The new species was collected during fieldwork carried out in the Serranía de los Yariguíes, vereda La Belleza, municipality of El Carmen de Chucurí, Santander Department, Colombia (06°34'N, 73°34'W, 731–789 m a.s.l.; Fig. 1), from 17 February to 1 May 2018, during the dry season. Specimens were found in two separate humid spots near the Río Cascajales, which drains Tropical moist broadleaf forests, within the ecoregion of the Magdalena valley montane forests, in the foothills of the Cordillera Oriental, Colombia (Dinerstein et al. 1995; Olson and Dinerstein 2002).

Previous fieldwork conducted between 1998–1999 by John Lynch in collaboration with the first author, successfully allowed the detection of microhabitats and several specimens of Microcaecilia nicefori (Lynch 1999); subsequently, between 2000 to date, fieldwork with caecilians such as Oscaecilia polyzona (Lynch and Acosta 2004), Caecilia sp., C. thompsoni, and C. isthmica (unpublished data) allowed successful detection of microhabitats and multiple specimens.

The collecting technique, which was used to obtain specimens of the new species, consists of first asking local people about the locations where they have spotted caecilians using the common names of “blind snakes”, or “captain worms” (“lombrices capitanas”), or “motolas” (this common name is specific for the Department of Santander). Subsequently, the reported sites are visited and inspected to select sites under the shade of vegetation, and where the soil is not compact and very humid (usually associated with water springs that form a mosaic of marshy and dry areas). Collecting efforts are focused in the selected damp microhabitats, digging with a hoe to a depth of approximately 20 cm (approximate sampling effort of 2-person-hour to collect five specimens). Coordinates and elevations were obtained with a Garmin GPSMAP 64SC (map datum WGS 84). Collected specimens were euthanized using 20% benzocaine (Chen and Combs 1999), fixed in 10% formalin, and preserved in 70% ethanol. Tissue samples from two individuals were obtained immediately after euthanasia and preserved in 96% ethanol. Specimens were deposited at the Biological Collections of the Instituto de Investigación de Recursos Biológicos Alexander von Humboldt, Villa de Leyva, Boyacá, Colombia (IAvH-Am and IAvH-CT) and the Amphibian Collection of the Universidad Industrial de Santander, Bucaramanga, Santander, Colombia (UIS-MHN-A).

To test the generic assignment of the new species and to explore the relationships of other endemic caecilians from Colombia, available mitochondrial DNA sequences of the genes 16S and CO1 from members of all Neotropical caecilian families (Caeciliidae, Typhlonectidae, Siphonopidae, Dermophiidae, and Rhinatrematidae) were analyzed (Table 1). The analysis included a fragment of COI (ca. 651 bp) and a fragment of 16S (ca. 510 bp). Sequences for most terminals were obtained from GenBank (Table 1). We added new sequences for eight Colombian terminals representing the new species, Caecilia thompsoni, C. isthmica, Typhlonectes natans, Epicrionops aff. parkeri, and Microcaecilia nicefori (Appendix 1). The cryptobranchid Cryptobranchus alleganiensis was used to root the tree. Laboratory protocols and primers are those of Palumbi (1996), Ivanova et al. (2006), and Carr et al.(2011). Bidirectional PCR products were used for Sanger sequencing at the Instituto de Genética of the Universidad Nacional de Colombia. Resulting sequences were visualized, assembled, checked for stop codons (COI), and edited in Geneious Pro v 10.2.3 (Kearse et al. 2012). All sequences were deposited in the Barcode of Life Data System (BOLD; Ratnasingham and Hebert 2007) and GenBank (Table 1). Sequences of each gene were aligned independently using the MAFFT plugin v 7.388 within Geneious, considering the secondary structure of RNA in 16S and implementing the G-INS-I algorithm. Subsequently, sequences of both genes were concatenated in a single dataset using Geneious, which was used to construct a Maximum Likelihood phylogeny using IQ-TREE (Nguyen et al. 2015), performing a partitioned analysis based on four partitions (16S, COI first codon position, COI second codon position, COI third codon position) using the partition finder algorithm (-m option TESTMERGE; Lanfear et al. 2012) in IQ-TREE and best fitting models for each partition selected by the same program (Chernomor et al. 2016; Kalyaanamoorthy et al. 2017). Each partition was allowed to have its own set of branch lengths (-sp option). Branch support analysis was performed with 1000 ultrafast bootstrap replicates (Hoang and Chernomor 2017).

Criteria and terminology for morphological descriptions, diagnostic characters, and data for other species of Caecilia follow Lynch (1999), Gower and Wilkinson (2002), Maciel et al. (2009), Maciel and Hoogmoed (2011), Kamei et al. (2009), Wilkinson and Kok (2010), Donnelly and Wake (2013), and Wilkinson et al. (2009, 2013, 2015). For comparative purposes, specimens of C. guntheri, C. isthmica, C. thompsoni, and C. subdermalis were examined (Appendix 1). Morphological observations were made using a stereoscope Nikon optical device SMZ-1B, with High Intensity Illuminator NI-150 Nikon and App Scope 3xSRA41. Measurements were taken using a Mitutoyo precision digital caliper to ± 0.1 mm; and using the following abbreviations for anatomical features and ratios of measurements:

ADD anal disc diameter;

AM anteromedial limit of the mouth on the upper jaw;

BH body height at midbody;

C1 first collar length;

C2 second collar length;

CM corner of the mouth;

CMB circumference at midbody;

D diameter at midbody;

ED eye diameter;

END distance between eye and naris;

HH head height at level with CM;

HL head length;

HW head width at CM;

HWNG1 head width at NG1;

IND distance between nares;

IOD interorbital distance;

TL total length;

TL/D TL divided by diameter at midbody (ratio of length/diameter);

LPOD distance between eye and lip;

ND naris diameter;

NG1 first nuchal groove;

NG2 second nuchal groove;

NG3 third nuchal groove;

PA primary annulus;

PAG primary annular groove;

PM premaxillary-maxillary tooth;

ST snout tip;

STD distance between snout tip and anterior margin of mouth;

STND distance between ST and naris;

STLPD distance between ST and lip;

STOD distance between ST and eye;

TA tentacular aperture;

INTA distance between TAs;

TAOD distance between TA and eye;

TALPD distance between TA and lip;

TANRD distance between TA and naris;

TASTD distance between TA and ST;

VP vomeropalatine tooth;

WC2 width at second collar;

WCH width of choanae;

WBV width of body at vent level;

WMB width at midbody;

TL/HL TL divided by HL;

TL/WMB TL divided by WMB;

TL/HW TL divided by HW;

HL/HW HL divided by HW.

Dermal scale pockets and subdermal scales were searched using the criteria proposed by Wilkinson et al. (2013) and sex and maturity were determined by examination of gonads. Live specimens were photographed with a digital camera model Canon EOS 70D and preserved specimens with a digital camera Canon EOS 5D Mark II.

The final concatenated molecular dataset consisted of a matrix of 1273 bp, 567 sites were parsimony-informative, 111 were singletons, and 595 were constant sites. The best fitting substitution model for both CO1 and 16S was TIM2+F+I+G4 after testing the large selection of models in IQ-TREE. The ML tree is shown in Fig. 2 (LnL: –15725.921). Our phylogenetic analysis recovered the new species nested within a moderately well-supported (84%) monophyletic Caecilia, in a maximally supported monophyletic Caeciliidae. The new species appears most closely related, of the sampled species, to C. volcani but support for this relationship is not strong (58%). Rhinatrema nigrum and R. bivittatum were recovered as monophyletic with the sister group Epicrionops. Siphonops was inferred to be paraphyletic with respect to Luetkenotyphlus (Siphonopidae), and Microcaecilia nicefori was recovered as the sister group of Dermophidae + Siphonophidae, the latter including the remaining Microcaecilia (with Brasilotyphlus guarantanus nested within it) and the paraphyletic Siphonops.

Generic assignment. The new species is assignable to the genus Caecilia because its eyes are not covered by bone and it has tentacles below the nostrils (Type D sensu Lynch, 1999, Fig. 3 D–E). In addition, the new species is nested within the Caecilia clade (Fig. 2) in our Maximum Likelihood phylogenetic analysis.

Our description of Caecilia pulchraserrana sp. nov. brings the number of known species of Caecilia to 35 (Frost 2018). Molecular data are currently available for only six of these species (including the three newly sequenced species analyzed here), which precludes a thorough analysis of the relationships within the genus. Consequently, our phylogenetic analysis (Fig. 2) was designed mainly to test the generic placement of C. pulchraserrana sp. nov. in addition to exploring the relationships of C. isthmica and C. thompsoni (two other species that are endemic to Colombia). Our results recovered C. pulchraserrana sp. nov., C. isthmica, and C. thompsoni within Caecilia (Fig. 2), corroborating the generic placement of the new species and the monophyly of the genus, as previously hypothesized by Wilkinson et al. (2011). Our analysis recovered Oscaecilia as the sister group of Caecilia, which agrees with the results of San Mauro et al. (2014) but disagrees with those of Pyron and Wiens (2011), who instead recovered Caecilia as paraphyletic with respect to Oscaecilia.

Our phylogenetic analysis only included two mitochondrial loci and a small number of species and should not be considered as a robust resolution of caecilian relationships. Nevertheless, our results highlight several potential cases of non-monophyletic taxa and suggest that a taxonomic revision, including a major generic rearrangement, is warranted. Our study includes, for the first time, the Colombian endemics Epicrionops aff. parkeri (Rhinatrematidae) and Microcaecilia nicefori in molecular phylogenetic analyses. On one hand, recent contributions (Maciel et al. 2018) have allowed taxonomic rearrangements within Rhinatrematidae, with Rhinatrema nigrum and R. bivittatum being recovered as monophyletic, supporting previous claims (Wilkinson and Gower 2010; Wilkinson et al. 2011; Pyron and Wiens 2011; San Mauro et al. 2014) that Epicrionops could be transferred to Rhinatrema. Our analysis recovers Epicrionops aff. parkeri nested within a monophyletic Epicrionops (E. marmoratus+ E. aff. parkeri with 89%), which was sister to Rhinatrema (Fig. 2), corroborating the results obtained by Maciel et al. (2018).

On the other hand, Microcaecilia nicefori was recovered as the sister taxon to a clade formed by the dermophiids Gymnophis multiplicata + Dermophis mexicanus and the remaining siphonopids, including Microcaecilia, Brasilotyphlus guarantanus, Siphonops, and Luetkenotyphlus. In addition, Microcaecilia and Siphonops were recovered as paraphyletic with respect to Brasilotyphlus guarantanus and Luetkenotyphlus brasiliensis, respectively (Fig. 2). Recently, Correia et al. (2018) also presented evidence that Microcaecilia is paraphyletic with respect to Brasilotyphlus. The placement of Luetkenotyphlus brasiliensis within Siphonops contrasts with results of Pyron and Wiens’ (2011) and Maciel et al.'s (2019) analyses that found Luetkenotyphlus and Siphonops to be sister taxa. Although analyses by San Mauro et al. (2006), San Mauro et al. (2014) and Correia et al. (2018) also recovered Luetkenotyphlus and Siphonops as sister groups, these studies only included one species of Siphonops (S. annulatus). Therefore, additional molecular data are needed to clarify the delimitation of these clades.

Consistent with previous findings (i.e., Correia et al. 2018), our phylogenetic analysis recovers Microcaecilia as non-monophyletic. Previously, based on evidence from dentition (relationship between VPs and rows of PM) and orbit (open versus closed orbit), Wilkinson et al. (2013) suggested that some Microcaecilia, including the type species of the genus (Dermophis albiceps Boulenger, 1882; not included herein), are more closely related to M. nicefori (Gymnophis nicefori Barbour, 1925, the type species of Parvicaecilia, currently in the synonymy of Microcaecilia; analyzed here for the first time) than to other species of Microcaecilia. That is, the position of trans-Andean Microcaecilia nicefori compared to other cis-Andean members of the genus suggests the revalidation of the genus Parvicaecilia. However, our analysis does not represent solid evidence due to several aspects, such as the low number of genes used, the low support values (a bootstrap value of only 45%), and the absence of key terminals, such as the type species of the Amazonian Microcaecilia (M. albiceps (Boulenger, 1882). Thus, inclusion of relevant taxa, such as M. albiceps, in future phylogenetic analyses is key to guiding taxonomic changes. At the interfamilial level, our results provide evidence for the first time that Shiphonopidae is paraphyletic with respect to Dermophiidae due to the placement of M. nicefori (Fig. 2). Additional, large scale phylogenetic studies are required to rigorously test this finding.

Lynch (1999) suggested that Caecilia degenerata is restricted to the Cordillera Oriental of Colombia (Departments of Boyacá, Cundinamarca and Santander). However, morphological and biogeographical evidence suggests that the cis- and trans-Andean populations are not conspecific. The type series was collected at two cis-Andean localities: Garagoa (Boyacá Department), the type locality, and Choachí (Cundinamarca Department), ca. 90 km southwest of the type locality (Dunn, 1942). Later, Ruiz-Carranza et al. (1996) and Lynch (1999) examined a series of trans-Andean specimens collected at Muzo (Boyacá Department), Tena and Sasaima (Cundinamarca Department), and Charalá (Santander Department), and referred them to C. degenerata, based on morphological similarity and (presumably) relative geographical proximity. Although the absence of secondary annular grooves, the number of primary annular grooves (127–138 in the cis-Andean populations vs. 123–137 in the trans-Andean populations), and the ratio of length/diameter (32–60 in the cis-Andean populations vs 48–58 in the trans-Andean populations; Ruiz-Carranza et al. 1996, Lynch 1999) are consistent with the hypothesis of conspecific populations. The cis- and trans-Andean populations are isolated by biogeographic barriers that includes high and steep mountains, xerophytic areas, and rainy environments, factors that usually play a fundamental role in the speciation of Andean amphibians (Lynch et al. 1997). To test the conspecificity of the populations of C. degenerata, a more extensive sampling of specimens, populations, and additional molecular data are required. Finally, although Taylor (1968) recorded specimens of C. degenerata in Tomaque (probably in Colombia or Peru) and Río Pache (probably in Peru), we agree with Lynch (1999) that C. degenerata is restricted to the (eastern) Cordillera Oriental of Colombia.