Department of Health, Life and Environmental Sciences, University of L'Aquila, Coppito, L'Aquila, Italy.
CIBIO-InBIO, Centro de Investigação em Biodiversidade e Recursos Genéticos, Universidade do Porto, Vairão, Portugal.
PLoS One. 2020 May 21;15(5):e0233573. doi: 10.1371/journal.pone.0233573. eCollection 2020.
The accuracy of the DNA barcoding tool depends on the existence of a comprehensive archived library of sequences reliably determined at species level by expert taxonomists. However, misidentifications are not infrequent, especially following large-scale DNA barcoding campaigns on diverse and taxonomically complex groups. In this study we used the species-rich flea beetle genus Longitarsus, that requires a high level of expertise for morphological species identification, as a case study to assess the accuracy of the DNA barcoding tool following several optimization procedures. We built a cox1 reference database of 1502 sequences representing 78 Longitarsus species, among which 117 sequences (32 species) were newly generated using a non-invasive DNA extraction method that allows keeping reference voucher specimens. Within this dataset we identified 69 taxonomic inconsistencies using barcoding gap analysis and tree topology methods. Threshold optimisation and a posteriori taxonomic revision based on newly generated reference sequences and metadata allowed resolving 44 sequences with ambiguous and incorrect identification and provided a significant improvement of the DNA barcoding accuracy and identification efficacy. Unresolved taxonomic uncertainties, due to overlapping intra- and inter-specific levels of divergences, mainly regards the Longitarsus pratensis species complex and polyphyletic groups L. melanocephalus, L. nigrofasciatus and L. erro. Such type of errors indicates either poorly established taxonomy or any biological processes that make mtDNA groups poorly predictive of species boundaries (e.g. recent speciation or interspecific hybridisation), thus providing directions for further integrative taxonomic and evolutionary studies. Overall, this study underlines the importance of reference vouchers and high-quality metadata associated to sequences in reference databases and corroborates, once again, the key role of taxonomists in any step of the DNA barcoding pipeline in order to generate and maintain a correct and functional reference library.
DNA 条形码工具的准确性取决于是否存在一个由专家分类学家可靠地确定为物种级别的综合存档序列库。然而，错误识别并不罕见，尤其是在对多样化和分类复杂的群体进行大规模 DNA 条形码普查之后。在这项研究中，我们使用了物种丰富的叶甲科 flea beetle 属 Longitarsus，该属需要高水平的形态学物种鉴定专业知识，作为一个案例研究，以评估经过多次优化程序后的 DNA 条形码工具的准确性。我们构建了一个 cox1 参考数据库，其中包含 1502 个序列，代表 78 种 Longitarsus 物种，其中 117 个序列（32 个物种）是使用一种非侵入性的 DNA 提取方法生成的，这种方法允许保留参考标本。在这个数据集内，我们使用条形码间隙分析和树拓扑方法发现了 69 个分类不一致。阈值优化和基于新生成的参考序列和元数据的分类修订解决了 44 个具有模糊和错误识别的序列，并显著提高了 DNA 条形码的准确性和鉴定效果。由于种内和种间的分化水平重叠，分类上的不确定性主要涉及 Longitarsus pratensis 物种复合体和多系群 L. melanocephalus、L. nigrofasciatus 和 L. erro。这种类型的错误表明分类学尚未建立，或者任何生物学过程使得 mtDNA 群体不能很好地预测物种界限（例如，最近的物种形成或种间杂交），从而为进一步的综合分类学和进化研究提供了方向。总体而言，这项研究强调了参考标本和与序列相关的高质量元数据在参考数据库中的重要性，并再次证实了分类学家在 DNA 条形码工作流程的任何步骤中的关键作用，以便生成和维护正确和功能齐全的参考库。
