With over 25,000 species, the drivers of diversity in the Orchidaceae remain to be fully understood. Here, we outline a multitiered sequence capture strategy aimed at capturing hundreds of loci to enable phylogenetic resolution from subtribe to subspecific levels in orchids of the tribe Diurideae. For the probe design, we mined subsets of 18 transcriptomes, to give five target sequence sets aimed at the tribe (Sets 1 & 2), subtribe (Set 3), and within subtribe levels (Sets 4 & 5). Analysis included alternative de novo and reference‐guided assembly, before target sequence extraction, annotation and alignment, and application of a homology‐aware k‐mer block phylogenomic approach, prior to maximum likelihood and coalescence‐based phylogenetic inference. Our evaluation considered 87 taxa in two test data sets: 67 samples spanning the tribe, and 72 samples involving 24 closely related Caladenia species. The tiered design achieved high target loci recovery (>89%), with the median number of recovered loci in Sets 1–5 as follows: 212, 219, 816, 1024, and 1009, respectively. Interestingly, as a first test of the homologous k‐mer approach for targeted sequence capture data, our study revealed its potential for enabling robust phylogenetic species tree inferences. Specifically, we found matching, and in one case improved phylogenetic resolution within species complexes, compared to conventional phylogenetic analysis involving target gene extraction. Our findings indicate that a customized multitiered sequence capture strategy, in combination with promising yet underutilized phylogenomic approaches, will be effective for groups where interspecific divergence is recent, but information on deeper phylogenetic relationships is also required.

中文翻译：

---

跨越广泛进化尺度的多层序列捕获策略：在兰花的系统发育和系统地理学研究中的应用

兰科植物拥有超过 25,000 个物种，仍有待充分了解其多样性的驱动因素。在这里，我们概述了一种多层次的序列捕获策略，旨在捕获数百个基因座，以实现 Diurideae 部落兰花从亚部落到亚特异性水平的系统发育分辨率。对于探针设计，我们挖掘了 18 个转录组的子集，以提供针对部落（第 1 组和第 2 组）、亚部落（第 3 组）和亚部落水平（第 4 组和第 5 组）内的五个目标序列集。分析包括替代的从头和参考引导组装，在目标序列提取之前，注释和比对，以及同源感知k - mer 的应用块系统发育方法，在最大似然和基于合并的系统发育推断之前。我们的评估考虑了两个测试数据集中的 87 个分类群：跨越部落的 67 个样本，以及涉及 24 个密切相关的Caladenia物种的72 个样本。分层设计实现了高目标位点回收率 (>89%)，第 1-5 组中回收位点的中位数如下：分别为 212、219、816、1024 和 1009。有趣的是，作为同源k的第一个测试‐mer 方法用于靶向序列捕获数据，我们的研究揭示了其实现强大的系统发育物种树推断的潜力。具体而言，与涉及目标基因提取的传统系统发育分析相比，我们发现匹配，并且在一个案例中提高了物种复合体内的系统发育分辨率。我们的研究结果表明，定制的多层序列捕获策略与有前途但未充分利用的系统发育方法相结合，对于最近出现种间差异的群体是有效的，但还需要有关更深层次系统发育关系的信息。