able to reject the null hypothesis of an ancestral polytomy giving rise to Vandeae, Epidendreae, and Cymbideae lineages using the ASTRAL‐III polytomy test. The relationships shown in the Angiosperms353 tree (Cymbideae + Vandeae clade) are consistent with the topology recovered using whole chloroplast genome sequences (Givnish et al., 2015). The plastid genome is typically inherited uniparentally without recombination and, therefore, not subject to incomplete lineage sorting. Our species tree analyses of both the Angiosperms353 and Orchidaceae963 targeted gene sets indicate extensive incomplete lineage sorting of nuclear genes and rapid radiation of the Vandeae, Epidendreae, and Cymbideae tribal lineages. The sequences recovered with the Orchidaceae963 bait set provided sufficient statistical power for resolving relationships among the three tribes. The analysis presented here using the 963 genes from the Orchidaceae‐specific bait set allows us to reconstruct a topology with greater support and resolves a polytomy in the case of a recent and rapid species radiation (Sayyari and Mirarab, 2018). In the future, examining which genes contribute to this resolution could be informative (Pease et al., 2016).

Additionally, we have demonstrated that the Orchidaceae963 bait set shows promise for characterizing species‐level relationships, especially in epidendroid orchids. The genus Acineta consists of 20 species and is distributed in the Neotropics (Christenson, 2006). These species are pollinated by orchid bees in the tribe Euglossini (Dressler, 1968a, b, 1982). Species boundaries have been difficult to characterize (Christenson, 2006), and no phylogenetic reconstruction of species‐level relationships has been established. Furthermore, studies of related taxa using a smaller number of genes have shown low resolution of species relationships (Whitten et al., 2000, 2014). Therefore, genomic resources such as the Orchidaceae963 bait set will provide the tools necessary to understand the complex relationships among orchid species and their pollinators in the face of recent and rapid diversification.

Lineage‐specific bait sets, such as the Orchidaceae963 set described here, provide the genomic resources to sequence hundreds of single‐copy nuclear genes for phylogenomics and population genomics while maintaining an overlapping set of genes with the broader‐scale Angiosperms353 probe set to allow for comparison across studies. We have demonstrated that the Orchidaceae963 bait set provides sufficient variation to resolve recent and rapid radiations in the backbone of the orchid tree of life and at shallow phylogenetic scales with an example from the genus Acineta. There was a degree of phylogenetic bias in gene recovery, where we recovered hundreds more genes from species in the Epidendroideae compared to species in the Orchidoideae and Cypripedioideae; this is likely because the baits were designed using sequences from the Phalaenopsis equestris genome, which is a member of the Epidendroideae. However, the number of genes in these two subfamilies was still very high, and utilizing a target file with diverse representative sequences during the assembly process may allow for greater gene recovery (McLay et al., 2021). Target sequence capture approaches show promise for population and conservation genomics (Villaverde et al., 2018; Hale et al., 2020), and lineage‐specific bait sets such as the Orchidaceae963 will provide sufficient variation to estimate population genetic diversity and structure. Finally, the potential to derive DNA barcoding markers that exhibit greater diversity than typical rbcL or matK markers can be useful in combating the illegal orchid trade, which has enormous negative impacts on orchid populations worldwide (de Boer et al., 2017; Hinsley et al., 2017).

AUTHOR CONTRIBUTIONS

S.K.T. and J.H.L.M. designed the probe set. L.A.E. and E.E.D.C. designed the sampling strategy for testing the bait set. L.A.E. generated genomic DNA libraries. L.A.E. and S.K.T. analyzed data and wrote the manuscript. All authors approved the final version.

APPENDIX S1. The exons used to generate the Orchidaceae963 bait set.

Click here for additional data file.

Species	Authority	Provenance	ABG Living Collection no.	ABG Biorepository no.
Paphiopedilum exul	(Ridl.) Rolfe	United States Fish and Wildlife Service	20150810	164
Phragmipedium longifolium	(Warsz. & Rchb. f.) Rolfe	Banos, Ecuador; Coll. Calaway H. Dodson	20111585	163
Anoectochilus chapaensis	Gagnep.	Coll. Douglas Goldman	20140840	185
Spiranthes sp.		Calhoun Co., Georgia; Coll. Ron Determann	20120193	161
Platanthera blephariglottis	(Willd.) Lindl.	Coll. Jeff Talbert; Deer Lake State Park, Florida	L‐20170054	158
Calopogon barbatus	(Walter) Ames	Coll. Jeff Talbert; Deer Lake State Park, Florida	L‐20160109	159
Coelogyne trinervis	Lindl.	Marie Selby Botanical Gardens	20063209	165
Gastrorchis humblotii	(Rchb. f.) Schltr.	Louisiana Orchid Connection	20163714	166
Catasetum integerrimum	Hook.	Rudolph Jenny, F1218	20031131	167
Lycomormium squalidum	(Poepp. & Endl.) Rchb. f.	Carchi, Ecuador; Coll. Mark Whitten	20111551	168
Cymbidium ensifolium	(L.) Sw.	United States Fish and Wildlife; 12580	20172133	169
Eulophia streptopetala	Lindl.	Orchid Seedbank Project	20172106	170
Anguloa uniflora	Ruiz & Pav.	Cordillera Azul, Peru; Coll. Mark Whitten, F‐1708	20111461	171
Maxillaria tenuifolia	Lindl.	Cheekwood Botanical Garden	20041060	186
Tolumnia bahamensis	(Nash ex Britton & Millsp.) Braem	Coll. Rob Rossmanith; Jonathan Dickinson State Park, Florida	L‐20160078	157
Acineta mireyae	G. Gerlach & M. H. Weber	Bill Goldner	20021863	96
Acineta sella‐turcica	Rchb. f.	Tropical Orchid Farm, TOF2291	20071149	77
Acineta superba	(Kunth) Rchb. f.	El Oro, Ecuador; Coll. Mark Whitten	19940903	98
Acineta superba	(Kunth) Rchb. f.	Paccha, Ecuador; Coll. F. L. Stevenson	19960940	99
Stanhopea nicaraguensis	G. Gerlach	Jinotega, Ecuador; Coll. John Atwood	19901127	106
Pescatoria lehmannii	Rchb. f.	JEM Orchids	19940657	172
Coelia bella	(Lem.) Rchb. f.	Monrovia de Chiriqui, Costa Rica; Robert Dressler	20111479	175
Encyclia polybulbon	(Sw.) Dressler	Cheekwood Botanical Garden; 1982‐0089	20041097	177
Liparis rhombea	J. J. Sm.	Ecuagenera Nursery	20150624	179
Eria ornata	Lindl.	Clyde Bramblett	20050036	180
Sobralia citrea	Dressler	Cerro Jefe, Panama; Coll. Mark Whitten	20111610	181
Phalaenopsis aphrodite	Rchb. f.	Louisiana Orchid Connection	20161292	182
Angraecum breve	Schltr.	Carter & Holmes Orchids	20031832	183
Polystachya concreta	(Jacq.) Garay & H. R. Sweet	Puerto Rico; Coll. Mark Laroque	20100385	184