A contribution to the understanding of phylogenetic relationships among species of the genus Octopus (Octopodidae: Cephalopoda)-Contribución al conocimiento de las relaciones filogenéticas entre especies del género Octopus (Octopodidae: Cephalopoda)

1 Cátedra de Ecología Marina, Facultad de Ciencias Exactas, Físicas y Naturales, Universidad Nacional de Córdoba, Argentina. 2 Cátedra de Genética de Poblaciones y Evolución, Facultad de Ciencias Exactas, Físicas y Naturales, Universidad Nacional de Córdoba, Argentina. E-mail: mchiappero@efn.unc.edu.ar 3 Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina. 4 Alfred Wegener Institute for Polar and Marine Research, Bremerhaven, Germany.


INTRODUCTION
The genus Octopus (Lamarck 1798) (Cephalopoda: Octopodidae: Octopodinae) includes approximately 200 species, many of which are important resources for fisheries worldwide. The genus is found in shallow waters of all oceans except in the polar regions. It displays a wide diversity in skin coloration, behaviour and life strategies but a strong similarity in structural morphology; as a consequence, phylogenetic relationships, species limits and identification are difficult to establish (Robson 1929, Roper and Hochberg 1988, Hochberg et al. 1992, Voight 1994, and reviewed in Guzik et al. 2005). In the last few years, molecular techniques have been increasingly applied to increase the accuracy of phylogenetic relationships assessments. For example, Barriga Sosa et al. (1995) examined five Octopus species from the northern Pacific Ocean using the mitochondrial cytochrome oxidase III (COIII) gene. They confirmed the species-level status of O. bimaculoides and O. bimaculatus, two two-spotted octopuses that occur sympatrically. Carlini et al. (2001), using cytochrome oxidase I, and Guzik et al. (2005), using one nuclear (Elongation Factor-1α) and two mitochondrial (cytochrome oxidase III and cytochrome b) genes, estimated the phylogenetic relationships among several Octopus species; these authors found that Octopus is not a monophyletic genus, although it includes monophyletic groups. However, Norman and Hochberg (2005), in a revision of the species-level taxonomy of the family Octopodidae, reassigned several of the species considered as belonging to the genus Octopus to other genera such as Amphioctopus, Callistoctopus and Enteroctopus, while others were considered as "unplaced" and left for the moment in the genus Octopus until a major revision is undertaken.
There are also taxonomic problems regarding the species O. vulgaris, whose status as a true cosmopolitan species or as a species complex is uncertain (Guerra et al. 2010). Warnke et al. (2004) used COIII and 16rRNA genes to clarify the limits and distribution of the species O. vulgaris. They revealed that populations from the Mediterranean, the western and eastern Atlantic, Venezuela, Japan and Taiwan form a monophyletic clade, confirming the presence of the species in the north-western Pacific. O. mimus was more closely related to O. bimaculoides than to O. vulgaris. However, as they included few species and specimens in their analyses, monophyly could be an artifact of poor species sampling. Guerra et al. (2010) Strugnell et al. (2005) and Teske et al. (2007)) included only a limited number of specimens from the southern part of South America.
The Gould octopus, O. mimus, is an important target of fisheries along the southern Pacific coasts of South America, from central Chile to northern Peru Vega 2003, Cardoso et al. 2004). Octopuses from these locations were considered as synonymous with the cosmopolitan species O. vulgaris, but were recently recognized again as a separate species by Guerra et al. (1999) and by Söller et al. (2000), who found more than 12% of nucleotide divergence between the two species. However, these authors only studied three specimens from a single locality (Iquique, northern Chile), all showing the same haplotype. On the Atlantic coast of southern South America, the commercially exploited O. tehuelchus or "pulpito" is distributed from southern Brazil to approximately 44°S in Argentina (Iribarne 2009). To date, the phylogenetic relationships of this species with others of the genus Octopus remain unknown.
The aim of the present study was to assess the relationships among representatives of O. mimus from localities not studied previously (the southeastern Pacific coast of Chile and Peru) and those previously sequenced, in order to estimate their phylogenetic relationships with other Octopus species, particularly O. vulgaris, and to assess the support for the monophyletic status of O. vulgaris through the use of Bayes factors. We also aimed to study the relationships of the poorly known species O. tehuelchus from the southwestern Atlantic Ocean (Argentina) with other species of the genus using the COIII gene.  Norman and Hochberg (2005) that are still identified as Octopus in GenBank have their accession numbers underlined; specimens marked as "unplaced" are those of uncertain position but left in the genus until further revision by the same authors. Species used as outgroups are listed at the bottom of the  Norman, 1992 Dudley Point, Northern Territory, Australia 1 AJ628207 e Amphioctopus exannulatus (Norman, 1993) Lizard Island, Queensland, Australia 1 AJ628223 e Amphioctopus kagoshimensis (Ortmann, 1888) One tree Island, Queensland, Australia 1 AJ628226 e Jogashima Island, Japan 1 AB573193 i Amphioctopus marginatus (Taki, 1964) Northern Sulawesi, Indonesia 1 AJ628232 e Nha Trang, Vietnam 1 AB573196 Amphioctopus mototi (Norman, 1993) New South Wales, Australia 1 AJ628233 e Amphioctopus ocellatus (Gray, 1849) Tokyo, Japan 1 NC007896 f Amphioctopus ovulum (Sasaki, 1917) East China Sea, Japan 1 AB573198 i Callistoctopus alpheus (Norman, 1993) One Tree Island, Queensland, Australia 1 AJ628215 e Callistoctopus aspilosomatis (Norman, 1993) One Tree Island, Queensland , Australia 1 AJ628216 e Miyagi Island, Okinawa, Japan 1 AB573205 i Callistoctopus bunurong (Stranks, 1990) St. Leonards Pier, Victoria, Australia 1 AJ628219 e Callistoctopus dierythraeus (Norman, 1993) Magnetic Island, Queensland, Australia 1 AJ628222 e Callistoctopus graptus (Norman, 1993) Townsville, Queensland, Australia 1 AJ628224 e Callistoctopus minor Sasaki, 1920 -UNPLACED East China Sea, Japan 1 AB573201 i Callistoctopus luteus (Sasaki, 1929) Kanagawa, Miura, Japan 1 AB573206 i Cistopus indicus (Rapp, 1835) Taichung Fish Market, Taiwan 1 AJ628208 e Enteroctopus dofleini (Wulker, 1910) British Columbia, Canada, Pacific Ocean 1 X83103 a Grimpella thaumastocheir Robson, 1928 Pt.
PCR conditions were as follows: 1 cycle of 2 min at 95°C, 30 cycles of denaturation at 94°C for 30 s, annealing at 48°C for 1 min, and extension at 72°C for 1 min; a post-treatment of 7 min at 72°C and a final cooling at 15°C were performed. PCR products were purified and sequenced at the facilities of Macrogen Inc. (Rockville, USA).

Data analysis
Sequences were inspected manually using the program CHROMAS version 2.23 (McCarthy 1998) and aligned using MUSCLE (Edgar 2004). All new sequences were submitted to GenBank; accession numbers are cited in Table 1. Phylogenetic relationships were assessed using the Bayesian approach implemented in MrBayes v3.1.2 (Ronquist and Huelsenbeck 2003). The model of sequence evolution that best fits our data set was chosen among 24 evolutionary models using MrModeltest 2.3 (Nylander 2004), considering both the complete data set and partitioning into the first, second, and third codon positions. The final model selected for the unpartitioned data set and for the first and third codon positions was GTR+I+G (Tavaré 1986). The selected model for the second codon position was GTR+G. The Bayesian analysis was conducted for the unpartitioned and codon position-partitioned data sets by running the Markov chain Monte Carlo with six chains for 3 million generations, on three independent runs. Trees were sampled every 100 generations and the first 25% of them were discarded as burn in. The robustness of clades was estimated by the Bayesian posterior probabilities (BPP). Results of the partitioned and unpartitioned analyses were compared by calculating Bayes factors ). The Bayes factor (B 10 ) is defined as the ratio of the likelihood of the alternative model to the null model. We used the ln of the harmonic mean of the likelihood values sampled from the stationary phase of the MCMC run (obtained from MrBayes) as an estimator of the model likelihoods (Newton and Raftery 1994). The Bayes factor in favour of a model was then calculated as 2log e B 10 and the resulting values were interpreted following Kass and Raftery (1995), where values between 2 and 6 indicate positive evidence, between 6 and 10 strong evidence, and greater than 10 very strong evidence against the null model (

RESULTS
The aligned sequences were 642 bp long. Four closely related haplotypes were found in O. mimus, differing by one or two mutations from each other. These new haplotypes differed by 2 to 4 substitutions from Table 1 (cont.). -Specimens, sampling regions, sample sizes and accession numbers of DNA sequences included in the phylogenetic inferences with COIII. Sequences reassigned to other genera by Norman and Hochberg (2005) that are still identified as Octopus in GenBank have their accession numbers underlined; specimens marked as "unplaced" are those of uncertain position but left in the genus until further revision by the same authors. Species used as outgroups are listed at the bottom of the that of Iquique published by Söller et al. (2000). Two haplotypes were found among the three O. tehuelchus sequenced, differing in two nucleotidic positions. The value of the Bayes factor comparing the unconstrained, non-partitioned tree (M 0 ) versus the unconstrained, codon-partitioned tree (M 1 ) indicated very strong evidence against the null model (Table 2). Therefore, we present the topology obtained by considering different evolutionary models for each codon. Subsequent Bayesian analyses (that is, the testing of alternative topologies) were performed considering the codon-partitioned data set.

O. tehuelchus (Argentina-Atlantic Ocean)
Callistoctopus alpheus ( O. tehuelchus appears more closely related to the genera Callistoctopus, Grimpella and Macroctopus, as well as to "unplaced" Octopus species sensu Norman and Hochberg (2005), than to the clade containing most of the valid species of Octopus (Fig. 1). Our O. mimus specimens from Chile and Peru form a well supported clade with O. mimus from other localities: those from Isla del Coco (Costa Rica) and Iquique (Chile). However, this clade also includes an O. vulgaris specimen from the Pacific coast of Costa Rica and O. oculifer from the Galapagos Islands with a BPP of 0.97. The sister group of these "Pacific O. mimus" are the two O. vulgaris from the Caribbean Sea, also with high support. All remaining O. vulgaris from various geographic origins cluster in a group that also includes O. tetricus specimens from Australia (BPP=1). The specimens named O. vulgaris from Recife (northern Brazil) by Söller et al. (2000) and Warnke et al. (2004) and identified as a new species, O. insularis, by Leite et al. (2008)  . DISCUSSION In the last few years, studies on the molecular systematics of octopuses have helped to clarify their confusing systematics and phylogenetic relationships and also to identify and describe new species (Norman and Hochberg 2005, Allcock et al. 2007, Leite et al. 2008, Strugnell et al. 2009). In the present study, we focused on the genus Octopus using a Bayesian approach to estimate the phylogenetic relationships among published sequences of the COIII gene of Octopus species, of O. tehuelchus from the southwestern Atlantic Ocean (a species never included in phylogenetic analyses of the genus before), and of individuals of O. mimus from newly sampled localities in the Pacific Ocean (Chile and Peru).
The phylogenetic trees showed two major groupings: one included representatives of the genera Hapalochlaena, Grimpella, Amphioctopus and Callistoctopus and the "unplaced" Octopus species and the other included valid species of Octopus.
Species of the genus Callistoctopus and O. kaurna formed a well-supported clade, while the "unplaced" O. pallidus, O. berrima and O. australis were more closely related to the genera Grimpella and Macroctopus than to the valid species of Octopus. Specimens of O. tehuelchus (a valid species of Octopus according to Norman and Hochberg 2005) were included in this major cluster, as the sister group to Grimpella and Callistoctopus with high support. Therefore, the generic placement of this species may need a revision. The second main cluster included the majority of the valid species of Octopus sensu Norman and Hochberg (2005). Guzik et al. (2005) estimated the phylogenetic relationships among species of the genus Octopus using one nuclear and two mitochondrial genes, and demonstrated the polyphyly of the genus. However, several species included in the work of Guzik et al. were soon afterwards reassigned to other genera by Norman and Hochberg (2005). Nevertheless, the genus still appears to be polyphyletic because species of Ameloctopus, Abdopus and Cistopus are included in our phylogenetic tree within the Octopus cluster, although with medium support (BPP=0.78).
O. vulgaris specimens are included in two different, well-supported clades, with O. insularis as their sister group. These specimens were considered by Söller et al. (2000) and Warnke et al. (2004) as part of the O. vulgaris species complex, but Leite et al. (2008) described them morphologically, established their molecular distinctiveness and phylogenetic position using the 16S rRNA gene, and assigned them the name O. insularis. In the papers by Söller et al. (2000) these specimens group with other O. vulgaris forming a monophyletic clade, but they do not include other species of Octopus. In the present work, the position of these specimens in the phylogenetic tree, using COIII and including more species of the subfamily Octopodinae, remains the same as in Leite et al. (2008).
The topology of the tree obtained in this study using Bayesian Analysis was very similar to that of Guerra et al. (2010) Söller et al. (2000), with the exception that they did not include O. oculifer in their study. Bayes factors showed that there is a strong support for the topology in Figure 1 against that obtained by enforcing O. vulgaris monophyly. Taken together, these results argue in favour of the presence of a cryptic species of Octopus in the Caribbean Sea different from O. vulgaris, O. insularis and O. maya (Fig. 1), and basal to O. mimus. As suggested by Söller et al. (2000), both "Caribbean O. vulgaris" and O. mimus could have originated from an ancestral species whose populations were separated by the rise of the Isthmus of Panama about 3 million years ago.
To resolve the incompletely described and wide geographic range of O. vulgaris, Warnke et al. (2004) estimated phylogenetic relationships among O. vulgaris from several localities, including South Africa, Tristan da Cunha (South Atlantic), West Africa, Japan, southern Brazil and the Caribbean, and compared the sequences of the COIII and 16S genes with those of O. mimus specimens from Chile and the Pacific coast of Costa Rica. They confirmed a previous result by Söller et al. (2000) that O. mimus and O. vulgaris are separate lineages, and concluded that the monophyly of O. vulgaris was supported. However, they included a very limited number of other Octopus species. Guerra et al. (2010) Warnke et al. 2004) with high bootstrap support, and proposed considering them as a monophyletic O. vulgaris species group. Bayes factors showed that there is strong support for the topology in Figure 1 against that obtained by enforcing O. vulgaris monophyly (that is, excluding O. mimus, O. tetricus and O. oculifer from the clade, and including the Costa Rican specimens), giving statistical support to the monophyly of an O. vulgaris s. str. + O. tetricus group sensu Guerra et al. (2010).
Our results emphasize the need for a revision of the generic status of O. tehuelchus, and more detailed population-level and/or phylogeographic studies in the O. mimus and O. vulgaris groups, in order to elucidate the number of species present, their limits and geographic ranges, and their phylogenetic relationships. This information will be useful for sustainable management of these important fisheries resources.