The sampling sites are located in: I) the Departments of Antioquia and Caldas in the central branch of the Andes, II) Boyacá and Santander in the centre and the eastern branch of the Andes, respectively, III) Serranía de Perijá Guajira, which is the furthest extension of the eastern branch of the Andes up north and IV) the Departments of Sucre and Córdoba, which belong to the Caribbean Region of Colombia (Table 1). Fieldwork was carried out from 2016 to 2019, mostly during the rainy season. A total of 26 sites were visited, which are located at different elevations.

Depending on the environmental conditions at the sampling localities, light-traps were installed on forest edges or on walking trails. The traps included a mixed light bulb, powered by a portable gas generator and a LepiLED UV lamp (Brehm 2017) installed on a white sheet. Specimens were selectively collected from the sheets and euthanised with killing jars of Ethyl Acetate. The moon phase was considered an important factor for night sampling and full moons were avoided. After collecting, one to three legs were removed from each specimen for DNA barcoding. The vouchers were subsequently pinned, labelled and left to dry.

Specimens were provisionally assigned to putative species by using morphological characters, for example, wing patterns, abdominal tympanic organs, then followed by identifications through DNA barcodes using the tools available on BOLD, such as BIN (Ratnasingham and Hebert 2013). The BIN approach was chosen because BINs have been shown to have a high concordance with traditional taxonomic species concepts and can be used as a reliable proxy for species (Ortiz et al. 2017). The specimens are deposited at the Museo de Zoología, Universidad de Sucre, Colombia. DNA barcodes from vouchers and metadata are publicly available on BOLD. Records of species distributions were submitted and are publicly available on GBIF (https://www.gbif.org) and SIB Colombia (https://sibcolombia.net, https://doi.org/10.15472/ucfmkh) (Murillo-Ramos 2021).

Molecular data: The DNA extraction process was carried out using a NucleoSpin Tissue Kit (MACHEREY-NAGEL), following the manufacturer’s protocol. The DNA barcode region of the mitochondrial gene cytochrome oxidase subunit I (COI) was sequenced for all the samples. In addition, two protein-coding nuclear gene regions, wingless (Wnt) and elongation factor 1 alpha (EF-1alpha), were sequenced for the specimens that did not match with any record on BOLD, based on the DNA barcode. Specimens with three sequences were subject to Maximum Likelihood (ML) analyses as explained below.

DNA amplification and sequencing were carried out following protocols proposed by Wahlberg and Wheat (2008) and Wahlberg et al. (2016). PCR products were visualised on agarose gels. Successful PCR products were cleaned enzymatically with Exonuclease I and FastAP Thermosensitive Alkaline Phosphatase (ThermoFisher Scientific) and sent to Macrogen Europe (Amsterdam, The Netherlands) for Sanger sequencing. Additionally, some samples were sent to the Canadian Center for DNA Barcoding, University of Guelph, Canada, where DNA extraction, PCR amplification and sequencing were performed, following standard high-throughput protocols (deWaard et al. 2008).

Multiple sequence alignments were carried out in MAFFT as implemented in Geneious v.11.0.2 (Biomatters, http://www.geneious.com/). To check for possible errors in alignments and potential contamination, we constructed Neighbour Joining (NJ) and Maximum Likelihood (ML) trees. Successful DNA barcode sequences were uploaded and compared to those on BOLD (Ratnasingham and Hebert 2007), where sequences of more than 21,000 geometrid specimen BINs are available. In total, 386 Colombian Geometridae specimens were processed, but only 284 were successfully sequenced. We assesed species identification with sequences > 500 bp by the Barcode Index Number (BIN) system as implemented on BOLD (Ratnasingham and Hebert 2013).

Those COI sequences without a match on BOLD were submitted to a follow-up analysis with two additional nuclear genes. We retrieved the dataset of Murillo-Ramos et al. (2019) stored in VoSeq (Peña and Malm 2012). This dataset includes more than 1,000 taxa of Geometridae with good sampling of Neotropical species. We tried to obtain further identifications by merging our dataset and the sequences generated by Murillo-Ramos et al. (2019). We ran Maximum Likelihood analyses with partitions by gene using RAxML-HPC2 v.8.2.12 (Stamatakis 2014) on the web-server CIPRES Science Gateway (Miller et al. 2010). Support for nodes was evaluated with 1,000 rapid bootstraps in RAxML.

The study sites are situated in two areas of Colombia: 1) the Caribbean Region and 2) the Andean Region (Fig. 2). Typical habitats in the Caribbean Region are dry and tropical rainforests and the sampling localities have elevations ranging from 0 to 600 m a.s.l. Typical habitats of localities sampled in Eastern and Central Andes are cloud forests and Paramo and the sampling sites have elevations ranging from 400 to 3800 m a.s.l. The Cerro Pintao site in the State of Guajira is covered by pre-montane forests, montane forests and paramo and the elevations of the sampling sites range from 2800 to 3000 m a.s.l.

Figure 2.  

Distribution map of the localities sampled in this study.

We report sequences of species belonging to the subfamilies Ennominae, Sterrhinae, Larentiinae and Geometrinae.

Results

We make available sequences of 281 specimens of Geometridae (dx.doi.org/10.5883/DS-GEOCO) of which 157 matched the BINs with previously-named species on BOLD (either from Colombia, but mostly from other countries, such as Ecuador; Suppl. material 1), while 115 sequences were assigned to BINs, which are identified only to genus or tribe level and more examinations are needed to reach species-level identifications. We checked (and modified accordingly) all existing identifications and provide further identifications, based on morphology and on the analysis of the two additional markers analysed using the Maximum Likelihood (ML) approach (Suppl. material 2). Identifications are provided in Suppl. materials 1, 3, 4. The current Colombian DNA barcode library contains sequences for species belonging to four out of eight subfamilies of Geometridae. Nearly 50% of the sequences are placed in the subfamily Ennominae, represented in our dataset by 159 BINs assigned to 55 genera (Suppl. material 3). Ennomines are very diverse in South America with more than 3,000 described Neotropical species (Pitkin 2002) and studies have suggested that Ennominae dominate particularly the lower elevational levels up to ca. 1,000 m (Brehm and Fiedler 2003). Although the estimation of elevational species richness gradients was beyond the scope of this study, one interesting pattern is that most of the Ennominae records correspond to the low elevation localities, which are mainly part of the Caribbean Region of Colombia, this Region being characterised by the presence of dry tropical forest.

Unsurprisingly, more than 50% of the barcodes which correspond to the subfamilies Sterrhinae and Larentiinae could not be identified at species level (Suppl. material 4), which reflects the lack of taxonomical revisions of those groups for South American species. Compared to available inventories in Ecuador and Costa Rica, the species list for Colombia is very far from being complete. Although the data release of this study is a significant contribution to the knowledge of Neotropical Geometridae, the current dataset stresses the gap in the knowledge of Colombian geometrids and it certainly represents only a small part of the fauna. Prior to the current study, only 99 public records were available on BOLD, which were assigned to 20 species. Unfortunately, there are no checklists of Colombian geometrids, although some species have been included in taxonomic works (Sullivan 2011, Brehm et al. 2011, Brehm 2018, Lindt et al. 2018). With the current dataset, we raise the number of sequences to 380 public records (including the previously-available data) of at least 177 species of Colombian geometrids (157 reported in this study).

Discussion

The occurrences, reported in this study, are based on specimens sampled in six Departments of Colombia, in which different habitat conditions prevail. Thus, the species composition of geometrids was very different in low elevation localities compared to the high-altitude sites and many species were narrowly distributed. Similar to the suggestion by Brehm and Fiedler (2003), the faunal composition of the geometrids found in this inventory seems to vary with regard to altitude, in which Sterrhinae, Geometrinae and Ennominae decrease in their proportions towards high elevations, while the opposite happens with Larentiinae. The species richness per locality is preliminary and it is clear that more studies are needed to understand the species distribution patterns and reveal the potential hotspot of geometrids in Colombia. Most probably the Andes host the highest diversity (Brehm et al. 2005, Brehm et al. 2016); however, far more sampling effort is required to consolidate the list. So far, the current species occurrences reported in databases from the GBIF portal contains 2,407 records of Geometridae in Colombia (GBIF Occurrence Download https://doi.org/10.15468/dl.b298x7). Those records have been provided from different collections of Colombian institutions and some observations from iNaturalist Research and the International Barcode of Life (iBOL) project. Of the 2,407 records for Geometridae in Colombia, only 453 records have been identified to species level (for a total of 131 species). Most records are identified only to the family (1,619 records), with some records to the genus level (335 records).

We provided further identifications of the specimens that were assigned to a BIN, but did not match with named species within Geometridae on the BOLD database. Based on the analysis of the two additional markers, using an ML approach and including the dataset retrieved from Murillo-Ramos et al. (2019), the phylogenetic tree confirmed the monophyly and taxonomic position of specimens in genera, such as Synchlora, Iridopsis, Glena and Physocleora. Even though it was not possible to reach species-level identification, the specimens clustered in their corresponding genera. In contrast, the results pointed out that genera like Idaea, Scopula, Nephodia, Isochromodes and Macaria require taxonomic revision, as they were recovered as being para- or polyphyletic. This could be explained by inadequate genetic information (three genes) or it could represent a true pattern phenomenon (poor taxonomy) or incomplete taxon sampling. We also found many independent lineages that, with more detailed studies, could possibly be assigned to undescribed genera within Boarmiini. Boarmiini is by far the most species-rich tribe-level clade of Ennominae, with ca. 200 genera and ca. 3,000 known species (Murillo-Ramos et al. 2021). However, the evolutionary relationships amongst boarmiines have been difficult to resolve, further complicated by numerous conflicting regional classifications. Boarmiini comprise many unidentified records in the Neotropical Region and this highlights the difficulties with working on species-rich groups that have not received much attention outside of Europe.

We could not assign more than 50% of the barcodes to species belonging to the subfamilies Sterrhinae and Larentiinae. Species in many genera, included in those subfamilies, are very difficult to identify, mostly due to the lack of taxonomic studies of Neotropical species. Most of the unassigned species correspond to the genera Idaea and Eupithecia, which are full of unidentified records, not only in this study, but also in public databases like BOLD. These two genera are amongst the species-rich radiations of Geometridae, widely distributed around the world and they are well-studied in the Palaearctic (Mironov 2003, Hausmann 2004), but poorly known in the Neotropics. Idaea comprises 669 species, while Eupithecia includes 1,360 described species (Hausmann 2004, Choi and Kim 2013, Mironov and Galsworthy 2014). The reasons behind their success have been attributed to the availability of suitable host plants, as well as to the ecological and morphological plasticity of the species (Brehm et al. 2005, Mironov and Galsworthy 2014). However, those hypotheses have not been tested, mainly because of insufficient biological information available for the vast majority of species in Eupithecia and Idaea. The latter prevents the analysis and interpretation of their diversification patterns. This is the same case for many lineages in Geometridae. Concerning the taxonomy and species number of those genera, our results underline the poor knowledge of Sterrhinae and Larentiinae in South America. There is no doubt that future taxonomic revisions will uncover many undescribed species and Neotropical taxa may, in fact, represent independent lineages from old-world genera.

Altogether, 157 geometrids identified in this study to the BIN-level are now barcoded from Colombia (dx.doi.org/10.5883/DS-GEOCO). This represents roughly 5% of the expected species richness of Colombia (1026 species have been reported from literature Scoble 1999, Scoble and Hausmann 2007 , while 2,407 records are reported in platforms like GBIF, https://doi.org/10.15468/dl.b298x7). This would neatly summarise the state of our knowledge and underlines the huge work that lies ahead before the fauna is well studied. The fact that we could only identify less than 50% of the specimens at species-level allows us to conclude that: I) a properly-curated DNA barcode reference library on BOLD is still a work in progress and it covers rather poorly the northern South American Geometridae fauna, II) there are not enough taxonomic experts on Colombian Geometridae, III) the available information for species identification is scarce and IV) more efforts are needed to develop our species list of Colombian Geometridae further towards a comprehensive checklist. Despite all the taxonomic uncertainties, we firmly believe that this barcode library will be a baseline reference for future research and will play an important role in monitoring and biodiversity studies. Although more inventories are needed to know more about the richness of Colombian geometrids, we also highlight that the type specimens described from Colombia should be barcoded in the future. This would make an important contribution to the database, by adding precision and giving Linnean names to barcodes already in the database and for new fresh material.