First molecular approach to the octopus fauna from the southern Caribbean

Specimens

A total of 38 tissue samples and 36 specimens were obtained between February and March of 2016 with the help of local fishermen by hand capture with aid of a hook or a shovel (Guerrero-Kommritz & Camelo-Guarin, 2016; Guerrero-Kommritz & Rodriguez-Bermudez, 2018). The sampling was performed in three areas with depths less than 7 m near the shore at the Caribbean Sea: San Andrés Island, Old Providence Island and Taganga Bay at Santa Marta, Colombia (Fig. 1). Mantle tissues were stored in 90% ethanol at −20 °C until DNA extraction. Voucher specimens were fixed in formalin 4% for 24 to 48 h, preserved in ethanol 70% and identified by morphometries and counts following Guerrero-Kommritz et al. (2016) catalogue for southern Caribbean octopod species. These were deposited in the ANDES Natural History Museum at Los Andes University, Bogota, Colombia (Table S1). The samples corresponded by morphological identifications (summarized in Table S2) to four genera and six described species. Namely, O. tayrona (Guerrero-Kommritz & Camelo-Guarin, 2016), O. taganga Guerrero-Kommritz & Camelo-Guarin, 2016, O. briareus Robson, 1929, O. hummelincki Adam, 1936, Amphioctopus burryi Voss, 1950 and M. beatrixi (Guerrero-Kommritz & Rodriguez-Bermudez, 2018). Specimens belonging to the O. aff. tayrona complex (O. spB and O. spD; Guerrero-Kommritz & Camelo-Guarin, 2016), which are currently being described as new species by J. G-K, and to the genus Callistoctopus Taki, 1964 were also found.

Research and collection were approved by the National Environmental Licensing Authority (ANLA, Spanish acronym): Collection Framework Agreement granted to Universidad de los Andes through resolution 1177 of October 9th, 2014 - IBD 0359.

Molecular techniques

DNA was obtained from tissues by phenol-chloroform extraction and ethanol precipitation following the Coffroth et al. (1992) protocol. Genomic DNA was stored in ultrapure water at −20 °C until polymerase chain reaction (PCR) amplification. The mitochondrial gene cytochrome c oxidase subunit I (COI) was amplified using the primers LCO1490 and HC02198 (Folmer et al., 1994) and the nuclear gene Rhodopsin (Rho) with the primers 5′-TTGCCTGTACTGTTTGCTAAAGC-3′ and 5′-GCCATCATTTCTTTCATTTGTGCGGCAT -3′(Strugnell, Norman & Drummond, 2004).

Each 15 µl PCR reaction contained the following reagents: 9.2 µl of ddH₂O, 1.5 µl of 10x buffer, 2.1 µl of 25 mM MgCl₂, 0.3 µl of 10 mM dNTPs, 0.3 µl of 5 U/ µl Taq DNA polymerase (Promega), 0.3 µl of 10 mM of both the forward and reverse primers and 1 µl of 20 ng/ µl DNA. PCR thermal cycling program for COI was: denaturation at 92 °C for 5 min; 35 cycles of 94 °C for 50 s, 54 °C for 1 min and 72 °C for 1 min; and final extension at 72 °C for 7 min completed each PCR. Thermal cycling program for Rho was: denaturation at 94 °C for 2 min; 35 cycles of 94 °C for 40 s, 50 °C for 40 s and 72 °C for 90s; and final extension at 72 °C for 10 min. PCR products were purified with the ExoSAP-IT protocol (Thermo Fisher Scientific) and sequenced by Macrogen Inc (Korea) in both directions using the same primers as for the PCR amplifications. Sequences generated in this study are available under the GenBank accession numbers MG778037–MG778111 (Table S1).

Taxon sampling

To assess relationships between the individuals found in the three sampling sites and their position within the Octopodidae, the dataset included COI and Rho sequences from different sources (Table S3) of species of the family as the ingroup, and Enteroctopus dofleini Wülker, 1910 and Muusoctopus Gleadall, 2004 sequences as outgroup. The ingroup was selected taking into account the identification to genus of the specimens collected, which belonged all to the family Octopodidae sensu Strugnell et al. (2014). The outgroup species were selected because these were found to be part of the Octopodidae sister group, following the latest Octobrachia phylogeny (Sanchez et al., 2018).

Phylogenetic analyses and species delimitation

Haplotypes of the COI and Rho datasets were firstly collapsed using the online fasta sequence toolbox FaBox 1.41 (http://www.birc.au.dk/software/fabox). Sequences were aligned with MAFFT v7 (multiple sequence alignment method based on fast Fourier transform (FFT); Katoh & Standley, 2013) with default parameters. Alignments were cleaned with trimAl v1.4 (automated trimming alignment tool; Capella-Gutierrez, Silla-Martinez & Gabaldon, 2009) using a 0.5 gap threshold, 0.25 residue overlap and 90% sequence overlap to erase non-informative sites and sequences. The best-fit model for each dataset was selected with ModelFinder (Kalyaanamoorthy et al., 2017) implemented in IQTree v1.6.2 (Nguyen et al., 2014) using the corrected Akaike Information criterion (AICc). For both COI and Rho datasets, the Generalized Time-Reversible plus Gamma (GTR + G) model of nucleotide substitution was assigned. Best scoring maximum-likelihood trees were constructed with RAxML v8.2.11 (Randomized Axelerated Maximum Likelihood; Stamatakis, 2014) using the rapid bootstrap algorithm, 1,000 bootstrap replicates and 20 alternative runs on distinct starting trees. These resulting trees were finally used as input for the species delimitation analysis with the maximum-likelihood Poisson Tree Process method (PTP; Zhang et al., 2013) as implemented in the web server (https://species.h-its.org/ptp/). Analyses were performed for each gene separately and bootstrap supports >70% were considered significant. Final datasets (sequence files with unique haplotypes), alignments and tree files are available in (Data S1).

Regional findings

Our study provides, for the first time, molecular data from octopods found in the Colombian Caribbean and newly described species from this area. The mitochondrial analysis (Fig. 2) resolved the monophyly (BS = 100%) and species delimitation (BS = 92%) of O. taganga, the third ocellated species (with O. hummelincki and O. maya) from the Atlantic. Morphologically, O. taganga is closely related to O. hummelincki and appears to be related to the Pacific Ocean species Octopus oculifer (Hoyle, 1904), Octopus bimaculoides Pickford and McConnaughey, 1994 and Octopus bimaculatus Verrill, 1883 (Guerrero-Kommritz & Camelo-Guarin, 2016). However, preliminary molecular results showed it as more similar to O. maya than any other species present in the NCBI database (Guerrero-Kommritz & Camelo-Guarin, 2016). We found in the COI tree (BS = 70%) O. taganga as sister lineage of the O. tayrona/O. insularis clade and O. maya (Fig. 2). This group, together with O. mimus and O. hubbsorum, showed a close relationship with the ocellated species O. bimaculoides and O. bimaculatus (BS = 100%).

The species O. tayrona formed a monophyletic group with O. insularis, Caribbean O. vulgaris, Octopus sp. B, and Octopus sp. sequences from Old Providence Island in the COI analysis (Fig. 2, BS = 100%) and was delimited as one single species with PTP (BS = 89%). Until its description, O. tayrona and any other large octopus in the Caribbean were identified as O. vulgaris (Guerrero-Kommritz & Camelo-Guarin, 2016). However, it can be morphologically distinguished from European O. vulgaris by its smaller ligula (O. tayrona: 0.6 –1.6 mm; O. vulgaris: 1.2 –2.1 mm) and the absence of enlarged suckers in all arms, both in males and females. O. tayrona can be additionally distinguished from other similar species, such as O. insularis, mainly by its smaller size, the presence of large warts on the arms’ base and the absence of enlarged suckers on arms II and III (Guerrero-Kommritz & Camelo-Guarin, 2016). These features, together with other reproduction-related characters (e.g., relative size of the hectocotylus, presence and form of the calamus) and internal structures related to digestion (e.g., size of liver, teeth on radula, form of salivary glands and structure of stomach), are the base for morphological octopod species delimitation due to their role on ecological adaptation, reproductive isolation and, consequently, speciation. Currently, O. tayrona is considered by morphology part of the O. vulgaris type I group (Guerrero-Kommritz & Camelo-Guarin, 2016), the species complex form representing the wider Caribbean area (Norman, Finn & Hochberg, 2014). The inclusion of sequences identified as O. vulgaris from Puerto Rico and the Lesser Antilles within this clade in our COI tree either supports this claim or, most probably, indicates misidentification of the specimens, as already suggested by Lima et al. (2017). Regardless, these phylogenetic results create a conflict with the morphological species delimitation based on reproduction-related characters of O. tayrona and O. insularis. Moreover, the Rho tree consistently resulted on a highly supported (BS = 85%) monophyletic group of O. tayrona, O. sp. B and O. sp. from Old Providence. An O. insularis sequence from (González-Gómez et al., 2018) was included in the analysis, but during trimming it was erased due to its low coverage in the alignment. Therefore, together with the lack of rhodopsin sequences for Caribbean O. vulgaris, the mitochondrial results regarding O. tayrona could not be confirmed with the nuclear marker. Future work using more molecular data would be necessary to follow up these results. This would help to further evaluate the evolutionary relationship between O. tayrona, O. insularis and Type I O. vulgaris and confirm or reject their taxonomical distinction.

The newly described species Macrotritopus beatrixi resulted as an independent lineage in our analyses (Figs. 2 and 3) and was delimited as a single species (COI, BS = 100%; Rho, BS = 93%). So far, it represents the first published sequence of its genus. Thus, data from other Macrotritopus species are needed to resolve the genetic distinction of this new species. Moreover, our COI phylogeny and PTP results suggest that Callistoctopus SM1 from Santa Marta represents one independent lineage and a single species (BS = 100%), different from all other representatives of this genus included in the analysis (Fig. 2). A new species of Callistoctopus is currently under description by J. G-K., expected to be closely related to C. macropus Risso, 1826 from the east Atlantic Ocean and Mediterranean Sea (due to morphological similarities; J G-K, pers. obs., 2018). Callistoctopus macropus is also reported for the southern Caribbean (Díaz, Ardila & Gracia, 2000; Guerrero-Kommritz & Camelo-Guarin, 2016). However, after analyzing specimens from France the morphometrics and counts were different, and we considered this a misidentification (research in process J. G-K). Unfortunately, the sequence generated here was taken from an individual whose whole body could not be collected. Therefore, its morphological identification to species level was not possible to achieve. Moreover, due to the lack of sequences of C. macropus in databases, their evolutionary relationship could not be molecularly assessed here and warrants further investigation.

Contributions to the Octopodidae systematics

Our COI and Rho phylogenies help to fulfill the taxonomic gap in the Octopodidae family of species previously lacking molecular information and thus not considered in the phylogenetic systematics of the family (i.e., O. tayrona, O taganga, A. burryi and M. beatrixi). For instance, the Macrotritopus beatrixi sample contributes the first sequences of its genus, providing new insights on the Octopodidae genus-level systematics. Additionally, the inclusion of O. tayrona in our COI analysis, considered morphologically as part of the Type I O. vulgaris group, allowed the inference of the evolutionary relationship of this species with other forms of the species complex. Our results suggest that O. tayrona is more closely related to other species also found in the New World (e.g., O. taganga from Colombia, O. hubbsorum, O. bimaculatus and O. maya from Mexico, O. mimus from Chile and Peru, O. bimaculoides from California, USA) than to the O. vulgaris forms from the Old World and Brazil (Fig. 2). Moreover, its genetic distinction with O. insularis remains unresolved.

Our mitochondrial analysis showed a better-resolved topology, with less polytomies and more significantly-supported branches (BS > 70%) than the Rho tree. This could be explained by the general smaller effective population size of mitochondrial DNA compared to nuclear genes (Moore, 1995). This leads to a higher probability of coalescence in short internodes and can result on a more accurate recovery of the species-tree topology. However, even a well-resolved gene tree can hardly resemble a species tree due to incomplete lineage sorting and DNA barcodes are known to decrease their performance on poorly sampled groups (Meyer & Paulay, 2005). Moreover, recent genome-wide findings on the O. vulgaris species complex (Amor et al., 2019) highlight that phylogenetic analyses using only mitochondrial DNA underestimate octopus species diversity. Thus, despite the better performance of our COI analysis, more informative nuclear markers are needed to complement our results.

The number of cephalopod species in the Caribbean should increase as more surveys are made (Judkins et al., 2010; Miloslavich et al., 2010). Here we show a first attempt to fulfill taxonomic gaps on the octopod species of an understudied area of the highly diverse Caribbean. However, more sampling and taxonomic efforts are still needed to contribute to the local and the world-wide picture of the Octopodidae systematics. Therefore, for future studies we suggest the inclusion of more taxa poorly represented in databases to account for incomplete taxon sampling and further tests using more molecular information to account for insufficient single-locus resolution.