Allopatric Speciation within a Cryptic Species Complex of Australasian Octopuses

Despite extensive revisions over recent decades, the taxonomy of benthic octopuses (Family Octopodidae) remains in a considerable flux. Among groups of unresolved status is a species complex of morphologically similar shallow-water octopods from subtropical Australasia, including: Allopatric populations of Octopus tetricus on the eastern and western coasts of Australia, of which the Western Australian form is speculated to be a distinct or sub-species; and Octopus gibbsi from New Zealand, a proposed synonym of Australian forms. This study employed a combination of molecular and morphological techniques to resolve the taxonomic status of the ‘tetricus complex’. Phylogenetic analyses (based on five mitochondrial genes: 12S rRNA, 16S rRNA, COI, COIII and Cytb) and Generalised Mixed Yule Coalescent (GMYC) analysis (based on COI, COIII and Cytb) distinguished eastern and Western Australian O. tetricus as distinct species, while O. gibbsi was found to be synonymous with the east Australian form (BS = >97, PP = 1; GMYC p = 0.01). Discrete morphological differences in mature male octopuses (based on sixteen morphological traits) provided further evidence of cryptic speciation between east (including New Zealand) and west coast populations; although females proved less useful in morphological distinction among members of the tetricus complex. In addition, phylogenetic analyses suggested populations of octopuses currently treated under the name Octopus vulgaris are paraphyletic; providing evidence of cryptic speciation among global populations of O. vulgaris, the most commercially valuable octopus species worldwide.


Introduction
Taxonomy within the benthic octopuses (Family Octopodidae) continues to be a source of confusion and controversy and despite extensive revisions in recent decades, the true taxonomy of this family remains unresolved [1,2,3]. The most widely studied and economically significant group of cephalopods worldwide is the 'Octopus vulgaris group' of octopods. The type species of this group is the common octopus, Octopus vulgaris Cuvier, 1797. Octopus vulgaris alone accounts for .50% of the world's total octopod fisheries catch, exceeding 380,000 tonnes and has an international export value of .US$1 billion [4]. The Octopus vulgaris species group is comprised of tropical, sub-tropical and temperate species from the Americas, Europe, Africa, Asia and Australasia. Members of this group are large muscular octopuses that display similar morphological and behavioural traits as well as occupying similar ecological niches.
Within the subtropical waters of Australasia there is a group of morphologically, behaviourally and functionally similar Octopus species, closely related to Octopus vulgaris [3,5]. These species, currently treated under the names Octopus tetricus on the east and west coasts of Australia and O. gibbsi in New Zealand, have been suggested to be a species complex; the taxonomy of which remains unresolved [3]. We treat these taxa collectively herein as the 'tetricus complex', after the first formally described species within this group, Octopus tetricus Gould, 1852; the common Sydney octopus.
The tetricus complex comprises three geographically distinct member taxa ( Figure 1). Octopus tetricus was originally described from New South Wales and occurs along the east Australian coastline, ranging from Eden in southern New South Wales to Moreton Bay in southern Queensland [6]. Octopus tetricus comprises a major portion of the small-scale commercial octopod fisheries landings in New South Wales [7], and is also often caught as bycatch in prawn and finfish trawls [8]. Recently O. tetricus has been reported in Tasmania, significantly south of its previous known range [9] although this has not been verified by molecular data.
A second taxon, known as the common Perth octopus, occurs in Western Australia from Esperance to Shark Bay. This population has extensively been treated under the name Octopus tetricus [10,11,12,13,14] due to close similarities in morphological, behavioural and functional attributes between east and west coast forms. More recently however, the common Perth octopus has been treated under the name O. cf. tetricus; a reflection of the proposal that disjunct east and west populations may be sufficiently isolated and therefore represent sub-or distinct species [2,15]. Joll [11] estimated that 250 tonnes of O. cf. tetricus were harvested annually from Western Australian waters, primarily as by-catch from lobster fisheries. Octopus cf. tetricus often preys upon lobsters caught in craypots, and is considered to negatively impact this economically important fisheries resource.
A third nominal species, Octopus gibbsi O'Shea, 1999, was coined to describe a benthic octopus of unknown relation found within the shallow coastal waters off northern New Zealand. Prior to description by O'Shea [16], O. gibbsi had been treated under the name O. tetricus [17], and more recently the validity of O. gibbsi as a distinct species has been questioned [2]. Examination of museum specimens showed strong morphological similarities between O. gibbsi and Australian forms, leading to the proposal that O. gibbsi is synonymous with O. tetricus [2].
A phylogenetic analysis of the sub-family Octopodinae using amino acid sequences from two mitochondrial (cytochrome oxidase subunit III and cytochrome b) and a single nuclear genetic marker (elongation factor-1a) assigned Octopus tetricus and O. cf. tetricus as sister taxa [3]. Analyses of genetic distance (Kimura 2 Parameter) between these two representatives showed 2.0% and 2.6% sequence divergence within each mitochondrial gene fragment respectively. However, only single representatives from both Western Australia and New South Wales were sequenced in this study. Consequently, analyses of Guzik et al., [3] were insufficient to detect the occurrence of speciation between disjunct east and west populations, and no traditional morphological based studies comparing the two populations have been conducted. Furthermore, no molecular work to date has investigated the phylogenetic status of O. gibbsi, thus its taxonomy remains unresolved.
This study aims to resolve the taxonomic status and phylogenetic relationships of the Octopus tetricus species complex, using a combination of molecular and morphological techniques. Due to the emerging fisheries value and the lack of species-level resolution within the tetricus complex, taxonomic resolution within this group will aid in the management of these marine resources.

Materials and Methods
All tissue samples and DNA extracts were loaned from existing museum/university collections. Thus, no animals were harmed or killed in conducting this study. All appropriate permissions were obtained from the relevant institutions prior to accessing their collections.

Molecular analyses
Sampling.  (Table S1 in File S1). Tissue samples (as arm or mantle tissue ,1 cm in length) were taken from individuals collected from the Australian mainland, Flinders Island (Tasmania) and New Zealand ( Figure 1). All tissue samples were stored at 220uC in 70-90% ethanol until processing.
Phylogenetic analyses. jModelTest v0.1.1 [24] was used to carry out statistical selection of best-fit models of nucleotide substitution on the concatenated alignments and also for the COI alignment alone. The appropriate model was selected on the basis of 'goodness of fit measure' via the Akaike Information Criterion (AIC) [25].
Maximum likelihood (ML) topologies were constructed using PhyML v3.1 [26]. Full heuristic searches were undertaken and model parameter values were treated as unknown and were estimated. Strength of support for internal nodes of ML construction was measured using 1000 bootstrap replicates. Bayesian marginal posterior probabilities were calculated using MrBayes v3.2 [27]. Model parameter values were treated as unknown and were estimated. Random starting trees were used and the analysis was run for 15 million generations, sampling the Markov chain every 1000 generations. The program Tracer v1.3 [28] was used to ensure Markov chains had reached stationarity, and to determine the correct 'burn-in' for the analysis (the number of additional generations that must be discarded before stationarity is reached).
Genetic distance. Molecular Evolutionary Genetic Analysis (MEGA) v5.2 [29] was used to calculate genetic distances for populations of Octopus tetricus, O. gibbsi and O. cf. tetricus using the Tamura-Nei model [30]. Genetic distance was calculated using MEGA default settings (with the exceptions of the model and 'pairwise deletion of missing data' option). Mean values 6 SE of interspecific and intraspecific variations in number of mutations per site were calculated for the barcoding mitochondrial gene COI to allow comparison with published literature.
Timing of divergence. Divergence time between clades were calculated based on an estimated rate of evolution of cephalopods; 3.81 substitutions per site per billion years (with 95% highest posterior density around this mean of 2.43-5.24; [31]) within a generalised molecular clock.
Coalescent delimitation. Potential species delimitation among Octopus tetricus, O. gibbsi and O. cf. tetricus was investigated using a Generalised Mixed Yule Coalescent (GMYC) model [32] applied to the molecular/phylogenetic data. Partitioned sequence data from the mitochondrial genes COI, COIII and Cytb were prepared into XML files using the software program BEAUti v1.7.5 [33]. 12S and 16S regions were excluded from the analysis due to low comparable sample representation (see Table S1 in File S1). A coalescent prior and relaxed molecular clock [34] were set as parameters before Bayesian analysis was performed using BEAST v1.7.5 [33]. Each analysis was performed independently twice and log/tree files were combined using LogCombiner v1.7.5 [33]. The data was then analysed via a single threshold model [35] in the software package Splits [36] available in R v3.0.1 [37], whereby clades with posterior probability values greater than 0.9 were acknowledged.

Morphological analyses
Morphological data was obtained from preserved whole specimens sourced from Museum Victoria, Australian Museum (Sydney) and the University of Tasmania. Samples were collected from south west (n = 15) and south east (n = 32) of the Australian mainland (between the years 1980-2007) as well as Flinders Island, Tasmania (n = 11; 2011) ( Table S3 in File S1). All specimens had been initially fixed in 10% formalin and transferred to 70-90% ethanol for preservation. Morphological data for O. gibbsi (n = 6) was sourced from the published work of O' Shea [16].
Specimens were sexed based on three factors which allowed confident classification: 1) presence of terminal organ in males, 2) presence of hectocotylised arm in males and 3) number of genital glands present within the mantle (1 = male, 2 = female) [38]. Maturity in males was determined on the basis of the presence or absence of enlarged suckers (for mature and immature specimens, respectively) [39]. Maturity in females was determined by the state of egg development [40]. All specimens were weighed using digital scales to the nearest 0.1 gram after being removed from ethanol and patted dry with absorbent tissue.
Standard morphological characters were measured following Norman and Sweeney [41] (Table 1). Dorsal mantle length (MLd), mantle width (MW), head width (HW), arm width (AW), and the greatest non-enlarged sucker diameter (SDn) were recorded using digital callipers to the nearest 0.1 mm. In males, the greatest enlarged sucker diameter (SDe), the length of hectocotylised arm components (i.e. ligula [LL] and calamus [CL]) and terminal organ length (TOL); following dissection of the mantle, were also measured using digital callipers to the nearest 0.1 mm. For all specimens, third right (ALR3) and third left (ALL3) arm lengths were measured from arm tip to the beak opening using non-stretch string to the nearest 1 mm. The numbers of suckers occurring on the third right (SCR3) and third left (SCL3) arms were counted with the aid of a dissecting microscope. In cases where damage to an arm was perceived to inhibit growth, suckers appeared damaged, or arm regeneration was evident, arm length and sucker counts were not recorded. Where sucker and arm damage was minor, and sucker scars or remnants were visible, suckers and arm lengths were recorded. All missing values for individual traits were replaced with the global mean of that trait across the whole dataset.
All morphological analyses were performed using Systat v13 [42]. Differences in morphological traits between tetricus complex taxa were investigated using a multivariate General Linear Model (GLM), in which location was treated as a fixed factor, morphological counts were all treated as dependent variables and MLd was entered as a co-variate [43]. Inclusion of MLd as a co-variate controlled for the effect of body size, and therefore allowed investigation of size free shape variation in morphological traits. MLd was considered an appropriate proxy for an individual's body size as it was found to be highly correlated with body mass (R 2 = 0.8467, data not shown), is more often provided in the literature compared to total body length, and is a standardized measurement when compared to body weight (which can be obtained from fresh or preserved specimens) [44]. The presence or absence of an interaction between locations and MLd was investigated via GLM. A non-significant or weak significant result indicated individuals across all locations were of a similar size class and were therefore comparable.
Males and females were analysed separately to allow the inclusion of male reproductive organs in morphological analyses. Mean scaling was performed on all dependant variables prior to analyses as per Berner [43] using the software package R v3.0.1 [37]. The co-variate (MLd) was either log transformed (male) or mean scaled (female) to conform with homogeneity of variance and linearity. Only a single female of appropriate size class/ maturity was available from New Zealand, which was excluded from female morphological analyses.
Following multivariate GLM analyses on each of the sexes, principle component (PC) loadings were calculated for each individual by multiplying the mean scaled raw data of each trait by the canonical loading of that trait (supplied by the GLM output) and summing the products for all traits [45]. Principle components were then plotted for visualisation and canonical correlations used to calculate the eigenvalues and proportion of variance explained by each PC (Tables S4-S12 in File S1).
The importance of each morphological character in delineation between tetricus complex taxa was further investigated by Roy-Bargman step-down analysis [46], which has the advantage of retaining information on correlations between multivariate variables compared with univariate F-tests. Following a significant result from GLM analysis morphological traits were ranked in theoretical order of importance by multiplying the first and second canonical loadings (CL1 and CL2) for each trait by the total variance explained by PC1 and PC2, respectively. The resulting values were added together, and traits displaying the highest joint CL were ranked as having the highest priority. Each trait was then investigated sequentially in order of descending 'importance' via regression analyses; in which location was a categorical predictor and MLd a co-variate (for size-correction) for all analyses. Higher priority traits were added as co-variates in each successive analysis. Tukey's post-hoc tests were performed for each significant stepdown analysis to determine differences in morphological traits among locations.
Step-down analysis was continued until tests yielded an insignificant effect. Probability values were adjusted via the Bonferroni correction method to account for multiple testing.
To further explore classification of tetricus populations into taxonomic groups, Discriminant Function Analysis (DFA) was performed. As DFA cannot incorporate co-variates, analyses were conducted on calculated principle component loadings for each sex. Principle components were used for DFAs as they were calculated from the original multivariate GLM, and were therefore size corrected. In addition, PCs are composite variables calculated for each individual, and consequently encompass any correlations between morphological traits [45]. For all DFAs, Jackknifed correlation matrices were used as they are considered a more reliable estimator of group membership assignment [47].

Molecular analyses
Phylogenetic analyses. The AIC indicated that TrN+G was the preferred evolutionary model for the concatenated alignment and this was utilised within ML and Bayesian phylogenetic analyses. Topologies resulting from ML and Bayesian analyses were identical, recovering a highly supported clade containing Octopus tetricus from east Australia and Tasmania, as well as O. gibbsi from New Zealand (bootstrap value [BS] = 97.6, posterior probability [PP] = 1; Figure 2). All individuals collected from Western Australia fell within a highly supported monophyletic clade (BS = 98.6, PP = 1). A sister-taxon relationship was supported between the Western Australian and east coast (east Australia, Tasmania and New Zealand) clades (BS = 92.6, PP = 1).
All Octopus vulgaris individuals collected from the waters off Japan and China formed a highly supported monophyletic clade (BS = 97.3, PP = 1 Timing of divergence. Based on TrN distances, a date of divergence of ,3.2-6.9 million years ago (ma) was estimated between Octopus tetricus from the east coast of Australia (inclusive of O. gibbsi) and O. cf. tetricus from Western Australia (Table S13 in File S1). Furthermore, the Australian tetricus complex clades and the Japanese/Chinese O. vulgaris clade were estimated as being separated by ,5.4-11.6 million years (Table S14 in File S1).
Coalescent delimitation. Two ML clusters and three entities (i.e. species) were supported via GMYC analysis (p = 0.01). All individuals from the east coast of Australia, Tasmania (Octopus tetricus) and New Zealand (previously O. gibbsi) comprised a single monophyletic clade, whilst the second monophyletic clade was comprised entirely of individuals from Western Australia (Figure 3). A third clade was supported by the GMYC analysis and comprised a single individual from Western Australia, although this clade was paraphyletic, forming a monophyletic clade with other Western Australian individuals.

Morphological analyses
Males. No strong interaction between the independent variable (coast) and the co-variate (MLd) was recorded (Pillai Trace = 1.937, F = 1.709, df = 48,45, p = 0.04), therefore the General Linear Model was run without the interaction. A significant difference was recorded among four coasts for the multivariate model based upon 16 morphological traits (and MLd as co-variate) measured from 36 mature male octopods (Pillai Trace = 2.070, F = 2.503, df = 48,54, p = 0.001; Table 2). Visualisation of the male PC biplot showed individuals from the east coast of Australia, Tasmania and New Zealand could not be distinguished from one another, and were characterised by relatively small SCR3 and ALR3 ( Figure 4). Western Australian individuals formed a distinct group separate from east coast individuals. Individuals from Western Australia were characterised as having greater SCR3 and ALR3 (PC1) in comparison to individuals from east Australia, Tasmania and New Zealand. No distinctions based upon WD and HW among locations were detected (PC2). DFA showed a significant difference among individuals from east Australia, Tasmania, New Zealand and Western Australia (Pillai Trace = 1.201, F = 16.020, df = 6, 64, p = ,0.001). DFA assigned 100% (n = 7) of male individuals from Western Australia to a single group comprised solely of Western Australian individuals (Table 3). DFA assigned 83% (n = 15) of east Australian individuals to the east Australian group, with 17% (n = 3) allocated to the Tasmanian group. Furthermore, 88% (n = 7) of Tasmanian individuals were assigned to the Tasmanian group, whilst 12% (n = 1) were grouped with east Australian individuals. All individuals from New Zealand (n = 3) were allocated into the east Australian group.
Ranking of CLs determined male SCR3 to be the most important variable in detecting variance among groups (Table S8 in File S1).
Step-down analysis performed on male SCR3 showed a significant difference among coasts (F = 41.775, df = 3, p = , 0.001). Tukey's post-hoc analysis showed no significant difference among east Australia, Tasmania and New Zealand (p = .0.6), however Western Australia differed significantly from all three of these locations (p = ,0.001). Analysis of ALR3 (second highest ranked variable) showed a significant difference among coast once the co-variate and SCR3 were included in the model (F = 5.333, df = 3, p = 0.01). Tukey's post-hoc analysis showed no significant difference between individuals from east Australia, Tasmania and Western Australia (p = .0.1), whilst individuals from New Zealand differed significantly from both eastern and Western Australia (p = 0.02 and 0.01 respectively). Analysis of SCL3 (third highest ranked trait) showed no significant difference among coasts once the co-variate, SCR3 and ALR3 were included in the model (F = 0.410, df = 3, p = 0.7). Due to a non-significant result, stepdown analysis was discontinued.
Females. No interaction between the independent variable (coast) and the co-variate (MLd) was recorded (Pillai Trace = 1.083, F = 1.574, df = 18, 24, p = .0.1), therefore the model was run without the interaction. No significant difference was recorded among three locations for the multivariate model based upon nine morphological traits (and MLd as co-variate) measured from 25 mature female octopods (Pillai Trace = 0.122, F = 1.989, df = 18, 28, p = 0.05; Table 4). Visualisation of the female PC biplot showed overlap of individuals from east Australia, Tasmania and Western Australia along PC1 and PC2, which were primarily driven by HW/SCL3 and SCR3/ALL3 respectively ( Figure 5). Although non-significant, female individuals from Western Australia generally possessed greater HW and SCL3 in relation to individuals from east Australia.
DFA showed a significant difference among individuals from east Australia, Tasmania and Western Australia (Pillai Trace = 0.678, F = 5.637, df = 4, 44, p = ,0.01). DFA assigned 93% (n = 13) of east Australian female individuals into the correct group, whilst 7% (n = 1) were placed into the Western Australian group ( Table 5). 67% of individuals from Tasmania (n = 2) were placed into the correct group, whilst 33% (n = 1) were considered to belong to the east Australian group. 38% (n = 3) of female individuals from Western Australia were correctly assigned, whilst 50% (n = 4) and 12% (n = 1) were assigned to east Australian and Tasmanian groups respectively.

Species level relationships
The main focus of this study was to resolve the taxonomic status of the Australasian tetricus complex. Molecular and morphological results are consistent with the hypothesis that disjunct populations of Octopus tetricus from Australia's east coast (including Tasmania), and from Western Australia are separate species. In addition, findings of this study support the hypothesis that O. gibbsi of New Zealand is synonymous with east Australian O. tetricus [2]. Consequently, we propose that the species name O. gibbsi be considered a junior synonym of O. tetricus Gould, 1852, and will hereafter be included in reference to O. tetricus.
In the present study, interspecific variation of COI between eastern Octopus tetricus and western O. cf. tetricus was over one order of magnitude (,18 times) greater than intraspecific variation within each of these populations; a marked 'barcoding gap' consistent with the 'ten times rule' of Hebert et al., [48]. This study estimated interspecific divergence of COI sequences between O. tetricus and O. cf. tetricus to be 3.4%, similar to congeneric differences previously reported for octopods [49,50]. For example, within the family Octopodidae interspecific variation was found to be 1-2% and 2-3.3% for the octopod genera Pareledone [51] and Thaumeledone [49] respectively. The interspecific variation found between O. tetricus and O. cf. tetricus (3.4%) displayed higher species-level differentiation than the 1.3% divergence recommended by Undheim et al., [50] for O. vulgaris. Low nucleotide sequence divergence between octopod species in this and previous studies contrasts with higher levels recorded among moths, butterflies and birds, which range from 5.8-9.1% [48,52,53].
GMYC analysis suggested Western Australian Octopus cf. tetricus is a distinct species from O. tetricus, as well as supporting the synonymy of O. gibbsi with O. tetricus. However, GMYC analysis detected a second cryptic Western Australian species, which conflicts with the phylogenetic and morphological results of this study (which show no such cryptic speciation). This may be due to gaps in knowledge (i.e. more species exist than is currently known), although more likely reflects the tendency for GMYC analyses to 'over-split' taxa [35].
Talavera et al., [35] investigated the ability of GMYC analysis to delineate species using the well resolved European butterflies. Their analysis revealed 16 unexpected cryptic species, which (although the authors acknowledged that at least some of these cryptic species may represent real entities) was considered to be a failure of the model due to the high levels of intraspecific variability recorded within butterflies. As interspecific variability between Octopus cf. tetricus and O. tetricus was far greater relative to the low intraspecific variability within each individual group (see above), the discovery of a second cryptic Western Australian species is considered likely to be an artefact of 'over-splitting' by GMYC analysis.
Multivariate morphological analyses showed congruence in detecting significant differences between individuals from east Australia/New Zealand and Western Australia; although females appear to be a less reliable morphological discriminator of species identity. Male morphology was able to successfully discriminate between Octopus tetricus and O. cf. tetricus. Sucker numbers on the   species) possess the hectocotylus, a copulatory organ used to pass sperm to the female during mating. The hectocotylus is comprised of the ligula and calamus, which provide a limit to the emergence of new suckers at a relatively early stage of ontogeny [54]. Toll [54] investigated sucker counts on the males hectocotylised arm (HASC) among 12 species of the sub-family Octopodinae, and demonstrated its value in identification and delimitation of otherwise morphologically similar octopods. Toll [54] showed sucker numbers on the hectocotylised arm to be relatively fixed, with different species appearing to be characterised by a narrow range of values for HASC, which he proposed were genetically defined. This assumption appears to be supported by congruence between molecular and HASC data obtained in this study. Consistency of sucker counts despite fixation, preservation [54] or environmental influence further reinforces the usefulness of male HASC in cryptic cephalopod taxonomy.

Biogeographic factors
Speciation between Octopus tetricus and O. cf. tetricus is likely the result of reproductive isolation due to allopatric eastern and western distributions. Divergence of O. tetricus (east Australian, Tasmania and New Zealand populations) and O. cf. tetricus (from Western Australia) were estimated to have occurred somewhere within the last 3.2-6.9 million years. This coincides with cooling of the previously tropical Miocene seas along the southern Australian coastline and the rising of the Bassian Isthmus (a historic landbridge joining Tasmania and mainland Australia) during the Pliocene era, potentially dividing populations of a common tetricus complex ancestor in two. Glacial-interglacial epochs during the early Pleistocene resulted in northward progression of cooler waters, initiating the retreat of numerous wide-spread subtropical species along the eastern and western coasts, isolating populations which allowed for genetic differentiation to commence [55].
More recently oceanographic, climatic and ecological factors have likely maintained contemporary disjunction following the final inundation of the Bassian Isthmus 14,000 years ago. For example, the southern coast of Australia possesses extensive expanses with limited reef habitat in the Great Australian Bight and east of Wilson's Promontory in south-east Victoria. Limited reef habitat has been proposed as a factor in genetic divergence of populations and speciation events in other southern marine taxa such as decapods, echinoderms [56,57], and gastropods [58,59]. However studies conducted on O. gibbsi (treated as O. tetricus) among reefs in Northern New Zealand found reef habitat was not essential for successful settlement [17], and O. tetricus were often found in lairs within sandy bottomed estuaries along the southern coast of New South Wales (M. Amor, personal observation). The Great Australian Bight is also associated with sharp drops in sea surface temperature (SST), which is a likely explanation for maintenance of allopatric distributions between east and west taxa.
The absence of significant genetic differentiation between New Zealand and east Australian Octopus tetricus populations suggests ongoing gene flow across the Tasman Sea; a 2000 km wide marine body separating the two landmasses. Due to the benthic shallowwater habit of Octopus tetricus adults [15], connectivity between New Zealand and east Australian populations is likely attributable to trans-Tasman dispersal during the planktonic larval stage; although adults of the genus Octopus can raft on floating wood or drifting macroalgae [60], which may function as a rare mode of passive trans-Tasman migration. A number of other southern Australasian marine taxa display similar trans-Tasman genetic homogeneity, including the southern rock lobster, Jasus edwardsii [61,62,63] and morwong (cheilodactylid) fishes [64,65]. Planktonic larval durations (PLD) for the Octopodinae appear much shorter (35-60 days; reviewed in Villanueva, [66]) than those of the lobster J. edwardsii (2 years [67]) and cheilodactylid fishes (1 year [68]). Octopus paralarvae appear to be active and often constant swimmers [13,69], potentially facilitating dispersal within surface currents. However, simulation based oceanographic modelling studies suggests that in the absence of rafting, a period of several months is required for even a low probability of successful trans-Tasman dispersal [70]. Octopod paralarvae have been observed rafting on macroalgal and other drift debris [71], which may function as habitat for postsettlement juveniles until arrival at suitable shallow-water habitat. Additionally, paralarvae of some octopods can delay settlement in the absence of suitable habitat [72]. These 'super-paralarvae' obtain larger sizes and more developed swimming capabilities, while retaining paralarval morphological characters (reviewed in Villanueva and Norman,[69]), and may facilitate trans-Tasman dispersal for Octopus tetricus. Further investigation into physiological, behavioural and ecological aspects of paralarval life histories would further our understanding of the dispersive capabilities of O. tetricus.

Evidence of range shifts and implications of climate change
This study is the first to verify the presence of Octopus tetricus in the temperate waters off Flinders Island, Tasmania. This suggests the southern distributional limit of O. tetricus along the Australian mainland (currently recognised as Eden, New South Wales) is underestimated and requires resurveying, in fact O. tetricus has been sighted as far south as Cape Conran, Victoria (M. Amor personal observation, 2013). Temperate coastal waters in eastern Tasmania appear to be warming at approximately four times the global ocean warming average due to climate change driven strengthening of the Eastern Australian Current [73]. This has  been linked to recent range expansions of a number of sub-tropical and tropical marine species in Tasmanian waters, including 22 fish species, eastern rock lobster, leatherback turtle and two species of box jellyfish [74]. Coastal warming in Tasmania may have resulted in current temperatures exceeding the lower thermal limits of O. tetricus paralarvae, potentially allowing population establishment outside of their previously known range, as has been suggested for the sea urchin Centrostephanus rodgersii [75]. Investigation of the potential impacts of O.tetricus range expansion on native ecosystems and commercial fisheries should be given high priority.

Broader phylogenetic relationships
Mitochondrial DNA analyses placed the Australasian tetricus complex within a monophyletic clade along with Japanese and Chinese Octopus vulgaris, supporting previous speculations that these taxa are closely related [76]. The current study estimated that the tetricus complex and Japanese/Chinese O. vulgaris arose from a common ancestor following an 'anti-tropical' divergence event that took place between ,5.4-11.7 ma. This estimated time of divergence is consistent with mid-Miocene climatic warming and the emergence of intervening tropical waters at lower latitudes [77]; suggesting vicariant isolation of a once common subtropical ancestor into Northern and Southern Hemisphere populations. Warming of equatorial waters during the mid-Miocene has also been implicated in trans-equatorial divergences for a number of marine taxa, especially reef fishes [78,79,80]. In addition, antitropical affinities between other subtropical Australasian-Japanese/Asian octopods have been noted. For example, Amphioctopus kagoshimensis Ortmann, 1888 from subtropical Japan and the morphologically indistinguishable taxon Amphioctopus cf. kagoshimensis recently discovered at similar latitudes in Australasian waters are predicted to represent closely related relicts of a wider distributed ancestry [76]. The ability of molecular analyses to detect cryptic species suggests that future molecular work would clarify the taxonomic, phylogenetic and palaeogeographical relationships between seemingly cryptic anti-tropical cephalopod species pairs. Paraphyletic relationships within the vulgaris complex revealed in this study directly question the purported cosmopolitan distribution of Octopus vulgaris, and supports hypotheses regarding the existence of numerous cryptic vulgaris-like species [2,76,81].  Norman and Kubodera [76] previously suggested the possibility of an Asian vulgaris-like species ranging from Taiwan to Japan that was distinctly separate from genuine O. vulgaris, originally described from the Mediterranean Sea and Atlantic Ocean. Findings of this study support this theory of speciation between Atlantic and Pacific vulgaris-like species. However, the results of this study were based on samples from extremes in the distribution of O. vulgaris. Future work aimed at resolving the taxonomy of this species complex should include individuals from a representative range of the entire O. vulgaris distribution.

Conclusions and future directions
This study is the first attempt to resolve the taxonomy of the Australasian Octopus tetricus species complex. Molecular and morphological results support east Australian Octopus tetricus as a distinct species from Western Australian O. cf. tetricus, which requires future formal taxonomic description. Additionally, New Zealand's O. gibbsi was found to be synonymous with east Australian and Tasmanian O. tetricus. Paraphyletic relationships within the Octopus vulgaris complex revealed in this study adds support to hypotheses regarding the existence of numerous cryptic vulgaris-like species, warranting taxonomic revision of the O. vulgaris species complex to aid in the management of this significant global marine resource.

Supporting Information
File S1 Supporting tables. Table S1, Specimen information for individuals of which molecular sequencing was undertaken during the present study. Table S2, Specimen information for individuals accessed via GenBank for use in the present study. Table S3, Specimen information for individuals of which morphological traits were recorded during the present study. Table S4, Canonical correlation (CC) output for male octopod multivariate analysis. Table S5, Canonical loadings (CL) output for male octopod multivariate analysis. Table S6, Principal components (PC) calculated from canonical correlation and canonical loading outputs from male octopod multivariate analysis.