Molecular phylogenetic analyses reveal both underestimation and overestimation of species diversity in northern rain frogs (Craugastor)

Among direct-developing rain frogs of the genus _Craugastor_ is a clade of 19 described species (_bocourti_ series) that occur in Mexico and northern Central America. Many of these 19 species have been described based on subtle morphological differences and have never been examined using molecular data. Here, we used a multilocus dataset (one mitochondrial (mtDNA) and four nuclear (nDNA) gene fragments; totaling 3,048 concatenated base pairs) to investigate species limits and phylogenetic relationships among 60 northern rain frogs referable to 12 species with a focus on species from Guatemala. We inferred phylogenies using maximum likelihood and Bayesian analyses on separate mtDNA and nDNA datasets. Concatenated and coalescent species-tree analyses support the monophyly of multiple species, with interspecific relationships mostly unresolved. These mtDNA and nDNA trees were often incongruent with morphology-based taxonomy. For example, two genetically shallow clades contained individuals referable to at least five described species, whereas a single described species contained deep divergences indicative of multiple cryptic species. These findings indicate that morphology-based taxonomy has both overestimated and underestimated actual species diversity (depending on the species), an interpretation supported by two molecular species-delimitation procedures. Based on these findings we synonymize _C. glaucus_ (Lynch, 1967) and _C. stuarti_ (Lynch, 1967) with _C. xucanebi_ (Stuart, 1941). We also synonymize _C. nefrens_ (Smith, 2005) and _C. cyanochthebius_ McCranie & Smith, 2006 with _C. campbelli_ (Smith, 2005). The molecular data also support multiple undescribed species, notably within _C. decoratus_ (Taylor, 1942). Overall, we show how morphology-based species delimitation can both underestimate and overestimate species richness in morphologically conservative groups.

在直颌雨蛙属（Craugastor）的直接发育型雨蛙类群中，存在一个包含19个已描述物种的演化支（bocourti类群），该类群分布于墨西哥及中美洲北部。这19个物种中，多数基于细微的形态差异被确立为独立物种，且从未有研究借助分子数据对其开展过检验。本研究采用多基因数据集（包含1个线粒体DNA（mtDNA）基因片段与4个核DNA（nDNA）基因片段，拼接后总长度达3048个碱基对），对隶属于12个物种的60只中美洲北部雨蛙个体展开物种界限与系统发育关系研究，重点聚焦危地马拉分布的类群。我们分别针对线粒体DNA与核DNA数据集，采用最大似然法与贝叶斯推断法构建系统发育树。拼接序列分析与溯祖物种树分析均支持多个物种的单系性，但物种间的系统发育关系大多未能得到有效解析。线粒体DNA与核DNA构建的系统发育树，时常与基于形态学的分类学结论存在冲突。例如，两个遗传分化程度较低的演化支中，包含了至少5个已描述物种的个体；而单个已描述物种内部则存在深度分化，暗示存在多个隐存物种。上述结果表明，基于形态学的物种分类既存在高估也存在低估实际物种多样性的情况（依物种而异），该结论得到两种分子物种界定方法的支持。基于本研究结果，我们将C. glaucus（Lynch, 1967）与C. stuarti（Lynch, 1967）归为C. xucanebi（Stuart, 1941）的同物异名；同时将C. nefrens（Smith, 2005）与C. cyanochthebius（McCranie & Smith, 2006）归为C. campbelli（Smith, 2005）的同物异名。分子数据同样支持多个未描述物种的存在，尤以C. decoratus（Taylor, 1942类群内的情况尤为显著。综上，本研究阐明了在形态保守类群中，基于形态学的物种界定方法如何同时存在低估与高估物种丰富度的问题。